<%BANNER%>

Green Roofs as an Urban Stormwater BMP for Water Quantity and Quality in Florida and Virginia

Permanent Link: http://ufdc.ufl.edu/UFE0024408/00001

Material Information

Title: Green Roofs as an Urban Stormwater BMP for Water Quantity and Quality in Florida and Virginia
Physical Description: 1 online resource (273 p.)
Language: english
Creator: Lang, Sylvia
Publisher: University of Florida
Place of Publication: Gainesville, Fla.
Publication Date: 2010

Subjects

Subjects / Keywords: bmp, evapotranspiration, florida, greenroofs, growing, media, nitrogen, phosphorus, stormwater, subtropics, urban, virginia, water
Soil and Water Science -- Dissertations, Academic -- UF
Genre: Soil and Water Science thesis, Ph.D.
bibliography   ( marcgt )
theses   ( marcgt )
government publication (state, provincial, terriorial, dependent)   ( marcgt )
born-digital   ( sobekcm )
Electronic Thesis or Dissertation

Notes

Abstract: GREEN ROOFS AS AN URBAN STORMWATER BEST MANAGEMENT PRACTICE FOR WATER QUANTITY AND QUALITY IN FLORIDA AND VIRGINIA Green roofs are well known as an urban stormwater volume Best Management Practice (BMP) in northern climates, but information regarding water quality benefits/impacts and optimal plant-growing media combinations for green roofs in the sub-tropics is lacking. The objectives of this study were to: 1) determine the optimal plant-growing medium combination for water and nutrient retention and 2) characterize green roofs capability to reduce stormwater volume and peak runoff and 3) determine whether green roofs behave similarly (as a sink or source) for nutrients and metals, in Florida and Virginia. Objectives were tested via i) a green roof bin study in Florida and ii) paired green roof studies in Florida and Virginia. The results of the green roof bin study showed that growing medium type affected water retention and nutrient leaching more than plant type. Water retention ranged from a low of 24% for Building Logics (B) medium with no vegetation to a maximum of 83% for UCF (U) growing medium with perennials. Differences among media were attributed to physical characteristics of the media: pore-size distribution and OM content. Plants increased water retention by 7-10% above bare medium, with perennials having the greatest, and succulents having the least effect. TP and TN loads for the establishment period (initial 6-weeks) ranged from 110 mg P m-2 (U-perennials) to 1800 mg P m-2 for (H-succulents or bare medium); and from 190 mg N m-2 (U-runners) to 1800 mg N m-2 (H-succulents). The majority (60-90%) of the nutrient load leached out in the establishment period of the 24-week study period. Green roofs monitored in Virginia and Florida behaved similarly for water retention and peak reduction. In both Florida and Virginia, small rain events ( < 0.254 cm), had significantly (p < 0.05) higher mean retention (79% and 98%, Florida and Virginia respectively) than large rain events (26% retention and 72%, Florida and Virginia, respectively). Green roofs significantly (p < 0.05) reduced the peak runoff in both Florida (94% for small and 60% for large events) and Virginia (100% for small and 79% for large events). Green roofs behaved similarly for nutrients (sources for phosphorus, sinks for N-NO3, and buffered pH) and differently for metals. Al and Fe levels were significantly higher (p < 0.05) in green roof (GR) runoff than conventional roof (CR) runoff in Florida; while Pb was significantly lower (p < 0.05) in Virginia GR runoff. In conclusion, this study found that green roofs in the subtropics are better suited as a stormwater volume control BMP, than a nutrient control BMP.
General Note: In the series University of Florida Digital Collections.
General Note: Includes vita.
Bibliography: Includes bibliographical references.
Source of Description: Description based on online resource; title from PDF title page.
Source of Description: This bibliographic record is available under the Creative Commons CC0 public domain dedication. The University of Florida Libraries, as creator of this bibliographic record, has waived all rights to it worldwide under copyright law, including all related and neighboring rights, to the extent allowed by law.
Statement of Responsibility: by Sylvia Lang.
Thesis: Thesis (Ph.D.)--University of Florida, 2010.
Local: Adviser: Clark, Mark W.
Local: Co-adviser: Daroub, Samira H.

Record Information

Source Institution: UFRGP
Rights Management: Applicable rights reserved.
Classification: lcc - LD1780 2010
System ID: UFE0024408:00001

Permanent Link: http://ufdc.ufl.edu/UFE0024408/00001

Material Information

Title: Green Roofs as an Urban Stormwater BMP for Water Quantity and Quality in Florida and Virginia
Physical Description: 1 online resource (273 p.)
Language: english
Creator: Lang, Sylvia
Publisher: University of Florida
Place of Publication: Gainesville, Fla.
Publication Date: 2010

Subjects

Subjects / Keywords: bmp, evapotranspiration, florida, greenroofs, growing, media, nitrogen, phosphorus, stormwater, subtropics, urban, virginia, water
Soil and Water Science -- Dissertations, Academic -- UF
Genre: Soil and Water Science thesis, Ph.D.
bibliography   ( marcgt )
theses   ( marcgt )
government publication (state, provincial, terriorial, dependent)   ( marcgt )
born-digital   ( sobekcm )
Electronic Thesis or Dissertation

Notes

Abstract: GREEN ROOFS AS AN URBAN STORMWATER BEST MANAGEMENT PRACTICE FOR WATER QUANTITY AND QUALITY IN FLORIDA AND VIRGINIA Green roofs are well known as an urban stormwater volume Best Management Practice (BMP) in northern climates, but information regarding water quality benefits/impacts and optimal plant-growing media combinations for green roofs in the sub-tropics is lacking. The objectives of this study were to: 1) determine the optimal plant-growing medium combination for water and nutrient retention and 2) characterize green roofs capability to reduce stormwater volume and peak runoff and 3) determine whether green roofs behave similarly (as a sink or source) for nutrients and metals, in Florida and Virginia. Objectives were tested via i) a green roof bin study in Florida and ii) paired green roof studies in Florida and Virginia. The results of the green roof bin study showed that growing medium type affected water retention and nutrient leaching more than plant type. Water retention ranged from a low of 24% for Building Logics (B) medium with no vegetation to a maximum of 83% for UCF (U) growing medium with perennials. Differences among media were attributed to physical characteristics of the media: pore-size distribution and OM content. Plants increased water retention by 7-10% above bare medium, with perennials having the greatest, and succulents having the least effect. TP and TN loads for the establishment period (initial 6-weeks) ranged from 110 mg P m-2 (U-perennials) to 1800 mg P m-2 for (H-succulents or bare medium); and from 190 mg N m-2 (U-runners) to 1800 mg N m-2 (H-succulents). The majority (60-90%) of the nutrient load leached out in the establishment period of the 24-week study period. Green roofs monitored in Virginia and Florida behaved similarly for water retention and peak reduction. In both Florida and Virginia, small rain events ( < 0.254 cm), had significantly (p < 0.05) higher mean retention (79% and 98%, Florida and Virginia respectively) than large rain events (26% retention and 72%, Florida and Virginia, respectively). Green roofs significantly (p < 0.05) reduced the peak runoff in both Florida (94% for small and 60% for large events) and Virginia (100% for small and 79% for large events). Green roofs behaved similarly for nutrients (sources for phosphorus, sinks for N-NO3, and buffered pH) and differently for metals. Al and Fe levels were significantly higher (p < 0.05) in green roof (GR) runoff than conventional roof (CR) runoff in Florida; while Pb was significantly lower (p < 0.05) in Virginia GR runoff. In conclusion, this study found that green roofs in the subtropics are better suited as a stormwater volume control BMP, than a nutrient control BMP.
General Note: In the series University of Florida Digital Collections.
General Note: Includes vita.
Bibliography: Includes bibliographical references.
Source of Description: Description based on online resource; title from PDF title page.
Source of Description: This bibliographic record is available under the Creative Commons CC0 public domain dedication. The University of Florida Libraries, as creator of this bibliographic record, has waived all rights to it worldwide under copyright law, including all related and neighboring rights, to the extent allowed by law.
Statement of Responsibility: by Sylvia Lang.
Thesis: Thesis (Ph.D.)--University of Florida, 2010.
Local: Adviser: Clark, Mark W.
Local: Co-adviser: Daroub, Samira H.

Record Information

Source Institution: UFRGP
Rights Management: Applicable rights reserved.
Classification: lcc - LD1780 2010
System ID: UFE0024408:00001


This item has the following downloads:


Full Text

PAGE 1

GREEN ROOFS AS AN URBAN STORMWATER BEST MANAGEMENT PRACTICE FOR WATER QUANTITY AND QUALITY IN FLORIDA AND VIRGINIA By SYLVIA BERTIE LANG A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLOR IDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY UNIVERSITY OF FLORIDA 2010 1

PAGE 2

2010 Sylvia B. Lang 2

PAGE 3

To my parents, husband and Simran 3

PAGE 4

ACKNOWLEDGMENTS I thank all my parents for their unending support and their underlying beli ef in me, that one can accomplish anything they set their mind to do. I am grateful to my husband, Manohardeep Singh Josan and my daughter Simran for all of their patience and loving kindness while I worked on this dissertation. I th ank my advisor, Mark Clark and my co-chair Samira Daroub, for their long hours of reviewing the writing and for all th e hours planning the project, reviewing the data and methods and discussing the details of the project and patience with all life changes during the research period. I appreciate all the support from my sister Christy and her family, who gladly housed me during the green roof research in Virginia, and took care of Simran while I worked. A special thanks to Kathleen for editing various chapter sections. I am grateful to my committee members: J.J. Delfino, Pierce Jones and Marty Wanielista, who supported this work by reviewing my writing and proposals and listened to slide presentations and are encouraging me to publish articles. I appreci ate the help of Dr. Chini and Bahar Amarghani and Matt Perry who helped co ordinate the installation of the monitoring system of the green roof at UF. I am grateful to Jeannette Stewart for her in spiration and hard work in the middle of the night collecting rain samples in the cold rain in Virginia. I ap preciate everyone who pushed me to keep going and wrap this up. My dear frie nds Andrea Albertin, Kath leen McKee, Kanika Sharma, Adrienne Frisbee, Marisa Tohver, Wend y-Lin and Laura Tyler, many of whom worked side by side with me, either in the lab, the libr ary or the home and also helped by watching little Simi at odd hours, so that I could work. I thank all those who dona ted their time and muscles to help install the cisterns at UF, lift heavy bins, build platformssuch as Ed Dunne, Todd Osborne, Jim Schultz, Barry, Andy and Scott in Hastings, and Italo Lent a and Nick Taylor. I am grateful for my supportive officemates, 4

PAGE 5

especially Daniel Moura, Liz Hodges, Augustine Obour, and Manmeet Waria, who worked late nights with me in the office and edited crucial pa rts on the spur of the moment. I thank the WBL, Yu Wang, Yubao and Xialwei, Gavin Wilson a nd Carolina and especially to Dr. Reddy for making it possible to analyze my samples. I am grateful to my funding sourceslike AEI, FDOT, working with Glenn Acomb, collaborati on with Fairfax County Stormwater, Tanya Amrhein, Deborah Severson for financing different pa rts of the project. Thanks to all my friends and extended family in America, Hungary and India for their care and support too. And above all, I thank God for completing the PhD a nd for Simran and the new baby on its way. 5

PAGE 6

TABLE OF CONTENTS page ACKNOWLEDGMENTS...............................................................................................................4 LIST OF TABLES................................................................................................................. ........12 ABSTRACT...................................................................................................................................26 CHAPTER 1 INTRODUCTION AND LITERATURE REVIEW..............................................................28 Introduction................................................................................................................... ..........28 Rooftops as a Source of Urban Stormwater....................................................................29 Green Roof Design: Intensiv e, Extensive and Florida............................................30 Stormwater Best Management Practices (BMPs)...........................................................31 Green Roofs Role in Stormwater Retenti on and Detention in the Hydrologic Cycle...32 Nutrients and Metals........................................................................................................33 Green Roofs in Florida....................................................................................................33 Justification.................................................................................................................. ...........34 Research Goals................................................................................................................. ......36 Hypotheses and Research Objectives.....................................................................................36 Study Approach......................................................................................................................38 Dissertation Format............................................................................................................ ....39 2 HYDROLOGIC DYNAMICS OF GREEN ROOF PLANT-GROWING MEDIUM COMBINATIONS FOR NORT H CENTRAL FLORIDA....................................................41 Introduction................................................................................................................... ..........41 Physical Characteristic s of Growing Media....................................................................41 Role of Plants in Retention and Detention of Stormwater..............................................44 Role of PlantGrowing Media Sy stem in Stormwater Control.......................................45 Objectives...............................................................................................................................46 Hypotheses..............................................................................................................................47 Materials and Methods...........................................................................................................48 Study Area.......................................................................................................................48 Experimental Design.......................................................................................................48 Experimental Set-Up.......................................................................................................50 Growing media.........................................................................................................52 Analysis of physical properties of the growing media.............................................54 Plant types................................................................................................................55 Irrigation regime and rainfall...................................................................................60 Rainfall.....................................................................................................................60 Sampling Regime............................................................................................................60 Composite water sampling.......................................................................................60 Water release characterizati ons (lysimetric sampling).............................................61 6

PAGE 7

Criteria for Evaluating Growing Media for North Central Florida Green Roofs for Stormwater BMPs........................................................................................................63 Criteria for Evaluating Plant Selection fo r North Central Florida Green Roofs for Stormwater BMPs........................................................................................................64 Statistical Analyses..........................................................................................................6 6 Results IWater Retention....................................................................................................67 Characterization of rainfall and irrigation.......................................................................67 Water retention................................................................................................................68 Effect of plant type on water retentio n irrespective of gr owing medium type.........71 Interactive effects of pl ant-growing medium combin ations on water retention over time...............................................................................................................71 Effect of plant type on water retention within growing media --all time periods combined...............................................................................................................74 Effect of plant type on water retentio n irrespective of gr owing medium type.........76 Results II-Water Release Curves............................................................................................76 Changes in Mass and Water Content over the Six Month Study Period.........................76 Effect of Plant Type on Water Content by Growing Media Type over Six Months.......77 Effect of Growing Media Type on Wate r Uptake and Release Directly after Irrigation......................................................................................................................80 Effect of Plant Type on Water Uptake/Release Characteristics......................................85 Evapotranspiration Rates.................................................................................................93 Discussion...............................................................................................................................94 Effect of Physical Characteristics of the Growing Media on Water Retention, Uptake and Release......................................................................................................94 Effect of Plant Type on the Physical Characteristics of the Growing Media..................98 Conclusions...........................................................................................................................103 3 NUTRIENT DYNAMICS OF PLANT-GROW ING MEDIUM COMBINATIONS FOR NORTH CENTRAL FLORIDA...........................................................................................109 Introduction................................................................................................................... ........109 Objectives.............................................................................................................................111 Hypotheses............................................................................................................................112 Materials and Methods.........................................................................................................112 Experimental Set-Up.....................................................................................................112 Experimental Procedure................................................................................................113 Sampling Protocol.........................................................................................................114 Irrigation regime.....................................................................................................114 Rainfall measurements...........................................................................................115 Water quality sampling..........................................................................................115 Statistical Analyses........................................................................................................117 Calculation of total loads........................................................................................117 Transformations of the data set..............................................................................118 Results and Discussion......................................................................................................... 119 Total Phosphorus Concentration and Load During 6-Week Establishment Period......119 TP Concentrations and Loads among Gr owing Media Types over 6 months...............119 Effect of Time on TP Load............................................................................................122 7

PAGE 8

Effect of Plant Type and Plant-Gr owing Medium Combinations on Total Phosphorus in Green Roof Bins.................................................................................124 Initial and Final TP Ash Concentrations in the Growing Medium...............................126 Nitrogen (TKN, NO3 and TN) Concentrations a nd Load in Leachate during Establishment.............................................................................................................130 Effect of Plant, Growing Media, Inte ractive Effects and Time on TKN Load.............135 Effect of Growing Media and Plan t Type on Nitrate Loads during the Establishment Period.................................................................................................138 Effect of Growing Media and Plant T ype on Nitrate Load over 6 Months...................139 Initial and Final Nitrogen C oncentrations and CN Ratios in the Growing Media........140 TSS Concentrations in Leachate...................................................................................141 Results of the Lysimetric MethodAnalysi s for concentration based first flush (CBFF).......................................................................................................................143 Correlations among Nutrients in Leac hate, Retention and Water Input.......................144 Conclusions...........................................................................................................................145 4 WATER QUANTITY AND QUALITY IN PA IRED GREEN ROOF STUDIES IN FLORIDA AND VIRGINIA................................................................................................150 Introduction................................................................................................................... ........150 Effect of Green Roof Growing Medium and Drainage Layer on Retention.................151 Effect of Climate and Rain Event Charac teristics on Green Roof Water Retention and BMP Design........................................................................................................153 Stormwater Detention BMPs and Critical Design Storms............................................155 Water Quality and Green Roofs....................................................................................156 Rooftops as a source of urban stormwater.............................................................156 Hypotheses and Objectives...................................................................................................160 Hypotheses....................................................................................................................161 Specific Objectives........................................................................................................161 Materials and Methods.........................................................................................................161 Study Site 1: Florida......................................................................................................161 Growing media.......................................................................................................163 Cisterns and pump station......................................................................................164 Automated irrigation system..................................................................................165 Study Site 2: Virginia...................................................................................................166 Data Collection and Analyses.......................................................................................167 Analysis of cumulative rain and runoff..................................................................168 Analysis of individual rain events..........................................................................169 Collection of water quality data.............................................................................172 Analysis of water quality data................................................................................174 Results and Discussion......................................................................................................... 175 Hydrology ResultsCharles R. Perry Green Roof in Gainesville, FL.........................175 Cumulative hydrographsCharles R. Perry green roof in Gainesville, Florida...176 Hydrology ResultsYorktowne Square Condominium (YSC) Green Roof in Merrifield, VA...........................................................................................................184 Differences in Water Retention among Seasons and by Rainstorm Size in Florida and VA.......................................................................................................................188 8

PAGE 9

Comparison of Hydrologic Dynamics of a Green Roof in FL and VA.........................190 Water Quality ResultsCharles R. Perry Green Roof in Gainesville, Florida............191 Concentrations and loads from individual storm events........................................191 Discussion--CRP-FL Water Quality Data.....................................................................194 Water Quality ResultsYSC Green Roof in Merrifield, Virginia...............................195 Comparisons in Concentrations of Nutrients and Metals between roof types in FL and VA and between locations..................................................................................197 Comparison of Areal Loading Rates between Florida and Virginia............................203 Conclusions...........................................................................................................................203 5 CONCLUSIONS.................................................................................................................. 208 Implications of Water Retent ion for Florida Green Roofs...................................................208 Implications of Nutrient Lo ads for Florida Green Roofs.....................................................211 Effect of Plant Type on Nutrient Loads................................................................................214 Role of Green Roofs in Stormwater Hydrology and Water Quality in Florida as compared to Virginia.........................................................................................................217 Summary...............................................................................................................................219 APPENDIX A IRRIGATION REGIME FOR GREEN ROOF BIN STUDY..............................................221 B SPECIFICATIONS OF LITE TOP GROWING MEDIUMHYDROTECH USA........222 C PHYSICAL PROPERTIES OF STALITE...........................................................................223 D MODEL FIT FOR WATER RETENTION ANOVAS........................................................225 E WATER CONTENT CURVES FOR WEEKS 12, 18 AND 24..........................................226 F WATER RETENTION CURVES FOR WEEKS 6, 12, 18 AND 24...................................230 G MEAN WATER UPTAKE/RELEASE RA TES FOR GROWING MEDIA AND PLANT-GROWING MEDIUM COMBINATIONS...........................................................234 H ET RATES FOR PLANT-GROWING ME DIUM COMBINATIONS BY TIME PERIOD......................................................................................................................... .......235 I CHANGES IN ORGANIC MATTER CONTENT AND GRAIN-SIZE DEISTRIBUTION IN THE GROWING MEDIA, BEFORE AND AFTER THE GREEN ROOF STUDY.......................................................................................................236 J PHYSICAL PROPERTIES AND MACRO AND MICRONUTRIENTS FOR HYDROTECH GROWING MEDIUM................................................................................237 K TN LOAD ANOVA MODELS USED FOR ANALYZING EACH 6-WEEK PERIOD....238 9

PAGE 10

L TKN LOAD MODEL DETAILS (F-VAL UES AND R-SQUARED VALUES) AND RESULTS FOR ANOVA ANALYSES...............................................................................240 M NITRATE LOAD DATA FOR TH E ESTABLISHMENT PERIOD..................................241 N NITRATE LOAD DATA FOR TH E 24-WEEK STUDY PERIOD....................................242 O DRAWINGS OF CISTERN AND PUMP SET-UP FOR CHARLES R. PERRY GREEN ROOF..................................................................................................................... 243 P CHARLES R. PERRY GREEN ROOR HYDROGLOGY DATA BY INDIVIDUAL STORMS..............................................................................................................................245 Q CUMULATIVE HYDROGRAPH AND SU MMARY TABLES FOR WINTER 20072008, CHARLES R. PERRY GREE N ROOF IN FLORIDA..............................................253 R HYDROLOGIC SUMMARY TABLES FO R SPRING 2008, CHARLES R. PERRY GREEN ROOF IN FLORIDA..............................................................................................254 S CUMULATIVE HYDROGRAPH AND HYDROLOGIC SUMMARY TABLES FOR SUMMER 2008, CRP GREEN ROOF IN FLORIDA.........................................................255 T CUMULATIVE HYDROGRAPH AND HYDROLOGIC SUMMARY TABLES FOR FALL/WINTER 2006-2007, YSC GREEN ROOF IN VIRGINIA.....................................256 U CUMULATIVE HYDROGRAPH AND HYDROLOGIC SUMMARY TABLES FOR SUMMER 2007, YSC GREEN ROOF IN VIRGINIA........................................................258 V CUMULATIVE HYDROGRAPH AND HYDROLOGIC SUMMARY TABLES FOR SPRING 2008, YSC GREEN ROOF IN VIRGINIA...........................................................259 W CUMULATIVE HYDROGRAPH AND HYDROLOGIC SUMMARY TABLES FOR LATE SUMMER 2008, YSC GREEN ROOF, IN VIRGINIA............................................260 X COMPARISON OF STORM DURATION/ SIZE, INTENSITY, STORM WATER RETENTION, AND REDUCTION IN RUNOFF INTENSITY FOR VA AND FL..........261 Y SUMMARY OF HYDROLOGIC CHARACTE RISTICS OF RAIN EVENTS AND CORRESPONDING GREEN ROOF RUNO FF FOR WATER QUALITY SAMPLING EVENTS FOR CRP GREEN ROOF IN GAINESVILLE, FLORIDA, 2008......................262 Z INDIVIDUAL STORM HYDROGRAPHS FOR WATER QUALITY SAMPLING EVENTS, CRP GREEN ROOF, FL, 2008...........................................................................263 AA WATER QUALITY DATA BY TIME INTERVAL FOR RAIN EVENTS SAMPLED IN FLORIDA AND VIRGINIA...........................................................................................266 LIST OF REFERENCES.............................................................................................................269 10

PAGE 11

BIOGRAPHICAL SKETCH.......................................................................................................273 11

PAGE 12

LIST OF TABLES Table page 2-1 Examples of variability of growing media constituents a nd physical properties among locations................................................................................................................ .43 2-2 Physical properties and organic matter content of Hydrotechs LiteTop growing medium from Hydrotechs specifications sheet.................................................................53 2-3 Composite water sampling date s and sources of water IN................................................61 2-4 Water Release Curves (Lysimeter Experiment) Sampling................................................63 2-5 Results of the type III tests of fixed effects for the PROC GLIMMIX model for water retention data for all bins over 24 weeks...........................................................................70 2-6 P-values for 2-way ANOVAs of water re tention by growing media and plant type and growing media-plant type inter actions for three 6-week periods................................70 2-7 Mean water retention (%) and standard error for each growing media type for each time period.................................................................................................................... .....71 2-8 Mean water retention (%) by plant type (irrespective of growing medium type) for each time period............................................................................................................... ..71 2-9 Results of PROC GLM comparing effect of plant type on retention within each growing medium for all time periods combined................................................................75 2-10 Bulk density and porosity of the thre e growing media before the study 7/3/08................99 3-1 Measured bulk density (BD) and porosity of the three growing media (B, H, U) before the study................................................................................................................114 3-2 Irrigation regime for the growing medium-plant bins.....................................................115 3-3 Composite sampling dates, time interval represented, water source and parameters measured....................................................................................................................... ...117 3-4 Lysimetric sampling dates, parame ters measured and leachate source...........................117 3-5 Results of the type III tests of fixe d effects for the PROC GLIMMIX model on TP load (mg P per bin). ........................................................................................................12 2 3-6 P-values for 2-way ANOVAs of growing media and plant type for four 6-week periods for TP load in bins (analyzed as mg per bin)......................................................123 12

PAGE 13

3-7 Growing media TP ash concentrations (ma ss basis) before and after green roof bin study, and after 2-years in situ on the CRP green roof in FL for H and 4-yrs on YSC green roof in VA for B growing medium........................................................................128 3-8 Growing media TP ash concentrations (are al basis) before and after the green roof bin study, and after being in situ on a green roof (for 4 years for B and 2 years for H growing medium).............................................................................................................128 3-9 Results of the type III tests of fixed effects for the PROC GLIMMIX model for TN areal load data from the 24 week green roof bin study....................................................132 3-10 P-values for 2-way ANOVAs of growing me dia and plant type for TN load for each of the 6-week periods.......................................................................................................132 3-11 Plant type effect on TN load by time period in mg N m-2...............................................133 3-12 Growing medium effect on TN load (mg N/m2) by time period.....................................134 3-13 Comparison values of TN load from ot her studies in Florida and the Mid-west............134 3-14 Results of the type III tests of fixed effects for the PROC GLIMMIX model for TKN load data from the 24 week study....................................................................................136 3-15 Results of water extractable -N analysis (1:10 (soil to wa ter) water extraction, shaken for 1 hour; N is reported on a concentrati on and mass basis) for growing medium samples taken before and after bin study.........................................................................140 3-16 Results of EA-TN analysis of air dr ied, ball milled soil, analyzed on the EA (Elemental Analyzer), reported on a mass ba sis for before and 1yr after green roof bin study and 2 yr and 4yr in situ on a green roof for B and H......................................140 3-17 CN ratios of the growing media before and after the study; and for B and H media, CN ratios after being on a green roof for 2 years and 4 years, respectively....................141 3-18 Pearson correlation coefficients among nutri ent loads in leachate, water retention and waterin...............................................................................................................................145 4-1 Range of runoff reduction for green roof s in various climates, using the KppenGeiger climate classification system (Peel et al 2007)....................................................151 4-2 Underlayment of the Hydrotech green roof. Source: Hydrotech Specification Sheet (2005)...............................................................................................................................162 4-3 Physical properties of Hydrotech LiteTop growing medium Source: Hydrotech Specification Sheet (2005)...............................................................................................163 4-4 Nutrients added to Hydrotech LiteT op growing medium. Source: Hydrotech Specification Sheet (2005)...............................................................................................163 13

PAGE 14

4-5 Example of time interval represente d by a sampling time and amount of runoff represented by the water sample during a storm (FL 6/23/08)........................................175 4-6 Example of calculation of load by time interval based on concentrations in grab samples taken at various time interv als during a rain event on 6/23/08..........................175 4-7 Characterization of 36 rain events captur ed in the wet season between July 7, 2007 to October 23, 2007, in terms of rain duration, rain amount (cm), green roof runoff duration, % volume storm water retention.......................................................................177 4-8 Results of analysis of 36 rain events in the wet season between July 7, 2007 and October 23, 2007 for delay in start and extens ion of end of runoff, the maximum rain intensity, maximum runoff intensity (Max RO) and the percent decrease in peak intensity............................................................................................................................178 4-9 Relationship between intensity (cm hr-1), duration (hr), total volume of rainfall or irrigation versus green roof runoff and rete ntion (%), for individual events shown in Figures 4-10 and 4-11, storms in the dryperiod/winter...............................................181 4-10 Characterization of 23 rain events ca ptured between July 12, 2006 and October 23, 2006 (Summer/Early Fall) in Merrifield, Virginia, for rain duration, rain amount (cm), maximum intensity, green roof runo ff duration, and maximum runoff intensity..186 4-11 Mean, median and range of stormwater re tention (%), delay in start and extension in end of runoff, reduction in peak intensity for 23 rain events between July 12, 2006 and October 23, 2006.......................................................................................................186 4-12 Differences in mean retention by rain event size for 91 storms between July 2007 and September 2008, in Gainesville, Florida...................................................................189 4-13 Differences in reduction in peak runoff by rain event size for 91 storms between July 2007 and September 2008, in Gainesville, Florida..........................................................189 4-14 Differences in mean retention by rain event size for 82 storms between December 2004 and September 2008 in Merrifield, Virginia...........................................................189 4-15 Differences in reduction in peak runo ff by rain event size for 82 storms between December 2004 and September 2008, in Merrifield, Virginia........................................189 4-16 Summary of mean concentra tions of nutrients and TSS for the six storms sampled in 2008 from the green roof (GR) and conventiona l roof (CR) runoff in Gainesville, FL..192 4-17 Summary of mean concentra tions of metals for six storms sampled in 2008 from the CRP green roof and conventional roof in Gainesville, FL..............................................192 4-18 Total load of nutrients measured in 6 storms 2008 from CRP green roof in Florida......194 14

PAGE 15

4-19 Total load of nutrients measured in 4 storms in 2007-2008 from YSC green roof in Virginia............................................................................................................................197 4-20 Total load of metals measured in 4 storms in 2007-2008 from YSC green roof in Virginia............................................................................................................................197 4-21 Results of the ANOVAs, pairwise t-test s for nutrients and TSS and TDS between conventional roof runoff and green roof runoff in Florida and Vi rginia; and pooled ttests between locations within c onventional or green roof type......................................201 4-22 Comparative values of green roofs, c onventional roofs and urban lawns from various locations...........................................................................................................................202 4-23 Areal loading rate (mg m-2) for nutrients sampled in gr een roof (gr) a nd conventional roof (cr) runoff from CRP roof in Gainesville, FL, 2008................................................203 4-24 Areal loading rate (mg/m2) for nutrients sampled in green roof (gr) and conventional roof (cr) runoff from YSC roofs in Merrifield, Virginia, 2007 and 2008.......................203 A-1 Irrigation regimen of the green roof bi ns for the hydrologic and nutrient green roof study in Gainesville, FL, 2007.........................................................................................221 D-1 Details of the PROC GLM model used for the ANOVA analyses for water retention for Weeks 1-6.................................................................................................................. .225 D-2 Effect of plant type on water retention within growing medium types. Results of individual ANOVAS within each growing medium type and time period......................225 G-1 Mean rates of water uptak e (% volumetric water conten t/hr) for different growing medium types (B, H, U)...................................................................................................234 G-2 Mean rates of water uptak e (% volumetric water content/ hr) for different plant types from water retention curves for Weeks 6, 12, 18 and 24 for B, H and U........................234 H-1 Evapotranspiration Rates in mm/day fo r Week 6Late Summer, based on change in mass over daylight hours on 9/6/07 and 9/7/07...............................................................235 H-2 Evapotranspiration Rates in mm/day fo r Week 12Fall, based on change in mass over daylight hours on 10/18/07 and 10/19/07................................................................235 H-3 Evapotranspiration Rates in mm/day fo r Week 18Early Winter, based on change in mass over daylight hours on 12/5/07 and 12/6/07.......................................................235 J-1 Physical properties reported for Hydrotech growing medium.........................................237 J-2 Macro and micronutrients reported for Hydrotech growing medium..............................237 K-1 Details of the PROC GLM model used for the ANOVA analyses for TN load data for Weeks 1-6.................................................................................................................. .238 15

PAGE 16

K-2 Details of the PROC GLM model used for the ANOVA analyses for TN load data for Weeks 7-12................................................................................................................. 238 K-3 Details of the PROC GLM model used for the ANOVA analyses for TN load data for Weeks 13-18...............................................................................................................2 38 K-4 Details of the PROC GLM model used for the ANOVA analyses for TN load data for Weeks 19-24...............................................................................................................2 39 K-5 Results of the ANOVA (degrees of free dom, sum of square, mean square and Fvalues) for TN load for weeks 1-6...................................................................................239 K-6 Results of the ANOVA (degrees of free dom, sum of square, mean square and Fvalues) for TN load for weeks 7-12.................................................................................239 K-7 Results of the ANOVA (degrees of free dom, sum of square, mean square and Fvalues) for TN load for weeks 13-18...............................................................................239 K-8 Results of the ANOVA (degrees of free dom, sum of square, mean square and Fvalues) for TN load for weeks 19-24...............................................................................239 L-1 Details of the PROC GLM model used for the ANOVA analyses for TKN load for Weeks 1-6........................................................................................................................240 L-2 Results of the ANOVA (degrees of freedom, sum of square, mean square and Fvalues) for TKN load for weeks 1-6................................................................................240 M-1 Nitrate loads per bin (mg per m2) for each week of the establishment period (Weeks 1-6)...................................................................................................................................241 N-1 NO3 load (mg m-2) for all plant-growing medium co mbinations and all time periods over the 24 week study....................................................................................................242 P-1 Summary of amount, duration, time of maximu m intensity (peak) for rain events and runoff measured between 7/9/07 and 7/2/08 for the Charles R. Perry Construction Yard..................................................................................................................................245 P-2 Charles R. Perry Construction Yard Green Roof--Summary of rain events and runoff volume and duration, and per cent of stormwater retention for 2007-2008 by season.....248 P-3 Delay in start of runoff after rain be gan, extension of runoff after rain stopped, increase in lag time to peak runoff and % reduction in peak runoff for rain events July 2007 to July 2008 for the Charles R. Perry Construction yard Green Roof............250 P-4 Summary table of mean, standard deviati on, median and range for all rain events and runoff from Charles R. Perry Constructi on Yard Green roof between July 2007 and July 2008..........................................................................................................................252 16

PAGE 17

Q-1 Mean, median and range of rain event size, hours since previous storm, and green roof runoff response for rain events between 12/20/07 and 3/18/08 for CRP roof, FL...253 Q-2 Mean, median and range of delay in st art of runoff, extension of runoff, and % reduction of peak intensity for rain ev ents between 12/20/07 and 3/18/08 for CRP roof in FL.........................................................................................................................253 R-1 Characterization of 14 irri gation events and 4 rain even ts captured between April 11, 2008 to June 7, 2008, for rainfall or irrigati on duration and volume, green roof runoff duration and volume, and % volume rainfall or irrigation retention in Florida...............254 R-2 Results of analysis of 14 irrigation events and 4 rain events between April 11, 2008 and June 7, 2008 for % decrease in peak inte nsity, increase in lag to peak, delay to start of runoff and extension of runoff dur ation past the end of storm event in Florida........................................................................................................................ ......254 S-1 Characterization of 17 rain events ca ptured between July 26, 2008 to September 15, 2008, in terms of rainfall duration, rainfall depth (cm), green r oof runoff duration and depth and percent rainfall retention in Florida..........................................................255 S-2 Mean, median and range of delay in start and extension in end of runoff, reduction in peak intensity for 17 rain events between July 26, 2008 and September 15, 2008 in Florida........................................................................................................................ ......255 T-1 Characterization of 36 rain events ca ptured between November 1, 2006 and March 8, 2007 (Late Fall/Winter) in Merr ifield, Virginia from the YSC green roof, for rain duration, rain amount (cm), maximum inte nsity, green roof runoff duration, and maximum runoff intensity................................................................................................256 T-2 Mean, median and range of stormwater re tention (%), delay in start and extension in end of runoff, reduction in peak intensity for 36 rain events captured between November 1, 2006 and March 8, 2007 in Virginia..........................................................256 U-1 Characterization of 12 rain events ca ptured in June 2007 and August 2007 (summer) in Merrifield, Virginia from the YSC green roof, for rain duration (hr), rain amount (cm), maximum intensity (cm hr-1), green roof runoff duration (hr), and maximum runoff intensity (cm hr-1)..................................................................................................258 U-2 Mean, median and range of stormwater re tention (%), delay in start and extension in end of runoff, reduction in peak intensity for 12 rain events collected in June and August 2007 in Virginia...................................................................................................258 V-1 Characterization of 10 rain events ca ptured in April and May 2008 (Spring) in Merrifield, Virginia from the YSC green roof, for rain duration (hr), rain amount (cm), maximum intensity (cm hr-1), green roof runoff duration (hr), and maximum runoff intensity (cm hr-1)..................................................................................................259 17

PAGE 18

V-2 Mean, median and range of stormwater re tention (%), delay in start and extension in end of runoff, reduction in peak intensity for 10 rain events collected in April and May of 2008 in Virginia..................................................................................................259 W-1 Characterization of 10 rain events ca ptured in August and September 2008 (Lat Summer) in Merrifield, Virginia from the YS C green roof, for rain duration (hr), rain amount (cm), maximum intensity (cm hr-1), green roof runoff duration (hr), and maximum runoff intensity (cm hr-1)................................................................................260 W-2 Mean, median and range of stormwater re tention (%), delay in start and extension in end of runoff, reduction in peak intensity for 12 rain events collected in August and September of 2008 in Virginia.........................................................................................260 X-1 Comparison of length of Virginia and Fl orida storms sampled during study period......261 X-2 Comparison of storm size of Virg inia and Florida storms sampled................................261 X-3 Comparison of average rain intensity of Virginia and Florida storms sampled..............261 X-4 Comparison of maximum rain intensity of Virginia and Florida storms sampled..........261 X-5 Comparison of storm water retention of Virginia a nd Florida storms sampled..............261 X-6 Comparison of max. stormw ater runoff intensity from green roofs in VA and FL.........261 Y-1 Characteristics of rain events sampleds torm start and finish times, duration, total rain (cm), mean intensity, maximum intensity and lag to peak (min).............................262 Y-2 Characteristics of green roof runoff for rain events sampled for water quality in 2008 in FloridaStart/Finish tim es of green roof runoff, runoff duration, runoff volume (cm depth across roof surface), mean runoff rate (cm hr-1), maximum runoff rate (cm hr-1) and lag to peak (min)...............................................................................................262 Y-3 Summary of reduction in maximum runoff rate, increase in lag to peak, delay in start of runoff, extension of runoff past the e nd of storm and % storm water retention by the green roof for storms used for wa ter quality sampling in 2008 in Florida................262 Z-6 Storm hydrograph of runoff from the green roof and a conventional roof of the same size based on rainfall and a runoff coefficient of 1. ........................................................265 AA-1 Concentrations of nitrate, ammonium, T SS and SRP from six rain events in 2008 in Virginia............................................................................................................................266 AA-2 Concentrations of nutrients, pH, TSS and TDS from rain events in 2007-08 in Virginia............................................................................................................................267 AA-3 Concentrations of metals sampled in 5 storms in Virginia from YSC green roof and conventional roof runoff in 2007 and 2008.....................................................................268 18

PAGE 19

LIST OF FIGURES Figure page 2-1 Complete Randomized Block Design of 3 growing media versus 3 plant types (and bare media) with 3 replicat es and 2 filter fabric contro ls and 2 empty bin controls (not depicted.)....................................................................................................................49 2-2 Photo of the 40 bins in a complete randomized block design............................................52 2-3 Rubric of plant groups in re lation to an assumed ET gradient..........................................56 2-4 Photos of Gaillardia grandiflora Goblin, Coreopsis lanceolata and Helianthus debilis Nutt. (photos by P.J. Alexander and K.Hill.).........................................................57 2-5 Photo of Sedum acre (photo S. Lang 2007), Delosperma cooperi and Portulaca grandiflora Hook...............................................................................................................58 2-6 Arachis glabrata (Perennial Peanut) photo by Roka 2004, Mimosa strigillosa (Sunshine mimosa) and Phyla nodiflora ............................................................................59 2-8 Photos on the left show s plants after 2 weeks of grow th compared to 5 weeks of growth................................................................................................................................65 2-9 Box plot distribution of water retention for Building L ogics (B), Hydrotech (H), and UCF (U) growing media for all weeks co mbined and all plant types included.................69 2-10 Differences in water retention among growing medium-plant type combinations for three of the 6-week time periods........................................................................................73 2-11 Amount of increase of wa ter retention by plant-growi ng medium type attributed solely to plant type, irrespectiv e of g-m type or time period.............................................76 2-12 Mean change in mass of bins for Hydrotech (upper line), Bu ilding Logics (middle line) and UCF (lower line) over 6 month study period......................................................78 2-13 Mean change in volumetric water cont ent in bins containing Building Logics (B), Hydrotech (H), and UCF (U) growing media over 6 months............................................79 2-14 Effect of plant type on water conten t in Building Logics growing medium over six months................................................................................................................................79 2-15 Effect of plant type on water content in Hydrotechs gr owing medium over 6 months................................................................................................................................80 2-16 Effect of plant type on water conten t in UCFs Black and Gold growing medium...........80 2-17 Water content curves for the thr ee growing media (B, H, U) over 72 hours post irrigation at the end of Week 6..........................................................................................82 19

PAGE 20

2-18 Comparison of change in volumet ric water content over 72 hours among three growing media types (B, H, U) with no vegetation in response to 1.27 cm ( ) irrigation during week 6 of the study (9/5/07-9/8/07).......................................................82 2-19 Top row shows rate of water uptake as increase of % of volumetric water content per hour for Weeks 6, 12, 18 and 24 (September 5-8, October 18-20, December 5-8 and February 2-4).....................................................................................................................83 2-20 Volumetric Water Content (%) over 72 hours for 3 plant types and bare medium (p, s, r, m) within Building Logics growing medium for Week 6...........................................86 2-21 Change in Volumetric Water Conten t (%) over 72 hours for 3 plant types and bare medium (p, s, r, m) within Hydr otech growing medium for Week 6................................86 2-22 Volumetric Water Content (%) over 72 hours for 3 plant types and bare medium (p, s, r, m) within UCF growing medium for Week 6.............................................................87 2-23 Change in Volumetric Water Conten t (%) over 72 hours for 3 plant types and bare medium (p, perennials; s, succulents; r, r unners; and m, bare media) within Building Logics growing medi um for Week 6.................................................................................87 2-24 Change in Volumetric Water Conten t (%) over 72 hours for 3 plant types and bare medium (p, s, r, and m) within Hydrotech growing medium for Week 6.........................87 2-25 Change in Volumetric Water Conten t (%) over 72 hours for 3 plant types and bare medium (p, s, r, and m) within UCF growing medium for Week 6...................................88 2-26 Mean rates of uptake and release (cha nge in volumetric water content per hour) for each plant-growing medium combinations for lysimetric sampling week 6. Error bars show standard deviation.....................................................................................................88 2-27 Plant Health Index (PHI) for perennial plants (Coreopsis lanceolata,Gaillardia pulchella and Helianthus debilis).......................................................................................92 2-28 Plant health index (PH I) of Runner Type species (Ar achis glabrata, Phyla nodiflora and Mimosa strigillosa)......................................................................................................92 2-29 Plant health index (PHI) of succulent type plants (Delosperma cooperii, Portulaca grandiflora, Sedum acre)....................................................................................................93 2-30 Grain size distribution of the th ree growing medium before planting...............................96 2-31 Organic matter by percent mass determin ed by Loss on Ignition for all three medium before the study, after 1 year and for Hydrotech after 2 years in situ on the Charles R. Perry Construction Yard green roof a nd for Building Logics, 4 years on the Yorktowne Square Condominiums in Virginia.................................................................96 20

PAGE 21

2-32 Bulk density of each growing medium-p lant type combination after the study period ended 11/8/08...................................................................................................................100 2-33 Photographs showing the root morphology of a runner type plants Mimosa strigillosa ........................................................................................................................101 2-34 Photographs showing the root morphology of a succulent, Delosperma cooperii ........102 2-35 Photographs showing the root morphology of a perennial type plant, Coreopsis lanceolata. ........................................................................................................................102 2-36 Photographs of the root systems of a self-recruited plant afte r bins were abandoned, Tropical Crab Grass.........................................................................................................103 3-1 Complete Randomized Block Design of 3 growing media vs. 4 plant types with 3 replicates and 2 filter fabric controls a nd 2 empty bin controls (not depicted)...............114 3-2 Weekly TP loads (lin es) and precipitation/irriga tion volumes (bars) for the establishment period (first six weeks of growth).............................................................119 3-3 TP concentrations in leachate from a ll 40 bins, collected over four 6-week periods......120 3-4 Precipitation for the four different 6-week time periods between 7/23/07 1/18/08......120 3-5 TP load as mg P/ kg dry soil from a ll 40 bins, based on leachate concentrations and volumes collected passively over 6-week periods up to 24 weeks..................................121 3-6 Cumulative hydrograph of precipitation during 24-week study period; sampling dates indicated by circles..........................................................................................................1 24 3-7 Mean values of TP load (with standard error) for each growing media-plant combination for the entire study period...........................................................................125 3-8 Plant effect on cumulative TP load in mg P m-2 for Building Logics growing medium......................................................................................................................... ...127 3-9 Plant effect on cumulative TP load in mg P m-2 for Hydrotech growing medium..........127 3-10 Plant effect on cumulative TP load in mg P m-2 for UCF growing medium...................127 3-11 Plant Health Index (PHI) for perennial plants (Coreopsis lanceolata,Gaillardia pulchella and Helianthus debilis).....................................................................................129 3-12 Plant health index (PH I) of Runner Type species (Ar achis glabrata, Phyla nodiflora and Mimosa strigillosa.....................................................................................................129 3-13 Plant health index (PH I) of succulent type plants (Delosperma cooperii, Portulaca grandiflora, Sedum acre)..................................................................................................129 21

PAGE 22

3-14 Cumulative TN load leached from va rious plant types with in B growing medium over 24 weeks..................................................................................................................130 3-15 Cumulative TN load leached from va rious plant types with in H growing medium over 24 weeks..................................................................................................................131 3-16 Cumulative TN load leached from the various plant types within U growing medium over 24 weeks..................................................................................................................131 3-17 Plant type effect on TN load by time pe riod (significant differences shown in Table 3-11).................................................................................................................................133 3-18 TN areal load by time period for growi ng media type (significant differences shown in Table 3-12)...................................................................................................................134 3-19 Cumulative TN load (mg N/m2) over 24 weeks for each of the plant-growing medIum combin ations. ....................................................................................................134 3-20 Effect of plant type within Buildi ng Logics growing medi um over time on TKN load...................................................................................................................................136 3-21 Effect of plant type within Hydrot echs growing medium over time on TKN load........137 3-22 Effect of plant type within UC Fs growing medium over time on TKN load.................137 3-23 TKN areal loading rates for all plant-grow ing medium combinations for total load at the end of establishment period (6 weeks).......................................................................137 3-24 Effect of plant type on nitrate load s during the establishment period for all plant growing medium combinations........................................................................................139 3-25 Photo of leachate from Week 5t op bucket contains leachate from U growing medium, bucket on left contains leacha te from B growing medium, and bucket on right is leachate from H growing medium.......................................................................142 3-26 TSS (mg/L) concentrations measured in composite sample from Week 3 (no algae).....142 3-27 TSS (mg/L) concentrations in leachat e from the lysimeter experiment collected Week 18 (no algae)..........................................................................................................143 3-28 Lysimeter Week 1-Effect of time and plant type on TP concentrations in leachate from H (Time 1 and 2), B and U growing media (Time 1, 2 and 3)................................144 4-1 Photo of the Charles R. Perry Constr uction Yard after bituminous water proof layer was cold applied, photo taken from the third floor conference room, March 2007........162 4-2 Photo of the installation of plants on the CRP green roof. Photo was taken from the third floor conference room of Rinker Hall in April 2007...............................................162 22

PAGE 23

4-3 Green roof installed by Building Logics in 2002, planted with 8,400 sedum plugs ( S. album, S. sexangular, and S. reflexum ). Photo taken May 2005.....................................166 4-4 Conventional Roof at Yorktowne Squa re Condominiums, built in 2002, at the same time as the green roof on the adjacent building...............................................................167 4-5 Example of analysis of individual hydrographs fro m a green roof and modeled conventional roof runoff based on rainfall to determine increase in delay in start of runoff and the increase in lag to peak and th e increase in the extension of runoff from the green roof past the end of the storm event.................................................................171 4-6 Example analysis of an individual hy drograph for storm water volume reduction due to the presence of a green roof. ......................................................................................171 4-7 Example analysis of individual rain event and green roof hydrographs for reduction in peak storm runoff.........................................................................................................17 2 4-8 Charles R. Perry Construction Yard Green Roof on June 11, 2007, 12 weeks after establishment.................................................................................................................. ..176 4-9 Cumulative green roof runoff hydrograph co mpared to cumulative rain and irrigation for wet season, July 7, 2007 to October 23, 2007........................................................178 4-10 Hydrologic response of green roof to individual irrigation and rain events on a weeklong time scale in the dry peri od/winter months, 1/6/08 to 1/14/08....................180 4-11 Hydrologic response to green roof to irrigation versus rain events in the dry period/ winter months, 2/ 6/08 to 2/14/08.....................................................................................180 4-12 Hydrologic response of green roof to irrigation (1.27 cm) and a small precipitation event (0.5 cm) with wet antecedent moisture conditions on 2/7/08................................181 4-13 Comparison of % water re tained directly af ter 1.2 cm of irri gation over 12 hours in the controlled bin studywith re tention after irrigation in situ on the Charles R. Perry Construction Yard green roof during the winter season 2007.........................................183 4-14 Cumulative stormwater runoff hydrograph for green roof and rainfall at YSC, Merrifield, VA green roof, Summer/Early Fall 2006 (7/12/2006-10/23/2006)...............185 4-15 Mean nitrate, ammonium, SRP and TSS concentrations per storm event for green roof runoff (6 storms) and conventional r oof runoff (2 storms) in summer 2008 from the CRP green roof in Gainesville, FL.............................................................................193 4-16 Metals concentrations (A) Aluminum, B) Iron, C) Copper and D) Zinc) in CRP green roof (gr) and conventional roof (cr) r unoff in 2008 in Gainesville, FL...........................194 4-17 Mean concentration and SD for nutrients and metals measured in YSC green roof (gr) and conventional roof (cr) r unoff in Merrifield, VA 2007 and 2008 ......................196 23

PAGE 24

C-1 Image of STALITE Light weight Aggregate specificati ons sheet for properties and gradations of the material fo r structural applications......................................................223 C-2 Image of STALITE Li ghtweight Aggregate specifications sheet for physical characteristics of the material for structural applications................................................224 E-1 Changes in water content for growing va rious plant types (p, r, s, m) in growing medium types B and H in Week 12, up to 72 hours post-irrigation with 1.27 cm water. Note 1.27 cm of rain fell between hours 48 and 60, causing an increase in water content....................................................................................................................226 E-2 Changes in water content for growing va rious plant types (p, r, s, m) in growing medium U in Week 12, up to 72 hours postirrigation with 1.27 cm water. ..................227 E-3 Changes in water content in A) Buil ding Logics, B) Hydrotec h and C) UCFs Black and Gold growing media over 72 hours post-ir rigation of 1.27 cm water in Week 18...228 E-4 Changes in water content in A) Buil ding Logics, B) Hydrotec h and C) UCFs Black and Gold growing media over 72 hours post-ir rigation of 1.27 cm water in Week 24...229 F-1 Changes in water in retention in A) Building Logics, B) Hydrotech and C) UCF Black & Gold growing media over 72 hour s post-irrigation of 1.27 cm water in Week 6 (9/5-9/8/2007).....................................................................................................230 F-2 Changes in water retention in A) Building Logics, B) Hydrotech and C) UCFs Black and Gold growing media over 72 hours post-irrigation of 1.27 cm water in Week 12...........................................................................................................................231 F-3 Changes is water retention in A) Building Logics, B) Hydrotech and C) UCF growing media over 72 hours post-irrigation as a percent of water applied in response to a 1.27 cm irrigation event in Week 18 (December 5th, 2007) of the study period...............................................................................................................................232 F-4 Water retention in A) Building Logics B) Hydrotech and C) U growing media over 72 hours post-irrigation (1.27 cm) as a % of water applied, in week 24 of the study.....233 I-1 Changes in total OM content before and after green roof study a nd in the top particle size fraction (> 2 mm)......................................................................................................236 I-2 Changes in grain size dist ribution of B, H and Us grow ing medium after 1 year in green roof bins and 4 years and 2 years on a roof top, for B and H, respectively...........236 Q-1 Cumulative hydrograph for 12/20/07 to 3/18/08 for CRP green roof in FL....................253 S-1 Cumulative green roof runoff compared to cumulative rainfall and irrigation for wet season months (June 7th to September 15th) in 2008, hyetograph shown in blue...........255 24

PAGE 25

T-1 Cumulative hydrograph of green roof runoff versus cumulative precipitation for YSC, Merrifield, VA for fall/early winter 2006 2007 (October 24, 2006 to February 7, 2007)............................................................................................................................256 U-1 Cumulative hydrograph of rainfall and runoff from the YSC green roof in Merrifield, VA for June and August 2007 (July data was unavailable due to technical difficulties).......................................................................................................................258 V-1 Cumulative hydrograph of green roof r unoff and rainfall for April and May 2008 for YSC green roof in Merrifield, VA...................................................................................259 W-1 Cumulative hydrograph of rainfall and green roof runoff for August-September 2008 for YSC greenroof in Merrifield, VA..............................................................................260 Z-1 Storm hydrograph for rain event on 6/23/08....................................................................263 Z-2 Storm hydrograph for rain event sampled on 6/25/08.....................................................263 Z-3 Storm hydrograph for rain event sampled on 6/26/08.....................................................264 Z-4 Storm hydrograph for rain event sampled on 6/30/08.....................................................264 Z-5 Storm hydrograph for rain event sampled on 7/8/08.......................................................265 25

PAGE 26

Abstract of Dissertation Pres ented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy GREEN ROOFS AS AN URBAN STORMWATER BEST MANAGEMENT PRACTICE FOR WATER QUANTITY AND QUALITY IN FLORIDA AND VIRGINIA By Sylvia Bertie Lang May 2010 Chair: Name Mark W. Clark Co-chair: Samira Daroub Major: Soil and Water Sciences Green roofs are well known as an urban st ormwater volume Best Management Practice (BMP) in northern climates, but information regarding water quality benefits/impacts and optimal plant-growing media combinations for gr een roofs in the sub-tropics is lacking. The objectives of this study were to: 1) determine th e optimal plant-growing medium combination for water and nutrient retention and 2) characteri ze green roofs capability to reduce stormwater volume and peak runoff and 3) determine whethe r green roofs behave similarly (as a sink or source) for nutrients and metals, in Florida and Virginia. Objectives were tested via i) a green roof bin study in Florida and ii) paired gr een roof studies in Florida and Virginia. The results of the green roof bin study showed that growing medium type affected water retention and nutrient leaching more than plant type. Water retent ion ranged from a low of 24% for Building Logics (B) medium with no vegeta tion to a maximum of 83% for UCF (U) growing medium with perennials. Differences among media were attributed to phy sical charact eristics of the media: pore-size distribution and OM content. Plants increased water retention by 7-10% above bare medium, with perennials having the gr eatest, and succulents having the least effect. 26

PAGE 27

TP and TN loads for the establishment peri od (initial 6-weeks) ranged from 110 mg P m-2 (Uperennials) to 1800 mg P m-2 for (H-succulents or bare medium); and from 190 mg N m-2 (Urunners) to 1800 mg N m-2 (H-succulents) The majority (60-90%) of the nutrient load leached out in the establishment peri od of the 24-week study period. Green roofs monitored in Virginia and Florid a behaved similarly for water retention and peak reduction. In both Florida and Virginia, sm all rain events (< 0.254 cm), had significantly (p<0.05) higher mean retention (79% and 98%, Flor ida and Virginia respectiv ely) than large rain events (26% retention and 72%, Florida and Virginia, respectively). Green roofs significantly (p<0.05) reduced the peak runoff in both Florida (94% for small and 60% for large events) and Virginia (100% for small and 79% for large events ). Green roofs behaved similarly for nutrients (sources for phosphorus, sinks for N-NO3, and buffered pH) and different ly for metals. Al and Fe levels were significantly higher (p <0.05) in green roof (GR) runo ff than conventional roof (CR) runoff in Florida; while Pb was significantly lower (p<0.05) in Virginia GR runoff. In conclusion, this study found that green roofs in the subtropics are better suited as a stormwater volume control BMP, than a nutrient control BMP. 27

PAGE 28

CHAPTER 1 INTRODUCTION AND LITERATURE REVIEW Introduction Green roofs have been in use in Europe for approximately 40 years as a stormwater best management practice, while in the United States they are just becoming recognized as a new way to reduce stormwater at the source. The eff ectiveness of extensive green roofs to reduce stormwater quantity and treat it for water quality has been studied primar ily in cool temperate climates such as Sweden, Germany, Michiga n, Ontario, Oregon, and Pennsylvania (Berndtsson et al., 2006; Van Woert et al., 2005; Liptan and Strecker, 2003; and DeNardo et al, 2003.). Energy savings gained by using green roofs has b een explored in warmer climates such as the Mediterranean and the tropics (Theodosiou, 2003; and Wong, 2003), but currently little information is available regarding plant hard iness, optimal substrate depth and type for subtropical regions such as Florida (Hardin, 2006). The subtropics present a unique situation for green roofs, because of higher precipitation rates and potential evaporation ra tes, higher temperatures, occa sional hurricanes, and even the occasional winter frost. This unique climate al so attracts humans, making Florida one of the fastest growing states in the United States. As urban areas become more densely populated, space for stormwater BMPs becomes limited, maki ng green roofs a good option as an alternative BMP (Moran, 2003; Liptan and Strecker, 2003). Green roofs take advantage of already existing rooftop space to reduce the source of urban stormwater by detaini ng and evaporating the rainwater. Green roofs also have the pot ential to reduce urban stormwater pollution by adsorbing particles from wet and dry atmos pheric deposition. Green roof water quality depends on the soil thickness, medium, vegetation and drainage sy stem in place (Berndtsson et al., 2006); green 28

PAGE 29

roofs have been cited both as a sink and as a source for nitrogen, phosphorus and metals. German researchers (Steuslof, 1998 in Berndtsson et al., 2006) found that green roofs mitigate for elevated levels of metals and nutrients largely based on their ability to detain the water, while Swedish studies (Berndtsson et al., 2006) f ound that nitrogen is re leased from roofs. Currently there are few studies regarding th e performance of green roofs as an urban stormwater BMP to mitigate water quality in Florida (Hardin et al., 2006), and limited information exists for the south-Atlantic region of the United States in ge neral (Florida, Georgia, North Carolina, South Carolina, Virginia, We st Virginia, Maryland, Washington, D.C. and Delaware) (Moran et al., 2005, De Nardo et al., 2005; and Carter et al., 2005). (Note: The midAtlantic states traditionally re fer to New York, Pennsylvania, Ne w Jersey, and usually Delaware and Maryland, Virginia are part of the south-At lantic region and techni cally lies within the northern extent of the Cfa Koppen system of climate classification) (Pickett et al., 2000). Rooftops as a Source of Urban Stormwater Urban areas contribute large amounts of st ormwater runoff and pollutants due to impervious surfaces. In a highly urbanized city setting in the USA, typically 72% of the land area is impervious (Schueler, 2001); 40% bei ng comprised of rooftops (Urbonas, 2001; and Liptan, 2003) and 60% consisting of car habitat. Rooftop runoff poses a greater threat to water quantity in urban watersheds than rural watersheds. This is because the runoff enters re ceiving water bodies more rapidly in urban areas than in rural settings due to the greater connec tivity of roofs to gutters and sewers. The presence of pavement impedes infiltration to groundwater increasing the proporti on of water going to surface overland flow and increasing the veloc ity of the runoff. When surfaces are paved, vegetation that originally provided interception and evapotranspiration is removed, and natural depressions in the landscape, which normally de tain 50% of the runoff, are eliminated (Dunne 29

PAGE 30

and Leopold, 1978). The volume and rate at which th e runoff is delivered to the receiving water body is greatly increased (Andoh, 1997) resulting in a re duction of the hydr ologic response time and greater recurrence of floods. Rooftops contribute to stormwater pollution via two mechanisms, one is through the release of constituents from the roofing materials usedsuch as zinc, copper, polyaromatic hydrocarbons (PAHs), cadmium or lead (Cla rk, 2001) and secondly from atmospheric depositionfor example nitrogen, phosphorus and ev en pesticides (Moran, 2003 and Dietz, et al., 2005). Researchers in Michigan looking for sources of stormwater contaminants found rooftops to be the largest source of dissolved metals, while parking lots contributed the PAHs (Clark, 2001). Green roofs reduce contamination from rooft ops by reducing the amount of water leaving the roof (De Nardo, 2003; Van Woert, 2005; Li ptan, 2003; and Villareal, 2004) and by plant uptake and transformations of N and P deposite d atmospherically or added by fertilization (Berndtsonn, et al., 2006). Green Roof Design: Intensive, Extensive and Florida Green roofs are generally categorized into tw o types: intensive green roofs and extensive green roofs. Intensive roofs are cap able of sustaining plants such as shrubs and small trees 1 m to 5 m in height, requiring soil de pths of 0.4 m and upward and must sustain a higher load bearing and may require more irrigation and fertilization (Moran, 2003). Extensive green roofs generally consist of a re gular roof covered with an asphalt coating (the bituminous layer), overlain by a protective root barrier layer (copp er wire may be woven through this layer or root resistan t polyester can be used to prev ent root penetration). Overlying the root barrier layer is the moisture retention mat or fabric which holds the excess precipitation that percolates through the growth medium and drai nage layer. The moisture retention fabric is 30

PAGE 31

often 0.75 cm thick and may have an eggshell carton type filter above it that also has retention capabilities. Above these laye rs is the growing medium. Growing media for green roofs vary, but are characteristically light in weight, low in organic matter, and preferably consist of local materials (Emilsson, 2005). The vegetation on extensive roofs needs to be both inundation and drought-tolerant, as well as tolerant to heat and cold. In temperate climates (Sweden, Germany, Michigan, Pennsylvania a nd Virginia) extensive roofs are often planted with a variety of sedums (Sedum album, S. reflexum, S. sexangulare, S. spurium, S. album, S. kamtschaticaum and S. pulchellum ) or may include Delosperma spp. (D. aberdeenense, D. basuticum, D. cooperi, D. nubigenum) (Snodgrass and Snodgrass, 2006; Berndstonn et al., 2006; Van Wo ert et al., 2005; DeNardo et al. 2005; and Moran, 2003). Extensive green roofs require little maintenance. They have shallow so il substrate depths usually between 5 cm to 15 cm and are often plan ted with native vegeta tion that can survive on precipitation alone, therefore they usually can survive without irrigation or fertilization. Vegetation heights are low, ranging from 5 cm to 8 cm (Moran, 2003). Extensive green roofs in Florida have been more successful in deeper substr ates than those typically used in cool climates, likely due to higher evapotranspi ration rates in the subtropics (personal communication Martin Wanielista and Mike Hardin, 2006). Stormwater Best Management Practices (BMPs) Stormwater BMPs can be struct ural and non-structural. Non-st ructural BMPs mimic nature and include techniques such as marking-off areas for conservation before building, keeping imperviousness and car habitat to a minimum and diminishing the footprint of buildings on the landscape. Green roofs help reduce the footprint of a bu ildingby placing the accompanying stormwater BMP on the roof and help replace a small portion of the original vegetation that once existed in the buildings place. 31

PAGE 32

Structural BMPs are man-made works that cont ribute to detaining flow of stormwater and treating the stormwater for water quality. They can be small distributed structures, such as vegetated swales, bioretention areas or rain gard ens, or mini-wetlands distributed through out the watershed that promote infiltration. Rainwater is evapotranspired and/or reused at the source by disconnecting impervious surfaces from stor mwater conveyance systems and by creating absorbent surfaces (Graham, 2004). Source control BMPs may include improved site design of new buildings and housing complexes, that include marking off and setti ng aside natural areas, re ducing impervious areas and footprints of the buildings and roads, re-inf iltrating rooftop runoff onsite, or collecting it in rain barrels or cisterns for later re-use, a nd minimizing runoff from roofs and pavements via green roofs and pervious pavement (Graham, 2004). Green roofs are bot h a source control as well as a structural control for stormwater. Green Roofs Role in Stormwater Retention and Detention in the Hydrologic Cycle Green roofs retain and detain water from rain fall events and assist in evaporation. High rates of evapotranspiration from a vegetated roof can reduce the annual ru noff to less than half the amount of incoming precipitation (Berndtsson, 2006). Rainfall detention is defined as water temporarily detained after a rainfall event to be la ter released it at a later time; and retention as the fraction of rainfall that is retained on the roof that eventually is evaporated from the growing medium or transpired by the plants. For exampl e, De Nardo (2003) found that green roofs played an important role in attenuati ng the peak runoff by detaining the water for a period of time. She found that green roofs reduced tota l runoff over a whole year by 40%. Villareal et al. (2004) in Swede n, tested the role of green ro ofs as one link in a chain of connected BMPs. The green roofs were retrofit on to municipal buildings and its stormwater runoff entered a chain of BMPs that included op en channels, inner courtyard detention ponds and 32

PAGE 33

miniature treatment wetlands in an urbanized setting. By simulating runo ff for different sized storms, Villareal et al. (2004) f ound that in the absence of green roofs, peak flows and the total inflow volumes to the last BMP in the series of BMPs increased significantly for the same storm event return periods. The authors concluded that to be able to offer the same level of retention and attenuation without green r oofs, the pond complex would have to be significantly larger (Villareal et al. 2004). They also found that gr een roofs played a more important role in mitigating small storms, while the stormwater pond had a greater relative importance in large storms. Nutrients and Metals In a second study in Sweden, Berndtsson et al (2006) examined green roof runoff quality from green roofs over bicycle parks in Malmo, Sweden and green roofs covering a school in Augustenborg, Sweden. They found that the green r oof on the school acted as a sink for nitrogen, reducing TN by 58%, but behaved as a source for phosphorus and potassium. They also found that the green roofs role as a sink or source varied either due to age, parent materials or input levels with regards to metals in runoff. The newer green roof covering the bicycle parks in Malmo acted as a sink for lead, while the older green roof on top of the school in Augustenborg behaved as a source for lead. These results co rroborated those of a Ge rman study on vegetated roof research plots at the Techni cal University of Berlin, where gr een roof plots were capable of acting as a sink when input levels were elevated (the green roof plots retained 95%, 88%, 80% and 68% of the loads of Pb, Cd, NO3 and PO4 over a 3-year period), however when input levels were low, then the green roof acted as a sour ce (Kohler et al., 2002 in Berndtsson et al., 2006). Green Roofs in Florida Green roofs in the subtropics are purported to need 15 cm (6 in) of soil to sustain the currently recommended native plan t for green roofs in central Fl orida (Wanielista and Hardin, 33

PAGE 34

2005) and plants native to Florid a or those that are naturalize d to the Florida climate were hypothesized to be the plants that would be best suited to surviving on a green roof in Central Florida with the least amount of maintenance necessary (limited irrigation and no fertilization). The list of plants below was originally created by the curator of the arboretum at UCF, Martin Quigley, and shows mainly native or plants adapte d to the Florida climate that were considered for use in the first extensive gree n roof in Central Florida create d at UCF; the first six plants were eventually chosen for UCFs green roof in 2006 and those plants succ essfully survived hot summers and cool winters with minimal irrigation (Hardin and Wanielista, 2006). Lonicera sempervirens (Coral Honeysuckle) Gaillardia pulchella (Firewheel Daisy) Myricanthes fragrans (Simpsons Stoppers) Muhlenbergia capillaries (Muhly Grass) Helianthus debilis (Beach or Dune Daisy) Salvia coccinea (Tropical Sage) Monarda punctata (Spotted Beebalm) Hamelia patens (Firebush) Erythrina herbacea (Coral Bean) Mimosa strigillosa (Powderpuff) Solidago spp (Goldenrod) Hypericum hypericoides (St. Andrews Cross) Oenothera laciniata (Cutleaf Primrose) Scoparia dulcis (Sweet Broom) Phyla nodiflora (Carpet Flower) Scutellaria integrifolia (Rough Scullcap) Justification Green roofs have potential in the sub-tropics to be a mechanism to reduce stormwater runoff at the source, attenuate the peak runoff rates and increase the lag time to peak of concentration in urban areas. Due to Floridas un ique climatic conditions, such as intense heat, drought, intense and frequent rains, its green roofs will depend on plants that need a deeper substrate than temperate green r oofs (Wong et al., 2003). Deeper s ubstrate requires roofs to have a higher structural load bearing capacity, resulting in increased co sts. A growing medium that is 34

PAGE 35

light-weight and has water retenti on capabilities that favor green r oof plant health with minimum irrigation and fertilization could ma ke the use of green r oofs feasible in Florid a at reduced costs. Another potential obstacle with green roofs in the subtropics is that heat tolerant plants occasionally die during an unusual frost, as was the case in experimental r oofs in Mexico (pers. comm.. Green Roof Conference, 2004.) More pub lished articles and co ntrolled studies with replicates or field studies with paired roofs are needed in th e sub-tropics to determine which plant types and respectiv e growing media types would yield the most viable green roof for Florida cost effectively. Additi onally quantitative data regardi ng the effects of green roofs on urban stormwater hydrology and water quality ar e needed to determine what would be the benefits or drawbacks of using green roofs as an urban stormwater BMP in north Central Florida or northern Virginia. Data in the literature regarding the effect of green roofs on water qua lity is contradictory and inconsistent. More research is needed with continuous flow measurements of precipitation events in the field with a pair ed roof study between a conventiona l roof and green roof of the same size in subtropical climates. Results could help suggest whether or not green roofs improve water quality by reducing the total volume of outflow; or whether green roofs have the capacity to actually treat the water passi ng through the growing medium and root zones of the plants; or whether green roofs will act as a nutrien t and metal source in this climate. Data generated from this study regarding wa ter quality and quantity, optimal growing medium type and plant type will be available fo r economic projections by land-use managers as to the monetary value of green roofs as a BMP in an urban area. Stormwater managers can use this data to answer questions such as: How mu ch BMP credit should be allocated for green roofs for water quantity and/or water quality? Th is research on green roofs aims to conduct 35

PAGE 36

experiments which will generate results that will help address the following questions regarding water quantity for Florida and Virginia: 1) What role do green roofs play in reducing stormwater runoff? 2) What role do green r oofs play in increasing the lag tim e of stormwater runoff entering its receiving water body? 3) What amount of covera ge of green roofs is necessary to achieve a significant reduction in stormwater volum e entering a receiving water body? For water quality, the research aims to addre ss these questions: 1) Do green roofs act as a source or sink for N and P? 2) How do green r oofs behave differently at two extremes of the same climatic zone, with regards to being a sink or source for N and P? 3) What amount of green roof coverage would be necessary in an urban watershed before a water quality change could be measured in the receiving water body? For Florida only, this research aims to answer the following questions: 1) Which plants can withstand the extreme drought conditions, heavy rains and high temperatures as well as occasional frost typical of the Florida sub-tropi cal zone the best, with the least amount of irrigation? Research Goals The goals of the research are to: 1. Determine the optimal plant-growing medium co mbination for Florida out of three growing media used in Florida and three plant types. 2. Characterize the effect of green roofs on urban stormwater hydrology for Florida and Virginia. 3. Characterize the effect of green roofs on the wa ter quality of runoff for Florida and Virginia. Hypotheses and Research Objectives The hypotheses specific to green roof growing media and plant types are: H1: Successful green roof plants in the sub-tropics (defined as those plants that can survive drought, high heat and intense ra in without irrigation after esta blishment) will consist of a 36

PAGE 37

mixture of (structural) plant types different than those used in cool temperate climates, because of the differences in climatic conditions between the two climates. H2: Plant type will have a greater affect than growing medium type on water retention and nutrient retention in the green r oof system, especially in the sub-tropics, because of the role of plants in nutrient and wa ter uptake during rapid plant growth and the ability of the plants to evapotranspire water out of the green roof system. H3: The highest rates of nutrient leaching will oc cur during the establishment period of a green roof as compared to later in the green roofs life, because th e growing medium at this time acts as a nutrient source with little influence of plants. H4: An optimal plant-growing medium combination exists for the sub-tropics that will minimize irrigation requirements and minimize nutrient leaching in green roofs. H5: The overall benefit of green roofs for wate r retention, peak runoff attenuation and increases in lag time to peak runoff will be less in green roofs in Florida than in Virginia, because of higher peak precipitation rates, greater total volume of rain events, and greater recurrence of convective storms in a humid subtropical climate (Florida) than in a transitional humid subtropical/continental climate (Virginia). H6: Green roofs influence on water quality in stormwater runoff will be similar in both in Virginia and Florida and will act as a si nk for nitrogen and source for phosphorus and sediment, and may be either a sink or source for metals, when compared to conventional roofs. Objectives specific to green roof growing me dia and plant types for Florida and regarding water quantity and quality in bot h Florida and Virginia are: 1. Quantify the effect of growing medium type and plant type on storm water retention capacity in Florida and determine whether growing me dium type or plant type plays a more significant role in water retention. 2. Quantify the amount of nutrient (TSS, TN [TKN+ NO3], TP) leaching from green roof bins with different types of growing media and pl ants and a) determin e which plant growing medium combination has the least amount of leaching (greatest amount of nutrient retention) and b) whether plant or growing media type is most influential in nutrient leaching in the subtropics. 3. Quantify the amount of nutrient leaching durin g the establishment period and after the establishment period, determine whether there is a trend of decrea sing leaching over time. 4. Observe the plant health of three different plant types growing in different growing media types and determine whether any plant-growi ng medium combination can survive without irrigation or fertilization after the establishm ent period and if so, which plant-combination has the best plant health wit hout irrigation or fertilization. 37

PAGE 38

5. Quantify the amount of stormwater retention, p eak runoff attenuation, and increase in lag time to peak, if any, due to the pr esence of the green roof in both Florida and Virginia and compare these rates. 6. Measure water qualityconcentra tion and load of nutrients (NO3/NO2, PO4, TDS, TSS) and metals (Cd, Zn, Al, Fe, Cu) in green roof and conventional roof runoff in Florida and Virginia to determine whether the green roofs are acting as a sink or source for N, P and metals, and whether they are behaving the same way (as a sink or source) for the same parameters in both regions (Florida and Virginia). Study Approach The hypotheses were tested through a three part study consisting of: a) a controlled green roof bin experiment in Florida, b) a green roof monitoring study in Florida, and c) a paired green roof study in Virginia. The controlled green roof bin study was used to test the first three hypotheses and objectives, which ar e those related to: 1) determining the affect of growing medium versus plant type on water retention and nutrient retention a nd identifying successful plants in the sub-tropics; 2) quantifying the ra tes of nutrient leaching during the establishment period of a green roof and determining any trends or differences over time in the beginning of the green roofs life; and 3) identifying the optimal gr owing medium-plant type combination for extensive green roofs in north central Florida (that will minimize irrigation requirements and minimize nutrient leaching in green roofs). Specifically, the contro lled green roof bin study was carried out using three commercial growing media types and several plants previously suggested for green roofs in Florida during a 6-week establishment period with regular irrigation, followed by 6 months of no-irrigation, to determine whic h plant types can potentially survive heat, drought, frost and inundation wit hout irrigation after becoming es tablished (defined as 6weeks) during the peak growing season. The paired green roof/conventi onal roof study in Virginia and the green roof monitoring in Florida provided data to meet Objectives 5 and 6, which were to 1) ch aracterize the effect of green roofs on urban stormwater hydrology and 2) characterize the effects of green roofs on 38

PAGE 39

water quality, for Florida and Virginia. Specifi cally, this was accomplished in Virginia by monitoring the volume of runoff from both a green roof and conventional roof of the same age and size in Merrifield, Virginia continuously between 2005 and 2008. In Florida, the Charles R. Perry Construction Yard Green Roof was monitore d continuously for stormwater runoff quantity throughout 2007 and 2008. The hydrology data was used to characterize the water detention and retention capabilities of these two green roofs. The water quality objective was met for these roofs by sampling runoff from disc rete storms over a 2-year peri od and analyzing the runoff for nutrients and metals and comparing total loads from each roof type. Dissertation Format The dissertation is written as a series of chapters on the in dividual studies in a manuscript format. Chapter 2 presents the results of the hydrologic component of the optimal plant-growing medium combination study for north Central Fl orida. The hydrologic component of the study characterizes the water release curves/water retention capacity of the various soil-plant combinations which was measured by weighing i ndividual bins regularly and measuring filtrate throughout the study period. The hydrologic compone nt also describes th e evapotranspiration rates of the various plant types. Chapter 3 cont ains the results of the optimal plant-growing media combination study with regards to nutri ent retention and focuse d on characterizing the nutrient levels in leachate of the various so il-plant combinations primarily during the establishment phase of the green roof, defined he re as the initial six (6) weeks of watering and growing of the plants, beginning in mid-summer, followed by additional samples taken at 12, 18 and 24 weeks. Leachate was analyzed for TSS, TP and TN in samples taken directly after controlled irrigation events co llected at weeks 1, 6, 12, 18, 24 and 52 as well as in composite samples representing 6 week periods of time up to 6 months. Based on the results of this study, a 39

PAGE 40

conceptual model was created to predict the r unoff quantity and quality from the Florida and Virginia green roofs that were studied in situ for the paired roof study. Chapter 4 contains the results of the green roof monitoring st udy in Florida, and the paired green roof study in Virginia, as well as a compar ison of the two roof systems in the two extremes of the sub-tropical climate. Chap ter 5, the conclusion chapter, fo cuses on the policy implications and the implications of load data and coverage models for Florida and Vi rginia and a prototype of green roof policies for Florida and Virginia an d contains the overall co nclusions regarding the management of green roofs in Virginia and north Central Florida as a stormwater BMP for water quantity and quality. 40

PAGE 41

CHAPTER 2 HYDROLOGIC DYNAMICS OF GREEN ROOF PLANT-GROWING MEDIUM COMBINATIONS FOR NORTH CENTRAL FLORIDA Introduction Due to the variability of the subtropical c limateuneven distribution of rain, high ET in windy months, occasional frost and droughtexten sive green roofs in Florida have been hypothesized to function better at depths of (>15 cm) of growing media, as compared to their temperate climate counterparts, which are typi cally designed with 5 cm 10 cm of growing medium (Mike Hardin and Martin Wanielis ta personal communication, 2006). The deeper medium is thought to provide a greater water st orage capacity, therefore buffering the variability in water demand and supply of water for plants on Florida green roofs. In general, green roofs are designed to retain and detain water from rainfall events and assist in evapotranspiring the water. Evapotranspiration rates fr om a vegetated roof can reduce the annual runoff to less than ha lf the precipitation (Berndtsson, 2006). Retention is defined as the fraction of rainfall that either evaporates from the growing me dium, or transpires out of the plants, or remains on the roof stored as (non-plan t available) soil moisture. Detention is defined as the water temporarily stored in the growing me diums pore space that eventually is released as runoff over a longer period of time. Retention and detention processes of green roofs reduce the peak volume entering the stormwater system, in crease the lag time to peak concentration and influences baseflow over a longer period of time (Bengtsson et al., 2005; Zimmer et al., 1997; DeNardo et al., 2004). Physical Characteristics of Growing Media Important physical characteristics of growing medi a which can directly or indirectly affect water retention include: bulk de nsity, grain-size distribution, sorting, packing arrangement, organic matter content, texture and mineralogy. For example bulk density is important for green 41

PAGE 42

roofs because it determines the overall mass that will be added to a roof and can reflect the amount of compaction. Compaction infl uences infiltration rate which influences the ability of the green roof to act as a water storage device. Or ganic matter and mineralogy are important because it influences both the pore size distribution and intra-aggregate pore space, which can affect moisture content (Bra dy and Weil, 2002). Content of growing media can vary vastly with the region, the ingredients chosen and intention of the roof. In Germ any, where green roofs have been used extensively for over 40 years, standards are in place regarding the proportion of aggregates to other particle size classes, % organic matter, type, maximum carbonate conten t and other regulations with regards to slope, substrate contents, bulk density and compactibili ty. The US has few standards for green roof growing media at this time ASTM E2396-05, 2398-05 and 2397-05. Currently, many of the green roof media used in the US have been patterned after growing media characteristics of those used and tested in Germany, a cool oceanic temperate climate, which are largely carbonate-free aggregates in the 0.95 cm (3/8 in.) range, with few finer particles and low in organic matter (1-3%). I hypothesize that the subtropics have a growing media that is optimal for this climate (maximizi ng plant available water be tween storms, but able to create enough void pore space so as to be us eful in stormwater detention) and may have characteristics that differ from a growing media that is optimal in a cool temperate climate. Examples of growing media from various parts of the world include ma terials varying from crushed roof tiles to volcanic pumice for the inorganic aggregate sized portion of the mix and from sphagnum peat moss to chicken manure fo r the organic portions of the growing media (Table 2-1). 42

PAGE 43

Green roof systems have other layers that influence water retenti on, such as drainage layers. Drainage layers vary in their form and wa ter retention characteristics, ranging from small cups to large with holes at the top of the cup for drainage, to an absorbent plastic mesh layer, to an aggregate layer with perforated PVC running through the aggregate for drainage. The details of these layers vary as well, such as the size of cups and density of holes for drainage in the corrugated plastic drainage layers type and the mesh thickness and absorbency varies for the plastic mesh layer and aggregate layers can vary in thickness and amount of piping available to carry the water to the downspout, all these factors further influe nce the water retention capacity of a green roof and the amount of water available to plants. Th is study characterizes the water retention properties of exclus ively the growing media, (and does not incorporate the water retention effects of the under layers). Table 2-1. Examples of variability of growi ng media constituents a nd physical properties among locations. Location Constituents Source Physical Properties Sweden 43% crushed ceramic roof tiles (8-12 mm) 37% sand 10% organics 5% clay 5% crushed limestone (8-12 mm) Villareal et al. (2004) Emilsson (2008) BD= 1.48 g cm-3 Porosity 43.34% Organic Content=1.6% Pennsylvani a 60% hydrolite 12.5% sphagnum peat moss 12.5% coir 15% perlite DeNardo et al. (2003) Saturated wt 4.9 kg m-3 (porosity 55%, field capacity 11%, Ksat 0.12 to 0.2 mm s-1) Michigan 40% heat expanded slate 40% USGA sand, 10% Michigan Peat 5% dolomite 3.33% composted yard waste 1.67% composted poultry manure Van Woert et al. (2005) Capillary pore space 20%, non-capillary pore space 21.4% Bulk density 1.3 kg m-3 In situ, each growing medium is underlain by a different materialthe Building Logics growing medium in Virginia has a corrugated plas tic cup underlayer that contains absorbent gel 43

PAGE 44

packs in the bottom of the cups. American Hydrotechs system on the Charles R. Perry Construction Yard also uses a corrugated cup un der layer which has a higher density of smaller cups than Building Logics and no gel packs; and UCFs mix a top the Student Union at UCF in Orlando was underlain with aggregate and perforat ed PVC piping. The aim of this study was not to analyze the whole gree n roof system, but isolate the role of the media in retention. For this reason, when the media were tested in the meso cosms, Floradrain drainage layer was employed in a reverse position, with the drainage holes that are normally situated at the top of the cups positioned on the bottom to readily transport pe rcolate water away. This study does not promote any brand over another; each company whose propr ietary mixes were used, have several growing media types for different situations. The growi ng media tested here, were chosen based on the fact that two were actually used in FloridaHydrotechs Litetop on the Charles R. Perry Construction Yard, and UCFs Black and Gold was used on the Student Union Building at UCF in Central Florida, and Building Logicss mix was chosen because it was used on the Yorktowne Square Condominium green roof that is monitored and described in Chapter 4, and because Building Logics intends to use their mi x in Florida and would like to adapt it to the subtropical climate. Role of Plants in Retention and Detention of Stormwater In addition to growing media, plants are also an integral component of green roofs. Plants are important in a green roof system because of aesthetics, cooling via transpiration, shading and creating a monolithic layer by holding the substrate in place with its root system (Cantor, 2008). In terms of green roofs as a stormwater manage ment practicethe main function of plants is their ability to reduce media moisture content via transpiration and increase interstitial pore space available for water storage. This is important because the amount of storage available for 44

PAGE 45

the next storm event depends on how much water was released via transpiration rate of the plants after drainage stops. After a storm, the available storage volume in the growing media depends on the percent of available volume of void pores. Before the next storm, the growing me dia plant system must release the water either via the plants or due to physical characteristics of the soil, to create new available pore space in the green roof media. Fu rther, to serve as a stormwater control for the local climate, ideally the cycl e of filling the pore spaces a nd voiding the pore spaces would match up to the periodicity and volume of the storms in the region. Additionally, since green roofs are defined as being living roofs, the plants must remain viable during the lifetime of the roof. To sustain plant life on a roof, either enough plant available water must remain in the soil betweens st orms for plants to survive or irrigation must be added. Plant available water is defined as the difference between field capacity and wilting point (Brady and Weil, 2002). Role of PlantGrowing Media System in Stormwater Control Zimmer and Geiger (1997) simulated and test ed small beds of green roofs by applying rainfall via sprinklers at in tensities ranging from 5 cm hr-1 (2 in hr-1) to 18 cm hr-1 (11 in hr-1). They found that for inte nsities up to 18.5 cm hr-1 (7.3 in hr-1) the green roofs showed positive retention effects (Zimmer and Geiger, 1997). De Na rdo et al. (2003) also found that green roofs played an important role in attenuating the pe ak runoff by detaining the water for a period of time. She found that green roofs reduced tota l runoff over a whole year by 40% in the midAtlantic region. Swedish scientists tested the role of green roofs as one BMP in a chain of connected BMPs. The green roofs in the study were retrofit at municipal buildings and runoff from these buildings then entered a chain of BMPs that included open cha nnels, inner courtyard ponds and 45

PAGE 46

miniature wetlands in an urbanized setting. The hydrological performance of the new stormwater system was assessed by modeling the individual BMP elements using th e unit hydrograph and design storm concepts to provide synthetic inflow hydrographs (Vil lareal et al., 2004). Simulations for the area that contained the green roof were run with and without the green roofs as part of the chain of BMPs (the green roofs covered 31% of a 4600 m2 area that drained to an elongated pond located between the building and the parking lot). They found that the BMPs significantly redu ced peak flow because they were very effective in retaining and detaining stormwater volumes. Their presence delayed the storm peaks in the unit hydrograph as well as lowered the peak flow. The upstream BMPs in the study, which included the green roofs, retained 50% of the stormwater produced by the -yr rain event, 37% for the 2-yr, 30% for the 5-yr and 21% for the 10-yr design storm. Their green roofs played a more important role in mitigating small storms, while the stormwater pond had a greater relative importance in large storms. Given that different plants have different transpiration rates and soils have different water retention capabilities and evaporation rates, I hypothesize that an optimal growing media-plant combination exists for each regional climate a nd that by varying medium composition, depth and plant selection it should be possible to construc t a green roof with little or no supplemental irrigation after the establishment period for north central Florida. The optimal plant-growing medium combination in a green roof for stormwat er management in north central Florida should have high water retention capacity and medium re-release characteristics. Objectives The objective of this study wa s to determine which plant-growing media combination at a 15 cm depth is optimal for water retention during storm events and re-release before the next storm, while maintaining healthy plants. Three (3) plant groupings and three (3) commercially 46

PAGE 47

available growing medium types were tested and compared to each other and to several controls over a period of 6 months in a mesoscale field experiment. The determina tion of optimal plantsoil combination for a stormwater BMP was base d on 1) plant viability and plant growth, 2) nutrient retention/release, and 3) water retention/release. This ch apter deals exclusively with the water retention/release portion of the study. The specific objectives of the hydrologi cal component of the study were to: 1. Determine water retention during a six (6 ) week plant establishment period with irrigation. 2. Quantify and compare the water retention of the experimental green roof bins after establishment with no irrigation over a longer period of time (6 months). 3. Determine an overall water balance for pl ant-growing media combinations, meaning determine amount of water leached, taken up by plants and growing medium after irrigation, and back-calculate ET rates ba sed on water loss for each plant-growing medium combination, based on these values. 4. Characterize the water release characteristics of the plant-growing media combinations and how they change over time. Hypotheses The null hypotheses related to the hydrologic study on the establishment period of the green roof (first six weeks of growth) and the 6 month study for a shallow north central Florida green roof (15 cm deep) were: H1o: There are no significant differences in wa ter retention among growi ng media types, plant types or any of the 9 growing media-pl ant combinations and/or bare media. H2o: There are no significant differences in evap otranspiration rates amon g the 3 plant types, 3 growing-media types or any of the 9 growing media-plant combin ations and/or bare media. H3o: There are no significant differences in water retention within plant-growing media combinations over time. 47

PAGE 48

Materials and Methods Study Area The climate of north central Florida is hum id with an average annual rainfall of 1300 mm (NOAA, 2007). Rainfall distri bution is uneven throughout the year and characterized by a relatively short wet season and relatively long dry season with variable beginning and end dates (Butson, 1958). For Gainesville, the averag e beginning date of the rainy season based on a 25-year record (1931-1955) is June 15th and the end date is September 5th (Butson, 1958). During this period over 50% of th e annual precipitation occurs. Rain events in the rainy season are due to convective st orms (NOAA, 2007; Brown, 1981; Irmak et al., 2002). The mean monthly precipitati on average between June and September is 178 mm and the mean monthly temperature for these months rang e from 18 to 35C. The winter mean monthly precipitation average is 80 mm and mean mont hly temperature ranging from 4C to a maximum 22C (NOAA, 2007; Brown, 1981; Irmak et al., 2002). Periods of drought can occur within the rainy part of the year and during the winter months there is also regular frost and occasional hard freeze events. The peak values of evaporation in north-central Florida occur in May when daily incoming solar radiation are at a maximum of 7 mm day-1 and are lower in June, July and August because of cloud cover associated with the summer rainy period. Minimum daily evaporation rates occur in December and January at a rate of 2 mm day-1 (Irmak et al, 2002). Experimental Design The hypotheses for hydrologic dynamics were test ed using 40 different mesocosms. The mesocosms were set up in a 3 x 4 factorial design with each treatment replicated 3 times, plus 2 filter fabric controls and 2 empty bin controls The mesocosms were set-up outdoors on the UF campus, 60 cm above the ground. The ground consiste d of white gravel that was devoid of vegetation. The study had two hydrol ogic phases, an establishment period where irrigation and 48

PAGE 49

rain occurred and a post-establishment phase wh ere only rainfall occurred (and no irrigation was administered). Soil moisture release curves we re evaluated at the be ginning and end of the establishment phase and several times during th e post-establishment phase to determine how moisture retention and the water release charac teristics changed over time in the mesocosms. Three different soil growing media, Buildi ng Logics, Hydrotechs LiteTop and Black and Gold were tested in 39.5 cm by 45.7 cm c ontainers, and planted with three different plant types (Perennials, Succulent s and Runners), shown in Figure 2-1. Three replicates of each plant type were planted in each growing medium. Controls for the growing medium consisted of three containers of medium without plants for each growing medium. Over all controls consisted of two containers containing only the corrugated material and filter fabric that underlain the growing medium in all container types, as we ll as two empty containers-devoid of plants, growing medium or underlayers. S1 / P4 S2 / P4 S3 / P4 S1 / P3 S2 / P3 S3 / P3 S1 / P2 S2 / P2 S3 / P2 S1 / P1 S2 / P1 S3 / P1Plant Group1: PERENNIALS Plant Group 2: SUCCULENTS Plant Group 3: RUNNERS Plant Group 4: BARE MEDIA Growing Media: 1 Building Logics 2 Hydrotech 3 UCF X 3 replicates Response Variables: Hydrology: Quantity,Timing Moisture release curve Water Quality: TN (TKN,NO3),TP, TSS) Plant Growth: lateral growth rate biomass 2 controls= empty bins w/ filter fabric 2 controls=empty bins Figure 2-1. Complete Randomized Block Design of 3 growing media versus 3 plant types (and bare media) with 3 replicates and 2 filter fabric controls and 2 empty bin controls (not depicted.) 49

PAGE 50

The green roof mesocosm bins were situated side by side in a ra ndomized block design. Five large wooden frames were built to hold 8 gr een roof bins side by side. The frames were constructed out of two 3 m long (3 m x 10 cm x 10 cm) landscape beams set parallel to each other on top of two cinder blocks located at both ends of the beam. The beams were placed into two beam holders sawed out of wood, created to hold the beams parallel to each other in such a manner that the top surface of the beam had a slop e of 1:12. The bins rested on top of the beams 60 cm above the ground at a slope of 1:12, mimicki ng the slope of the green roof on top of both the Charles R. Perry Construction Yard a nd Yorktowne Square Condominiums. The cinder blocks were leveled before the beams we re placed on top of them. (Figure 2-2). Filtrate generated by the bins after rain events and irrigation flowed passively into lidded buckets that were placed on the ground undernea th each of the 40 bins. The volume of filtrate was measured weekly for the initial six weeks after planting and then every six weeks between weeks 6 and 24. Additionally, the vo lume of filtrate was collected and measured directly after a 1.27 cm controlled irrigation test administered on weeks 1, 6, 12, 18 and 24. Water samples of the filtrate from the 40 green roof bins were co llected weekly for six weeks and then once every six weeks up to 24 weeks. Sampling times and methods are described in detail the section Sampling Protocol. Experimental Set-Up All bins were packed with equal volumes of growing media between July 4th and July 23rd, 2007 and packed following ASTM standards fo r green roofs used by American Hydrotech, which is based on German FLL standards for th e installation of green roofs in Germany. The bins were filled with 17.8 cm (7 inches) of l oosely packed growing medium then compacted to 15.2 cm (6 inches) of depth using a 4.5 kg (10 lb) weight which was dropped on to the surface of the bins from a height of 30 cm (1 ft) with 20 blows per surface area of the bin (Figure 2-2). 50

PAGE 51

After packing, the bins were c overed with heavy plastic to ke ep rain and moisture out of the growing media until planting. All bins were planted on 7/22/07 and were moistened preplanting and then immediately irrigated after pl anting for 32 minutes, the equivalent of 2.54 cm (1 inch) of irrigation. All bins were irrigated uni formly using a mister nozzle at a flow rate of (280 mL min-1). Bins were irrigated regularly for 1.27 cm (0.5 inch) over 16 minutes following the irrigation regime shown in Appendix A. During the establishment period, defined as the first six weeks of plant growth, all filtrate resulting from both irrigation a nd rain was funneled into an 1 8.9 L (5 gallon) bucket through a tube attached to a hole drilled in the bottom of the bin. The buckets were opaque white with opaque lids and were placed dir ectly under the bins to provide shading to reduce algal growth. Before the study began, the buckets were scrubbed with soapy water, rins ed, then acid washed and triple rinsed. Additionally, the buckets we re scrubbed weekly the first six weeks and thereafter were scrubbed every six weeks until the study was completed at the end of week 24. The water that funneled into the buckets from th e bins remained in the buckets for up to one week at a time during the establishment phase of the study. After the initial 6 week establishment period ended, the filtrate would remain in the bu cket for 6 weeks of time, and was sampled from on weeks 6, 12, 18, and 24. Water samples from the bins were considered to be composite samples representative of the time period during which the filtrate accumu lated. Samples taken weekly during the first 6 weeks (weeks 1, 2, 3, 4, 5 and 6), represented leach ate resulting from both rainfall and irrigation. While for weeks 12, 18 and 24, the composite sa mples represented leach ate generated from rainfall over 6 weeks time with no irrigation. Bins were irrigated only during the first 6 weeks of the study. 51

PAGE 52

To create the water release curves for the th ree growing media, and the three plant types and their combinations, the volum e of filtrate collected directly after a controlled 1.27 cm irrigation event was measured at 20 minutes, 60 minutes, 2 hours, 6 hours, 12 hours and 24 hours after irrigation. The buckets were scrubbed i mmediately before the controlled irrigation experiment and kept directly under the bins at all times during this portion of the experiment. Water collected during the water release charact erization experiment was measured for volume using a 500 mL and 100 mL graduated cylinder. Figure 2-2. Photo of the 40 bins in a comp lete randomized block design. Succulents ( Sedum acre, Delosperma cooperii and Portulaca grandiflora ) are shown planted in UCFs Black and Gold in the forefront, next to it are succulents planted in Hydrotechs LiteTop mix, followed by perennials ( Helianthus debilis Coreopsis lanceolata and Gaillardia pulchella ) in UCFs growing media. Growing media The three growing media selected for this study have been used in Florida or will be used in Florida in the near future. The first grow ing medium was UCFs Black and Gold growing medium (referred to as U in this chapter) wh ich was engineered by UCF/STE and used in 52

PAGE 53

Orlando at UCF. The second medium, Building L ogics growing medium (referred to as B in this chapter), will be used in Florida in the near future and was utilized on the green roof at Yorktowne Square Condominiums in Merrifield Virginia. The third medium, Hydrotechs Hydrolite growing medium (referred to as H in this chapter), is manufactured by American Hydrotech, Inc., a Chicago based green roofing co mpany, which installed the green roof atop the Charles R. Perry Construction Yard in Gainesvi lle, Florida. American Hydrotechs LiteTop Intensive growing media meets the German FL L Standards for the Guidelines for Planning, Performance, and Maintenance of Vegetated Roof tops. Hydrotechs LiteTop Intensive growing medium consists of 45-70% LiteTop Lightweight aggregate (0.15 cm to 0.95 cm [1/163/8] aggregate), 0-30% Coarse to Medium Sand, 0-30% Perlite, Sphagnum, or Other Lightweight Soil Additive and 5-30% Approved Compost and Nutrient additives as needed; Hydrotech adjusts their performance specification valu es in accordance to the availa bility of local materials and special project conditions relate d to plant selection and/or en vironmental conditions. Appendix B contains the Specifications of LiteTop Intens ive growing media and th e General Description and LiteTop Components. Table 2-2 shows the ra nge of density and saturated water and air content, OM content of Hydrotechs LiteTop Intensive growing medium. Table 2-2. Physical properties and organic ma tter content of Hydrotechs LiteTop growing medium from Hydrotechs specifications sheet. Property Dry bulk density 0.6-1.1 g/cm3 Saturated bulk density 1.0-1.5 g/cm3 Saturated water capacity >40% Saturated air content >10% Organic matter content (mass %) 6-12% Bins packed with UCFs growing media c ontained 12.7 cm of an expanded clay mix which by volume, was 60% expanded clay, 15% peat moss, 15% perlite and 10% vermiculite 53

PAGE 54

and was underlain by 2.54 cm of Black and Gold pollution control layer consisting of 40% tire crumb from recycled automobile tires, 20% expanded clay, 15% peat moss, 15% perlite and 10% vermiculite. Building Logics growing medi um consists of 90% Stalite (Rotary Kiln Expanded Lightweight Aggregate) and 10% mulch (See Appendix C for the specifications of Stalite). Analysis of physical propert ies of the growing media As the reported values of physical and chem ical properties for each growing media based on their specification sheets had wide ranges for bu lk density, organic matter and particle sizes, soil samples taken before and after the study was completed were analyzed for BD, OM, particle size distribution, porosity, as well as chemical properties (pH, TP and TN reported in Chapter 3). Each growing medium was sampled before the st udy began and sampled again at the end of the study. Soil cores (40 cm long x 10.2 cm dia.) were taken from the bulk mixes before packing the bins and subsamples were taken from the cores, air dried, ball milled. After the study was completed, all 40 bins were placed in a green house to air dry for 8 weeks and were weighed weekly. When the mass of the bin no longer changed for three cons ecutive weeks, the bins were stirred to homogenize th e soil and (15 cm long x 10.2 cm dia.) cores were taken from the bins and used for organic matter content (OM) analys is and for grain-size distribution analysis and porosity measurements. OM was measured through loss on ignition (LOI) in a muffle furnace g of ball milled air dried soil was oven dried for 24 hours at 105C, reweighed, then baked in the muffle furnace at 250C for 30 minutes and then the temperatur e was raised to 550C for 2 hours. The samples were reweighed after cooling in a desiccator. The loss on ignition values from the muffle furnace method were compared to LOI values using a Thermogravimetric Analyzer (951 TGA) by DuPont, Thermal Analyzer Analytical Instrument s Division. The graphs of loss on ignition from 54

PAGE 55

the 951 TGA were analyzed using the TA.PC for Windows Acquisition Program, Version 3.2 by Instrument Specialists, Inc. The air dried soils samples were sieved using four USA Standard Testing Sieves which meet ASTM E-11 specifications. Approximately 100 g of air dried soil was weighed out in triplicate from the soil core taken from before the study began as well as the soil samples from the three replicate bins of bare media for each growing medium type. The soil was sieved using Sieve #10 (2 mm), Sieve #18 (1 mm), Sieve #35 (500 um) and Sieve #60 (250 um) and shaken on a Ro-Tap Testing Sieve Shaker Model B by CE Tyler for 5 minutes. Bulk density for each bin was calculated based on the volume (23,000 cm3) of soil packed into the bin and the mass of th e soil after 8 weeks of air drying in the green house. Particle density was measured by water displacement. Poros ity was calculated as 1-BD/PD, where BD is bulk density and PD is particle density. Plant types The three plant groups were selected based on several factors: a) assumed position along an evapotranspirative scale, b) nativeness, and c) growth and reproductive mechanisms. The three plant groups tested are categorized as follo ws: 1) Native Florida Dune habitat plants, 2) Succulent plants, and 3) Runne r type plants, (Figure 2-3). Native Florida dune habitat plants: selected for this study were Helianthus debilis Coreopsis lanceolata and Gaillardia grandiflora (Figure 2-4). They consist of clump-forming upright perennials that reseed th emselves, which make them useful in laterally covering a green roof over a period of time (several seasons). Th ey are considered to have medium to low ET rates as compared to other flowering ornament als in the Florida landscape. They are known to survive periods of drought and heat in dune habita ts and several species of these genera evolved in this climate. 55

PAGE 56

High ETP1: Succulent plants- Similar in structure to N. Green Roof Plants,Low ETP2: Native Dune Habitat Plants:Low -Med ET Alkali conditions High heat Low irrigationP3: Runners: Med ET Low growing Low maintenance Low irrigationLow ET Figure 2-3. Rubric of plant groups in relation to an assumed ET gradient. The structure and function of these plants is that they are medium low growing selfseeding Asteraceae, that spread by both vegeta tive growth and reseed ing and rebound after high temperature and drought. They have low irrigation requirements, ha ve high heat tolerance, like alkali conditions and are adapted to the native clim ate--thin medium sized leaves that wilt easily under low water conditions, however are resilient. Helianthus debilis Nutt. is one of 62 species of Helianthus that exist world-wide, Helianthus are facultative upland plants and are planted most commonly by placing the root mass 1.27 cm to 2.54 cm below the surface; this was the method of planting was used for both the bins and the Charles R. Perry Construction Yard roof. Adequate water up to 6 months is recommended by the USDA for Helianthus debilis after which point irriga tion is not needed. In our study, the plants were irrigated for only 6 weeks. Gaillardia pulchella (one of 13 spp. of Gaillardia ), commonly know as Firewheel Daisy or Blanket Flower, is plan ted in a similar manner to Helianthus and reseeds itself by dropping seeds on a well-drained firm soil. Coreopsis lanceolata is a clump-forming perennial herb with 56

PAGE 57

short rhizomes. Coreopsi s prefer full sun in well-drained soil and are natura lly found in dry infertile sites, making them ideal of Florida roof tops. Coreopsis require a firm seed-bed for establishment. Having a thick layer of plant residue on the surface in terferes with seed germination. Seed germination occurs by fall an d during the winter plan ts remain low growing rosettes. Lanceleaf can tolerate regular mowing in summer and fall and one fall mowing is recommended (USDA/NRCS 2009). Figure 2-4. Photos of Gaillardia grandiflora Goblin, Coreopsis lanceolata and Helianthus debilis Nutt. (photos by P.J. Alexander and K.Hill.) Succulent plants: were selected for this study becau se of their well-known performance on northern green roofs during dr ought, heavy rain and cool temperatures (DeNardo, 2004; Van Woert, 2006). Three species selected for this study were Sedum acre, Delosperma cooperii and Portulaca grandiflora (Figure 25) and were purchased from a local nursery, Grandiflora in Alachua, FL. The species and varieties used were adapted to the Florida climate. The succulents have low ET rates, because, Sedums and Delosper ma use Crassulacean Acid Metabolism (CAM) to reduce moisture loss during drought. While Port ulaca grandiflora, while a succulent C4 dicot, it also can exhibit acid metabolism similar to CA M plants in certain environments (Ku et al., 1981). The way succulents generally reduce moisture loss, is by having thick leaves with a low area-to-volume ratio and sunken stomata and thick cuticles. The CAM metabolism in allows the plants to close their stomata during the day, open them by night and absorb low amounts of CO2 57

PAGE 58

in the night, which the plant transforms into malic acid and store in their vacuoles, to use as an energy source in the day time (Keeley and R undel, 2003). Therefore succulents with CAM metabolism, release very little wa ter through their stomata, additionally they can store water in vacuoles, making them resistant to drought and heat Succulent plants are chosen for cool climate green roofs, based on their rapid lateral c overage, low moisture and low fertilization requirements (because CAM plants are very efficient in using nitrogen), and ability to grow in a thin soil substrate (5 cm to 10 cm). While Sedums and Delosperma are known to tolerate freezing temperatures, snow coverage and droughts, Purslane can tolerate heat and drought for short periods as well as periodic inu ndated conditions (htt p://plants.usda.gov). Figure 2-5. Photo of Sedum acre (photo S. Lang 2007), Delosperma cooperi and Portulaca grandiflora Hook. Runner type plants : include those plants which are suggested by IFAS at UF based on decades of study to be Floridian alternatives to turf because of their low maintenanceno mowing, little fertilizing. For example, many varieties of perennial peanut ( Arachis spp. ) have been tested by IFAS at UF as an alternative to lawn that require less irrigation and fertilizer than a normal lawn. Arachis glabrata (Figure 2-7), while non-native, has adapted well to the Florida climate; it is native to South America between latitudes of 13S and 28S, is found naturally in climates with 1200 to 1600 mm yr-1 of rain, however it is extr emely drought tolerant and can survive once established on 600-700 mm yr-1 rain with a 5 month wet season and also can 58

PAGE 59

tolerate wet conditions and flooding on occas ion, such as a climate with 4000 mm yr-1. Arachis glabrata can grow laterally at a rate of 2 m per y ear with no competition and with other plants present grows at a rate of 5 cm 20 cm a year In one year the plant can fully fill in an area where it is planted at a rate of 1 rhizome per 0.5 m with adequate water. However, generally it takes 2-3 years to fully establish. As a nitrogen fixer, N fertiliza tion rates are low, and the plant is usually propogated by rhizome versus plantl ets or seed, growing one woody tap root and spreading laterally (French et al ., 2006 and Freire et al. 2000). The two other runners that grow well in Florida under low moisture conditions are Mimosa strigillosa and Phyla nodiflora and are native to the region (Fi gure 2-6). An adaptation to the sub-tropical climate that both pl ants possess is the ability to become dormant under high heat. They are able to survive on little irrigation a nd have the ability to reseed themselves. The structure and function of this plant group is to spread primarily through lateral vegetative growth and place energy into creating a deep woody root sy stem first, for which reason the plant has the capacity to grow back if it dies back from a frost or drought. Mimosa and Perennial peanut have the ability to close their leaves to reduce the amount of ET during hot periods. Figure 2-6. Arachis glabrata (Perennial Peanut) photo by Roka 2004, Mimosa strigillosa (Sunshine mimosa) and Phyla nodiflora 59

PAGE 60

Irrigation regime and rainfall The irrigation regime during the establishment period consisted of watering daily the first week and then 3 times a week in subsequent weeks until week 6 at which time the irrigation was discontinued (Appendix A). This irrigation regimen is the same as the irrigation regime used on the Charles R. Perry Construction Yard (CRP) green roof, so that comparisons can later be made between runoff data from Hydrotech bins in th is study with runoff from Hydrotech LiteTop mix used on the CRP green roof. A green roof study on optimal irrigation for Central Florida green roofs at UCF 2006 found that 1.27 cm () or irrigation twice a week was superior to overwatering defined as 2.54 cm (1) twice a week (Hardin, 2006). Rainfall Rainfall for the site was measured using the University of Florida station W4DFU located at N 29 38 (29.639) and W 82 20 (-82.345 ) at an elevation of 42.5 m (140 ft), using Peet Bros Ultimeter 2000. The 30-yr average ra infall pattern for Gainesville, FL from NOAA (2007) was used for comparisons of rainfall from the study period. Rainfall measured during the different time periods is shown in reported in the Results and Disc ussion section of this chapter. Sampling Regime Composite water sampling Weekly composite samples of drainage wate r were taken on Mondays on Weeks 1, 2, 3, 4, 5, and 6 and Weeks 12, 18, and 24. Samples were com posite over time, meaning that all drainage from individual rain events and separate irrigation events from the week were passively collected through a tube and funneled into a bucket for each bin. The stage hei ght of the water was measured in the bucket before sampling and multiplied by the cross-sectional area of the bucket to determine the volume of drainage water for the time period. The volume of drainage water was used in the calculation of water retained by the growing media-plant system. Water retained 60

PAGE 61

was defined as that water which did not exit the system as filtrate at any point. The volume of water retained was assumed to be either stored in the growing media, or taken up by plants and/or evapotranspired out of th e bins by the plants. The amount of retention was determined by subtracting the filtrate from the quantity of water in and dividing this by the volume of water applied to the system via rain or irriga tion: Vol. Water Retained = (Vol. WaterIN Vol. Filtrate)/Vol. WaterIN The source of water and sampling dates are shown in Table 2-3. Table 2-3. Composite water sampli ng dates and sources of water IN. Week Sampling Date Time Period of Composite Sample Source of WaterIN Parameters Measured Week 1 7/30/2007 1 week Rain/Irrigation Volume Week 2 8/5/2007 1 week Rain/Irrigation Volume Week 3 8/13/2007 1 week Rain/Irrigation Volume Week 4 8/20/2007 1 week Rain/Irrigation Volume Week 5 8/27/2007 1 week Rain/Irrigation Volume Week 6 9/3/2007 1 week Rain/Irrigation Volume Week 12 10/14/2008 6 weeks Rain Volume Week 18 11/30/2007 6 weeks Rain Volume Week 24 1/17/2008 6 weeks Rain Volume Water release characterizations (lysimetric sampling) Water release curves for the different media types and growing medium-plant combinations were determined via a lysimeter experiment. The lysimeter experiment consisted of weighing the bins, both before and after irriga ting the bins with a 1.27 cm () of water, at different intervals up to 72 hours (T able 2-4) every 6 weeks. Bins were weighed the morning of the first day of the experiment using a CP Wplus Parcel Scale (Adam Equipment Company 2005), with a maximum capacity of 75 kg and readab ility to the 0.01 kg. The e rror in readings of 0.01 kg represents 10 cubic centimeters of wate r or 0.06 mm of precipitation or ET across the surface of the bin. The bins were irrigated with 1. 27 cm () of water dire ctly after the initial weighing, and then reweighed at 1 hour post-i rrigation, 6 hours post-irrigation, 12 hours post61

PAGE 62

irrigation, 24 hours post-irrigation, 36 hours post-irrigation, 48 hours post-irrigation, 60 hours post-irrigation and 72 hours post-irri gation. Filtrate was collected from the bins at 20 minutes, 1 hour and 6 hours after irrigati ng and 12 hours post-irrigation, (and 24 hours after irrigation in the event that water was still seeping out 24 hours af ter irrigation). A recipient was kept under all bins at all time periods to co llect filtrate at any time period. The objective of the lysimeter e xperiment was to create water release curves for the three different growing media and for the different plant types in re sponse to controlled irrigation events over 72 hour periods at different points of the study period. The results of the experiment also reveal how water content changes in the gr een roof bins over 6 months in response to differences in the climate during different periods of the year. Water inputs were determined in two ways. On e, by pre-calculating the value of expected irrigation volume based on the duration of irrigation in minutes multiplied by the flow rate of the individual nozzles and secondl y, by verifying this value by su mming the volume of filtrate collected directed after irrigation with the increase in mass of the bin after irrigation. The flow rate of the nozzles was determined before the bi ns were planted during an initial irrigation test where empty buckets were placed directly under the nozzles for 1 hour; the flow rate was 280 mL min-1. The nozzle flow rates were measured periodically during intermittent calibration tests throughout the study period by placing a small plas tic bottle under the nozzle and irrigating for 5 minutes and then measuring the volume of water emitted. Water retention at times 1 hour, 6 hours, 12 hours, 24 hours, 36 hours, 48 hours, 60 hours and 72 hours post-irrigation were calc ulated as the change in the mass of the bins minus leachate volume divided by waterIN. For example to calculate water retention at Time = 6 hours postirrigation: 62

PAGE 63

Water Retention Time 6 hr = ( Mass of Bin 0hr to 6hr Mass of filtrate 0 to 6hr)/ WaterIN Where Mass of Bin 0 hr to 6 hr = (Mass of bin 6 hours Mass of bin Time 0) Evapotranspiration rates were determined base d on the change in mass of the bins over 12 hour periods of the 72 lysimeter experiment when no leaching occurred, no rain occurred and solar radiation was optimum (sunny versus overcast days), for ex ample between hours 36 and 48 and/or hours 60 to 72, which correspond to 7am to 7pm of days two and three of the experiment: ETTime 48 to 60 hours = Mass of Bin 60 hr Mass of Bin 48 hours The lysimeter experiments were planned for ev ery sixth week, if rain was forecast for the experiment days, the experiment was postpone d by several days until sunny weather was forecast. Table 2-4. Water Release Curves (Lysimeter Experiment) Sampling Week Sampling Dates Parameters Sampled Leachate Source Week 1-UCF 7/26/07-7/29/07 Bin mass/leachate volume at 0hr, 1hr, 6hr, 12hr, 24hr, 36hr, 48hr, 60hr, 72hr 1.27 cm irrigation event (16 min) Week 1-H 7/30/07-8/2/07 Bin mass/leachate volume at 0hr, 1hr, 6hr, 12hr, 24hr, 36hr, 48hr, 60hr, 72hr 1.27 cm irrigation event (16 min) Week 1-BL 8/5/07-8/8/07 Bin mass/leachate volume at 0hr, 1hr, 6hr, 12hr, 24hr, 36hr, 48hr, 60hr, 72hr 1.27 cm irrigation event (16 min) Week 6 9/5/07-9/8/07 Bin mass/leachate volume at 0hr, 1hr, 6hr, 12hr, 24hr, 36hr, 48hr, 60hr, 72hr 1.27 cm irrigation event (16 min) Week 12 10/17/0710/20/07 Bin mass/leachate volume at 0hr, 1hr, 6hr, 12hr, 24hr, 36hr, 48hr, 60hr, 72hr 1.27 cm irrigation event (16 min) Week 18 12/5/07-12/8/07 Bin mass/volume at 0hr, 1hr, 6hr, 12hr, 24hr, 36hr, 48hr, 60hr, 72hr 1.27 cm irrigation event (16 min) Week 24 2/3/08-2/5/08 Bin mass/volume at 0hr, 1hr, 6hr, 12hr, 24hr, 36hr, 48hr, 60hr, 72hr 1.27 cm irrigation event (16 min) Week 60 9/8/2008--Bare Media Bin mass/volume at 0hr, 1hr, 6hr, 12hr, 24hr, 36hr, 48hr, 60hr, 72hr 2.54 cm irrigation event (32 min) Criteria for Evaluating Growing Media fo r North Central Florida Green Roofs for Stormwater BMPs The criteria by which the growing media are being evaluated in their selection as a growing medium for use on a green roof as a stormwater BMP in N. Central Florida are: 63

PAGE 64

Ability to support plant life Water Retention/Evaporation Rate Nutrient Release The growing media were evaluated for their ab ility to support plants during establishment with irrigation and after the establishment period without irri gation, to test their potential to survive in a low maintenance setting. This chap ter deals exclusively with the influence of growing media; plants; and pl ant-growing media combinations on water retention and water release characteristics in a green roof system. Criteria for Evaluating Plant Selection for North Central Florida Green Roofs for Stormwater BMPs The 3 selections of plant mixes with differe nt ET rates were evaluated for usefulness as plants for green roofs as a BMP for north central Florida on the basis of: a) survivability in FL climate on a green roof (low substrate, low water), b) growth rate and lateral coverage rate (time it takes to reach 60% coverage), c) water efficiencywater usage and evapotranspiration rates. Plant health was assessed several times during the first six weeks and then regularly every six weeks. Plant health was measured by exam ining the plant, noting the succulence, rigor, greenness, condition of leaves, presence/absen ce of insects and diseases, and yellowing. A healthy thriving plant was given a 5 on a scale of 1-5. Thriving was de fined as having green leaves fully intact (no holes, crumbling or yellowing), blooms when in season, noticeable biomass growth (evaluated with photos taken from the same distance every sampling period). A plant in good health (Figure 2-7 B) but with little increase in bi omass or less turgor was given a 4 on a scale of 1-5. Pl ants that were not affected by disease or had no yellowing on leaves, but had not grown much since the previous eval uation, were rated as having fair health, and given a 3 (Figure 2-7 C). A plant in poor he althdefined as presence of insects or partial yellowing of leaves was given a 2; a plant that was beginning to wither, but still alive was given 64

PAGE 65

a 1; and dead plants received a 0. The rati ng system was tested by three different people independently using photos and the same nu mbers were given to 39 of the 40 plants. Lateral growth of the plant was determined by measuring percent coverage based on taking a photo from 1 m directly above the plant and an alyzing the coverage on computer. Ideally, a healthy bin would end up with 33% coverage by each of the three species present and a plant health of 5 (thriving). A) B) C) D) Figure 2-8. Photos on the left shows plants after 2 weeks of gr owth compared to 5 weeks of growth; photos above (A and B) show runne rs planted in U, while photos below (C and D) show them in B growing medium. 65

PAGE 66

The Plant Health Index (PHI) incorporated bo th the % coverage and plant health and was scaled to 1-100. A bin covered equally by speci es (33% coverage by each) and a plant health rating of 5 would receive a 35, therefore graphs demonstrating the PHI are shown on a scale of 1-35. Statistical Analyses Water retention was calculated by subtracting the volume of wa ter that drained out from the volume of water in and dividing it by the tota l volumeIN, which was the total mm of rain and irrigation. Irrigation depth was cal culated based on minutes of actual irrigation during the time period multiplied by the mean flow rate of the mister nozzle heads (280 mL/min +/10 mL). Differences in water retention data among gr owing media types, then among plant types irrespective of growing media t ype, then and finally between time periods was analyzed using mixed models in GLIMMIX (in SAS 9.2 by SAS Institute). Differences were considered significant at the p 0.05 level. Tukeys HSD tests were used to detect significant differences in water retention among plant types within each growing media type and time period. Two types of water retention/re lease curves were created fr om the lysimeter experiment data. One curve showing water content in the bins calculated on a dry mass basis and a second graph showing the percent retention of waterIN (which was normally 1.27 cm) during the lysimeter experiment. The water content was dete rmined by subtracting out the dry mass of the bin and the dry biomass of the plants from the mass of the bin taken at a given time period, for example at Time = 12 weeks, the equation for this would be: kg = Mass of Bin 12 weeks Dry Mass of Bin Dry Biomass The percent of water applied retained was calcu lated for each time step that the mass of the bins was recorded between 0 and 72 hours and was plotted against time. To compare the characteristics of water retention and water re lease between growing me dia types and growing 66

PAGE 67

medium-plant combinations, the slope was calculate d for the uptake side (initial water retention between time 0 and time 1 hr) and the slope on the release side, between time 1 hr and time 72 hr. The decrease in water in the growing medium between time 1 hr and 12 hr is attributed to plant uptake/evapotranspiration and drainage, while water loss after drainage stops is attributed solely to plant uptake/evapotrans piration. At some point the wate r release curve decreases below the original water content, at this point the water lo ss is attributed to the effect of the plant in the growing medium versus the medium alone. Results IWater Retention Characterization of rainfall and irrigation The six week establishment period of the green roof bins began on July 23, 2007 and ended September 3, 2007, total rain and irrigation per 6-week time period is shown in Figure 2-8. During this time period the bins were irrigated regularly follo wing the irrigation regime in Appendix A. Rainfall occurred on 21 different da ys during the 43 day establishment period, and on 22 of the days there was no rain. Precipitati on events ranged from 0.25 mm to 72 mm of rain per day. The total amount of rainfall for this pe riod was 286 mm, which represents 20% of the mean annual precipitation for Gainesville, FL (NOAA 2007). The distribution of the rainfall was uneven. The number of dry days between storms ranged from 1 day to 9 days. The number of consecutive days in a row with rain ranged from 1 to 5 days. Rainfall per day ranged from a mini mum of 0.25 mm to a maximum of 72 mm. The 4 days with the highest amount of precipitati on (72 mm, 61 mm, 38 mm and 24 mm) accounted for 70% of the total rainfall for th e whole period, while the other 17 days of rain (80% of the rainy days) accounted for 30% of the total rainfall. This pattern of 20 30% of the days of rain accounting for 70 80% of the rain fall for a whole year is seen in many parts of the US and world (e.g. Boston, Seattle, Florida), for which reason Low Impact Development techniques and 67

PAGE 68

urban stormwater BMPs usually focus on capturing the small events (< 2.54 cm) as these represent the majority of preci pitation events in a year (F rance 2002, Urbonas et al. 2002). The average pan evaporation rates for May, June, July, August and September are 132 mm mo-1 (Irmak 2002). 0 2 4 6 8 10 12 Weeks 1-6Weeks 7-12Weeks 13-18Weeks 19-24 Time periodPrecipitation (cm) Figure 2-8. Precipitatio n/irrigation for the f our different 6-week time periods (7/23/07 1/18/08). Water retention Water retention in the indivi dual bins ranged from a minimum of 14% to a maximum of 96% (Figure 2-9) for the various 6-week tim e periods over the 6 month study period. When comparing mean water retention among growing me dia types, irrespective of time periods or plant type, Building Logics growing medium had significantly lower overall mean water retention (32% 1.4%) than H ydrotechs LiteTop Intensiv e and UCFs Black and Gold (47% 1.4% and 52% 1.4%) growing medi a respectively (Figure 2-10) (at an = 0.05, p < 0.0001 level, using a generalized linea r mixed model GLIMMIX in SAS 9.2). 68

PAGE 69

Figure 2-9. Box plot dist ribution of water retention for Buildi ng Logics (B), Hydrotech (H), and UCF (U) growing media for all weeks comb ined and all plant types included (box indicates the interquartile range, horizontal line indicat es median and diamond shows mean and whiskers the range). The water retention data for each bin for three time periods (Weeks 1-6, Weeks 13-18 and Weeks 19-24) were entered into PROC GLIMMIX in SAS 9.2 and analyzed in a generalized linear mixed model using an AR(1 ) structure subject to bin (mode l parameters are in Appendix D), with water retention as the response variab le and growing media and plant types and time as factors (see Table 2-5). (Note that the time period for Weeks 7-12 was not used because overflow occurred in the buckets during that tim e period.) There was a significant effect of growing media, plant type and time on water re tention, however there were no significant interactive effects on water rete ntion due to growing media-plan t type (p = 0.3967) (Table 2-5). 69

PAGE 70

Table 2-5. Results of the type III tests of fixe d effects for the PROC GLIMMIX model for water retention data for all bins over 24 weeks. Effect Num Den F Pr>F G_media 2 24 118.7 <.0001 P_type 3 24 12.9 <.0001 G_media*P_type 6 24 1.09 0.3967 Week 1 70 518.02 <.0001 Week*week 1 70 497.42 <.0001 Since the effect of time on water retention was significant, two-wa y ANOVA analyses of growing media and plant type were carried out fo r each six week period separately using PROC GLM in SAS 9.2. Table 2-6 shows the p-values of these processes and how the p-values of growing media and plant type and growing me dium-plant combination change over time. Table 2-6. P-values for 2-way ANOVAs of water retention by gr owing media and plant type and growing media-plant type interactions for three 6-week periods (analyses performed in PROC GLM in SAS 9.2; model fit, F-values shown in Appendix D). 1-6 weeks 13-18 weeks 19-24 weeks Growing media <.0001 <.0001 <.0001 Plant type <.0001 .7951 0.0016 G*P combination 0.0385 .0003 0.1607 The water retention values among growing medi a types were significantly different from each other in all time periods (p < 0.0001, us ing PROC GLM to perform ANOVAs by time period at an = 0.05 level), Table 2-6. Mean water retent ion values by growing medium type are shown for each 6 week period in Table 2-7. The basic pattern of Building Logics having less retention and Hydrotech and UCF having comparable retention rate s was exhibited over all time periods (Table 2-7). The percent of retention in the different media fluctuated over time. These fluctuations are attributed to changes in the ac tual inputs of water from rain and irrigation, as there was a weak negative corr elation between the amount of water input and the amount of water retention (Pearson Correlation coeffici ent of -0.41, with p < 0.0001, n = 108), the same pattern was seen in Moran (2005). For example, in time periods with lower rainfall, such as 70

PAGE 71

weeks 13-18 where only 20 mm fell, proportionally more water was retained (57% retention for B), than in a time period with less rain, such as weeks 19-24 (with 38 mm of rainfall), where only 28% was retained for B, even though the absolu te amounts of water reta ined in the two time periods was the same. Table 2-7. Mean water retention (%) and standa rd error for each growing media type for each time period (results of a ge neralized linear mixed model in GLIMMIX, SAS 9.2. Levels not connected by a lett er denote significant differe nces at a p <0.005 level, within time periods. Weeks 1-6 Weeks 13-18 Weeks 19-24 Mean SE Mean SE Mean SE B 27.8 1.3% a 57 1.3% a 28.2 1.3% a H 42.7 1.3% b 71.9 1.3% b 43.1 1.3% b U 49.7 1.3% c 78.9 1.3% c 50.1 1.3% c Effect of plant type on water retention irrespective of growing medium type The presence of plants augmented water rete ntion, irrespective of growing media type, significantly (p < 0.005) in all three time periods (Table 2-8) Bins containing bare media showed significantly lower water retention than bins containing vegetation, irrespective of growing media type (p < 0.001, = 0.05, GLIMMIX in SAS 9.2). Table 2-8. Mean water retention (%) by plant type (irrespective of growing medium type) for each time period; results of a generalized linear mixed model in GLIMMIX, SAS 9.2. Levels not connected by the same letter within a time period, denote significant differences within that time period only at the p<0.005 level, among plant types. (Weeks 7-12 overflowed). Plant Weeks 1-6 Weeks 13-18 Weeks 19-24 Type Mean SE Mean SE Mean SE m 33 1.4% a 63.1 1.4 % a 34.3 1.4 % a p 43.2 1.4% b 72.5 1.4 % b 43.6 1.4 % b r 42.7 1.4% b 71.9 1.4 % b 43.0 1.4 % b s 40.5 1.4 % b 69.7 1.4 % b 40.9 1.4 % b Interactive effects of plant-growing medium combinations on water retention over time Tukeys test, a more conservative method to detect signific ant differences among means, was implemented to determine the interactive eff ects of plant type-growing media type on water 71

PAGE 72

retention within each time step (Figure 2-10) Differences between plant-growing media combinations were determined using the mean of each of the 12 plant-growing medium combinations and a pooled variance, resulting in lo wer type I error (less pr obability of the test detecting a significant difference, when in actualit y there is not). This section analyzes the data in a way that shows the least amount of significan t differences of any other analyses used and discusses how plant-growing media combinations affected water retention over time. In this section, significant difference refers to a di fference between means of two growing mediumplant combinations that is significantly greater than a predicted value de termined by Tukeys test at a p < 0.05 level. Abbreviations used are U for UCF, B for Building Logics, H for Hydrotech and p for perennial, r for runners, s for succulents and m for bare media and these letters are used in combinations (for example Hp or Br or Us ) to represent the indivi dual growing medium-plant combinations. The results in this section all refer to Figure 2-10. In weeks 1-6 : UCF bins planted with perennial s had significantly higher retention (Tukeys test between means, p< 0.05), than a ny other growing media-plant combination (Figure 2-10); and Bm (Building Logics bare growing media) had the lowest retention among all plantgrowing media combinations. In weeks 1-6, perenni als also had a positive effect on retention in B and H media, with significantly higher water re tention (p <0.05, Tukeys test) than those bins containing bare media. However no significant differences in retention were found among the three plant types (perennials, runners and succule nts) in B or H growing medium (Figure 2-10). Nor was retention in bins planted with runne rs or succulents signifi cantly different from bins containing bare media in e ither B or U (Figure 2-10), only in H growing media did bins planted with runners have significantly higher rete ntion than those with bare media. Bp was not 72

PAGE 73

significantly different from H and U bins planted with succu lents or left bare, but had significantly less retention that H and U bi ns planted with perennials or runners. Weeks 1-60% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% BHU Growing MediaW ater Retention p r s m de f fefe cd bcd bc bb a bcd bcd Weeks 13-180% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% BHU Growing MediaWater Retention (%) p r s m cde bcd de e abcd abc abcd a bcd ab ab abcd Weeks 19-240% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% BHU Growing MediaWater Retetion (%) p r s m c c c c a b ab a b a ab a Figure 2-10. Differences in water retention amo ng growing medium-plant type combinations for three of the 6-week time periods. Levels not connected by the same letter signify significant differences in retenti on based on Tukeys test at an = 0.05 level using PROC GLM in SAS 9.2. 73

PAGE 74

In weeks 13-18: Perennials planted in U growing me dium again had the highest retention value of all plant-growing media combinations and Bm had the lo west retention values. However in this time period, Up was only significantly higher (p < 0.05, Tukeys test) than Hm (bare media in Hydrotech) and all B bins; Up was not significantly higher than other plant types in U growing media, nor bare U medium. Among bins containing B medium, Br had the highest retention in this time step (ver sus those planted with perennials as seen in the previous time step). Br was significantly higher than B bare media, but not significan tly different from the other plant types within B. By weeks 13-18 di fferences in retention among all plant-growing media combinations was lessening, in fact, retention exhibited by Br was not significantly different from any H bins, nor fr om U bins planted with runners or succulents or left bare, Br was only significantly lower than Up. In weeks 19-24 : U bins again had the highest retenti on values and B bins had the lowest, but there were no longer any signi ficant differences in retenti on among plant types and/or with bare media within any of the growing media ty pes (Figure 2-10). Retention of B bins was significantly less than that of H bins or U bins. There were few significant differences in retention between H bins and U bins. H bins planted with runners or succulents were not significantly different from any of the U bins; only retention in H bins plan ted with perennials or left bare were significantly lo wer than all types of U bins. Effect of plant type on water retention with in growing media --all time periods combined The results of PROC GLM with in each growing media type, when all time periods were combined, showed that perennials had a positive effect on retention. Within Building Logics growing media, bins planted with perennials exhibited significantly higher retention rates (38% 2% over all time periods combined) as compared bare media (27.5 %) (p = 0.0003), as well as compared to runners (32% 2% ) and succulents (32% 1.9%) to a lesser degree (p = 0.029 and 74

PAGE 75

p = 0.031, respectively). Table 29 shows the significant differen ces in retention among plant types within Building Logics gr owing media for all time periods. Table 2-9. Results of PROC GLM comparing e ffect of plant type on retention within each growing medium for all time periods comb ined. Levels not connected by the same letter (within a growing medium ty pe) are not significantly different. Plant Type Sig. Diff. B (Mean SE) p a 38% 2% r b 32% 2% s b 32% 2% m b 28% 2% Plant Type H (Mean SE) p a b 50% 2.8% r a 52% 2.8% s a b 46% 2.8% m b 42% 2.8% Plant Type U (Mean SE) p a 60% 3% r a b 51% 3% s b 49% 3% m b 47% 3% Within Hydrotech growing media, bins plan ted with perennials (50% 2.8%) or runners (52% 2.8%) had significantly more retention fo r all time periods combined than bare media (42% 2%) to a p < 0.05 level. There were no significant differences in retention among runners, succulents and perennials within Hydrotech growing media. Within UCFs Black and Gold growing media ty pe, bins planted with perennials exhibited significantly higher water retention rates than either succulents or bare media. Bins planted with runners in this media type onl y had significantly higher retenti on rates than bare media (p < 0.05). UCF bins planted with pere nnials had the highest mean re tention over all time periods combined (60% 3.3%), followed by U bins planted with runners (51% 3.3%), then succulents (49% 3%) and las tly bare media (47% 3%). 75

PAGE 76

The analysis of water retention for all time pe riods combined shows th at bins planted with perennials had significantly higher water retention than those pl anted with runners, succulents and bare media in B medium (Table 2-9); perennia ls had significantly higher water retention than those planted with succulents and bare media in H medium; and pe rennials only had significantly higher water retention than bare media in U medium. Effect of plant type on water retention irrespective of growing medium type The overall mean retention of bare growing medi a, irrespective of type of media, was 43%, the extra retention of the green r oof bin attributed to plant type, irrespective of growing media is shown in Figure 2-11. Perennials, on average fo r all time periods combined, increased water retention by 9.6 0.5% above having just bare media. Runners increased water retention by 9.1 0.5% and succulents increased water retention on average by 6.9 0.5%. 0% 2% 4% 6% 8% 10% Water Retention ( % ) Perennials Runners p Succulents p Figure 2-11. Amount of increase of water retent ion by plant-growing medium type attributed solely to plant type, irrespectiv e of g-m type or time period. Results II-Water Release Curves Changes in Mass and Water Content over the Six Month Study Period The three different growing media followed simi lar general patterns of changes in absolute mass and water content in the green roof bins irrespective of growing media type (Figure 2-12 76

PAGE 77

and 2-13). In general, water content decrea sed sharply during the summer months (Julybeginning of October) when ET was respectively higher than the rest of the study period; and water content increased over the winter months (between October and February), when ET was lower. The increase in water content in the cool er season are attributed to changes in ET, a decrease in plant growth and increas e in plant senescence (Figure 2-13). Viewing water content and mass together (Fi gures 2-12 and 2-13), one can see that while U and H have the similar water content levels at all time periods (p < 0.05 t-test SAS 9.2), the absolute mass of the media is significantly different at each time period, except 9/5/07, with U being significantly lighter (p< 0.05, t-test, SAS 9.2) than H. The importance of the absolute mass of the medium from a green roof design standpoint, relates to the cost im plications involved in the building design with respect to the implied increased load bear ing capacity of the roof for a heavier medium. UCFs Black and Gold medium and Hydrotechs Lite Top mix showed more overall fluctuation in water content over the study peri od than Building Logics, ranging from 0.17 0.33 for H and 0.21 0.35 for U, while water content in B fluctuated less (0.12 0.18), had a narrower range of values, and lower mean water content at all time periods irresp ective of plant types (Figure 2-12). The fluctuation in the mass of bins is at tributed to changes in: a) water uptake in plants, as well as, b) biomass grow th and c) water storage in the porous medium of the green roof system over different periods of the year. Effect of Plant Type on Water Content by Growing Media Type over Six Months Figures 2-14, 2-15 and 2-16, show the changes in water content within each growing medium due to plant types; water content was isolated by subtracting out the mass of the dry biomass mass of the bin as well as the dry mass of the soil. Betw een weeks 1 and 6, the 77

PAGE 78

difference in water content between plant types wi thin a growing media ty pe increased (Figure 214, 15 and 16). For example, by Week 6, the water content of bins planted with perennial plants were significantly lower than those planted with succule nts or contained bare media, in all growing media types, due to a successive loss in soil moisture, caused by the higher rate of ET of the plant type. Figure 2-12. Mean change in mass of bins fo r Hydrotech (upper line), Building Logics (middle line) and UCF (lower line) over 6 month study period. Plant growth for all plant types at week 6 was characterized by 20-30% coverage by individual species in both plan t types and no senescence had occu rred. In H growing media, the water content of bins planted w ith runners was also si gnificantly lower than those planted with succulents or bare media (Figure 2-15). 78

PAGE 79

0.00 0.05 0.10 0.15 0.20 0.25 0.30 0.35 0.40 6/17/20078/6/20079/25/200711/14/20071/3/20082/22/2008 DateVolumetric Water Content B H U Figure 2-13. Mean change in volumetric water c ontent in bins containing Building Logics (B), Hydrotech (H), and UCF (U) growing medi a over 6 months. Comparing this water content graph in Figure 2-12, of absolute mass of the bins, one can see that despite an average 8 kg difference in mass per bin be tween U and H bins, the water content at each point in time is similar between U and H. 0.00 0.10 0.20 0.30 0.40 0.50 6/17/20078/6/20079/25/200711/14/200 7 1/3/20082/22/2008 P avg S avg R avg BM avg Volumetric Water Content Figure 2-14. Effect of plant type on water cont ent in Building Logics growing medium over six months. 79

PAGE 80

Volumetric Water Content Figure 2-15. Effect of plant type on water content in Hydrotechs growing medium over 6 months. Water Content (UCF)0.00 0.05 0.10 0.15 0.20 0.25 0.30 0.35 0.40 0.456/17/20078/6/20079/25/200711/14/20071/3/20082/22/2008 P avg S avg R avg BM av g Volumetric Water Content Figure 2-16. Effect of plant type on water content in UCFs Black and Gold growing medium. Effect of Growing Media Type on Water Upta ke and Release Directly after Irrigation Water uptake and release was measured via the lysimetric method. Water content and % retention directly after irrigation with 1.27 cm was measured by weighing the bins before, 1 hr 6 hrs, 12 hrs, 24 hrs, 36 hrs, 48 hrs and 72 hour s after irrigation. The results were 12 water content graphs for each of the growing medium-p lant combinations (Appendix E) and 12 graphs of change in % water retention over time (Appendix E). The water content graphs for the growing media in Week 6, the end of the establ ishment period, are shown below in Figure 2-17, as an example of water content curves for th e other time periods, which are in Appendix E. 80

PAGE 81

Water contents curves highlighting the effect of plant type on growing medium type, are shown later in the Effect of pl ant type on water uptake and releas e section of this chapter, in Figures 2-20, 21 and 22. To visually better compare the absolute differences in water uptake and release among the growing media types and plant types within the growing media types, the amount of change in water content per time step was plotted against time. In this manner, the change in water content for each growing medium can be viewed superimposed and normalizes the initial water content of the plant-growing medium combination for the start of each lysimeter experiment (Figure 2-18). The process of norma lizing the data by plotting amount of change in volumetric water content against time (number of hours post-irrigation) made it possible to visually compare the effect of plants on cha nge in water content over time within growing medium types, as well, (shown later in Fi gures 4-23, 24, and 25.) To numerically compare differences in rates among the various plantgrowing medium combinations, the amount of change of water content per time step was then divided by the hours in the time step to yield the rate of uptake in change in water content per hour after irrigation, and th e rate of release was calculated as the change in volumetric water co ntent per hour for the de scending limb of the water content curves. 81

PAGE 82

Figure 2-17. Water content curv es for the three growing media (B, H, U) over 72 hours post irrigation at the end of Week 6. Figure 2-18. Comparison of change in volum etric water content over 72 hours among three growing media types (B, H, U) with no vegetation in response to 1.27 cm ( ) irrigation during week 6 of the study (9/5/07-9/8/07). The same analytical process was used to calcula te the water uptake and release rates for all the other weeks as well (Weeks 12, 18 and 24) for all plant-growing medi um combinations. The mean water uptake and release rate s for the growing media are repor ted in a table in Appendix G, and shown below in Figure 2-18. (Uptake and release rates for the plant-growing media combinations for the different weeks are also reported in Appendix G, and discussed in the plant effect section of the chapter). The uptake and release rates were analyzed in a generalized mixed linear model in SAS to identify whether th ere were significant differences in water uptake among the growing media and among all plantgrowing medium combinations, within time periods(in Appendix G). The water content in the bins containing UC Fs growing medium increased most quickly after irrigation and also decrea sed the most quickly as the bi ns dried out for 72 hours after irrigation, in four of the five time periods (Figure 2-22 and 2-25Week 6, and Appendix E 82

PAGE 83

Weeks 12, 18 and 24). The rate of uptake of water in the first hour after irrigation was significantly greater in UCF growing medium th an in Hydrotech or Building Logics growing medium in Weeks 12, 18 and 24 (Appendix G). In week 6 there were no significant differences between U and H. Hydrotech growing medium also had a quicker rate of uptake than BL in every time period (Figure 2-19). The differences in water retention among th e growing media are attributed to the differences in pore size distri bution and organic matter content among the media, with media with finer pores and greater OM content taking up more water after irrigation at a greater rate, also noted in Getter (2007) in Wolf and Lundholm (2006). Wo lf and Lundholm (2006) noticed in their study of water uptake by different plan t forms under varying watering regimes, that the plants (under the driest regime) had greater uptake of water than the controls. They assumed that both the planted and unplanted pots had been ta king up the same amount of water after irrigation, as they would weigh their plants after it reach ed field capacity (drainage stopped). They were surprised that planted microcosms lost more water than unp lanted ones and stated that Change in Volumetric Water Content per hour Figure 2-19. Top row shows rate of water uptake as increase of % of volumetric water content per hour for Weeks 6, 12, 18 and 24 (Septe mber 5-8, October 18-20, December 5-8 and February 2-4). Bottom row shows the ra te of water release from the mesocosms as % volumetric water content per hour on the y-axis. The x-axis indicates growing medium type. 83

PAGE 84

their finding may indicate that planted microc osms actually retained more water immediately after watering. In my study, I found that the planted mesocosms in the beginning of the study (Week 6) did indeed retain more water than unpl anted ones directly afte r watering (Figures 2-20, 21 and 22). Furthermore, variability in the amount of water uptake between plant types, seemed to be influenced by the plant physiology, water content at the time of watering, and season. I propose that one reason the uptake of water wa s greater in planted mesocosms relates to increases in organic matter content due to belo wground growth of plants and changes in porosity due to the increased OM conten t in the medium (Getter et al, 2007; Wolf and Lundholm, 2008). The initial water content of the soil at the ti me of irrigation played a role in the water uptake rate, regardless of the season. For example, the water content was similar in two different seasons (Week 6-late summer and Week 24-earl y spring), with a value of approximately 0.20, the respective rates of water uptake for th e same mesocosms (UCF and H planted with Perennials) were also similar for those weeks. Additionally, the water uptake rate was greater when the water content started out at this level (0 .20) then when the soil wa s very dry. Such as in weeks 12 and 18, where the water co ntent in the growing media was about half that (0.1) for both UCF and H for perennial plants, but the water uptake rate was signi ficantly lower. It may seem intuitive that when the soil is the most dry (has the lowest water content, it would absorb the most amount of water most quickly), however th is was not the case in this study. The reduced uptake of water seen in October and December may be due to reduced plant health or growth or season. For example, late summer plants and ea rly spring, the plant-comb inations are taking up water and utilizing it, while in late October and ea rly December, the plants are starting to use less water and are not putting any of the water into biomass production and unh ealthy plants are not 84

PAGE 85

able to remove as much water through transpirat ion as viable late summ er plants, or new spring growth plants. Effect of Plant Type on Water Uptake/Release Characteristics The effect of plant type on the water reten tion and water release characteristics of the growing medium was most obvious in week six (Figures 2-20, 21 and 22), where it was evident that bins planted with perennials and runners were uptaking water a qu icker rate than those planted with succulents or were left bare. Furthermore, the accumulated potential water loss grew over time, creating a wider spread between th e water release curves of bins planted with perennials at the end of the study than in the beginning. Consecutively over time, the perennial bin dried out more, so at the beginning of each ly simeter experiment the water content of the bin was less than the previous time. The greatest amount of retention and overall re-r elease of water over time is visible in the bins planted with perennials and runners. Leach ing occurred up to 12 hours in both soils, after which all decreases in mass are at tributed to evaporat ion, and transpiration. Bins planted with succulents and bare media had similar retenti on and release patterns in both growing media types. No significant differences in amount reta ined or total amount released was detected between succulents and bare media (Table in Appendix G). Perennials growing in UCFs Black and Gold took up water the quickes t after irrigation, as well as released the most amount of water w ithin 72 hours post-irrigation (Figure 2-26).The increased uptake of water in the Up plant-grow ing medium combination was attributed to the physiology of perennial plants with their high quantity of medium fine roots (see photo in Figure 2-35) relative to th e two other plant types tested in this study (runners and su cculents) (Figure 235), coupled with their inability to close the stom ata as in the case of su cculents or fold their leaves closed such as leguminous runners are able to do, to prevent water loss, that makes them 85

PAGE 86

ideal for a finer grain size di stribution growing medium with a higher water retention capacity. Paradoxically, the perennials ability to both take up water more quickly and release it more quickly and even release water to a water cont ent lower than the content measured at the beginning of the 72 hr lysimeter experiment, result ed in a greater soil moisture deficiency over time. The sequential lowering of the water conten t of the soil containing perennials (as well as runners), irrespective of the soil type, widened the gap between the water content levels in the soils growing perennials and runners as compar ed to those with succulents and bare media (Appendix E), over time as the growing season advanced and then less ened as the growing season ended. Figure 2-20. Volumetric Water Content (%) over 72 hours for 3 plan t types and bare medium (p, s, r, m) within Building Logics growing medium for Week 6. Figure 2-21. Change in Volume tric Water Content (%) over 72 hour s for 3 plant types and bare medium (p, s, r, m) within Hydrotech growing medium for Week 6. 86

PAGE 87

Figure 2-22. Volumetric Water Content (%) over 72 hours for 3 plan t types and bare medium (p, s, r, m) within UCF gr owing medium for Week 6. Figure 2-23. Change in Volume tric Water Content (%) over 72 hour s for 3 plant types and bare medium (p, perennials; s, succulents; r, r unners; and m, bare media) within Building Logics growing medium for Week 6. Figure 2-24. Change in Volume tric Water Content (%) over 72 hour s for 3 plant types and bare medium (p, s, r, and m) within H ydrotech growing medium for Week 6. 87

PAGE 88

Figure 2-25. Change in Volume tric Water Content (%) over 72 hour s for 3 plant types and bare medium (p, s, r, and m) within UCF growing medium for Week 6. 0% 2% 4% 6% 8% BpBsBrBbm HpHsHrHm UpUsUrUm -8% -6% -4% -2% 0% Change in Volumetric Water Content/hour Figure 2-26. Mean rates of uptak e and release (change in volumetric water content per hour) for each plant-growing medium combinations for lysimetric sampling week 6. Error bars show standard deviation. Mean rates of uptake and rele ase for other weeks are in Appendix G. (B, Building Logics; H, Hydrot ech; U, UCF; p, pere nnial; s, succulent; r, runner; and m, bare media.) The difference in water content between plan ts containing perennial s and those containing succulents or bare media was greatest in the Late Summer/Early Fall (weeks 12 and 18). The water content in these plant types was the l east different at the be ginning of the study, immediately after pre-moistening soils and plan ting them; and at the e nd of the study in the dry/winter season, characterized by low ET, and the presence of some dead plants. While the rapid uptake and release of wate r from perennials can be seen as advantageous to opening pore 88

PAGE 89

space up for water storage before the occurrence of a subsequent storm, this lowered water content in the soils growing the perennials (irrespective of plant type) jeopardizes the plants health by creating water stress on the plant due to the high accumulated potential water loss that plant-soil combination is experiencing and ma kes the plant susceptibl e to diseaseswhether insects, root rot or other disease, that in optimum health woul d be less. For example, 2 weeks after irrigation ended (Week 8) the perennial plan ts had an infestation of insects, while other plant types did not. Perennial pe anut suffered yellowingwhich could have been due to a) low iron, b) high pH, c) improper inocculation of the soil with Rhizobium due to dislodging all soil from the roots during transplanting of the plant and a resultant N-deficiency in the plant (USDANRCS, 2009; Sartain personal communication, 2007). Bins planted with succulents or no vegetation had the highest and l east fluctuating amount of water content over the six month study period. Within the Building Logics growing medium type, there were no significant di fferences in water content plante d with succulents as compared to those with no vegetation at any point in time during the study period. Nor were there any significant differences in water content between bi ns planted with runners and perennials within the Building Logics growing media (Figure 2-20). There were however significant differences in water content between both the perennials when compared to succulents or bare media and significant differences between runners and succule nts or bare media at the end of the 6 week establishment period, and week 12 and week 18, but not week 24 (Appendix E). These differences were attributed to the greater capac ity of runners and perenni als to transpire than succulents. Durham et al. (2006) also noticed a higher water content in microcosms containing the succulent plant Sedum acre as well, and attributed it in part to a shading effect of thick-even 89

PAGE 90

coverage of Sedum acre, reduci ng the amount of evapor ation from the soil itself by reducing the surface temperature of the soil (in Wolf and Lundholm, 2008.) Growing media without plants had a higher water content be cause water stored in the lower part of the soil profile had no mechanism by which to leave the soil profile after leaching stopped other, than through evapor ation and evaporation is controll ed by pore size and continuity of the pores. This phenomenon of higher soil moisture in areas without vegetation has been well documented in watershed studies conducted in th e Coweeta Watershed in North Carolina where whole catchments were deforested in order to in crease soil moisture and baseflow to streams (Douglass and Swank, 1972). Media with smaller por es are able to evaporate over a longer period of time, because of the connectivity of the pores, however at some point the smaller pores hold on to the residual water more tightly, maki ng this water unavailable to plants. The higher saturated water content and lower residual water cont ent levels in finer soils yields a larger range of plant available water than that of a coarse soil (Brady and Weil, 2002), as was seen in the case of the coarser Building Logics growing medium which had a lower water content in every season and smaller increases in water content during the 72 hour lysimeter experiments. The differences in the steepness of the slope reflect the effect of plant type due to differences in ET rate among plant types, and up take of water as these two factors control changes in water storage within the same medi a type. Perennials have the steepest slope or fastest decrease in mass over time The mass in all bins decrease sharply (have the steepest slope) between weeks 6 and 12 (September 5 and Oct ober 14), which corresponds to the time period directly after the six week establishment pe riod when irrigation was suspended, immediately after which there were 9 consecutive days withou t precipitation. The distri bution of the rain in time is an important factor in designing the ir rigation regime for a green roof. The rainfall 90

PAGE 91

pattern for the second six weeks (Sept 5 to Oct 14, 2007), highlights how th e distribution of the rain, not the total volume can be important, often more so than the total volume in deciding plant health or mortality. For example the total ra in for this period was 60% above the 30 year average, yet the distribution of the rain in time, having nine days without rain consecutively be followed by six days for rain that contained 80% of the total rain of October, underscores the need to deal with uneven distribut ion of rain in time and the need for a either cistern or a soil moisture sensor to regulat e irrigation on a green roof. The UCF soil had the most change in soil moisture over the study period and was the least affected by plant typei.e. had the lower vari ability of change in soil moisture among succulents, runners and bare media than Hydrotech. Also bins planted with succulents and runners had a higher water content than those cont aining bare media, which is similar to what was noticed in Wolf and Lundholm (2008) and Durh am et al. (2006), where they attributed the higher water content of planted bins as compared to bare media to the surface cooling effect of the shading from vegetation and therefore reduced evaporation rates from the soil itself. In the UCF medium, the water content of bins planted w ith perennials was significantly less than any other plant type in Week 6 and always lower th an succulents and bare media in all subsequent weeks as well. The general trend of a steeply decreasing ma ss during the summer period and a sharp rise in mass that correlates to the period of year wh en ET rates decline (Irmak, 2002) was seen in all bins. The increase in mass during the cooler part of the year is attributed to an increase in soil moisture due to both the death of plants during th e cool part of the year and the reduction of ET due to seasonal changes. Figures 2-27, 28 and 29 show how plant health changed by species over time. The absence of certain plant species starti ng in Weeks 13-18, corresponds to the increase of 91

PAGE 92

soil moisture and change in seasons to a cooler time of year shown in Figures 2-14, 15 and 16. The death/decline of the perennials at week 18, ab sence of one species of succulents (Portulaca) and the decline of runners by week 18, coupled with the season (declining ET due to lower relative humidity, less solar radiation and lower te mperatures) may explain the stabilization of the mass and increase in mass due to increasing soil moisture for those dates coinciding with early winter (Week 18) and the subsequent rise in soil moisture throughout late winter/early spring (Week 24). Figure 2-27. Plant Health Inde x (PHI) for perennial plants ( Coreopsis lanceolata,Gaillardia pulchella and Helianthus debilis ). PHI combines % coverage by species with plant health (1-5); PHI of 35= 33% coverage of bin and ma ximum (5) plant health). Weeks 1-6 Weeks 7-12 Weeks 13-18 Weeks 19-24 Figure 2-28. Plant health index (PHI) of Runner Type species ( Arachis glabrata, Phyla nodiflora and Mimosa strigillosa ); PHI combines % coverage by species with plant health (1-5); PHI of 35= 33% coverage of bin and ma ximum (5) plant health). 92

PAGE 93

Figure 2-29. Plant health index (P HI) of succulent type plants ( Delosperma cooperii, Portulaca grandiflora, Sedum acre); PHI combines % coverage by species with plant health (15); PHI of 35= 33% coverage of bi n and maximum (5) plant health). It is hypothesized that the physio logy of perennials root system in conjunction with the leaf structure may have allowed for more water uptake, as well as ET when planted in a finer soil with greater pore connectivity and smaller pore sizes, such as in H and UCF. Growing-media with pore size distributions that have a more even gradation usually have more fine pores. Additionally growing media with higher OM c ontent usually have more micropores and higher water content as a result. Therefore finer por es with good connectivity between pores, organic matter content or actual mineralogy may be af fecting why there was less difference in water retention between plant types within UCFs growing medium. The lessening capacity of bins originally plante d with perennials to retain water over time, may be attributed to the fact th at most perennials died either from disease after suffering from drought in September, or from frost in late December/early January. Succulents however did not die from the drought or frost; while, some species of runners, such as Perennial Peanut were dead by late fall. Evapotranspiration Rates Evapotranspiration rates measured during the lysimeter experiment at the end of the establishment period, Week 6 (Table in Appendix H) are within the range of mean daily ET rates 93

PAGE 94

calculated for September for the Gainesville re gion using the FAO56-PM method based on Kpan values in Irmak 2002. In Hydrotech and UCFs gr owing media, the ET rates for bins planted with perennials and runners ar e higher than those planted with succulents (Appendix H). In all growing media, bins containing pl ants had higher ET rates than those without plants, except for Hydrotech being planted with succulents. Eva poration rates reported fo r Building Logics are lower than in the other porous media. This ma y be due to larger text ure not being able to evaporate out the water, and less water being reta ined initially to be evapotranspired out. Large pores in a coarse layer are not be able to support capillary move ment up from smaller pores of a finer layer (Brady and Weil, 2002). As shown in Figure 2-30, the grai n size distribution of Building Logics media is comprised of 90% grav el size particles in the beginning of the study. As expected, as the growing season ended a nd winter started, ET rates decreased for all growing medium-plant combinations (Appendix H). This was attrib uted to decrease in plant health, increase in plant senescence (shown in Figures 2-27, 28 and 29), as well as less hours of sunlight, cooler temperatures and lower PET rates associated with the change of seasons. Discussion Effect of Physical Characteristics of the Growing Media on Water Retention, Uptake and Release The differences in water retention among growi ng media was attributed to differences in pore-size distributions an d organic matter content between th e media, rather than the total porosity and total void space availa ble for storage in the various media. For example, Building Logics growing medium had a porosity similar to that of Hydrotech (of approximately 34%), but a different particle size distribution (Figure 2-30). Therefore the pore size distribution, which is a function of grain size distribu tion and porosity, of these two media, Hydrotech and Building Logics, were dissimilar as well. Hydrotechs grain size distribution had a lower percentage of 94

PAGE 95

gravel sized particles and higher percentage of very coarse, co arse and medium and fine sand particles than Building Logics (Figure 2-30). Building Logics growing medium consists of 90% Stalite (rotary kiln expanded shale) and 10% compost. Stalite cont ains 90% gravel sized particles of expanded shale with a porosity of 43% and 10% coarse sand sized pa rticles of expanded shal e (Stalite Specification sheets 2007Appendix C). As a result, Bs particle size distribution is comp rised of 90% gravel size (>2 mm) particles and has a total of 1.6% OM with the majo rity (85%) of the OM residing in the top particle size fracti on (the >2 mm) (Appendix I) In contrast Hydrotech has 28.6 % gravel sized particles (and 71.4% sand sized pa rticles); it has 4.4% 0.6% OM content and the majority (76%) of the organic matter is in th e smaller particle sizes < 2 mm (Appendix I). Having more organic matter in the smaller si ze fractions can positively influence aggregate formation and hence water retention (Thomas 2006). Therefore, despite the similarity in porosity, bins containing Building Logics growing medium consis tently had the lowest water content available to plants irre spective of the time of year, as evidenced by the water retention curves (Appendix F) and water content cu rves in Figure 2-23 in the Water Release Characteristics section of this Chapter. Futhermore, even in growing media which had more similar grain size distributions to begin with, such as that of H ydrotech and UCF (Figure 2-30), it is likely that the greater amount of organic matter (Figure 2-31) in U growing medium positively influenced water retention. According to Brady and Weil (2002), the availa ble water holding capacity of two well-drained mineral soils that are similar in every way but organic matter (OM) content, even with a difference as low as 2% OM, will find that the wa ter holding capacity is higher in the one with more OM, due to direct and indirect factors. 95

PAGE 96

A direct factor of OM that positively influences water rete ntion includes the OMs very high water-holding capacity, which exceeds that of mineral soil of an equal volume (whether at field capacity or wilting point) (Brady and Weil, 2002). Indirect factors include OMs ability to increase plant available water by stabilizing soil structure and increasing total pore space, both through increasing the volume of individual pores and the amount of pores (Brady and Weil, 2002). Higher OM content therefore results in an increase in wa ter filtration and water-holding capacity along with an increase in the amount of water held at the wilting coefficient. 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% UBH Gravel (>2mm) V. Coarse Sand (1-2mm) Coarse Sand (0.5-1.0mm) Medium Sand (0.25-0.5mm) Fine Sand or smaller (<0.25mm) % Mass ( dr y wei g ht of soil ) Figure 2-30. Grain size dist ribution of the three growing medium before planting. 0.0% 1.0% 2.0% 3.0% 4.0% 5.0% 6.0% 7.0% 8.0% 9.0% BL before B 1yrBL4yr H before H 1yrH2yr U before U 1yr BL before B 1yr BL4yr H before H 1yr H2yr U before U 1yr % Mass ( dr y wei g ht of soil ) Figure 2-31. Organic matter by percent mass determined by Loss on Ignition for all three media before and after 1 year; and for H medium, af ter 2 years in situ on the CRP green roof in FL and for B medium, after 4 year s on the YSC green roof in Virginia. 96

PAGE 97

The particle size distribution fo r U growing medium, of an even gradation of particles from gravel to fines (Figure 2-30) combined with el evated OM content (6.5% 0.8%) as compared to other growing media, indicates th at U growing medium likely has more finer pores in the soil than B and H and may have more micropores with in the organic matter, which have a stronger affinity to hold water. While Building Logi cs growing medium may also have micropores within the intraggregate pores of the heat expanded shale, the va por pressure differences at the edge of the gravel pieces and higher affinity of water to be held by the micropores within the gravel, may be immobilizing water present inside the Stalite, and keeping that water unavailable to plants. The large amount of energy needed to overcome the vapor pressure differences inside and outside the porous gravel pieces for water uptake or release to occur, may explain why the rates of water uptake and water release are significantly lower in B growing medium than in the other media (p = 0.05); and why the water content of B does not shift much throughout the six month study period. From one point of view, less variab ility in water content throughout the year could be beneficial in a green roof setting if a homeost atic condition of water is positive for the plant types health, neither much water is lost during the drying out process, nor gained during the wetting process. However since the AET rates of plants planted in Building Logics appear to be half that of the AET rates of the same plants in other soils at the end of the establishment period, such as UCFs growing medium (ETperennials = 6.0 mm day-1 in UCF and ETperennial = 3.1 mm day1 in BL), the water in the soil appears not to be plant available and plants with higher transpiration rates will be stressed in this growin g media as compared to one with finer pore sizes and more organic matter. 97

PAGE 98

Succulents, however, which have naturally adap ted in xeric climates and are able to close their stomata throughout the day and store water in their vacuoles to later use this moisture during periods of drought (Keeley and Rundel, 2003), do not seem to be deterred from growth by the lesser water content apparent in B growing medium. Succulent s are ideal for this growing mediumwhich is why they were chosen initially during 40 years of development of green roofs after being observed to naturally colonize gravel roof s in Germany. A loosely packed gravel soil with low water retention actually mimics the natural setting in which succulents are found rocky, porous and well-drained or dry landscapes Adding organic matter or a substance with finer pores to Building Logics gr owing medium for use in the subt ropics, especially as a growing medium for native Floridian plants, could increa se the water holding capacity of Building Logics growing medium significantly and suppor t Florida plant life more easily. Effect of Plant Type on the Physical Characteristics of the Growing Media Plant type affected the physical characteristic s of the growing media such as bulk density and porosity over six months growth. Table 2-10 shows th e initial bulk densities of the various growing media at the time of packing (before plants were added) and Figure 2-32, shows the final average bulk densities (BD) of the different growing medium-p lant combinations at the end of the study period. The bulk density of the medi um changed by varying degrees over 6 months of growth depending on the type of plant present. These changes in BD and porosity are likely due to the formation of aggregates as OM was broken down from larger particles to finer particles, and as organi c matter was incorporated into the gr owing medium after the senescence of finer root hairs (or via orga nic acids exuded by the roots). Th e stabilization of soil through the creation of aggregates and resul ting increase in pore sizes and pore volume and decreased BD is well described by Brady and Weil (2002), Mosadde ghi et al. (2000) and Thomas et al. (1996). The greatest increases in porosity and decreases in BD were seen in bins planted with runners, 98

PAGE 99

which had the finest root hairs of the three plan t types, and next in perennials, which had the greatest quantity of medium sized roots. For exam ple, the bulk density of UCFs bare media at the end of the study was 0.74 0.01 g cm-3, while the bin planted with runners the BD was 0.69 0.01 g cm-3, this same trend of decreasing BD and increasing porosity was noticed in bins planted with runners in Building Logi cs growing medium (BD of 1.010.01 g cm-3 in bare medium at the end of the study and BD 0.96 0.01 g cm-3 for bins with runners). BD and porosity did not vary significantly between those bins planted with succule nts or left bare over the study period. Table 2-10. Bulk density and porosity of the three growing media before the study 7/3/08. Growing Medium BD (g/cm3) SD Porosity SD U 0.71 0.01 0.41 0.01 H 1.07 0.02 0.31 0.01 B 0.99 0.02 0.31 0.01 It is likely that organic ma tter below ground increased more significantly for runner and perennial plants than succulents, based on the physiology of their root systems (see Figure 2-33 and 34). The runners utilized generally had one tap root with finer root hairs emanating from the main root. Perennials, such as Coreopsis lanceolata had a large quantity of medium roots emanating from one point below the surface (Figure 2-35), while succulents had a smoother engorged root, as in the case of Delosperma cooperii with relatively few root hairs (Figure 234). According to Bernard et al. (2003) 60-70% of plant carbon entering the soil from plant residue is derived from the root system a nd includes soluble (amino acids, organic acids, carbohydrates) and insoluble (slough ed off cells, etc.) material. The soluble carbon originating from the incorporation of plant residue below the surfacesuch as polysaccharides and organic acids facilitate the formation of aggregates (Tisdall and Oades, 2006), which enhances water retention. 99

PAGE 100

Bins planted with runners in UCF and Build ing Logics showed a decrease in BD and increase in porosity that is likely due to additions of OM in the finer particle sizes. The hypothesis that there was an increase in aggregat es is supported by evid ence of an increase of particles in the >2mm size fraction that did not br eak apart after shaking during the sieving test in soils collected after 1yr in the bins and 2 yrs in situ on the CRP green roof. Bulk Densit y g / cm3 Figure 2-32. Bulk density of each growing me dium-plant type combination after the study period ended 11/8/08. Figure I-1 in Appendix I shows how the top size fraction (>2 mm) increased in H and U growing media, while in B growing medium, the top size fraction decreased over time. In the case of B medium, it began with 90% gravel sized (>2 mm) particle s which contained the majority of the OM present in the medium, in the form of mulch. Over time the proportion of OM in this top fraction decreased, likely due to the OM in the large size classes breaking down into smaller particles, increasing the proportion of finer OM in the smaller particle size ranges. The increase in finer particles of OM from senescing root material in B may have increased the likelihood of aggregate formation. Before the study began, over 85% of the OM for BL was located in the top particle-size fraction; after 1year, only half of the organic matter was located in the top size fraction (the othe r half of the OM being distribu ted among the lower size fractions (<2 mm)); and by year 4, the majority of orga nic matter in B growing medium was found in the 100

PAGE 101

smallest size fractions (Appendi x I). Total OM content increa sed over time in B growing medium, increasing from 1.1% 0.1% from befo re the study to 3.5 % after 1 yr and to 3.65 % after 4 yrs in situ on a green r oof (YSC in VA). In contrast, the OM content decreased in H and UCF. This trend of increasing OM in a highly mineral medium was seen in Emilsson (2006), as well as the trend of a growing medium with high OM content decreasing initially due to oxidation and then over time building back up again. When below ground plant residue enters the soil, initially two-thirds of the plant residue decomposes in one year (Cresser et al., 1995 in Bernard et al., 2003), followed by a slower stea dy breakdown of OM. In the case of growing mediaoften there is peat moss and compost presen t to begin with. The 3-fold increase in OM in B growing medium after 4 years of being planted with Sedums spp. on the YSC green roof (Figure 2-31) was attributed to th e incorporation of senesced plant material from the root system. The increase in OM in the B medium in situ sugg ests that an originally low OM mix is not only adequate for succulents, but that water retention may increase with time on such a green roof, as OM increases time. Furthermore, the presence of OM, and increase in aggregate formation due to exudation of organic acids during the decomp osition of added plant residue (Bernard et al., 2003), may result in greater retentio n of heavy metals on a green roof. Figure 2-33. Photographs showing the root morphology of a runner type plants Mimosa strigillosa. 101

PAGE 102

Figure 2-34. Photographs showing the root morphology of a succulent, Delosperma cooperii Figure 2-35. Photographs showing the root morphology of a perennial type plant, Coreopsis lanceolata. 102

PAGE 103

Figure 2-36. Photographs of the root systems of a self-recruited plant af ter bins were abandoned, Tropical Crab Grass. Conclusions When designing a green roof for stormwater reduction control in Florida, the most important design factor influenci ng retention is the initial sel ection of growing medium type. This decision, when interested in greatest wa ter retention, should be made based on the pore size-distribution of the media, pr esence of organic matter or cons tituents such as perlite and vermiculite and total porosity. In this study, retention rates were comparable to those found in studies in both temperate climates (DeNardo, 2005) and in the subtropics eg. Texas (Timm and Rasmussen, 2008). The retention rates ranged from a low of 24% for Building Logics medium with no vegetation to a maximum of 83% for UCF growing medium planted with perennial plants. Within each growing media type--re tention ranged from 24-53% BL, 30-72% H and 31103

PAGE 104

85% for the four different 6-week study peri ods, depending on the plant types present and amount of rainfall in the indi vidual 6-week study period. Plants were the next most important factor in fluencing water retention rates on the selected plant-growing media combinations used in this study. Evapotranspiration ra tes at the end of the 6-week establishment period, ranged from 1.5 mm day-1 for Building Logics-unvegetated medium to a maximum of 3.6 mm day-1 for UCF bins planted with perennials. There was a significant difference in the amount of water rele ased by transpiration ve rsus that of just evaporation within a growing media. Having pl ants present in the northern Florida climate definitely increased the amount of void pores available for water storage from the subsequent stormretention increased by 7-10 % (above bare medium alone) due to plants. It was however noticed that the plant type most capable of transpiring the greatest amount of water, the perennials, also suffered the most from heat stress during droughts and were consequently more susceptible to plant diseases and succumbe d more easily to death than succulents. The succulents, while surviving the largest range of climatic variation during the study and exhibitin best overall plant healt h, contributed less to increasing re tention of water directly after storms. Bins containing succulents did have the highest residual (non-plant available) water content in the medium it was planted in (irrespectiv e of media type) and hea lthiest overall plants during the entire study. In terms of the water release curves, water content was generally the lowest and water retention directly after irrigating was the lowest in Building Logics grow ing medium and highest in UCFs Black and Gold mix. As stated earlier, the ability of the perennials to evapotranspire out more water over the same period of time as the other plants resulted in a larger differential between the water content over time in those bi ns containing perennials and those containing no 104

PAGE 105

vegetation or succulents. Essentially, the water c ontent in those bins planted with perennials and runners became successively less, to the point that the plants themselves (which were responsible for transpiring the water out) were in greater peril of suffering from plant diseases. It is recommended that for a roof with no irrigation in the north central Florida climate succulents be used, however in a situation where irrigation can be used, then it is recommended that a mix of succulents, perennials and runners be used together as each plant structure has a different adaptation which may make them compatible to be planted together. Wolf and Lundholm (2008) also found that planting species with different morphologies together can maximize the plants different capabilities of util izing water at different moisture levels along a moisture gradient, increasing the amount of total transpiration, total soil moisture utilization and maximimizing the clearing of pore spaces before a subsequent storm. Furthermore, it is suggested that subtropical grasses be tested in vegetated rooftops in Florida, especially those turf grasses such as Bahia, Sunstar and tropical crabgrass, which thrive on low water inputs. Perennials in this study appeared to best suited for finer gr ained, higher porosity soils, such as UCF and Hydrotechs growing media and need ed more water than was being applied for proper plant health (eventhough a recommended irrigation level th at worked else where was being implemented). Succulents, a hearty plant t ype in all three media, were the best suited plants for B growing medium. If Building Logics would like to make a mix for FL that will support native vegetation, the addition of finer particles, organic matter (composed of finer particles) or an additive such as WTR (wastewa ter treatment residuals), could vastly increase the water retention capabilities for this climate and may improve plant growth of other forms than just succulents. Hydrotech may consider addi ng more OM and/or reducing the medium-sand 105

PAGE 106

portion of its mix, e.g. reducing the quantity of USGA sand that comprises the medium-sand portion to a different ingredient with greater water holding cap acity. There is a need to study change in OM over time in roofs in situ. This study briefly analyzed some changes in OM content and changes in particle size-distribution over time after being in-situ on a green roof for 2 years and 4 years (H on the CRP roof and B on the YSC roof, respectively), but a systematic study of OM changing over time would be useful, as a trend has been noticed cursorily in other studies, but a well thought out study has not been carried out with regards to natural accumulation of OM over time (versus noting a reduction in OM due to decomposition and oxidation over a short period, whic h has been noticed over 1 year studies, such as in Emilsson and Rolf (2005). A mix that starts with high OM seems to oxidize and decrease at first; while a mix that is very mineral with gravel size OM, li ke in the case of B medium, appears to degrade into finer OM particles, increasing the water co ntent and retention capaci ty of the medium and building up the overall OM content over time corollary to plant senescence. In the case of a green roof that is situated out of sight, well inoculat ed Perennial Peanut in a lower pH medium and Phyla nodiflora may have the potential to pe rform well with little to no irrigation. A mix of Mimosa strigilosa and Perennial Peanut grown intermingled with tropical crab grass, may also have the potential to fill a roof quickly, create a monolithic layer with the formation of a large quantity of fine root hairs, based on the fact that tropical crabgrass colonized the abandoned bins in the hotte st, driest part of the seas on, during a time period with no irrigation and survived for one ye ar without irrigation. Further study is needed to corroborate or disprove this theory. Tropical crab grass seemed to accumulate the most amount of fine root hairs the quickest, upon visual insp ection at the time of deconstruc ting the bins and investigating which plants survived without ir rigation a year after planting. The Sedum and Delosperma were 106

PAGE 107

noted to grow back from below-ground biomass the following spring afte r the study, as did the Coreopsis and in some cases the Gaillardia Helianthus did not grow back as frequently or healthily as Coreopsis or Gaillardia in this study. While the Sedums and Delosperma did not do much to reduce stormwater runoff through increase d water retention direct ly after irrigation, the green roof bins planted with thos e two species did stay alive as a whole over one year with and without irrigation and did shade the soil, possibl y increasing the amount of water content in the soil, due to less evaporation of moisture, similar to Getter (2007). Less ET due to plant shading has the potential of allowing more moisture to be present for another plant type when grown in combination with other plants, a possible benef it of growing a continuum of plant forms in conjunction with each other to utilize the soil mois ture at all points in the water gradient, as noted by Wolf and Lundholm (2008). In summary, the first hypothesis that there would be no significant differences in water retention among growing media was rejected and the alternate hypothesis that growing medium type does significantly effect water retention afte r irrigation or rainfall, was accepted. Infact, it was found that growing medium type was the si ngle most important factor governing retention and water uptake and plants were secondary. The second hypothesis that there would be no significant differences in water retention among pl ant types was also reje cted and the alternate hypothesis that plant type does affect water re tention was accepted. The perennial plant type increased water retention more than any other pl ant type in all media. The analysis of water retention for all time periods combined shows that bins planted with perenn ials had significantly higher water retention than those planted with runners, succulents and bare media in B medium; perennials had significantly higher water retention than those planted with succulents and bare 107

PAGE 108

media in H medium; and perennials had significantly higher water retention than bare media in U medium, but not other plant types. The third hypothesis that there would be no significant differences in water retention among the 12 plant-growing medium combinati ons was also rejected, and the alternative hypotheses was accepted that ther e were significant differences among plant-growing medium combinations, with Up having the highest retenti on rate and water uptake/ rerelease rates and Bs, Br, Bm, having the lowest. The fourth hypothesis that there would be no significant differences in ET rates among growing medium types, plant types or any comb ination was rejected, as well as the hypothesis that there were no significant differences in re tention over time was reje cted. It was found that, % water retention fluctuated mainly in relation to water input, which varied over time, as well as varied with respect to changes in the season and related changes in the plants morphology and the ET rates. There was however no specific decreas ing or increasing trend of retention or water uptake and release over time, it wa s most highly associated with changes in the season and rainfall patterns. 108

PAGE 109

CHAPTER 3 NUTRIENT DYNAMICS OF PLANT-GROWING MEDIUM COMBINATIONS FOR NORTH CENTRAL FLORIDA Introduction While green roofs have been proven to provide stormwater quantity benefits (Denardo et al., 2005; Van Woert, 2006; Berghage, 2007), their role in influencing wa ter quality has been questionablesometimes being cited to improve water quality (Berghage, 2007; and Berndtsson, 2006) and other times shown to increase nutrient concentrations in runoff (Van Setter et al., 2006; Berndtsonn, 2006). Berghage et al. (2008) summarizes the curr ent consensus of green roof researchers, that: Nutrient load of a green roof is to a large extent determined by the organic component of the medium and th e fertilization practice and goes on to say that it is clearly necessary to maintain suffici ent nutrient content in the gr een roof to support the plant community, however it is not clear what deficient, sufficient or excessive nutrient content is, or what test should be used to evaluate nutrient content. Certainly, an integral component of the green roof is plants; and for plants to remain viable they depend on water and the ava ilability of nutrients. Nutrients are present in the medium depending on the mineralogy of the growing-medi um, or they must be added to support plant life. Ideally, for a green roof to function as a stormwater quality BMP, it will contain sufficient nutrient levels to support healthy plant life, but little or no excess that will enter stormwater through leaching. Excess nutrients are being defined here, as those leaching out of the growing media and not being taken up by plants. Most hydrologic and water quality st udies related to green roofs begin after the green roof has already been es tablished, this time period ranges from several weeks to several months depending on the study (Van Woert, 2006; Berndtsson, 2005). Therefore, most green roof water quality studie s do not account for the load of nutrients leaving 109

PAGE 110

the roof before the study even begins, when the roof is initially becomi ng established, which is the period that has the highest nut rient load leaving the roof a nd entering stormwater (Emilsson et al., 2007). This study characterized the load of nutrients exiting the roof during this time period. In the case of Florida, th e establishment period was define d as 6 weeks, this time period was chosen based on crop-nutrien t studies in Florida for plants with similar life-cycles and phenology as the green roof plants being tested here (IFAS, 2009 and personal communication Sartain, 2007). This study characterized the nutri ent release from three differe nt growing media being used or with intended use in Florida, vegetated by thre e different plant types, which provides data for modeling the expected nutrient levels in leachate from a green roof in a sub-tropical climatic setting (similar to N. Central Florida), for th ese growing media types at 15 cm of depth. As stated in Chapter 2, to qualify as an eff ective urban stormwater BMP in the subtropics, it is hypothesized that a green roof must have high water rete ntion, medium water re-release characteristics, low nutrient le aching and ability to support hea lthy plant life. Whichever plantsoil combination meets these criteria, will be deemed suitable for FL green roofs from a stormwater Best Management Practice perspective. Any plants that are not viable within the 1 year study period will be deemed unsuitable for nort h central Florida. The plant-soil combination which has plants that rate the highest on the plant health scale described in Chapter 2, (with the least amount of nutrient leaching, and most amount of initial water rete ntion balanced with a mid-range of new available pore space before th e next storm), will be considered a good plant growing media combination for north central Florida green roofs. This study does not endorse any product. The study intends to help identify which of three known growing media (and three known plant types) help keep nutri ent leaching to a minimum in gr een roofs runoff, and aims to 110

PAGE 111

provide the reader insight into any physical and chemical components of the three known growing media that may or may not affect nutri ent leaching. In this way, any green roofer, landuse planner, or policy maker could use this inform ation in a positive way to help tailor green roof media mixes for this region (by increasing certain physical/chemical charac teristics that may be identified here) to be more beneficial to wa ter retention, low nutrient leaching and positive plant health for the north Central Florida region. Land-use planners and policy makers can also use the information in a beneficial way in policy maki ng regarding the management of green roofs and green roof runoff that will most be neficial to receiving water bodies. Objectives The objective of this study wa s to determine which plant-growing media combination at 15 cm (6 inch) depth had the lowest amount of nutrient leaching and how the nutrient concentrations changed over time. Three plant gr oups and three growing media types were tested and compared to each other and to several controls over a period of 6 months in a mesoscale field experiment at the University of Florida, Gainesville, FL. The determination of optimal plant-growing medium combination for a stormwat er BMP was based on 1) plant viability and plant growth, 2) nutrient retentio n/release, and 3) water retenti on/release. This chapter deals exclusively with the nutrient leachin g aspect of the green roof bin study. The specific objectives of the nutrien t component of the study were to: 1. Characterize the total load of N and P leached during the establishment phase of a shallow green roof in Florida (defined as the first 6 weeks after planting). 2. Quantify and compare the load and concentr ation of nutrients in leachate from the experimental green roof bins after establis hment with no irrigation over a longer period of time (6 months). 3. Characterize the nutrient release characteristic s of the plant-growing media combinations and determine which growing medium, plant and or growing medium-plant combination had the least amount of nutrient (TSS, TN and TP) release. 111

PAGE 112

4. Determine if there was a (concentration based) first flush effect afte r controlled irrigation events, where the concentrati on of nutrients (TN and TP) wa s higher soon after the storm ended and lower as time proceeded. Hypotheses The null hypotheses related to nut rients for the establishment period of the green roof (first six weeks of growth) in a shallo w north central Florida green roof 15 cm (6 inches) deep were: H1o: There are no significant diffe rences in the total load of N and P in the leachate among the three growing media types, three plant t ypes or any of the growing media-plant combinations. H2o: There are no significant differences betw een total load of N a nd P among the individual time periods during the six week establishment period. The hypotheses regarding nutrien t leaching for the 6 month study period were the same as for the 6 week period. The alternate hypotheses regarding the cont rolled irrigation portion of the study (also called the Lysimeter Experiment in the previous chapter), where water samples were collected directly after irrigation every 6 weeks up to 24 weeks were: H1a: There is a first flush effect of higher conc entrations of nutrients in leachate immediately after irrigating and lower concentrations later. H2a: Nutrient levels in leachate diminish over time (the 6 month study period) as excess nutrients are washed out of the system. Materials and Methods Experimental Set-Up Plant-growing media combinations were packed into bins using ASTM standard methods described in Chapter 2 in Experimental Set-Up. The growing media characteristics of Hydrotechs LiteTop mix (H), Building Logics medium (B) and UCFs Black and Gold (U) are also described in detail in the Experimental Se t-Up section of Chapter 2, as is the statistical design of the study. Bulk density and porosity for all three media are shown in Table 3-1. Additional tables about the physical and chemi cal properties of Hydrotech medium are in 112

PAGE 113

Appendix J. Refer to Chapter 2 for details of how the bins were packed, set-up and for details about the growing media types, plant types a nd planting techniques; a nd Figure 3-1 shows the complete randomized block design of the 3 x 4 factor ial design, plus controls for this experiment. Experimental Procedure The optimal plant-growing media combinati on study had two main collection methods of the leachate. The first was composite sampling, where leachate collected passively into 5 gallon buckets over 1 week intervals during the first 6 weeks of the expe riment, and then was collected passively over 6 week increm ents up to week 24. The second method was a lysimetric method, consisting of a 1.27 cm controlled irrigation event with the bins being weighed and samples of leachate being taken direc tly after irrigation at va rying time intervals0 minutes, 1 hr, 4 hrs, 12 hrs, and 24 hrs after irrigation. The composite leachate samples were produced over varying time periods. Water samples were taken from the buckets weekly during the first 6 weeks (weeks 1, 2, 3, 4, 5 and 6); these samples of the leachate were the result of both ra infall and irrigation that percolated through the growing media. For weeks 12, 18 and 24 the com posite samples represented leachate from 6 weeks of time and were the result of rainfall only (no irrigation). Irri gation was discontinued after the first 6 weeks of the st udy (i.e. the establishment period). The composite samples were analyzed for TP TN and TSS and measured for volume. The load data from the composite samples were used to characterize: a) the total load of nutrients (TP, TN and TSS) in leachate of all the soil-p lant combinations during the establishment phase of the green roof (defined here as the initial six (6) weeks of wate ring and growing of the plants); b) the total load for three additional six-week time periods, with samples being taken at 12, 18 and 24 weeks; and c) the concentration data from the composite sampling provided data to show 113

PAGE 114

at what rate nutrient leaching diminished in the various growing media-plant combinations over time, if at all. The lysimetric methods primary purpose was to a) characterize changes in nutrient (TN, TP, TSS) concentrations directly after irrigati on at varying intervals (20 min, 60 min, 6 hrs, and 24 hrs), and b) to identify whether a first flush effect was apparent in the runoff. Table 3-1. Measured bulk density (BD) and porosity of the three growing media (B, H, U) before the study. BD (g/cm3) SD Porosity SD B 0.99 0.02 0.31 0.01 H 1.07 0.02 0.31 0.01 U 0.71 0.01 0.41 0.01 S1 / P4 S2 / P4 S3 / P4 S1 / P3 S2 / P3 S3 / P3 S1 / P2 S2 / P2 S3 / P2 S1 / P1 S2 / P1 S3 / P1Plant Group 2: Succulents Plant Group 3:RunnersGroup 4:No PlantsUCFHydrotechBuilding Logics X 3 replicatesContainer StudyGrowing Media 1:Plant Group1:Perennials Figure 3-1. Complete Randomized Block Design of 3 growing media vs. 4 plant types with 3 replicates and 2 filter fabric controls and 2 empty bin controls (not depicted). Sampling Protocol Irrigation regime The irrigation regime (Table 3-2) during th e establishment period was the same as the irrigation regime used on the Charles R. Perry ( CRP) green roof, so that comparisons could later be made between runoff data from Hydrotech bins in this study with runoff from Hydrotech Lite top mix used on the CRP green roof. This irrigation regime was an adaptation based on the 114

PAGE 115

results of Hardins green roof study on optimal ir rigation for Central Flor ida green roofs at UCF 2006 and personal communication with Wanielista UCF (2006). Table 3-2. Irrigation regime for the growing medium-plant bins. Days of Establishment Period (6 weeks) Amount and Frequency of Irrigation Day 1 to 3 1.27 cm 2xs a day Days 4 to 7 1.27 cm 1x per day Day 7 to 14 1.27 cm every other day Day 14 to 21 1.27 cm every other day Day 21 to 28 1.27 cm every third day Day 28 to 35 1.27 cm every third day Day 35 to 42 1.27 cm every third day Rainfall measurements Rainfall for the site was measured using a UF weather station located on top of the UF Dental Tower approximately 500 m from the site at N 29 38 22 and W 82 20, at an elevation of 140 ft, using Peet Bros Ultimeter 2000. Rainfall patterns from the study period are compared to the 30-year average for Gaines ville, Florida from NOAAs records for 1971-2000 (www.ncdc.noaa.gov/oa/climate/online/ccd/nrmpcp.txt). Water quality sampling Weekly composite samples of TSS, TN and TP were taken on Mondays on Week 1, Week 2, Week 3, Week 4, Week 5, Week 6 and Week 12, Week 18, Week 24 (Table 3-3). Samples were composite over time, meaning that all leachate from rain events and irrigation events from the week were passivel y collected through a tube and funne l into a 19 L (5 gallon) bucket covered with a lid that had a raised spout. The volume of leachate in the bucket was measured immediately before sampling. Water quality sample s were taken directly from the bucket after stirring the leachate rapidly, scraping down the por tions of the side of the bucket that were submerged (i.e. if an algal film or slime formed during the weeks time, this film was scraped down and stirred vigorously into the water in the bucket). The sample bottles were acid washed 115

PAGE 116

in the lab, and triple rinsed with leachate in th e field, before being filled with the sample and capped on-site. Water samples taken for TP anal yses were immediately acidified after sampling, and refrigerated to 4C, following USEPA sta ndard procedures for TP samples. In the laboratory, 10 mL aliquots were sub-sampled from the field samples and digested using USEPA standard methods for a persulfate digestion (E PA Method 365.1) and then analyzed on an AQ2+ automated discrete analyzer (Seal Analytical, Inc. Me quon, WI 53092) following USEPA standard methods for TP analyses (EPA Method 365.1). Before water samples for the lysimetric expe riment were taken, the buckets were always emptied and scrubbed and before irrigating the plants and collecting the fresh leachate. The leachate from the irrigation was collected 20 minutes, 1 hour, 6 hours, 12 hours and 24 hours post-irrigation. The volume of leachate was meas ured at each of thos e intervals and water samples were sub-sampled (120mL) and analyzed for TP and TN (TKN and NO3) using the same methods described for the composite samples. Since sampling for the lysimetric experiment occurred only when there were three da ys of sunshine in a row (as the water content monitoring portion of the lysimetr ic study was weather dependent) the sampling days occurred in or near the 1st week, 6th, 12th, 18th and 24th weeks of the study. However, in the event that a rain storm occurred on the scheduled collecti on days, the sampling was postponed by several days, hence the lysimetric sampling dates (Table 3-4) are different from the composite sampling dates. 116

PAGE 117

Table 3-3. Composite sampling dates, time inte rval represented, water source and parameters measured. Week Sampling Date Time Period Leachate Source Parameters Measured Week 1 7/30/2007 1 week Rain/Irrigation Volume, TSS, TKN, NO3, TP, Plant Health Week 2 8/5/2007 1 week Rain/Irrigation Volume, TSS, TKN, NO3, TP, Plant Health Week 3 8/13/2007 1 week Rain/Irrigation Volume, TSS, TKN, NO3, TP Week 4 8/20/2007 1 week Rain/Irrigation Volume, TSS, TKN, NO3, TP, Plant Health Week 5 8/27/2007 1 week Rain/Irrigation Volume, TSS, TKN, NO3, TP Week 6 9/3/2007 1 week Rain/Irrigation Volume, TSS, TKN, NO3, TP, Plant Health Week 12 10/14/2007 6 weeks Rain Volume, TSS, TKN, NO3, TP Week 18 11/30/2007 6 weeks Rain Volume, TSS, TKN, NO3, TP, Plant Health Week 24 1/17/2008 6 weeks Rain Volume, TSS, TKN, NO3, TP, Plant Health Table 3-4. Lysimetric sampling dates, pa rameters measured and leachate source. Week Sampling Dates Parameters Sampled Leachate Source Week 1-U 7/26/07-7/29/07 Leachate volume and TSS, TP, TKN, NO3 at 20 min, 60 min, 6 hr 1.27 cm irrigation event (16 min) Week 1-H 7/30/07-8/2/07 Leachate volume and TSS, TP, TKN, NO3 at 20 min, 60 min, 6 hr 1.27 cm irrigation event (16 min) Week 1-B 8/5/07-8/8/07 Leachate volume and TSS, TP, TKN, NO3 at 20 min, 60 min, 6 hr 1.27 cm irrigation event (16 min) Week 6 9/5/07-9/8/07 Leachate volume and TSS, TP, TKN, NO3 at 20 min, 60 min, 6 hr 1.27 cm irrigation event (16 min) Week 12 10/17/07-10/20/07 Leachate volume TSS, TP, TKN, NO3 at 1 hour post-irrigation 1.27 cm irrigation event (16 min) Week 18 12/5/07-12/8/07 Leachate volume and TSS, TP, TKN, NO3 at 1 hour post-irrigation 1.27 cm irrigation event (16 min) Week 24 2/3/08-2/5/08 Leachate volume and TSS, TP, TKN, NO3 1 hour post-irrigation 1.27 cm irrigation event (16 min) Week 60 9/8/2008--Bare Media only Leachate volume and 2.54 irrigation event (32 min) TSS, TP, TKN, NO3 1 hour post-irrigation Statistical Analyses Calculation of total loads Total phosphorus and tota l nitrogen (sum of NO3 and TKN) concentrations for each bin were multiplied by the leachate volume for each bin, resulting in total mg of P or N leached out of each bin for each time period (weeks 1, 2, 3, 4, 5, 6, 7-12, 13-18 and 19-24). The total phosphorus (TP) and total nitroge n (TN) loads for the establis hment period (weeks 1, 2, 3, 4, 5, 117

PAGE 118

and 6) were compared to each ot her and correlated to the waterIN (rainfall + precipitation) for those weeks. Differences in trends among the gr owing media types were analyzed using the Senslope estimator (Gilbert, 1987), the same method was used for comparing trends in the total load of nutrients (TP, TN) as well as, the individual nitrogen species (NO3 and TKN), among plant types within a growing-media type. For comparisons of load of nutrients over the 24 week time period--the TP loads were compared between 6 week periods, therefore for th e entire establishment period, loads from each week in weeks 1-6 were summed t ogether to give a total load fo r the period week 1-6 and is referred to hereafter as Weeks 1-6 and is co mpared to the loads of Week 7-12, Week 1318 and Weeks 19-24. Transformations of the data set A log transformation was applied to the TP a nd TN load data sets (as well as for the individual nitrogen species) and an AR(1) cubic structure was used, after which the data set met underlying assumptions of normality necessary to use it in a linear mixed model, PROC GLIMMIX using SAS 9.2 and PROC MIXED for the ANOVAs. The actual loads of TP leaching from the grow ing media-plant combinations in each bin as mg per bin were used as the input values fo r the PROC GLM model to test for significant differences in load over time and among plant-gr owing medium combinations. Using true loads (mg P per bin) was chosen over mg P/kg soil for the model input, because the mass of the bins varied with the growing media bulk density; so there was not an equal mass of soil in each bin, just an equal volume of soil. 118

PAGE 119

Results and Discussion Total Phosphorus Concentration and Load during 6-Week Establishment Period During the initial 6 week establishment peri od, there was no definitive decreasing trend in TP concentration during the establ ishment period (first six weeks of growth) for any of the plantgrowing medium combinations. No r was there a trend of decreasing load; TP Load varied more with water inputs than time (Figure 3-2). Figure 3-2. Weekly TP loads (lines) and prec ipitation/irrigation volumes (bars) for the establishment period (first six weeks of growth). TP Concentrations and Loads among Growing Media Types over 6 months TP concentrations in the composite samples of leachate representi ng runoff from 6 week periods, did not follow a decreasi ng trend over the 24 weeks (Figure 3-3), rather they seemed to vary with rain patterns and increases in concentr ations at certain points in time may have been influenced by dry periods preceding heavy rains (Figure 3-4 and Figures 3-5). For example, the increase in TP concentrations for Weeks 7-12 ma y have been due to an accumulation of TP on the surface of the bins during the dry period as was noted in Moilleron et al. (2002) and Gromaire-Mertz et al. (1999), who noticed a distinct increase in pa rticulates in runoff from roofs after a dry spell as compared to wet periods Weeks 7-12 began with a dry 9 day period, 119

PAGE 120

followed by heavy rains (17 cm over several days ) that caused an overflow of leachate in the buckets (Figure 3-6). Infact, normally it is assumed that with more dilution, concentrations will be lower, as described by Gromaire-Mertz et al. (1999) who studied roof r unoff in an urban area, where during high flows from excessive ra in, concentrations were low, while light rain occurring after dry weather resulted in high con centrations, but similar loads. 0.0 1.0 2.0 3.0 4.0 5.0 6.0 7.0 8.0 9.0 10.0 0510152025 WeeksTP Concentration (mg P/L) H conc U conc BL conc Figure 3-3. TP concentrations in leachate from all 40 bins, collected over four 6-week periods. 0 2 4 6 8 10 12 Weeks 1-6Weeks 7-12Weeks 13-18Weeks 19-24 Time periodPrecipitation (cm) Figure 3-4. Precipitation for th e four different 6-week time periods between 7/23/07 1/18/08. 120

PAGE 121

Figure 3-5. TP load as mg P/ kg dry soil from all 40 bins, based on leachate concentrations and volumes collected passively over 6-week periods up to 24 weeks. Regression equations for H: y = 0.0333x2 1.5542x + 17.983; for B: y = 0.0038x2 0.2694x + 3.8546 and for U: y = 0.0065x2 0.2827x + 3.0667. In our case, the TP load leaching each medi um type decreased over the 24-week study period (Figure 3-5), despite slight increases in co ncentration during periods of heavy rain (Weeks 7-12). Concentrations of TP in leachate from bins containing Building Logics (B) and UCF (U) growing media over the 24 week time period ranged between a mi nimum of below detection (0 mg L-1) (measured in Weeks 19-24) to a maximum of 1.0 mg L-1 (measured in Weeks 13-18), which are similar to TP concentrations in green roof runoff reported by Kim et al. (2006) in Korea (1.00-1.18 mg P L-1), slightly lower than values repor ted in Moran (2005), and higher than TP concentrations in Berndtss on et al. (2009), who reported mean TP concentrations of 0.02 0.31 mg L-1 for a range of extensive and intens ive green roofs in Sweden and Japan. The range of concentrations of TP in Hydrotech leachate (0.04 mg L-1 in Week 18 to 10.60 mg L-1 in Week 4) were generally higher than othe r TP concentrations re ported in the literature for green roof runoff (Berndtss on et al., 2009; Berndt sson et al., 2006; Kim et al., 2006); though some values were in the range of TP concentrat ions reported by Berghage et al. (2008) using the 121

PAGE 122

SME (saturated media extract) method to test for TP levels directly in growing media (2.15 5 mg L-1). Hydrotech TP values were, however, lo wer than those in urban street runoff from intensively developed residen tial and commercial watersheds in Korea (Lee and Bang, 2000). The loads of TP per kg of growing media for all bins are shown in Figure 3-5; and the results of the ANOVA on with TP load as the response variable and growing media, plant types and time as factors are shown in Table 3-5. Table 3-5. Results of the type III tests of fi xed effects for the PROC GLIMMIX model on TP load (mg P per bin). (G_media, grow ing media; P_type, Plant type.) Effect Num DF Den DF F Value Pr>F G_media 2 24 121.74 <.0001 P_type 3 24 6.45 0.0023 G_media*P_type 6 24 5.00 0.0019 Week 1 105 5.92 0.0167 Week*week 1 105 18.73 <.0001 Wk*wk*wk 1 105 27.35 <.0001 Effect of Time on TP Load Total P load decreased over time for all growi ng media, as can be seen in Figure 3-5 and in week 18 there were no significant differences in TP load among the three media. Differences among growing media types were more signifi cant than differences between plant types, although there was a plant effect on TP leaching within growing media types. There was a significant effect of time on TP load (p-value for week of 0.0167) (Table 35), therefore two-way ANOVA analyses of growi ng media and plant type were carried out for each six week period separately. Table 3-6 shows the p-values of these processes and how the pvalues of growing media and plant t ype combinations changed over time. 122

PAGE 123

Table 3-6. P-values for 2-wa y ANOVAs of growing media and plant type for four 6-week periods for TP load in bins (analyzed as mg per bin). Weeks 1-6 Weeks 7-12 Weeks 13-18 Weeks 19-24 Growing media <.0001 <.0001 .7951 <.0001 Plant type <.0001 .0120 <.0001 <.0001 G*P combination .5864 .0008 .0003 <.0001 The results of the PROC GLIMMIX model with all weeks combined showed that the three growing medium types had TP loads that were si gnificantly different (p<0.001) from each other. More specifically, based on the 2-way ANOVAs (Table 3-6), the TP load values in leachate among growing media were significantly differe nt from each other in weeks 6 and 12 (p< 0.0001) and in week 18 there were no significan t differences among the 3 growing media. In week 24, while the mean value of TP load per bin in H growing medium had dropped more than 12 fold from its original value (from 13.2 mg P kg-1 soil to 1.01 mg P kg-1 soil), the TP load values had risen slightly from the previous tim e period and were again significantly higher than that of both U and B growing media. The TP load was affected directly by the am ount of rain that fell on the bin. During heavy intense rains of large volume, mo re TP was leached out of the so il. The initial 6 week period was between the end of July and the beginning of September in 2007, which was a period of heavy rain and regular irrigation, and several brief dry spells. Week 6, irrigation was turned off and a hot dry spell occurred (Figure 3-6), stressing ma ny of the plants (signs of heat stress were noticed on the leaves, followed by ins ect infestation on the perennials). The period of time between weeks 7-12 (Septemb er-October) started out as relatively dry compared to the first six weeks of the study, but just before the 12 week sampling event, the bins were hit with heavy rains (end of Septembe r and mid-October) (Figure 3-6). Weeks 13-18 (October-December) was the driest period of the 4 time periods and weeks 19-24 (DecemberFebruary) were also dry and successively cooler th an the other time periods. The effect of these 123

PAGE 124

changes in weather on the plants could have affected the amount of particulate P leaching from parts of the plants that we re degrading or senescing. Figure 3-6. Cumulative hydrograph of precipitation during 24-week study period; sampling dates indicated by circles. Effect of Plant Type and Plant-Growing Medium Combinations on Total Phosphorus in Green Roof Bins The effect of plant type on TP load in leach ate in the overall analysis, irrespective of growing media type, was that runners (r) had TP load s that were significantly lower than that of the two other plant types (succulents (s) and pe rennials (p)) and were si gnificantly lower than bare media (m). Results of the LS-means analysis of the mean areal TP load over 24 weeks for all 12 growing media-plant combinations by GLIMMIX in SAS 9.2 are shown in Figure 3-7. Results of the LS means method for compari ng interactions among plantgrowing medium combinations on TP load over all time periods co mbined show that Um, Up, Ur, Us and Bp are not significantly different from the overall mean TP load for all plants over all time periods, while Bm, Br, Bs and Hm, Hp, Hr and Hs are si gnificantly higher than th e other plant-growing media combinations (Figure 3-7). 124

PAGE 125

While the bins planted with runners had signifi cantly less TP load in its leachate overall, the effect of plants is most si gnificant in week s 6 and 12, when plants ar e fully established and not yet severely affected by frost or lack of wate r. In time periods where the plants are dead or severely impaired, the difference in TP load can not necessarily be attributed directly to the effect of the original plant type in the bin. However, differences in P levels from bins with dead plants may relate to senescence and differences in loss of TP from the dead tissue as below ground biomass becomes incorporated into th e soil or as plants above ground decay and decompose. Figures 3-8, 3-9 and 3-10 show the effect of plant type on the cumulative loads of TP leaching from bins over time within each growing medium as an areal loading rate. From these figures one can see that U grow ing medium had the lowest range of TP load (159 mg P m-2 to 358 mg P m-2), which was just slightly lower than the cumulative TP loads from B growing medium (339 mg P m-2 to 461 mg P m-2) and both B and U had cumulative TP loads significantly less th an that of H (2073 3288 mg P m-2). TP Load m g m-2 Figure 3-7. Mean values of TP load (with standard error) for each growing media-plant combination for the entire study period. (m, p, r, s represent plant types: bare media, perennials, runners and succulents, respectively; B, H, U represent growing media typesBuilding Logics, Hydrot ech and UCF, respectively.) 125

PAGE 126

Within B and U growing media, the plant with the highest cumulative TP load were the succulents. In the case of B growing me dia, succulents TP load (461 mg P m-2) was significantly higher than TP levels from bins planted with runners or pere nnials or bare media (with no significant differences among r, p or m.) Within U growing media, the cumulative load of TP from perennials was significantly lower than all other plant types or bare media (w ith no significant differences among r, s or m.) For Hydrotech growing medium, the cumulative lo ads of TP were highest in bins planted with bare media (3288 mg P m-2), followed by succulents (2850 mg P m-2); the difference in TP load between succulents and bare media was not significant. However the succulents and bare media both had TP loads significantly higher than bins planted with runners or perennials (2087 mg P m-2 and 2073 mg P m-2, respectively). Initial and Final TP Ash Concentrations in the Growing Medium The TP ash growing medium concentration data shown in Table 3-7, indicate that growing media that started out relatively high in TP, such as the H and U medium (with respect to B), terminated with a respectively lower TP ash con centration in the media, as the TP washed out over time. The media behaved as a net source of TP in these cases. In terestingly, the TP ash growing medium concentration data over time for Building Logics indicates that TP is actually accreting in the soil (Table 3-7) as plant material accumulates, yet TP load data from the leachate collected from B bins shows that these growing medium-plant combinations also behaved as a source of TP like U and H plant-growing medium combinations, and not a sink (Figure 3-4). The amount of TP leaching out of the bins cumula tively is higher than the TP ash levels in the growing medium before the study and before planting (Table 3-8)this may indicate that part of the source of the TP load in the leachate is from a) fertilizers on the plant material, b) dry and wet deposition and c) plant senescen ce or organic P from the plant itself. 126

PAGE 127

Figure 3-8. Plant effect on cu mulative TP load in mg P m-2 for Building Logics growing medium. Figure 3-9. Plant effect on cu mulative TP load in mg P m-2 for Hydrotech growing medium. Figure 3-10. Plant effect on cu mulative TP load in mg P m-2 for UCF growing medium. 127

PAGE 128

In week 6, the plants were still growing a nd establishing themselves, and in weeks 6-12 they were also actively growing, as shown in th e plant health graph s hown in Figures 3-11, 3-12 and 3-13. At that point in time, the TP load from bins planted with succu lents, regardless of soil type, were significantly higher than that of runners and perennia ls (p<0.005 value). There were no significant differences in the TP load between succulents and bare media, while bins planted with perennials have significantly lower TP loads than bare media. The results from the two-way ANOVA for week 12 indicate, that when the plants were fully established and a large portion of the initia l available TP had leached out of all the bins (regardless of soil or plant type s), the only significant differen ce in TP load among any of the plant types or bare media, was between runners (which had a significantly lower TP load in its leachate) than succulents. Table 3-7. Growing media TP ash concentrations (mass basis) before and after green roof bin study, and after 2-years in situ on the CRP green roof in FL for H and 4-yrs on YSC green roof in VA for B growing medium. TP mg P/kg soil TP mg P/kg soil TP mg P/kg soil B before 0.18 0.02 B 1yr 0.31 0.07 B 4yr 0.84 0.06 H before 0.94 0.01 H 1yr 0.84 0.35 H 2yr 0.68 0.003 U before 0.66 0.04 U 1yr 0.56 0.01 Table 3-8. Growing media TP ash concentrations (areal basis) before an d after the green roof bin study, and after being in situ on a green roof (for 4 years for B and 2 years for H growing medium). Measured via TP ash on ball milled growing medium samples. TP mg P/m2 TP mg P/m2 TP mg P/m2 B before 32 3 B 1yr 55 12 B 4yr 149 10 H before 167 2 H 1yr 148 61 H 2yr 120 3 U before 117 8 U 1yr 99 30 During weeks 13-18, an early frost affected runne r plants and perennial s; and another very cold period in weeks 19-24, killed all plants bu t two succulent species. In these time periods, bins with bare media had a significantly higher TP leachate load than any other plant type. In week 18 runners had significantly le ss TP in leachate than perennials. 128

PAGE 129

Figure 3-11. Plant Health Index (PHI) for pere nnial plants (Coreopsis lanceolata,Gaillardia pulchella and Helianthus debilis). PHI comb ines % coverage by species with plant health (1-5); PHI value of 35 indicates optimal species coverage (33% coverage of bin) and maximum (5) plant health. Plant Health Index ( PHI ) Plant Health Index ( PHI ) Figure 3-12. Plant health index (PHI) of Runner Type species (Arachis glabrata, Phyla nodiflora and Mimosa strigillosa. PHI combines % coverage by species with plant health (1-5);, with 35 indicating maximum desired covera ge by the species (33% coverage of the bin) and maximum (5) plant health.) Wee k s 7 -12 Weeks 13-18 Weeks 19-24 Weeks 1 6 ) ( PHI Plant Health Index Figure 3-13. Plant health index (PHI) of succule nt type plants (Delosperma cooperii, Portulaca grandiflora, Sedum acre). PHI combines % c overage by species with plant health (15);, with 35 indicating maximum desired coverage by the species (33% coverage of the bin) and maximum (5) plant health. 129

PAGE 130

Nitrogen (TKN, NO3 and TN) Concentrations and Load in Leachate during Establishment Total nitrogen (TN) loads in this study were determined by summing the TKN loads with the nitrate loads. The nitrogen measured as TKN represents all reduced forms of nitrogen, such as organic-N, ammonia-N and ammonium-N; while nitrate-N represents the oxidized forms of N (NO2-N and NO3-N). Total N refers to all the form s of nitrogen summed together and is presented first in this section. Cumulative areal loads of TN leached over the entire study period for the various plantgrowing media combinations ranged from 700 mg m-2 for Up (UCF perennials) to 40,000 mg m-2 for Hs (Hydrotech succulents) (Figures 3-14, 3-1 5, and 3-16). The TN load by week and over all time periods combined varied more among growing medium types than plant types. After log transforming the total nitrogen lo ad data per bin (mg TN per bin), PROC GLIMMIX in SAS 9.2 was used with AR(1) structure subject to bin, with TN load as the respon se variable and growing media and plant types and time as factors (see Table 3-9). 0 500 1000 1500 2000 2500 051015202530 WeekTN (mg N/ m2) Bp Bs Br Bm Figure 3-14. Cumulative TN load leached from various plant types wi thin B growing medium over 24 weeks. B, Building Logics medium; p, perennial; s, succulents; r, runners; and m, bare medium. 130

PAGE 131

0 5000 10000 15000 20000 25000 30000 35000 40000 45000 50000 051015202530 Week TN (mg N/ m2) Hp Hs Hr Hm Figure 3-15. Cumulative TN load leached from various plant types with in H growing medium over 24 weeks. H, Hydrotech growing medium; p, perennial; s, succulents; r, runners; and m, bare medium. 0 200 400 600 800 1000 1200 1400 1600 051015202530 Weekmg N/m2 Up Us Ur Um Figure 3-16. Cumulative TN load leached from the various plant types within U growing medium over 24 weeks. U, UCF growing medium. Among the B growing medium bins, Bs cumulatively leached significantly more TN than Br and Bp (results of ANOVA within B medium bins); by week 18 ther e were no significant differences in load leached among the bins. For cumulative TN load leached from the U bins for various plant types, Us was signi ficantly greater than other plan t types and bare media. Up was significantly lower than Ur and Us (results of ANOVA within U medium bins). No significant differences in load are detected in any medium after week 12. The results of the PROC GLIMMIX model for TN (details of model parameters and fit in Appendix K) was that over all weeks combined, th e three soil types were significantly different 131

PAGE 132

Table 3-9. Results of the type III tests of fi xed effects for the PROC GLIMMIX model for TN areal load data from the 24 week green roof bin study. Effect Num DF Den DF F Value Pr>F G_media 2 24 90.43 <.0001 P_type 3 24 0.47 0.7090 G_media*P_type 6 24 5.00 0.8888 Week 1 106 181.92 <.0001 week*week 1 106 86.34 <.0001 (p<0.001) (Table 3-9). There was a significant effect of time on TN load (p-value for week of <0.0001, Table 3-9), therefore two-way ANOVA analys es of growing media and plant type were carried out for each six week period separatel y. Table 3-10 shows the p-values resulting from these ANOVAs; Appendix L shows the details of th e analyses (F-values and model fit). Plant type had a significant effect on TN load in each time period (Table 3-10), while growing media only had a significant effect on TN load in the first six week period (Table 3-10). There was a slight growing media-plant inte raction in the first six week time period (p=0.06) and no interactive effect of plant-grow ing media combinations on TN load was detected in subsequent time periods (Table 3-10). Table 3-10. P-values for 2-way ANOVAs of growing media and plan t type for TN load for each of the 6-week periods. Weeks 1-6 Weeks 7-12 Weeks 13-18 Weeks 19-24 Growing media <.0001 0.6013 0.1629 0.4153 Plant type <.0001 <.0001 <.0001 <.0001 G*P combination 0.0633 0.1384 0.7527 0.1055 Table 3-11 and Figure 3-17 show how succule nts had the highest TN load, followed by bare media, perennials and runners in every time step. Mean TN load values for plant type are shown by time step in Table 3-11. The cumulative TN areal loading values for B and H (Table 312 and Figure 3-18) were high when compared to lysimeter studies on 50 m-2 plots in Ft. Lauderdale testing nitrogen leaching in St. A ugustinegrass versus a mixed species landscape 132

PAGE 133

(Erickson, 2005). They reported values of cumulative TN loss of 140 mgN m-2 and 250 mgN m-2 averages over a 1 year period, respectively fo r St. Augustinegrass and mixed species landscape (Table 3-13). In this study, the minimum mean cumulative TN loss for perennials was 170 mg N m-2 in U growing medium and occurred over on ly a 6 month study peri od. While significant losses of TN were not expected for U growing me dium after week 18, because past this point in time there were no longer any significant differe nces in load between the time steps, it was unknown precisely what the 1 year load would have been; therefore an es timate of load based on the assumption that the load remained constant for the subsequent 6 months past the end of the study period was calculated. This estima te yielded a minimum of 206 mg N m-2 for the year, which falls between the annual average loads re ported for St. Augustinegrass and mixed species landscape in Florida by Eric kson (2005) in Table 3-13. Table 3-11. Plant type effect on TN load by time period in mg N m-2. Plant Type Weeks 1-6 Mean SE Weeks 7-12 Mean SE Weeks 13-18 Mean SE Weeks 19-24 Mean SE m 497 0.18 42.6 0.16 11.2 0.16 8.99 0.18 p 459 0.18 39.4 0.16 10.3 0.16 8.31 0.18 r 430 0.18 36.9 0.16 9.69 0.16 7.78 0.18 s 537 0.18 46.1 0.16 12.1 0.16 9.72 0.18 Figure 3-17. Plant type effect on TN load by time period (significant differences shown in Table 3-11). 133

PAGE 134

Table 3-12. Growing medium effect on TN load (mg N/m2) by time period. TN Load (mg N m-2) Weeks 1-6 SE Weeks 7-12 SE Weeks 13-18 SE Weeks 1924 SE B 384 1.2 33 1.2 9 1.2 7 1.2 H 1694 1.2 145 1.2 38 1.2 31 1.2 U 169 1.2 14 1.2 4 1.2 3 1.2 Figure 3-18. TN areal load by time period for grow ing media type (signifi cant differences shown in Table 3-12). Table 3-13. Comparison values of TN load from other studies in Flor ida and the Mid-west. TN Load St. Augustinegrass-FL 1.4 kg/ha 140 mg/m2 Mixed Species-FL 2.5 kg/ha 250 mg/m2 Prairie 0.15 kg/ha 15 mg/m2 No tillage corn 50.3 kg/ha 5030 mg/m2 Chisel plowed corn 44.8 kg/ha 4480 mg/m2 Unfert Poa pratensis 1.88 kg/ha 188 mg/m2 Source: Erickson et al. (2005) Figure 3-19. Cumulative TN load (mg N/m2) over 24 weeks for each of the plant-growing medium combinations. LS-Means with Tuke ys test for comparison of significant differences among combinations. Bars not connected by the same letter are significantly different. Standard error bars are shown. 134

PAGE 135

Effect of Plant, Growing Media, Interactive Effects and Time on TKN Load TKN, a measure of the organic-N, ammoni a and ammonium forms of N, comprised 8098% of the total N in leachate for B bins, 4-6% of TN in H bins and 79-99% of TN in U bins in the first week of the study. In the case of the B and U bins, succulents were preferentially taking up TKN (NH4 and organic N) over NO3, as the leachate from Br bins contained 12-19% NO3, while B bins with bare media had leachate with only 2-9% NO3. In the case of the U bins, in those planted with succulents, the total N in the leachate consisted of 80-84% TKN, versus 9098% TKN comprising the TN detected in the leachate from the ba re media bins. This trend of greater uptake of reduced forms of N by CAM succulents was noted in Lttge (2004). He states that the ability of CAM succulents to use N is highly species specific and varies with age and environmental conditions according to Wekmann et al. (1995) and Baatrup-Pedersen and Madsen (1999) in Lttge (2004). Additionally he notes that while, various studies contrast each other with regards to preferential NO3 versus NH4 uptake in CAM plants, Fernances et al. (2002), and Nievola et al. (2001) and Endres and Mercier (20 01 and 2003) all found that the CAM plants they tested preferred NH4 and organic-N over NO3 and glycine. The importance of the preference of reduced forms of N in a green roof system where irrigation comes from a cistern, is that since the water in the tanks during long periods of storage can become anaerobic, most of the N will be in reduced N forms when irrigation occurs, and will be plant available to succulents or other plants that prefer reduced forms on N over NO3. Helianthus, in a study by Kurvits and Kir kby (1979), was found to uptake both NO3 and NH4, with slight preference of NO3 (along with enhanced cation uptake in NO3 fertilized Helianthus plants), but enhanced P uptake was noticed in NH4 fertilized plants NH4. The mean TKN loads in leachate were signifi cantly different among growing media types (p<0.0001 at an = 0.05 level) in the 24 week study, ba sed on the results of a mixed model 135

PAGE 136

(GLIMMIX, SAS 9.2) using log transformed TKN load in mg per bin as an input and plants, growing media, interactive eff ects between plants a nd growing media and time as factors. Table 3-14. Results of the type III tests of fixed effects for the PROC GLIMMIX model for TKN load data from the 24 week study. (Model fit shown in Appendix L.) Effect Num DF Den DF F Value Pr>F G_media 2 24 115.97 <.0001 P_type 3 24 1.83 0.1695 G_m*P 6 24 0.92 0.4960 Week 1 105 0.01 0.9190 Week*Week 1 105 3.71 0.0568 The trends of the cumulative loads of TKN fo llow that of TN presented in the previous section (Figures 3-20, 3-21 and 3-22), with one ex ception: the final cumulative load of Hr was significantly less than Hs and Hm for TKN. In the case of tota l-N, there were no significant differences in the overall cumulative load of TN among the 3 plant types and bare media within H growing medium (Figure 3-21). The significan tly lower amount of TKN leached as compared to bare media and succulents, may indicate a preference of TKN uptake. Additionally, of the three plant types, succulents appear to have been the most heavily over fertilized and have the lowest N requirements and uptake of all the plants (Figure 3-23). 0 50 100 150 200 250 300 350 051015202530 WeeksTKN mgN/m2 Bp Bs Br Bm Figure 3-20. Effect of plant type within Bu ilding Logics growing medium over time on TKN load. 136

PAGE 137

0 100 200 300 400 500 600 700 800 900 051015202530 WeeksTKN (mg N/m2) Hp Hs Hr Hm Figure 3-21. Effect of plant type within Hydr otechs growing medium over time on TKN load. 0 20 40 60 80 100 120 140 160 180 051015202530 WeeksTKN (mg N/m2) Up Us Ur Um Figure 3-22. Effect of plant type within UCFs growing medium over time on TKN load. Figure 3-23. TKN areal loading rates for all plantgrowing medium combinations for total load at the end of establishment period (6 weeks). Bars not connected by the same letter are significantly different. 137

PAGE 138

Effect of Growing Media and Plant Type on Nitrate Loads during the Establishment Period The largest differences in nitrate load among plant-growing medium combinations are due to growing medium type, rather th an an effect of plant type. Sim ilarly to all nutrients measured, bins containing Hydrotech growing medium had th e highest nitrate load in leachate during the establishment period, ranging from a low of 83 mg per m2 (for runners in week 4) and 18,000 mg per m2 (for runners in week 1) (Table M-1 in Appe ndix M). In contrast, bi ns containing B and U media had nitrate loads in the ranges of 3.04 mg N m-2 to 86 mg N m-2 and 0.18 mg N m-2 to 45 mg N m-2, respectively (Table M-1 in Appendix M). Th e minimum levels of nitrate in leachate for both B and U were in bins planted with perennials in week 4. The maximum nitrate loads for B and U growing media originated from bins planted with succulents in week 1. Nitrate levels in H decreased dramatically between weeks 3 and 4, from an average of 9150 mg N m-2 to an average of 190 mg N m-2 irrespective of plant type. For weeks 4, 5 and 6 the maximum nitrate load in leachate from H was 670 mg m-2, which was 200 times lower than nitrate loads measured for weeks 1-3 (Table M-1 in Appendix M). Desp ite the dramatic reduction in nitrate loads emanating from H growing medium by week 4, it st ill had nitrate loads 3 to 50-fold higher than that of bins containi ng B growing medium and 20 to 600 ti mes higher than nitrate loads of leachate from U bins, depending on plant type. In bins planted with U and B growing medi a, the most noticeable plant effect was associated with those bins planted with succulent plants, which often showed significantly higher (p<0.005) nitrate loads in leachate (Figure 3-24) th an other plant types. The trend of elevated nitrate loads emanating from bins planted with succulents was not as apparent in H growing medium (Figure 3-24). 138

PAGE 139

Effect of Growing Media and Plant Type on Nitrate Load over 6 Months For weeks 7-12 nitrate levels are below detect ion in leachate from bins containing UCF and Building Logics growing medium irrespective of plant type; for bins containing Hydrotech growing medium nitrate loads are in the range of 0 1150 800 mg N m-2 in leachate (Table N1 in Appendix N). Figure 3-24. Effect of plant t ype on nitrate loads during the es tablishment period for all plant growing medium combinations, bars connected by the same letter are not significantly different. By weeks 13-18 nitrate loads in all growing me dia types approach similar levels to each other and to the empty bins (Table N-1 in Appe ndix N). The mean nitrate loads in B, Empty, H, U bins respectively are 7.17 3.8 mg N m-2, 0.79 .3 mg N m-2, 1.05 3.5 mg N m-2 and 0.20 3.5 mg N m-2. By week 24 the mean nitrate loads in leachate from B (16.8 7.82 mg N m-2), H (6.00 4.66 mg N m-2) are not significantly different from each other (Table 3-19). However the nitrate load in leachate from bi ns containing U (U = 0.26 0.1 mg N m-2) and empty bins (non-detectable) are significantly lower than the nitrate load from B leachate. 139

PAGE 140

Initial and Final Nitrogen Concentrations and CN Ratios in the Growing Media Nitrogen levels directly in th e growing media were measured before and after the 24 week study in two ways, by elemental analysis in ai r dried soil that was ball milled and by water extraction. The results of these analyses are show n in Tables 3-15 and 3-16. As expected, only a small fraction of the TN in the growing media (as measured by the EA in dried ball milled samples) was water extractable N. For example, in all cases water extractable N is less than 1/10th the amount of N measured in the ball milled samples. This indicates that only a fraction of the total nitrogen present is available to plants. Over time, in growing media that initially ha d high levels of TN in the growing media (H=958 mg N kg-1 soil and U=1867 mg N kg-1 soil), ended with lower levels of TN in the growing medium, as measured directly in the dry ball milled soil, as well as in the water extractable form. This decrease in TN in the medi a correlates to the initi al decrease in organic matter in H and U growing media that was shown and discussed in Chapter 2. Additionally, B which experienced an increase in OM over 1 year (Figure 2-31 in Chapter 2), similarly had an increase in TN in the growing medium (Table 3-16). Table 3-15. Results of water extr actable-N analysis (1:10 (soil to water) water extraction, shaken for 1 hour; N is reported on a concentrati on and mass basis) for growing medium samples taken before and after bin study. Before Study N (mg N/kg soil) After Study N (mg N/kg soil) B 11.13 11.90 H 81.30 16.23 U 8.57 12.70 Table 3-16. Results of EA-TN analysis of ai r dried, ball milled soil, analyzed on the EA (Elemental Analyzer), reported on a mass basi s for before and 1yr after green roof bin study and 2 yr and 4yr in situ on a green roof for B and H. TN (mg N/kg soil) mean SD TN (mg N/kg soil) mean SD TN (mg N/kg soil) mean SD B before 349 23 B 1yr 586 431 B 2yr 1426 134 H before 958 9 H 1yr 775 102 H 4yr 1324 313 U before 1867 19 U 1yr 1313 130 140

PAGE 141

Table 3-17. CN ratios of the growing media befo re and after the study; and for B and H media, CN ratios after being on a green roof for 2 years and 4 years, respectively. Before After 1 year in green roof bins After being in situ on a green roof Media C:N SD C:N SD C:N SD B 26:1 1.4 32:1 1.2 BL 2yr 15:1 1.2 H 39:1 1.7 43:1 0.2 H 4yr 16:1 2.4 U 15:1 1.1 15:1 2.5 Over time the CN ratio of the various growing medium types began to converge on 15:1 (Table 3-17). This was expected for a soil in Flor ida, where soils that have a CN ratio over 25 to 30 have N in the soil that generally becomes i mmobilized by the bacteria population and is not plant available until the ratio begins to drop below 25. The N becomes the most available to plants when the ratio begins to approximate 15:1 over time; organic N in Fl orida mineralizes at a rate of approximately 2% a year. Therefore ex cess N applied early to growing medium is not truly plant available and either becomes immobili zed or exits in leachate. The growing media U began with a CN ratio of 15:1 and remained so over one year, meaning this medium had the most plant available N present in the soil and still had the lowest water extractable N to begin with of the three growing media (Table 3-15). TSS Concentrations in Leachate Total Suspended Solid measurements garnished from composite samples that accumulated in buckets over 6 weeks were indicative of susp ended solids from algal gr owth during that time, rather than sediments leaching di rectly out of the growing medium therefore TSS loads were not calculated for the various time periods. Algal gr owth, surprisingly, was not inhibited by the fact that buckets were opaque, placed in the shade an d had lids, their white color still allowed for algal growth over time. If TSS is referred to as a measure of algal growth then one can note, that more algae grew in the buckets with higher nutrientsHydrotech versus Building Logics, but the data as an indicator of suspended solids not re lated to algal growth was not usable for all time 141

PAGE 142

periods. Figure 3-25 shows the color/turbidity of leachate (composite samples) from the three soil types during a week algae was not present. BL had a reddish tinge, Hydrotech had a brown color and UCF had a yellowish aspect to the le achate. During the one-w eek intervals that the composite samples accumulated in the bucket no free floating algal growth was detected in the water column, however a thin film di d form on the sides of the bucket. Figure 3-25. Photo of leachate from Week 5top bucket contains leachate from U growing medium, bucket on left contains leacha te from B growing medium, and bucket on right is leachate fr om H growing medium. 0.00 10.00 20.00 30.00 40.00 50.00 60.00 70.00 80.00 90.00 100.00 BL H UCF Growing medium typeTSS (m g/ L) Perennials Succulents Runners Bare Media Figure 3-26. TSS (mg/L) concentrations measured in composite sample from Week 3 (no algae). 142

PAGE 143

Lys 18 TSS0.00 2.00 4.00 6.00 8.00 10.00 12.00 14.00 BL H UCF Growing MediumTSS (mg/L) Perennials Succulents Runners Bare Media Figure 3-27. TSS (mg/L) concentrations in leach ate from the lysimeter experiment collected Week 18 (no algae). Week 3 TSS concentration data is shown in Figure 3-26 as an example of a week in the middle of the establishment period where no alga e was noted in the water column and may be compared to Week 18 lysimetric water sample s TSS measurements, where algal growth was also absent (Figure 3-27). By comparing week 3 and week 18, one can notice TSS did decrease over time. Since the buckets were scrubbed each w eek during the first six weeks directly after composite sampling, which was directly before collecting leachate from the lysimeter experiment, TSS concentrations measured during lysimeter experiments did reflect the presence of normal suspended sediments rather than suspended particles of algae. Results of the Lysimetric MethodAnalysis for concentration based first flush (CBFF) The results of the lysimetric method of samplingcollecting water quality samples at intervals of 20 min, 60 min, and 6 hrs after irrigation-for Week 1 showed that TP concentrations did not change significantly ove r time (within the first 6 hours) for any plantgrowing medium type, indicating th at there was not a concentration based first flush effect in the leachate (Figure 3-28). Among media types, B and U growing media have significantly lower TP concentrations than Hydrotech for all plant type s. Also, within Hydrotech and BL growing media 143

PAGE 144

type, there is a visible effect of plants on concen tration: those planted with Perennials have lower TP concentrations than Bare Me dia (p=0.05) within Hydrotech soils for both Time 1 and Time 2. For Building Logics growing media, succulent s appeared to have slightly higher TP concentrations at Times 1, 2 and 3 in Week 1 as compared to other plant types (p=0.05) (Figure 3-28). 12.0 10.0 Perennials Succulents 8.0 Runners Bare Media6.0 TP (mg/L) 4.0 2.0 0.0 H Tim H Tim UCF e 1 e 2 BL BL BL UCF UCF Time 2 Time 1 Time 2 Time 3 Time 1 Time 3 Growing Medium and Time Figure 3-28. Lysimeter Week 1-Effect of time a nd plant type on TP concentrations in leachate from H (Time 1 and 2), B and U growing media (Time 1, 2 and 3). Time 1=20 minutes, Time 2=1 hour and Ti me 3 = 6 hours post-irrigation. Correlations among Nutrients in Le achate, Retention and Water Input TP load in leachate was most highly corre lated with TKN, then TN and then NO3 (Table 318). Nitrate and TN leachate lo ads were very highly correlated (0.99, p < 0.001). There was a slight negative correlation between TP/TKN and the amount of water retained. The more water that was received by the bins via rain and irri gation, the more quantity of nutrient was leached, this was seen more clearly for TP and TKN (r2 = 0.72 and 0.73, respectively with p<0.001 each.) The input of water and the higher correlation of TP leaving may relate to organic P leaching due to particulates leaving the system and/ or increased contact time with the soil. 144

PAGE 145

Table 3-18. Pearson correlation coefficients am ong nutrient loads in leach ate, water retention and waterin. Pearson Correlation Coefficients, Prob > |r| under H0: Rho=0, Number of Observations TP TKN NO3 TN Ret Water_in TP 0.97 <.0001 144 0.84 <.0001 144 0.86 <.0001 144 -0.31 0.0008 108 0.72 <.0001 144 TKN 0.97 <.0001 144 0.90 <.0001 144 0.92 <.0001 144 -0.28 0.0025 108 0.73 <.0001 144 NO3 0.84 <.0001 144 0.90 <.0001 144 0.99 <.0001 144 -0.14 0.1334 108 0.51 <.0001 144 TN 0.86 <.0001 144 0.92 <.0001 144 0.99 <.0001 144 -0.15 0.0990 108 0.53 <.0001 144 Ret -0.31 0.0008 108 -0.28 0.0025 108 -0.14 0.13 108 -0.15 0.09 108 -0.41 <.0001 108 Water_in 0.72 <.0001 144 0.73 <.0001 144 0.51 <.0001 144 0.53 <.0001 144 -0.41 <.0001 108 Conclusions The first goal of the study, to characterize the total load of TN and TP leached during the establishment phase (defined as the first 6 weeks after planting) of a shal low green roof in north central Florida, was met. The range of TP lo ads for the establishment period ranged from 110 mg P m-2 for Up to 1800 mg P m-2 for Hs and Hm for total phosphor us. For total nitrogen the range was 190 mg N m-2 (Ur) to 1800 mg N m-2 (Hs). It was found that the bulk of the nutrients, irrespective of nutrient type (TP, TNTKN, NO3), were leached out of all plant-growing medium combination in the first six week period, as compared to any other time period in the 24 week study period. For example, TN loads in leachate from the fi rst 6 weeks of establishment comprised 89% of the cumulative load of TN fo r the entire 24 week study period for all growing media types. The differences in nutrient loads (for all nu trient types) among grow ing media types were more significant than among plant types. For example, the TN lo ad in the establishment period for H was 1694 mg N m-2, which was 10 times higher than that of U (169 mg N m-2) and 5 times 145

PAGE 146

higher than in B (384 mg N m-2); for the subsequent time periods the difference in load among the growing medium types was less, and over time, the absolute values dro pped ten fold for each growing medium to a range of 145 mg N m-2 to 31 mg N m-2 for weeks 12 to 24 respectively for H, 33 mg N m-2 to 7 mg N m-2 for B, and 1403 mg N m-2 for U, for weeks 12 to 24 respectively. The null hypothesis that there would be no signi ficant differences in to tal load of TN and TP among growing media for the establishment period or for the entire 24-week study period was rejected and the alternative hypothesis wa s accepted. Nutrient loads varied the most significantly among growing me dium types and secondarily among plant types. For all nutrients there was an e ffect of time, this was more slight in TKN loads (p= 0.056), and seen most strongly in TP loads (p < 0.001). TP concentrations from this study for leachate from B and U (ranging from b.d. to 1 mg/L over all time periods), were in a similar range to TP concentrations found in other green roof studies, slightly higher than those from Berndstonn et al. (2009) and (2006) and slightly lower than TP concentrations in Moran (2005). TP concentrations from H were much higher than va lues reported in the lit erature for green roofs (Berndstonn (2006 and 2009), Kim (2006) and Mo ran (2005). The hypothesis that there would be no significant differences in TP load among plant-growing medium combinations was rejected and the alternative hypothesis was accep ted. Mean TP loads for each plant-growing medium combination were analyzed using th e least squares means method for comparing interactions among plantgrowi ng medium combinations on TP load over all time periods combined showed that for all plant types in U bins, as well as the plant-growing medium combination, Bp, were not significantly different from the overall mean TP load, while Bm, Br, Bs and all plant types in H bins were significantly higher than the overall mean. Mean TP loads for plant-growing medium combinations for the entire study period ranged from 18 mg P m-2 146

PAGE 147

for U (with Ur being the lowest and Us the highest within U), 30 42 mg m-2 for B (with Bp being the lowest and Bs the highest) and 90-140 mg m-2 for Hydrotech (with Hr being the lowest and Hm being the highest. Plant effect on cumulative TP load over the whole study period was that perennials had the lowest load and succulents ha d the highest load within ea ch growing media type. Total cumulative TP load for the various combinations ranged from a low of 150 mg m-2 for Up to a high of 3200 mg m-2 for Hs. Cumulative TP loads from leachate from Building Logics growing medium was in the middle with a range from 310 mg m-2 (Bm, Bp, Br) to 450 mg m-2 (Bs). Plant effect on TN was that irregardless of growing medium type, TN loads from runners in each time period were significantly lower than TN loads from succulents, which always had the highest TN load in every time step (though no t always significantly hi gher than bare media). With regards to a concentration based first flush effect, it was found that for TP there was no CBFF within the first six hour s (concentrations were measured 20 minutes post-irrigation, 1 hour post-irrigation and 6-hours pos t-irrigation). There were no significant differences in concentrations in these intervals, however there was a mass based first flush effect for the runoff in the first 20 minutes, simply because the majo rity of the leachate volume percolated through the medium in the first 20 minutes generating th e most volume of any of the time increments and hence resulting in the greatest mass load of nutrients of any of the time increments. The hypothesis that concentrations in leachat e for all nutrients would diminish over the whole study period was accepted, as concentrations of TP, TN, NO3, TKN did decrease in the lysimetric samples over the w hole study period (24 weeks). Implications of the findings are that since a) the nutrient loads in leachate varied most significantly among the growing medium types, but plant health did not vary significantly 147

PAGE 148

between media with even ten-fold differences in nutrient levels in l eachate (for example TN loads in H versus U), and b) the nutrient levels in all the media approach the same levels between weeks 12 and 18, it is recommendable to not pre-ferti lize or pre-mix excessive fertilizers into the growing medium before planting. The findings indi cate that all excess nutrients were not plant available and were not taken up by the plants, but rather would have entered stormwater runoff, posing an ecological threat to receiving water bodi es. (Excessive nutrients were being defined as any nutrient that washed out and wa s not taken up by the plants.) It is suggested that green roofing media manuf acturers in the sub-tropics (Florida) add the minimum levels of nutrients necessary for plant growth to growing medi a and test that those levels are not exceeded at any time. The initial nutrient content and make-up of the forms of N and P in U growing medium (high organic matter, ev en gradation of the material) resulted in less nutrient leaching for all nutrient types and excellent plant he alth. The plant health in bins containing U were comparable to plant health in bins containing H gr owing medium, which had the highest amount of excess nutrien ts in the leachate. It is una dvisable to connect green roof directly to receiving waterbodies before further treatmentsuch as bioretention, vegetative filter strips or wet detention at the ground level. It is recommendable to attach a cistern to the green roof system to recirculate the collected green roof runoff to re-utilize nutrients in the runoff either on the green roof itself or at the ground-level landscape. In summary, the three growi ng media initially had nutrient loads in leachate that were significantly different, with H ydrotech having the highest nutr ient loads and UCF with the lowest. Within 18 weeks all growing media had leach ed out the majority of their excess nutrients and began to have comparable levels of nutrien ts in the leachate (no significant differences among growing medium types). Desp ite the fact that Hydrotechs nut rient levels in leachate had 148

PAGE 149

dropped many-fold by the end of the 24 week study period, it was sti ll the overall highest nutrient load contributor to leac hate among the three growing medium types, irrespective of plant type. Regarding plant types, it is interesting to not e that two genera of succulents, Sedum spp. and Delosperma, which are known to do well in North Carolina (Moran 2004), but anecdotally have been said to do poorly in Florida, constitu ted two of three genera that survived all time periods (drought, frost and heavy rains). However, plant effect-wise thes e succulents did not improve water quality, infact leachate from bins containing succulents had nutrient loads that were significantly higher than leachate from other plant types and even bare media in 3 of the 4 time periods for TP. While runners overall had the lowest TP load s among plant types, th ey were adversely affected by frost and perished in the cold. Th erefore for future studies, instead of testing leguminous runners again, the plan ts that eventually colonized the abandone d bins on their own and survived without irrigation and survived fros t should be tested, such as subtropical grasses which also have the same morphological tendencies as the runners in this study. Finally, in review, since 1) all the nutrients that leached out of the during the study would have entered directly to the storm water system had they been on a roof 2) all growing media regardless of initial nutrient levels, eventually reached the same comparable nutrient loading levels in leachate by 18 weeks and 3) there were no noticeable differe nces in plant health between soil types U and H, despite differences in soil P and N levels, it is recommended for green roofs in Florida to begin with a low nutrient growing media from the start and to investigate the use of slow release fertilizer s in growing media mixtures in the future. 149

PAGE 150

CHAPTER 4 WATER QUANTITY AND QUALITY IN PAIRED GREEN ROOF STUDIES IN FLORIDA AND VIRGINIA Introduction Stormwater Best Management Practices (BMP s) can reduce runoff in various ways, such as interception, soil infiltrati on, evaporation, transp iration, rainfall harvesting, engineered infiltration or extended exfiltration (CWP, 2008). In the case of green roofs, differences in runoff retention are affected mainly by 1) the green roof design (growing medi um depth, roof slope) and 2) climate (annual rainfall, size of the stor ms and duration since the previous storm) (CWP & CSN, 2008; Van Woert et al., 2006; Simmons et al., 2006). To review, from top down, green roofs consist of plants, growing medium, followed by a filter cloth, then drainage layer, than root barrier and impermeable membrane, then roof deck. The constituents chosen to create the growing medium can affect water retention; for example, the addition of organic matter and perlite to the medium can increase porosity a nd lower bulk density and affect water retention directly or by making the pore size distribution favorable to wa ter retention (Brady and Weil, 2002). Additionally, micro and meso st ructures in the growing medi um, as well as the drainage layer (pores, irregularities, hi gh surface area) can contribute to the retention characteristics (Simmons et al., 2008). As a result of all these factors that can influen ce rainfall retention in green roofs, the range of volumetric runoff re duction reported for northern climates varies tremendously (Table 4-1). Green roofs in northern climates are reported in the literature to retain between 30% (Getter et al., 2007 in CWP, 2008) and 94% (Ga ngnes, 2007) of cumulative rainfall for the time period monitored. Simmons et al. (2008) reported rete ntion rates for green r oofs in the subtropics (Austin, Texas) where six different green roof designs with 10 cm of growing media planted with 150

PAGE 151

native perennial plants were tested with 12 mm, 28 mm and 49 mm events. The mean retention ranged from 100% (for <10 mm events), 28%-88% retention for the 12 mm event and 8% to 44% (for the 28 mm and 49 mm events) depending on the design of the green roof. Table 4-1. Range of runoff reduction for green roofs in various climates, using the KppenGeiger climate classification system (Peel et al 2007). Location Climate/ Rainfall Runoff Reduction Reference USA 40-45% Jarrett et al (2007) Germany Temperate Oceanic Temperate Continental 54% Mentens et al (2005) in CWP Michigan Temperate Continental 3085% Getter et al (2007) in CWP Oregon Temperate Mediterranean 68-69% Hutchinson (2003) in CWP N. Carolina Warm oceanic climat e 55-63% Moran and Hunt (2005) Pennsylvania Warm continental cl imate 45% Denardo et al (2005) Michigan Temperate continental 60-70% Van Woert et al (2005) Ontario Temperate continental 54-76% Banting et al (2005) in CWP Georgia Warm oceanic climate 43-60% Carter and Jackson (2007) Texas Warm semi-arid climate 33% Simmons et al (2008) Seattle, WA Cool oceanic climate 65%-94% Gangnes (2007) Climates listed are based on the Kppen-Geiger climate cla ssification from Peel, M. C. and Finlayson, B. L. and McMahon, T. A. (2007). Information in the table is adapted from Center for Watershed Protection (CWP) (2003). Effect of Green Roof Growing Medium and Drainage Layer on Retention Another factor that influences water retention in green roofs is the design of the drainage layer. Differences in retention found in Simmons et al. (2008) were largely attributed to drainage layer differences and secondarily, to the differences in constituen ts of the growing media used. Drainage layers can vary vastly in design, ranging from an aggregat e layer with perforated pipes (UCF green roof), to a geotextile with nylon coils (eg. XeroFlor XF158 used by Van Woert et al. (2006) in Michigan), to egg car ton like cups in a hard corruga ted plastic, such as Zincos Floradrain FD 25 drainage layer (used by Hydrotech, Inc. in this study). Differences in water retention due to the drainage layer can vary gr eatly between studies, fo r example, from 2 L m-2 for Xeroflor XF 158 to 3 L m-2 for Floradrain FD25 used in the Florida roof in this monitoring study, to even greater volumes for aggr egate drainage layers (Gangnes, 2007). 151

PAGE 152

As for the influence of individual constituents used in the growing me dium, Simmons et al. (2006) noted that the green roof type that consistently had th e highest retention had a higher proportion of perlite in the substance (95%) by volume.. In Portland, Oregon, Beyerlein (2005) found that two halves of a roof (built at the sa me time, with the same depth, same slope, and similar porosities) had different overall water reten tion due to differences in the constituents. The author found that the roof half containing 20% digested fiber, 10% compost, 22% coarse perlite and 28% sandy loam, retained significantly more water than the roof half consisting of 15% digested fiber, 25% encapsulated Styrofoam, 15% perlite and 15% coarse peat moss and 15% compost. In other words, the medium that cont ained sandy loama constituent with a finer pore size distribution than other compone nts, such coarse peat moss, a nd more perlite rather than any Styrofoam had greater retention. While both Styrofoam and perlite have similar bulk densities of approximately 0.04 g/cm3, perlite has a much higher water adsorption capacity than Styrofoam. The latter having a water adsorption capacity of 0.25%, while perlite, an expanded volcanic glass, has a 200% 600% water content by weight. Sandy loam in the mix may have accounted for a higher proportion of finer pores than the coarser mixture with the same porosity, since fine pores allow for more evapotranspiration to occur (as compared to a coarse mixture of gravel), releasing more water from the medium between storms and consequently making more pore spaces available for water storage for the next storm. While greater medium depth can be beneficial in providing more overall pore space for the storage of stormwater, Gangnes (200 7) in Seattle, found that grow ing medium that is too thick will inhibit evaporation out of the lower layers of the growing medium stratum. He compared the decrease of soil moisture in 5.1 cm, 10.2 cm, 15. 2 cm and 20.3 cm depths of growing medium after a 1.27 cm (1/2-inch) storm, and found that the soil moisture decreased between 0.5 cm and 152

PAGE 153

1.27 cm in the growing media that were betw een 5.1 cm and 15.2 cm deep over 48 hours, while the soil moisture in the 20.3 cm deep growing medium decreased less than 0.25 cm over a period of 60 hours. Effect of Climate and Rain Event Charact eristics on Green Roof Water Retention and BMP Design Climate and geographical location determine the rainfall pattern, rainfall intensity, evapotranspiration rates and vegeta tion of a region, therefore wate r retention of green roofs is highly influenced by the climate and is variable fr om climatic region to region (Table 4-1). The study areas, Gainesville, Florida and Merrifield, Virginia, both technically belong to the Humid Subtropical Climate (Cfa) according to the Kppen -Geiger climate classification system (Peel et al., 2007); with the Gainesville r oof being located at the sout hern most extent of this classification, and the Virginia roof located at th e northern most extent of this climate. Humid Subtropical (Cfa) is characteri zed by hot, humid summers (warme st mean monthly temperature >22C) and cool winters (coldest mean mont hly temperature between -3C and 18C). Humid Subtropical Climates are defined as receiving si gnificant amounts of preci pitation in all seasons. Winter rainfall in a Humid Subtropical Climate is attributed to large storms blown across the content by the westerlies and summer rainfa ll is largely associated with convective thunderstorms and occasional tropical storms or hurricanes. While Virginia, technically also meets th e Kppen-Geigers definition of humid subtropical climate without a dry season, base d on its latitude, mean warm and cool month temperatures and rainfall patterns, American climatologists often adapt Kppens definition of Humid Continental to include Virginia, by loweri ng the mean cool month temperatures (Peel et al., 2007). 153

PAGE 154

Subtropical and tropical green roofs experience elevated ET ra tes, rainfall rates and peak intensity during rain storms th at are often higher than in te mperate regions (Wong et al., 2002; Kohler et al., 2001). Variation in storm volume and intensity can affect different attributes of the rainfall, such things as the erosivity of the rainfall and influences sediment yield and water quality of the runoff (Trimble, 2007), as well as wh en the peak of concen tration occurs and the lag time to peak of the runoff. Similarly design ed green roofs placed in different climates may behave very differently (for example, due to di fferences in ET rates of the same plants in different climates, and due to influence of storm cycles and length of dry periods on soil moisture and water retention), the location of th e green roof may affect the performance of the green roof for water retention and may also affect water quality of the runoff. For this reason, the present study monitored two roofs in different extremes of one climatic region. I investigated how a 15 cm deep green roof beha ved hydrologically in a sub-trop ical climate in Gainesville, Florida, as well as how a 7 cm d eep green roof in Virginia. I ch aracterized to what degree these green roofs increased the peak to lag time, increas ed the delay to the start of runoff after rainfall begins and increased the duration of return flow and reduced the tota l volume of runoff, if at all. Quantification of these characteri stics of a green roof should a ssist stormwater engineers in designing and rating green roofs as a possible BMP to control volum e and peak runoff rates that can meet regional criteria for stormwater BMPs. Stormwater BMPs in Florida are based on a critical design storm that has a specified peak intensity, duration and total volume depe nding on the region in question. Additionally, stormwater BMPs are regulated to retain a specified portion of the stormwater water, and attain a specified outflow rate within a set period of time, depending on the region and regulating body. For example, in the case of the Charles R. Pe rry Construction Yard (CRPCY) green roof; it is 154

PAGE 155

located within the Ocklawaha River Basin and regulated by th e St. Johns River Management District. Stormwater detention faci lities in this watershed must be designed such that the postdevelopment total runoff volume does not exceed the pre-development runoff for the 25-yr, 24hr storm. Additionally, the rate of discharge fr om a water detention f acility may not exceed the peak pre-development discharge rate for a 24-hour duration storm. Since the CRPCY green roof may be assisting in attenuating the peak runoff rate of a wide variety of storms and may be lowering the total of volume of runoff from stor ms (by detaining a portion of the runoff in the growing media and transpiring a fr action of that rain back to th e atmosphere); green roofs in Gainesville may be able to partially fulfill the water quantity requirements for stormwater detention BMPs specified by the SJWMD for devel opments in the region and therefore reduce the need or lessen the size of on-the-ground stor mwater detention facilities down-slope from a building covered with a green roof. The results of this project can be used by stormwater managers, such as the SJWMD or the Stormwater Planning Division of Fairfax County, Virginia and economists to determine the possible economic benefits of green roofs from a stormwater storage perspective. Stormwater Detention BMPs a nd Critical Design Storms In many regions stormwater detention in ne w developments are based on a large rain event, such as the 25-yr, 24-hr storm event in the Ocklawaha Basin, yet in a humid subtropical climate and humid continental climate (FL and VA) 80% of the storm events of the year are small storms (<2.54 cm in FL, and <2.4 cm in VA) (Casey Trees, 2007). For example, in Alabama, also located in the humid subtropical climate, over a nine ye ar period, rain size and runoff was monitored by Pitt et al. (2007). They f ound that the majority of the storm events (65% of the rain events) were less than 1.27 cm and generated only 10% of the runoff for the time period, and the storms between 1.27 cm and 7.62 cm accounted for 30% of the rain events and 155

PAGE 156

generated the majority (75%) of the runoff volume. The larger storms (7.6 20 cm) made up only 3% of the storms and generated 13% of th e runoff and storms greater than 20 cm were infrequent (less than <0.1% of the rain even ts) and generated <2% of the runoff for the study period (Pitt et al., 2007). Since small and medium sized storms both account for the majority of the storms during the year, as well as the majority of the runoff, Low Impact Developments (LID) techniques typically focus on detaining an d treating the water from these storms, making large, costly and unsightly conve ntional stormwater f acilities obsolete. By capturing 80% of the years rain events, or at leas t attenuating these events, LID te chniques, such as green roofs, reduce the need for as many large BMPs on the ground, such as wet detention, dry detention, and exfiltration trenches and even bioretention cells which can save both space and money for the developer and the county. Coupled with cisterns, the reduction in total stormwater volume can be even greater than that of the green roof alone, reaching as much as 87% as reported by Hardin (2006) for Central Florida. Additio nally, this water can be reused on site for either irrigation of the vegetation at ground level or reused for irri gation of the green roof during periods of low precipitation or high heat and ET. Water Quality and Green Roofs Rooftops as a source of urban stormwater Urban areas contribute large amounts of st ormwater runoff and pollutants due to impervious surfaces. In a highly urbanized city setting, approximately 72% of the land area is impervious (Schueler, 2001), and 40% of the impe rviousness is comprised of rooftops (Urbonas, 2001; and Liptan, 2003); the ot her 60% of the impervious ness is car habitat. Rooftop runoff poses a greater threat to water quantity in urban watersheds than rural watersheds. This is because the runoff enters re ceiving water bodies more rapidly in urban areas 156

PAGE 157

than in rural settings due to the greater connec tivity of roofs to gutters and storm sewers. The presence of pavement impedes infiltration to groundwater, increasing th e proportion of water going to surface overland flow and increasing the velocity of the runoff. When surfaces are paved, vegetation that originally provided inte rception and evapotranspi ration is removed, and natural depressions in the landscape, which norma lly detain 50% of the runoff, are eliminated (Dunne and Leopold, 1978). The volume and rate at which the runoff is delivered to the receiving water body is greatly increased (A ndoh, 1997), resulting in a reduction of the hydrologic response time and gr eater recurren ce of floods. Deterioration of the receiving water bodies in urban areas is usually first noticed visually as channel erosion and degradation (Liptan and Strecker, 2003). The effects of stormwater runoff on water quality become evident later, when biological imbalance relative to predisturbance conditions occur (Stribling and Leppo, 2001). Since sediment concentration is a function of a combination of rainfall intensity and runoff rate (Trimble, 2007), green roofs, that can eff ectively lower the runoff rate and runoff amount from a developed area, have the potential of lowe ring the amount of soil particles that it detaches and transports enroute to a receiving water body, or in the receiving water body itself. Secondly, because of the importance in the relationship between rainfall intensity, the kinetic energy of the rain drops formed and their erosive power, it is important that a green roof be vegetated, for positive effects of water quality with relation to sediment yield and adsorbed P associated with an increased sediment yield. Rooftops contribute to stormwater pollution via two mechanisms, one is through the release of constituents from the roofing materials usedsuch as zinc, copper, polyaromatic hydrocarbons (PAHs), cadmium or lead (Cla rk, 2001) and secondly from atmospheric 157

PAGE 158

depositionfor example nitrogen, phosphorus and ev en pesticides (Moran, 2003; Dietz et al., 2005). Researchers in Michigan l ooking for sources of stormwater contaminants found rooftops to be the largest source of dissolved metals, wh ile parking lots contributed the PAHs (Clark, 2001). Another study comparing old metal roofs with new built-up roofs and old wood roofs with tar, found the new built up r oof to contribute 10 times as much total copper (0.13 mg L-1) than the other two types; and found that the old me tal roof contributed the highest amount of lead (0.035 mg L-1) and zinc (11.9 mg L-1) (Clark, 2001). Copper is the 3rd most utilized metal by man after aluminum and steel (Jolly, 2000 in Arnold, 2005), it is used in residential and commercial architecture because it is attractive, durable and a fire retardant. Currently copper products used in exterior architectural appl ications average 168.3 million kg/yr between 2001 and 2004 (Copper Development Association, 2005 in Arnold, 2005). Since copper is used in exterior applications, it weathe rs due to all forms of precipita tion; the quantity of Cu that dissolves and is transported is a function of atmospheric chemistry precipitation and roof orientation. Estimated copper load ing rate for roof runoff for th e US based on 179 locations and the factors mentioned above (pre cipitation chemistry, amount of pr ecipitation and roof aspect) is 2.12 g Cu/m2/yr with a range from 1.05 g Cu /m2/yr to 4.85 g/m2/yr, depending on location (Arnold, 2005). Copper in small quantities can be benign or beneficial, however even at low doses in the (ug/L) range, copper can become toxi c to aquatic organisms and other forms of life (Arnold, 2005). In Charlotte, North Carolina, atmospheri c deposition constituted 10-30% of total phosphorus and nitrogen as nitrat e, 30-50% of orthophosphorus, a nd 70-90% of total Kjeldahl nitrogen and ammonia in stormwater runoff (W u, 1998 in Moran, 2003). Rural roofs, especially those located within a 1-3 mile area of agricultural areas receive atmospheric deposits of nitrogen 158

PAGE 159

(N) and phosphorus (P), as well as pesticides in residential area s (Moran, 2003). In one Florida example, total inorganic nitrogen de position was estimated to be 17 kg N ha-1 yr-1 or 9,700 metric tons yr-1 in the Tampa Bay Watershed. The ratio of dry to wet deposition rates for inorganic nitrogen was 2.3. Ammonia (NH3) and nitrogen oxides (NOx ) contributed the most to the total N flux with 4.6 kg-N ha1 yr1 and 5.1 kg-N ha1 yr 1, respectively. Average wet deposition rates were 2.3 and 2.7 kg-N ha 1 yr 1 for NH4 and NO3, respectively (Poor et al., 2006). There are three likely mechanism by which gr een roofs could reduce nutrient loads 1) retain or transform contaminant, 2) reduce volu me of water thereby reducing contaminant load assuming concentration does not change, and 3) reducing volume thereby reducing velocities in downstream channels thereby reducing scour and sediment transport. Green roofs can reduce contamination from rooftops by reducing the amount of water leaving the roof (De Nardo, 2003; Van Woert, 2005; Liptan, 2003; Vill areal, 2004) and, by plant uptake and transformations of N and P deposited atmospherically or added by fertilization (Berndtsonn, 2006). In a second study in Sweden, Berndtsson et al. (2006) examined green roof runoff quality from green roofs over bicycle park s in Malmo, Sweden and green roofs covering a school in Augustenborg, Sweden. They found that the green roofs in Augustenborg acted as a sink for nitrogen, reducing TN by 58% (rain water contained 909 mg TN m-2 yr-1 and effluent from the roofs contained 378 mg TN m-2 yr-1); and the roof behaved as a source for phosphorus and potassium. They also found that the green roofs behaved differently either due to age, parent materials or input levels with re gards to metals in runoff. The newer green roof in Malmo acted as a sink for lead, while the older green roof on top of the school in Augustenborg behaved as a source for lead. The results of German study on ve getated roof research plots at the Technical 159

PAGE 160

University of Berlin in contrast, reported that over a 3-year period the green roof plots were capable of retaining 95%, 88%, 80% and 68% of the loads of Pb, Cd, NO3 and PO4 (in Berndtsson et al., 2006) applied to the plots. The authors did notice that at low level inputs the roof could act as a source, but at high level inputs the green roof acted as a sink. Since similarly designed green roofs placed in different climates may behave very differently, I am studying how a green roof behave s in a sub-tropical climate in Gainesville, Florida, as well as one in the tr ansitional zone out of sub-tropi cal into humid continental, to characterize to what degree different green roofs in this climate will increase the peak to lag time, increase the delay to the start of runoff afte r rainfall begins and increase the duration of return flow and reduce the total volume of runo ff. Quantifying these characteristics of a green roof will assist stormwater engineers in design ing and rating green roofs as a possible BMP to control volume and peak runoff rates that can me et existing state-wide, county-wide or Water Management District-wide criteria for stormwat er BMPs. Urban areas are one of the largest contributors of runoff and pollution to waterways. Hypotheses and Objectives Green roofs have been studied extensively in temperate climates for their effectiveness in reducing stormwater quantity. However, little is known about how effectively green roofs: function in subtropical climates, such as Florida and Virginia act as a source, sink or transformer of nutrients plant type influences green roof function The general objectives of this por tion of the research are to: Compare water retention capabilities and aff ect on lag time and peak runoff attenuation of green roofs as compared to modeled conve ntional roof runoff at two extremes of the subtropical climate type Determine if green roofs act as a source or sink for nutrients and metals in these same two subtropical locations 160

PAGE 161

Hypotheses The hypotheses for the two paired green roof/c onventional roof mon itoring studies were: H1: The overall benefit of green roofs for wate r retention, peak runoff attenuation and increases in lag time to peak runoff will be less in green roofs in Florida than in Virginia, because of higher peak precipitation rates, greater total volume of rain events, and greater recurrence of convective storms in a humi d subtropical climate (Florida) than in the transitional zone of the humid subtropical climate into a humid continental climate (Virginia). H2: Green roofs influence on water quality in st ormwater runoff will be similar in Virginia and Florida and will act as a si nk for nitrogen and source for phosphorus and sediment, and may be either a sink or source for metals when compared to conventional roofs. Specific Objectives The specific objectives of the monitoring were to: 1. Determine whether green roofs can mitigate for stormwater quantity in the same manner at the two geographic extremes of one c limatic zone, by quantifying the amount of stormwater retention, peak runoff attenuation, and increase in lag time to peak, if any, due to the presence of a green roof in both Florida and Virginia. 2. Measure water qualityconcentratio n and load of nutrients (NO3/NO2, PO4, TDS, TSS) and metals (Cd, Zn, Al, Fe, Cu) in green r oof and conventional roof runoff in Florida and Virginia to determine whether the green roofs are acting as a sink or source for N, P and metals, and whether they are behaving th e same way (as a sink or source) for the same parameters in both extremes of one climatic region (Florida and Virginia). Materials and Methods Study Site 1: Florida The Charles R. Perry Yard structure is a single story building with a green roof of approximately 241.5 m2 (2600 square feet) with a 1:12 slop e. The building is adjunct to the existing Rinker Hall. The green roof consists of 15 cm depth of Hydrolite growing medium underlain by 5 cm of green roof underlayment material provided by American Hydrotech, Inc. (Table 4-2). The green roof was installed and planted in March 2007. It is visible from the second and third floor windows of the east si de of Rinker Hall (Figures 4-1 and 4-2). 161

PAGE 162

Table 4-2. Underlayment of the Hydrotech green roof. Source: Hydrotech Specification Sheet (2005). Material Purpose MM 6125 Waterproof membrane Surface Conditioner concrete conditioner FlexFlash F membrane reinforcement FlexFlash UN flashings 12" x 100' Hydroflex RB protection/root barrier Floradrain FD25 drainage/H2O retention SystemFilter SF Filter Fabric LiteTop Intensive 6" LiteTop Intensive Soil Figure 4-1. Photo of the Charles R. Perry Cons truction Yard after bituminous water proof layer was cold applied, photo taken from the third floor conference room, March 2007. Figure 4-2. Photo of the installa tion of plants on the CRP green roof. Photo was taken from the third floor conference room of Rinker Hall in April 2007. 162

PAGE 163

Growing media The growing media used was Hydrotechs Litet op which consists of 45-70% Lite Top Lightweight aggregate (0.015 cm to 0.95 cm aggr egate), 0-30% Coarse to Medium Sand, 0-30% Perlite, Sphagnum, or Other Lightweight Soil Additive and 5-30% Approved Compost and Nutrient additives as needed. Chapter 2 shows the results of our own analysis of grain size distribution of the bulk growing medium deliver ed by Hydrotech to UF, used on the CRP green roof and Chapter 3 contains the results of the TP/TN analysis of the medium. Table 4-3 shows the density and saturated water a nd air content, the OM and C:N ratio reported by Hydrotech and Table 4-4 show the nutrients added, also from the Hydrotech Specification sheet. Table 4-3. Physical properties of Hydrotech LiteTop grow ing medium. Source: Hydrotech Specification Sheet (2005). Property Value 0.6-1.1 g cm-3 Dry bulk density 1.0-1.5 g cm-3 Saturated bulk density Saturated water capacity > 40% Saturated air content > 10% Organic matter content (mass %) 6 12% C/N ratio < 20 CEC > 6 Table 4-4. Nutrients added to Hydrotech Li teTop growing medium. Source: Hydrotech Specification Sheet (2005). Concentration Mass basis Areal Loading Rate Min Max Min Max Nutrient (Plant Available) g kg-1 g kg-1 g m-2 g m-2 Nitrogen(NO3,NH4) 0.047 0.236 7.7 38.3 Phosphorus 0.016 0.110 2.6 17.9 Potassium 0.094 0.236 15.3 38.3 Calcium 0.298 1.021 48.5 166.0 Magnesium 0.047 0.236 7.7 38.3 Sulfur 0.016 0.055 2.6 8.9 Iron 0.016 0.047 2.6 7.7 Manganese 0.016 0.047 2.6 7.7 Copper 0.004 0.008 0.6 1.3 Boron 0.004 0.008 0.6 1.3 Zinc 0.0002 0.0004 0.0 0.1 163

PAGE 164

Cisterns and pump station Two 5670 L cisterns, 1.6 m diameter by 2.9 m high were installed above ground located approximately 9.1 m from the building. Black polyet hylene tanks were chos en over clear tanks to inhibit possible growth of algae. The water from these cisterns was delivered back to the roof through a 1 hp pump (see Appendix O). Cistern si zing was determined based on storage volume sufficient for 3 weeks of irrigation at a rate of 1.27 cm of water two times per week to reduce the need for supplemental water. The cisterns were located away from the building with pipes from the downspouts located underground (Appendix O). Firm, even, compacted beds of sand and gravel were constructed in February 2007 for the tanks. Water from the cisterns (which consisted of rain runoff and irrigati on runoff) were pumped from the cisterns to the roof via 3.175 cm (1 ) PVC pipes. These pipes ran horizontally underground to the edge of the CRP Construction Yard building and then vertically up the side of the building. Once on the roof, the irrigation water was transported laterally through one 1.9 cm () PVC pipe along the edge of the gree n roof, and from there entered through connectors into each of forty-seven 1.27 cm dia. () Ne tafim lines covering the roof surface evenly, laying perpendicular to the main PVC pipeline. The pump station by Resources Recovery, Inc. co nsisted of a control pa nel, a filter box, a reservoir tank, a pump and junction box, a switch c onduit, a pump intake pipe, irrigation tap, the purge tap and the primary tap. The control pa nel housed the electronics that operated the switches and pumps. A backflow preventer was installed on the incoming potable water line, however originally there was no backflow prev enter on the incoming UF reclaimed irrigation line, which had allowed for sudden drops in wate r due to irrigation occurring on the UF campus located at an elevation lower than the cisterns A check valve was added to the UF reclaimed irrigation line in April 2008 to prevent a siphon effect from occurring when UF Physical Plant 164

PAGE 165

Department (PPD) would either 1) shut off its pump stations during heavy storms, causing the line to no longer be pressurized or 2) irrigate at an elevation lower than the cisterns, or 3) experience a break in an irri gation pipe at an elevati on lower than the cisterns. After the water was pumped to the roof and the green roof was irrigated, the runoff was collected by two 10.2 cm (4) drai ns, one for each half of the roof, which merged into one 10.2 cm (4) pipe underneath the eaves of the building and that turned vertical and carried the water underground to a Y-junction, where a valve contro lled whether the runoff flowed to an existing stormwater drain or instead to th e cisterns to be monitored. For the period of the study, the valve was switched to the cisterns. Automated irrigation system Netafim micro-irrigation system, donated by Rainbird, Inc., was installed in March 2007. Tubing, 1.27 cm () in diameter was spaced 30.5 cm (12) apart in parallel rows, covering the entire roof. Irrigation times we re set for 3 times weekly for 1.27 cm (34 min) at 5 am on Mondays, Wednesdays and Saturdays from March 2007 to July 2007 and later reset to 2 times weekly on Mondays and Thursdays at 5 am. Becau se of excessive runoff after irrigation, the duration was further reduced to 26 min in March 2008, the equivalent of 0.91 cm of water. When it was noticed that water was insufficient (plants browning slightly), the irrigation frequency was raised to 4 times weekly but the duration wa s lowered to 17 minutes (or 0.635 cm irrigation depth) starting June 27, 2008 (Monday, Wednesday, Friday, Sunday at 5 am). This new regime of more frequent, yet shorter du ration irrigation events resulted in less runoff and greener plants. The final irrigation arrangement remained 0.635 cm of water, 4 times a week (for a total of 2.54 cm per week). 165

PAGE 166

Study Site 2: Virginia Yorktowne Square Condominium green r oof was retrofit onto a 1968 condominium building in 2002 by Building Logics, Inc. (Figure 43) at the same time that a new conventional roof (bituminous asphalt shingle) was installed on a different section of the same building (Figure 4-4). The green roof was an EnviroTech system, construction of the roof consisted of a base sheet of tar covered with a 2-ply membrane with copper foil root barr ier, which consisted of one ply Famobit P4 and another ply of Fa mogreen RET CU-P4, as the top ply. The two membranes consist of a high-grape polymeric bitumen sheet and was modified with age stabilizing amorphous polyalphaolefin (APAO). H ydrogel packs are laminated to the top surface of the APAO membrane. Drainage occurs through and around the gel packs and eliminates the need for a separate drainage layer. The storage of water in the underlayment is specified as 3L m-2. The Famobit layers were followed by a filter cloth and 6.35 cm of Building Logics growing medium planted with 8,400 plugs of Sedum album S. sexangular and S. reflexum Figure 4-3. Green roof installe d by Building Logics in 2002, planted with 8,400 sedum plugs (S. album, S. sexangular, and S. reflexum ). Photo taken May 2005. 166

PAGE 167

Figure 4-4. Conventional Roof at Yorktowne S quare Condominiums, built in 2002, at the same time as the green roof on the adjacent build ing. Note puddling after rainfall, shown in foreground of photo. Runoff was collected at the Yorktowne Square green roof in Virginia from May 2006 to September 2008 in two 3426 L (905 gallon) cisterns from 1/8th of each roof surface or 48.8 m2 (525 sq. ft.). Two Global Water WL15 Level Logg ers pressure transducers monitored the stage height inside the cisterns and a rain gauge on top of the roof outfitted with RainLoader 2.1 software recorded rainfall. Data Collection and Analyses Rainfall data in Florida was collected by the UF Physics Department weather station using Texas Weather Instruments WRL-25 weather mon itor, on campus located 400 m from the green roof. Rainfall and runoff data were collected in 5 minute increments and paired together to create cumulative and individual storm hydrographs. In Virginia, rainfa ll was monitored using a rain gage located on top of the green roof and data was collected using RainLoader2.1 software. 167

PAGE 168

Green roof runoff flow rates and volume measurements were calculated based on change in stage height in the cisterns. The stage height was monitored using a Global Water (Gold River, CA) pressure transducer. Global Logger II (V er. 2.1.2) software prog ram was employed to download data bimonthly from July 2007 to Se ptember 2008 in Florida and from July 2006 to August 2008 in Virginia. In Florida, the pressure transducer was placed on the bottom of one of the two cisterns, which were connected to each ot her and had the same base level, so that the water levels in both cisterns were always equal. The measured water level was multiplied by the cross-sectional area of the tw o cisterns to yield the volum e of runoff. Runoff volume was consequently divided by total roof area (415 m2), resulting in centimeters of runoff per roof area, so that it could easily be compared to precipitation, which was also reported in cm. In Virginia, one pressure transducer was pl aced in each cisternone for the green roof and one for the conventional roof a nd stage height was multiplied by the cross-sectional area of the cistern and divided by the roof ar ea represented by the runoff (48.8 m2), which was one-eighth of the green roof. The identical amount of roof ar ea was monitored for the conventional roof as for the green roof, also the conventi onal roof had the same aspect a nd slope as the green roof and were the same age, as they were both rebuilt in 2002. Analysis of cumulative rain and runoff Cumulative runoff data from the Florida green ro of was paired with cumulative rain data for each seasonal period where data existed. Ther e were certain brief periods of time before March 2008, where leakages were occurring in the downspout from the roof, and the incoming irrigation line was acting as a siphon and emptying the cistern in Florida. Data from those periods of time were dealt with as follows: 1) for sudden draw-downs in cistern volumes during a storm or UF irrigation event due to the lack of check valve in the incoming UF irrigation line the exact water lost from the cist ern was added back to the data set; 2) for periods of time when 168

PAGE 169

the cisterns were overflowing, runoff was simulate d (and indicated as such in the results) based on the relationship of rain intensity and runoff flow directly before and after the overflow which yields a conservative estima te for green roof retention. The or ifice of the input pipe to the cisterns and the orifice of the overflow pipe leaving the cister ns are both 10 cm in diameter, therefore in cases of overflow where saturation was already reach ed on the green roof, the exit flow rate was assumed to equal the inflow rate (rain rate + a delay). In Virginia, cumulative runoff data from the gr een roof were origina lly paired with runoff from the conventional roof, however to be able to qualitatively compare retention rates to those measured in Florida, ultimately rainfall rates we re chosen for comparison instead, so that the method in Virginia be identical to the method used in Florida. Analysis of individual rain events Individual rain events were analyzed by pairi ng the rain data with runoff data at 5 minute intervals to create storm hydrogr aphs. The storm hydrographs were analyzed for 1) a reduction in total runoff volume, 2) a reduction in maximum runoff rate in cm hr-1, 3) an increase in lag time between peak of rainfall and peak roof runoff, an d 4) an increase in return flow period, (Figure 4-5). The reduction of total runoff volume was the difference in runoff due to rainfall with a runoff coefficient of 1.0 compared to the measur ed runoff volume from the green roof. A runoff coefficient of 1.0 assumed that all rainfall b ecomes runoff instantaneously (Figure 4-6). The reduction in maximum runoff rate is the % di fference in the maximum rain rate and the maximum green roof runoff rate, (Figure 4-7). This represents the reduction in the peak runoff rate that the green roof coul d provide for a worst-case scenar io conventional roof, where the runoff coefficient equals 1 and where the ma ximum runoff rate for the conventional roof approximates the maximum rain rate. The lag time to peak concentration for a conventional roof 169

PAGE 170

under this assumption means that the peak runoff with a runoff coefficient of 1 would occur at the time of peak rainfall. Therefore, the increase in lag to peak here is defined as the time difference between when the maximum rainfall rate occurs and the time of maximum green roof runoff occurs; both values are reported in cm/hr. The return flow period of the green roof, also called the extension in green roof runoff, refers to that period of time after which rainfall had ended and runoff continued and is reported in hour s and minutes. The delay in the start of runoff is the delay between the beginning of rainfall a nd the beginning of runoff. Water retention by the green roof, was characterized to be the decrease in total volume betw een rain fall and runoff; this value is reported as cm (of depth across the roof surface), Figure 4-6. The value reported as depth of rainfall or runoff across the roof su rface can be multiplied by any other green roofs areal extent for quick comparisons in water reten tion. Percent water retention was defined as the quantity of water retained by the gr een roof system (media and drai nage layer), calculated to be the difference between measured rainfall and m easured runoff, divided by the total input: % Water RetentionGreen roof syst em = (Rainfall [cm]-Runoff [cm])/ Rainfall [cm]). Hydrographs reporting modeled conventional roof runoff refers to a runoff coefficient of 1, implying that all rain was transformed directly into runoff to be used for comparison to the green roof runoff. 170

PAGE 171

Rainfall/Runoff (cm) Figure 4-5. Example of analysis of individual hydrographs from a green roof (pink line) and modeled conventional roof runoff based on ra infall (blue line) to determine increase in delay in start of runoff and the increas e in lag to peak and the increase in the extension of runoff from the green roof past the end of the storm event. Rainfall/Runoff (cm) Figure 4-6. Example analysis of an individual hydrograph for storm water volume reduction due to the presence of a green roof. The differe nce in the areas under the curves, the blue line represents the amount of rainfall; th e pink line is the green roof runoff. The difference between rainfall volume and runoff volume is the amount retained by the roof. 171

PAGE 172

Rainfall/Runoff (cm) Figure 4-7. Example analysis of individual rain event and green roof hydrographs for reduction in peak storm runoff. Collection of water quality data Water quality was monitored in the runoff of the green roof and conventional roof by taking grab samples during 5 diffe rent storms over a two year period, with 5 samples/storm x 2 roofs. Water samples were analyzed for nutrients (NO3/NO2, TP, OP) and metals (Cu, Fe, Cd), as well as TSS, TDS and pH. Water quality samples were taken as grab sa mples at various interv als during individual storm events with the goal of obtaining five samp les per storm. The sample s were taken in acid washed bottles in both Florida and Virginia. At th e CRP green roof in Gainesville, FL, the runoff was collected by opening a Y-junction that was spliced into the iron 10.2 cm (4) downspout pipe in June 2008 for the express purpose of water sampling during storms. The PVC Y-Junction was spliced into the downspout pipe downstream of two indivi dual 10.2 cm (4) roof drains merged into one 10.2 cm (4) drain. During sampling, sample bottles were triple rinsed with the runoff, then held gently against the edge of the in side of the pipe in such a way that it would fill 172

PAGE 173

up as quickly as possible. Since the ultimate duration of the storm was unknown at the time when the first sample was taken (which was the first moment that there was en ough runoff to sample in the down spout), 5 samples were only obtained in storms that lasted several hours. For storms that were less than an hour long, generally only 2 to 3 samples were obtained before the rain subsided. Water samples from the green roof were compared to water samples the conventional roof on top of Newell Hall on the University of Florida Campus in Gainesville, Florida. Water samples were analyzed for nutrients (TP, SRP, NH4, NO3) on the AQ2+ (Spectrophotometer) semi-auto analyzer, total suspended so lids (TSS) in the Wetlands Biogeochemistry Laboratory and metals usi ng an ICP (Inductive Coupled Argon Plasma) Spectrophotometer in the Soil Tes ting Lab of Belle Glade, FL. The nutrients measured were TKN, NO3, NH3, SRP and TP. Metals analyzed were Fe, Cu, Zn, Cd, and Al, these metals were chosen for analysis for two reasons: 1) in ur ban areas, streets and r ooftops are the top two contributors of Zn and Al, therefor e we were interested in charact erizing the levels coming off of a conventional roof and green roof and noting any differences that may be attributable to the green roof, and 2) in previous studies (Ur bonas, 1998; and Kohler and Schmidt, 2003), green roofs have been shown to lower concentrations of metals than conventiona l roofs, therefore we wanted to see whether or not this trend would hold true in Florida and Virginia. The same parameters were analyzed in Virginia by the Water Quality Laboratory of the Fairfax County Department of Health by Debor ah Severson using the same EPA methods as used in the WBL laboratory. Concentration data were transformed into load data by multiplying the concentrations of the measured parameter at a given time by the to tal volume of runoff that had occurred in the time increment represented by that measurement. Instantaneous flow rates used in load 173

PAGE 174

calculations were based on the measured changes in the stage height in the cisterns during the storms at the times when the grab samples we re taken. The same met hod of measuring flow rates, based on continuous stage height measurem ent over time was used in Virginia as well. Analysis of water quality data Water quality data was analyzed by reporting total load for individual storms sampling. The conventional roof in Florida was sampled less frequently than the green roof and for storm events where comparisons were made without conve ntional roof pairs, the average concentration of all conventional roof levels measured during two storms was used for comparison. Total load for each storm was calculated by multiplying the concentration of the parameter sampled by the volume of runoff generated in the time period repr esented by the sample. The start of the time period represented by a sampling time is the mid-point between the present sampling time and the previous sampling time; and the end of the ti me period represented is the mid-point between the current sampling time and the next sampling time (Table 4-5). The amount of runoff in cm generated during the time period was multip lied by the surface area of the roof (214.5 m2 for the CRP roof in Florida and 48.8 m2 for the YSC green roof in Virgin ia) to yield liter s of runoff for each time interval. For example, for the rain event sampled on 6/23/08 in FL, time intervals and water yield for each interval are shown in Table 4-5. The volume of runoff was consequently multipli ed by the concentration of each parameter to give the load for each time interval and then summed to yield the total load for each parameter for that rain event being analyzed. For example, for the rain event on 6/23/08, Table 4-6 shows how concentrations for time interval and summed for total loads per parameter for the rain event. 174

PAGE 175

Table 4-5. Example of time interval represented by a sampling time and amount of runoff represented by the water sample during a storm (FL 6/23/08). Time Interval represented by sample Runoff (cm) for the time interval Vol (L) Sampling Time 11:10 AM (Start of Runoff) 10:45am 11:15am 0 0 11:20 AM 11:15am -11:31am 0.010 24 11:43 AM 11:31am -12:15pm 0.204 493 12:46 PM 12:15pm 2:20pm 0.110 266 4:00 PM 2:20pm 6:00 pm (End of runoff) 0.037 89 Total Volume 872 Table 4-6. Example of calculati on of load by time interval ba sed on concentrations in grab samples taken at various time intervals dur ing a rain event on 6/23/08. This method was applied to all storms in both Florida and Virginia. Concentration Load Time NOx-N (mg L-1) NH3-N (mg L-1) TSS (mg L-1) SRP (mg L-1) Vol (L) NOx-N (mg) NH3-N (mg) TSS (mg) SRP (mg) 11:10 AM 0 0.0 0.0 0 0 11:20 AM 0.04 0.04 2.3 0.85 24 1.0 0.9 56 21 11:43 AM 0.03 0.05 2.6 0.59 493 15.2 22.2 1281 291 12:46 PM 0.03 0.05 2.2 0.84 266 8.6 12.0 585 224 4:00 PM 0.03 0.06 2.5 0.88 89 2.5 5.5 223 79 Load 27 41 2145 614 For the determination of load from a conven tional roof, the concen tration in the grab sample taken from the conventional roof was multiplied by the volume of rain (calculated as depth of rain that fell during the time interval multiplied by the contributing roof area), Table 46. The area of the conventional r oof contributing to the runoff sampled in Florida, was 148.5 m2. In the case of the YSC roof in Virginia, the conventional roof had an area of 48.8 m2. Results and Discussion Hydrology ResultsCharles R. Perry Green Roof in Gainesville, FL Cumulative hydrographs of green roof runoff in response to irrigation and rainfall are shown in approximately three month intervals re presenting the various se asons. Individual storm hydrographs, are only shown for the water quality sampling rain even ts to highlight the response 175

PAGE 176

of the green roof to an individua l rain event within a certain s eason. Comparisons to Virginia are discussed later in the Hydrology secti on on the YSC green roof in Virginia. Figure 4-8. Charles R. Perry Construction Ya rd Green Roof on June 11, 2007, 12 weeks after establishment.5 m2 (2600 sq. ft.), with 15 cm (6 inches) of Hydrotech LiteTop growing medium. Cumulative hydrographsCharles R. Perry green roof in Gainesville, Florida Cumulative hydrographs were created for thr ee time periods over th e entire study period for the Charles R. Perry Construction Ya rd: Summer-Early Fall 2007, Winter 2007-2008 and Spring 2008. The cumulative hydrograph and character izations of the rain events and runoff for the Summer-Early Fall 2007 are shown below in Figure 4-9 and Tables 4-7 and 4-8. The cumulative hydrographs and related summary ta bles for Winter 2007-2008 and Spring 2008 are presented in Appendices O and P. Summer-Early Fall 2007: The CRP roof was establishe d in April 2007 and hydrological monitoring of the runoff began on July 7, 2007. Th e months of July, August, September and October, were considered to be representative of the wet season, having a mean monthly rainfall of 148 mm. The cumulative hydrograph fo r the time period between July 7 and October 176

PAGE 177

19, 2007 in Figure 4-9 shows the water added to the system by both rain and irrigation, and the response of the green roof in terms of runoff. The hyetograph shows the individual rain events that occurred during the time pe riod, which allows the viewer to differentiate when the green roof runoff is increasing due to rain rather than irrigation. Thir ty-six storms from this time period, ranging in size from 0.08 7.34 cm, and 32 irrigation events (0.83 cm each), were analyzed individually to: 1) dete rmine the green roofs affect on peak rainfall intensity reduction and 2) determine the mean increase in lag time to peak rainfall due to the presence of the green roof, and 3) characterize the role of the green roof in creating a de lay in the start of runoff, as well as, extending the duration of runo ff past the time of storm cessation. The characteristics of the 36 storms analyzed are shown in Table 4-7 and the results of the analysis for % decrease in peak intensity, increase in lag to peak, delay to start of runoff and extension of runoff duration past the end of storm event are shown in Table 4-8. Overall, the green roof retained 41% of the rainfall that occurred in the we t season. The median sized storm for this season had a duration of 1 hour and depth of 0.4 cm. On average, the green roof lowered the peak intensity of runoff due to rainfall by 71%. The range of de lay in the start of runoff after the start of the storm was 0 hr to 3hr 30min, w ith a median value of 10 minutes. Runoff after the storm ended was extended by a mean of 6 hr 40 minutes. Table 4-7. Characterization of 36 rain events captured in the wet season between July 7, 2007 to October 23, 2007, in terms of rain duration, rain amount (cm), green roof runoff duration, % volume storm water retention. Type Time since previous rain (hr) Rain Duration (hr) Rain depth (cm) Runoff Duration (hr) Runoff depth (cm) Rainfall retention (%) Total 45.2 26.7 41 Mean 17.0 1 hr 50 min 1.2 7 hr 50 min 0.7 52 Stdev 12.9 2 hr 1.6 4 hr 20 min 1.2 40 Median 14.3 1 hr 0.4 7 hr 30 min 0.2 50 Min 0.7 10 min 0.1 5 min 0.0 0 Max 48.3 6 hr 30 min 7.3 17 hr 4.2 100 177

PAGE 178

Figure 4-9. Cumulative green roof runoff hydrograph compared to cu mulative rain and irrigation for wet season, July 7, 2007 to Octobe r 23, 2007. Hyetograph, shown in blue to distinguish when the cumulative runoff is running off due to irrigation versus rain. Table 4-8. Results of analysis of 36 rain events in the wet season between July 7, 2007 and October 23, 2007 for delay in start and extens ion of end of runoff, the maximum rain intensity, maximum runoff intensity (Max RO) and the percent decrease in peak intensity. Type Delay in Start RO (hr) Extension of RO (hr) Max Rain Intensity (cm/hr) Max RO (cm/hr) Reduction of Peak (%) Mean 40 min 6 hr 40 min 3.3 1.1 71 Stdev 50 min 3 hr 20 min 3.8 1.6 27 Median 10 min 6 hr 40 min 1.7 0.3 79 Min 0 min 30 min 0.3 0.0 7 Max 3 hr 30 min 14 hr 20 min 15.8 5.1 100 Winter 2007-2008 : The winter (dry) period began in late December and ended in late March with relatively low rain fall (82 mm per m onth), as compared to the wet season. Rainfall for this time period was characterized by medium-duration low-intensity storms (Appendix Q) emanating from storms blown across the conten t by the westerlies. The median duration and 178

PAGE 179

volume for rain events measured during the 2007-08 winter season was 1 hr 40 min and 0.5 cm depth (Table Q-2 in Appendix Q), slightly longer than the wet season storms. During this season, the roof was irrigated 2 times a week (Monday and Thursday mornings at 5 am for 34 minutes) for a total of 2.4 cm (0.9 inches) per week. In the cumulative hydrograph for this period, shown in Figure Q-1 in Appendix Q, the overall storm wa ter retention for the winter period (dry season) was 34%, which was slightly lower than the retention during the w et season, which was unexpected. Relationship of rain and irrigation to runo ff within the dry seas on (Winter 2007-08): Two cumulative hydrographs are also shown for two s horter periods of time within the dry season of 2007 2008, to show more closely how the roof responds to irrigation versus rainfall. The response of the green roof to irrigation for the weeks of January 6 15, 2008 (Figure 4-10) and January 31 February 12, 2008 (Figure 411) were chosen as demonstrations. The total irrigation and rainfall for the weeks of 1/7 to 1/14/08 and 2/7 to 2/14/08 was 9.14 cm (3.6) or 20200 L (5332 gallons) of water, and of this, 44% was retained by the green roof. The water that flowed off the roof as runoff was de tained in the cisterns without overflow, so all the rainwater for the runoff for this time period wa s kept out of the UF stormwater system. The relationships between storm intensity (cm/hr) and retention, total irriga tion + rain fall and intensity and duration vers us intensity are weak R2 < 0.05. The strongest relationship between retention and another parameter fo r these two weeks is with hours since last rain or irrigation event with an R2 = 0.47. The hours since the last rain or irrigation event influences the amount of pore space that will be available fo r the volume of the subsequent rain event or irrigation event to be stored. As the moisture condition at the time of the subsequent event depends on solar radiation and windspeed between the rain events, as this controls ET and 179

PAGE 180

indirectly the amount of pore space that will be available at the subsequent rain event. Figure 4-10. Hydrologic response of green roof to individual ir rigation and rain events on a weeklong time scale in the dry peri od/winter months, 1/6/08 to 1/14/08. Figure 4-11. Hydrologic response to green roof to irrigation vers us rain events in the dry period/ winter months, 2/6/08 to 2/14/08. 180

PAGE 181

Table 4-9. Relationship between intensity (cm hr-1), duration (hr), total volume of rainfall or irrigation versus green roof runoff and rete ntion (%), for individual events shown in Figures 4-10 and 4-11, storms in the dryperiod/winter. Intensity (cm/hr) Duration (hr) Total Rainfall/ Irrigation (cm) Total Vol (L) Response Duration (hr) Rainfall/ irrigation retained (L) Rainfall/ irrigation retained (%) Irrig 1/7/08 2.54 0.6 1.5 3066 12 1499 49% Irrig 1/10/08 2.54 0.6 1.5 3066 13 1158 38% Rain 1/12/08 0.18 3.0 0.5 1351 >8 602 45% Rain 1/13/08 0.33 3.0 1.0 1960 17 837 43% Irrig 2/7/08 2.54 0.6 1.3 3066 6 1548 50% Rain 2/7/08 0.15 3.0 0.6 1044 5 462 44% Irrig 2/11/08 2.54 0.6 1.3 3066 6 1669 55% Rain 2/12/08 0.14 9.0 1.5 3558 30 1052 30% Total 20177 8826 44% An irrigation event and a storm shown in th e cumulative hydrograph in Figure 4-11, that occurred on 2/7/08 are broken down into individu al storm hydrographs (non-cumulative) in Figure 4-12. The peak flow from irrigation on 2/7/08 was reduced by 67% (from 1.27 cm hr-1 to 0.42 cm hr-1), and the lag time was increased by one hour and the duration of return flow was extended by 6 hours. In the case of the small (3 hour, 0.5 cm) rain event on 2/7/08, the peak flow was also reduced by 74%, from a maximum of 0.35 cm h-1 to 0.09 cm h-1, there was essentially no lag time as the roof had just been irrigate d and was finishing running off when this storm started and the extension of r unoff period was extended by 14 hours. Figure 4-12. Hydrologic response of green roof to irrigation ( 1.27 cm) and a small precipitation event (0.5 cm) with wet anteceden t moisture conditions on 2/7/08. 181

PAGE 182

Validity of Excel spreadsheet water balance model and/or green roof bins as a model for calculating green roof runoff (Winter 2007-2008): As was expected for a green roof with wet antecedent moisture conditions, runoff began to occur much sooner than with dry antecedent moisture conditions. Most hydrologic models in existence (Cite R oofscapes, Villareal UH model) allocate waterIN to recharging soil moisture first and then water surplus to runoff/leachate, however in the case of the Charle s R. Perry Construction yard, saturation did not occur before roof runoff began. In a Water Balance approach, using Excel, green roof runoff was modeled by having rainfall satisfy soil moisture first, then having surplus available for runoff; and initial moisture content of the soil at the time of the rain was calculated based on pr evious storm size and duration since previous storm and residual water content of the so il, using this approach, runoff was underestimated for all storms. I found that in actuality, unsaturated subsurface flow occurred from the green roof well before the growing me dium was saturated. Runoff was occurring before the growing medium was saturate d, most likely due to flow thr ough preferential flow paths as described by Brady and Weil (2005) Water naturally flows severa l orders of magnitude higher through preferential flow paths such as burrows worm holes and in channels left by decaying debris or old roots that have rotted away. In th e case of this green roof, the size of the pores may not fine enough to hold the moisture, so wate r flows through by gravity drainage before the medium approaches saturation. When comparing water retention of the Char les R. Perry Construction Yard green roof during the 12 hour period after irrigation with va lues measured during the lysimeter experiment on 15 cm of Hydrotech growing medium duri ng the same season (dry/winter season), the 182

PAGE 183

retention curves are similar, implying that the measured water retention from the container study does reflect the behavior of water in the act ual Charles R. Perry green roof in situ. The water retention measured in the bin study, for a bin filled with 15 cm of Hydrotech growing medium and planted with perennials, measured in Week 12 (December) after planting, matches up with the water retention curves measur ed in situ for the CRP green roof in January. Figure 4-13. Comparison of % water retained directly after 1.2 cm of irrigation over 12 hours in the controlled bin study (pink line)--perennial plants in Hydrotech medium, Week 12/ (early Winter)with reten tion after irrigation in situ on the Charles R. Perry Construction Yard green roof (blu e line) during the winter season 2007. Spring 2008: The median rainfall and irrigation event during the spring season was a 25 min event with 0.87 cm of water depth (Table in Appendix R), while the median rain event without including irrigation, was 0.42 cm of rain. Median retention for events in this period was 94% and there was a median delay time in the st art of runoff and reductio n of peak flow of 35 minutes and 94% (Appendix R). The fact that th e water was being received from irrigation may affect the retention either due to lower anteced ent moisture conditions of the growing medium, higher evapotranspiration rates du e to lower ambient relative humi dity at the time of watering. 183

PAGE 184

Summer 2008: The overall retention for the 2008 we t season (June through the beginning of September) was 43%, which was similar to the overall retention reported for the wet season of 2007 (of 41%). A total of 38.7 cm of rain fell during this time pe riod and 22.1 cm of this rain entered the cisterns as runo ff (Figure in Appendix S). The summer period of 2008 was characterized by 16 rainfall events ranging from 0.0254 cm to 9.09 cm and water retention by the roof ranging from 100% to 17%, respectively. The mean and median rain events were 2 hr, 1.2 cm and 35 min, 0.58 cm, respectively. On average, 53% of rainfall was retained, irrespective of st orm size (Table S-1 in Appendix S). Cumulatively for the season, 25% of rain was retained (Table S-2 in Appendix S). Hydrology ResultsYorktowne Square Condomin ium (YSC) Green Roof in Merrifield, VA The cumulative hydrographs of runoff, as well as of the analyses of 82 individual storms, are shown in groupings of season, similar to the Florida data: Summer (late June, July, Aug, early Sept), Fall (late Sept, Oct, Nov, early Dec) Winter (late Dec, Ja n, Feb, early Mar), Spring (late Mar, April, May, early June) for each year, followed by a comparison of the retention of the green roof for Virginia (YSC) w ith the Florida green roof (CRP). The cumulative hydrograph for Summer-Early Fall 2006 is shown below, along with summary tables characterizing rain events and green roof runoff; cumula tive hydrographs for subseque nt time periods are in the appendices. SummerEarly Fall 2006 : The 23 storms sampled in Virginia in Summer/Early Fall 2006 (between July 12 and October 23, 2006) had a median duration of 3.8 hours and the median amount of rain was 0.74 cm (Table 4-10). The maximum intensity of the rain storms for this period were 4.87 cm hr-1 in Virginia, which is 3 times less than the maximum intensity measured in Florida (15.8 cm hr-1) for the same time period (July to October ). Median duration 184

PAGE 185

was 1 hour for Florida storms for the same seas on, meaning Florida storms were also on average three times shorter than the storms measured in Virginia. Total rainfall for the same period (July 7Oct 23) in Gainesville, FL was twice as mu ch as in Virginia, 45.2 cm of rain (with an additional 23 cm of irrigation), while total rain fall in Merrifield, VA was 24 cm (Figure 4-14). Overall retention, based on cumulative hydrographs for the same time period for Florida and Virginia were 50% and 64% respectively (Figures 4-9 and 4-14). Differences in retention may be due to the storms being shorter and more intense, with maximum intensities being three times higher in Florida than in Virginia. Additiona lly, the age of the growing medium may have influenced the formation of aggregates and addi tion of OM over time, wh ich could also have a bearing on retention. Finally, the presence of irrigation, increased the antecedent moisture conditions in the Florida green roof, which coul d also have increased the runoff during storms, by reducing available pore space for storage. Figure 4-14. Cumulative stormwat er runoff hydrograph for green roof and rainfall at YSC, Merrifield, VA green roof, Summer/Ea rly Fall 2006 (7/12/2006-10/23/2006). 185

PAGE 186

Table 4-10. Characterization of 23 rain events captured be tween July 12, 2006 and October 23, 2006 (Summer/Early Fall) in Merrifield, Virginia, for rain duration, rain amount (cm), maximum intensity, green roof runo ff duration, and maximum runoff intensity. Type Duration (hr) Rain (cm) Max int (cm hr-1) Duration RO (hr) Amt RO(cm) Max RO (cm hr-1) Total 118.1 19.6 79.9 9.88 Mean 6.22 1.31 1.12 8.88 0.52 0.17 Stdev 8.15 1.37 1.14 5.82 0.94 0.38 Median 3.58 0.74 1.21 8.38 0.00 0.00 Min 0.05 0.10 0.00 0.60 0.00 0.00 Max 27.5 4.72 4.87 21.2 3.48 2.08 Table 4-11. Mean, median and range of stormwater retention (%), delay in start and extension in end of runoff, reduction in peak intensity for 23 rain events between July 12, 2006 and October 23, 2006. Type %Retention Delay in Start RO (hr) Extension in end RO (hr) Increase in Lag to Peak (hr) % Reduction of Peak Mean 85% 3.19 2.71 1.03 92% Stdev 19% 1.98 2.22 0.95 13% Median 99% 3.30 1.60 0.63 100% Min 39% 0.15 0.90 0.25 57% Max 100% 6.50 7.30 2.80 100% Late Fall/Winter 2006: The late fall/winter time period consisted of 36 storms captured between November and March 2006-2007 (11/1/06 and 3/8/07) from the YSC green roof. The total amount of rainfall for this period was 31 cm and of that approximately 70% was retained by the green roof for this period of time. This rete ntion rate for Virginia is comparable to values reported for North Carolina and Pennsylvania (Moran 2005 and Berghage et al. 2009), who report 63% overall retention for North Carolina and greater than 50% for Pennsylvania. Rain storms for this season were characterized as having a median length of 5 hrs and a median amount of rain of 0.33 cm. The median duration a nd volume for rain events measured during the 2007-08 winter season in Florida (Dec 20Mar 18) was 1 hr 40 min and 0.6 cm depth (Table T-1 in Appendix T), which indicates that on average storms in Virginia were approximately three times longer with half as much volume than Florid a storms. The lower intensity of the storms in Virginia may explain why a larger proportion of st orms were fully detained in Virginia than Florida (44% versus 28% of the storms). 186

PAGE 187

Summer 2007 : Summer 2007 includes 7 storms captured in June and 5 storms captured in August 2007. The summer 2007 period was a relatively dry summer with 94 mm of rain in June, 82 mm of rainfall in July and 81 mm in August wh ich is 4%, 25% and 29% below the historical normal rainfall for those months, respectively. On average there were 3.5 dry days in a row before a rainfall with an average of 2 wet days in a row in June and August. Overall retention for June and August was 76%. The retention of rain fall per storm ranged from 45% to 100% with a median of 94% retention per storm. During a re latively dry period in Fl orida (April-June, 2008), where only 4 rain events occurred in 2 months, the median rete ntion was 94% (despite regular irrigation during that time period ) (Figure U-1 in Appendix U). Hi gher ET rates during the warm dry part of the year, may account for greater wate r uptake by the growing medium and plants and lower antecedent moisture conditions in the growing medium at the time of rainfall, which may explain the higher retention rates during th ese time periods in both locations. Spring 2008: Spring 2008 was represented by 10 stor ms collected between April and May (note that corruption of pressure transducer files caused a loss of data prior to and post these dates, therefore spring data were limited to th ese 10 storms). Rain events during this period ranged from 0.05 cm to 9.75 cm, with the median storm size for this time period being 0.47 cm over 5.8 hours. Rainfall retention ranged from 41 % to 100% with a median retention of above 90% (Table V-1, Appendix V). The median pe ak reduction was similarly high for this time period (94%), Table V-2, which was comparable to the values of retention and peak reduction found the previous summer, June and August 2007 in Virginia, and similar to values found in Florida for the time period between April and June for this same year (2008), Table S-2. Late Summer 2008: Late summer 2008, August-Septembe r, was a relatively dry period (the summer 2008 was categorized as moderate drought) followed by two large storms a few 187

PAGE 188

weeks apart-the remnants of Tropical Storm Fay on August 28th and rain from Tropical Storm Hanna on September 6th. The latter resulted in th e cisterns overflowing. Overall retention from the first week of August to the middle of Tr opical Storm Hanna was 65% (Appendix W). During Tropical Storm Hanna 17.2 cm of rain fell over 18 hours, up to the first 6 hours 65% of the incoming rain thus far was detained, after 6 hours the cisterns overflowed and measurements became invalid. Differences in Water Retention among Seasons and by Rainstorm Size in Florida and VA Differences in mean water retention by rain event size and reducti on of peak runoff for Florida are shown in Tables 4-12 and 4-13. The 91 rain events were divided into terciles. Small rain events (< 0.254 cm, n =31) in Florida, regardless of season, had a mean retention of 0.79% 0.32%, which was significantly greater than retention for medium events (0.254 cm 1.00 cm, n =30) and large events (>1.0 cm, n = 31) at a p < 0.05 level using Tuke ys HSD (Table 4-12). There were no significant differe nces in retention between medi um and large rain events in Florida (43% and 26% retenti on, respectively). There were si gnificant differences in the reduction of the peak intensity among all size cl asses of rain events (Table 4-13), but no significant differences in extension of runoff past the end of a rain event (small = 7.8 hr, medium = 11 hr, large = 8.7 hr) in Florida. In Virginia, 82 storms were sampled between December 2004 and September 2008 and analyzed for % stormwater retention (Table 414) and % decrease in peak runoff (Table 4-15. Storms during the sampling period in Virginia ranged between 0.0508 cm and 17.47 cm; 66% of these storms were less than 1.27 cm, 87% were < 2.54 cm and 97% were less than 7.62 cm in size. In Florida, the 94 different storms ranged fr om 0.0254 cm to 9.09 cm in size between July 2007 and September 2008. The median sized storm in Florida was 0.44 cm and 8.75 hours long 188

PAGE 189

Table 4-12. Differences in mean retention by ra in event size for 91 storms between July 2007 and September 2008, in Gainesville, Florida. Rain event sizes not connected by the same letter are significantly different at the = 0.05 level. Rain Event Size Mean Retention SD Significant Difference N Small (<0.254 cm) 0.79 0.3 a 30 Medium (0.254 cm 1.00 cm) 0.43 0.4 b 31 Large (>1.00 cm) 0.26 0.2 b 30 Overall (all size classes) 0.49 0.4 91 Table 4-13. Differences in reduction in peak runo ff by rain event size for 91 storms between July 2007 and September 2008, in Gainesville, Flor ida. Rain event sizes not connected by the same letter are significantly different at the = 0.05 level. Rain Event Size Mean Peak Reduction SD Significant Difference N Small (<0.254 cm) 0.94 0.1 a 30 Medium (0.254 cm 1.00 cm) 0.79 0.2 b 31 Large (>1.00 cm) 0.60 0.3 c 30 Overall (all size classes) 0.78 0.2 91 Table 4-14. Differences in mean retention by ra in event size for 82 storms between December 2004 and September 2008 in Merrifield, Virgin ia. Rain event sizes not connected by the same letter are significantly different at the = 0.05 level. Rain Event Size Mean Retention SD Significant Difference N Small (<0.254 cm) 0.98 0.1 a 26 Medium (0.254 cm 1.3 cm) 0.84 0.1 b 28 Large (> 1.4 cm) 0.72 0.2 c 28 Overall (all size classes) 0.83 0.2 82 Table 4-15. Differences in reduction in peak r unoff by rain event size for 82 storms between December 2004 and September 2008, in Merrif ield, Virginia. Rain event sizes not connected by the same letter are significantly different at the = 0.05 level. Rain Event Size Mean Peak Reduction SD Significant Difference N Small (<0.254 cm) 1.0 0.1 a 26 Medium (0.254 cm 1.3 cm) 0.93 0.2 b 28 Large (> 1.4 cm) 0.79 0.3 c 28 Overall (all size classes) 0.2 82 while in Virginia, the median sized stor m was 0.58 cm and 9 hours long. The storm size distribution was fairly similar in the two regions, in Florida, 76% of these storms were less than 1.27 cm,, 87% were < 2.54 cm, and 99% were less than 7.62 cm in size. 189

PAGE 190

Refer to Appendix X for comparisons of stor m duration, storm volume, average intensity, maximum intensity, stormwater retention and maximum storm water runoff intensity between Virginia and Florida. Comparison of Hydrologic Dynamics of a Green Roof in FL and VA Despite the greater depth and pore space availabl e in the thicker Florid a green roof, storms of the similar volume delivered over a longer pe riod of time in Virginia, had higher rates of stormwater retention than in Florida. The reason for the higher retention rates in Virginia may be due to the fact that the storms sampled in Virginia were longer in duration than those in Floridafor example, the study period extending fr om July to October was averaged together for Virginia and storms lasted on average 9 hours (Table 4-10) while in Florida the storms from the same study period lasted 2 hours or less on average (Table 4-8), additionally Table V-1 shows that the median and mean duration values fo r Virginia for all storms combined were three times longer than those in Florida over the enti re sampling period. The peak intensity was threefold higher for the storms in Florida (for example, 3.3 cm hr-1 for storms between July and October 2007 on the CRP roof, versus 1.05 cm hr-1 for July to December 2007 in Virginia); the mean intensity of rain storms was also higher in Florida than Virginia (Table X-3, Appendix X). The variation in the dura tion of the storms and the intensity of the storms may explain why a roof with a thinner substrate (4.5 cm thick) could have greater retention than one that is 15 cm thick. The results of greater retention in the shallow Virginia roof we re originally counterintuitive, taking into account that the green roof in Virginia was a) sha llower and b) consisted of an equally porous medium as that in Florida, but with a larger por e size distribution and c) had less organic matter initially a nd d) contained succulent plants versus perennials. Explanations for the differences in water rete ntion between the two roofs may relate to: 1. Differences in the storm characteristics between the two regionswith Virginia having storms 190

PAGE 191

on average three times longer than those in Florid a, as well as a lower mean storm intensity and lower maximum storm intensity, in any given season, 2. Changes in the VA growing medium over time, increase in OM and plant matter; 3. Di fferences in the underlayment in Florida and Virginia (Floradrain 25 versus Hydrogel packs) and unexpected changes in drainage of the underlayment over time that could possibly impede drainage and cause a pe rched water table, 4. Presence of Irrigation in Florida which may affect antecedent moistu re conditions at the time of a rain event and available pore space. The underlayment of the green roof in Gaines ville, FL, consisted of Floradrain FD 25, a corrugated plastic layer with small cups that held 3 L m-2. In the case of Virginia, the underlayment consisted of superabsorbent hydrog el packs adhered to the surface of the APAO layer with the same re tention rating of 3 L m-2. However the mechanisms by which the two underlayments function are quite different, for example the superabsorbent hydrogel packs function by absorbing 500% its dry weight in wa ter and once saturated rills form between lines of gel packs to allow for drainage between the r ills. One possibility that could lead to greater retention in this system, would be if rills are blocked and a perche d water table were to form and more water were in turn detained and made available for ET over time. Water Quality ResultsCharles R. Perry Green Roof in Gainesville, Florida Concentrations and loads from individual storm events Nutrient and metals were sampled seven times from the Charles R. Perry (CRP) Green Roof; once, one week after plan ting in 2007 and then six times one year after establishment, during the summer of 2008, for nutrients (NO3, NH4, SRP, TP, and TSS) and metals (Fe, Cu, Cd, Zn, and Al). Sampling dates and size of storm events selected for water quality sampling are shown in Table Y-1, Appendix Y; characteristics of the green roof runoff for the same events are shown in Table Y-2. 191

PAGE 192

Grab samples were taken at various interv als during the storm to create storm chemographs, exact times of sampling and their relations hip to flow volumes are indicated by markers on the individual storm hydrographs for each storm shown in Figures Z-1 through Z-5 in Appendix Z. The individual storm hydrographs were analyzed for the effect of the green roof on hydrology as evaluated based on the reduction in ma ximum runoff rate, the increase in lag time to peak of concentration and the delay of the start of rain to st art of runoff and the increase in duration of return flow of the water temporarily detained in the green roof medium in Table Z-2. Concentrations by time interval for nutrients measured are shown in Appendix AA, while means and standard deviations for entire stor ms for Florida are shown in Table 4-16. Table 4-16. Summary of mean conc entrations of nutrients and TSS for the six storms sampled in 2008 from the green roof (GR) and conventiona l roof (CR) runoff in Gainesville, FL. Date NOx-N(mg L-1) NH3-N(mg L-1) TSS (mg L-1) SRP (mg L-1) Mean SD Mean SD Mean SD Mean SD 6/23/2008 GR 0.03 0.01 0.05 0.01 2.46 0.29 0.70 0.14 6/25/2008 GR 0.11 0.03 0.07 0.02 1.12 0.35 0.64 0.22 6/26/2008 GR 0.08 0.04 0.12 0.09 3.70 2.36 0.55 0.04 6/30/2008 GR 0.04 0.01 0.22 0.05 7.75 3.89 0.56 0.12 7/8/2008 GR 0.05 0.02 0.05 0.23 1.27 3.48 0.00 0.08 7/8/2008 CR 0.07 0.05 1.29 0.04 7.41 1.44 1.29 0.02 8/21/2008 GR 0.11 0.04 0.16 0.04 0.05 0.03 0.03 0.20 8/21/2008 CR 0.01 0.01 0.10 0.07 0.06 0.04 0.47 0.00 Table 4-17. Summary of mean conc entrations of metals for six storms sampled in 2008 from the CRP green roof and conventional roof in Gainesville, FL. Date Cu (mg/L) Fe (mg/L) Cd (mg/L) Zn (mg/L) Al (mg/L) Mean SD Mean SD Mean SD Mean SD Mean SD 6/23/2008 CR 0.001 0.005 0.205 0.048 0.15 0.07 0.457 0.057 6/25/2008 CR 0.001 0.005 0.402 0.143 0.000 0.000 0.055 0.03 0.423 0.046 6/26/2008 CR 0.176 0.06 0.2747 0.102 0.0001 0.00062 0.047 0.2687 0.24 6/30/2008 CR 0.012 0.013 0.444 0.033 0.0001 5.6E-05 0.25 0.05 0.6841 0.33 7/8/2008 GR 0.000 0.002 0.282 0.025 0.00 0.00022 0.055 0.03 0.623 0.27 7/8/2008 CR 0.012 0.02 0.003 0.007 0.0001 0.00149 0.001 0.003 0.250 0.11 8/21/2008 GR 0.00 0.004 0.34 0.059 0.00 0.00027 0.47 0.9 0.671 0.13 8/21/2008 CR 0.15 0.084 0.01 0.017 0.00 0.00051 0.03 0.037 0.28 0.031 192

PAGE 193

NO3 0.00 0.20 0.40 0.60 0.80 1.00 1.20 1.40 Storm 1 Storm 2 Storm 3 Storm 4 Storm 5 GR Storm 5 CR Storm 6 GR Storm 6 CRmg/L NH30.00 0.20 0.40 0.60 0.80 1.00 1.20 1.40 Storm 1 Storm 2 Storm 3 Storm 4 Storm 5 GR Storm 5 CR Storm 6 GR Storm 6 CRmg/L SRP0.00 0.20 0.40 0.60 0.80 1.00 1.20 1.40 1.60 1.80 Storm 1 Storm 2 Storm 3 Storm 4 Storm 5 GR Storm 5 CR Storm 6 GR Storm 6 CRmg/L TSS0.00 2.00 4.00 6.00 8.00 10.00 12.00 Storm 1 Storm 2 Storm 3 Storm 4 Storm 5 GR Storm 5 CR Storm 6 GR Storm 6 CRmg/L Figure 4-15. Mean nitrate, ammonium, SRP and TSS concentrations per storm event for green roof runoff (6 storms) and conventional roof runoff (2 storms) in summer 2008 from the CRP green roof in Gainesville, FL. 193

PAGE 194

Al 0.000 0.200 0.400 0.600 0.800 1.000 1.2006/2 3 /2008 6 /25 /20 08 6/ 26 /2008 6/30 /2 008 7/ 8/20 0 8 gr 7/8/2008 cr 8/ 2 1/2008 gr 8/21/2008 crStorm EventAl mg/L 0.000 0.050 0.100 0.150 0.200 0.250 0.3006 /2 3/20 0 8 6/25/2008 6/2 6 / 2 00 8 6 /3 0 / 2 00 8 7 /8 / 2 0 08 g r 7/8/2008 cr 8/21/2008 g r 8/21/2008 crStorm Event Cu mg/L Fe -0.100 0.000 0.100 0.200 0.300 0.400 0.500 0.6006/23/2008 6/25/2008 6/26/ 20 08 6 /30/ 200 8 7/8/2008 gr 7/8/2008 cr 8/21 / 2008 gr 8/21/2008 c rStorm EventFe mg/L Zn0.000 0.200 0.400 0.600 0.800 1.000 1.200 1.400 1.6006/ 2 3 /2 0 08 6/ 2 5 /2 0 08 6/ 2 6/ 2 008 6/30/2008 7/8/ 2 008 gr 7/ 8 /2 0 0 8 cr 8 / 21/2008 gr 8/21/2008 crStorm EventZn mg/LFigures 4-16. Metals concentrations (A) Aluminum, B) Iron, C) Copper and D) Zinc) in CRP green roof (gr) and conventional roof (cr) runoff in 2008 in Gainesville, FL. Table 4-18. Total load of nutrients measured in 6 storms 2008 from CRP green roof in Florida. Date Vol (L) NOx-N (mg) NH3-N (mg) TSS (mg) SRP (mg) Total 6/23 870 27 41 2150 610 Total 6/25 100 11 7 110 64 Total 6/26 215 17 25 800 120 Total 6/30 190 7 43 1500 110 Total CR 7/8 11470 890 870 23800 82 Total GR 7/8 11100 780 14270 82200 14300 Total CR 8/21 26500 3030 4180 1380 700 Total GR 8/21 15400 200 1510 960 7300 Discussion--CRP-FL Water Quality Data The values of SRP from the Charles R. Perry Construction Yard green roof runoff ranged from 0.25 to 1.44 mg L-1, with a median value of 0.72 mg L-1 (Figure 4-15). In comparison to other green roofs, such as in Sweden (Berndt sonn, et al. 2006), our median SRP level (0.72 mg 194

PAGE 195

L-1) was slightly lower than the mean values reported for a green roof above the Bicycle Parking in Lund (1.4 mg L-1 and 0.9 mg L-1), but greater than the mean SRP reported for the Canoe Club green roof in Malmo, Sweden (0.2 mg L-1). The SRP concentrations from our green roof runoff were lower than SRP levels from reclaimed water in Florida, that of C onservII in Orlando, which is pumped at a rate of 46 million liters per day directly through sandy so il to the superficial aquifer. The mean SRP value for ConservII is 1.67 mg L-1 (Moura, 2009), for all rain events except 7/8/08, the median SRP level in green roof runoff was half that (0.72 mg L-1). The implication is that reclaimed water sends 40 tons of P to the aquifer, while green roofs are sending loads on the order of grams to kg per year to surface water. The conventional roof runoff SRP levels (0 0.04 mg L-1) were lower than background levels of SRP found in groundwater in Central Florida (0.05 mg L-1). Conventional roof runoff of SRP ranges from 0.00 to 0.04 mg L-1 with a median value of 0.03 mg L-1. In North Carolina (Moran, 2005), both concentr ations and amounts of total nitrogen and total phosphorus increased from ra infall to green roof outflow and from the control roof outflow to green roof outflow. It was determined that the growing medium, composed of 15% compost, was leaching nitrogen and phosphorus into the green roof outflow. Water Quality ResultsYSC Green Roof in Merrifield, Virginia Stormwater runoff from five storms were sampled from both the conventional roof and the green roof at Yorktowne Square Condominiums (YSC) in Merrifield, Virginia between 2007 and 2008; 2 storms represented the spring time and 3 storms represented the summer time. The stormwater runoff was sampled between 3 and 5 times each storm. Concentrations of the nutrients/suspended and dissolved solid (NO3, OP, TP, TDS and TSS) and metals (Cu, Cd, Fe and Pb) sampled are shown by time increment by storm in Appendices BB. Means and standard 195

PAGE 196

deviations are shown in Figure 417 and the total load for each r oof is shown in Table 4-19 and 4-20. Nutrient wise, TP and OP concentrations were always higher in the green roof runoff than in the conventional roof runoff. In general, the TP consisted of 80-90% orthophosphorus (SRP) from all the green roof runoff, while in CR runoff it was only 75% made of OP. There was greater variability between storms for both nutrients and metals than within storms. First flush effects were noticed for some parameters, where first flush can be define d on a concentration or mass load basis. When defined on a mass basis, th e load of a nutrient or metal exiting the system is disproportional to the amount of flow occurring in the initial portion of the storm (initial portion can be defined as an i ndefinite portion of time closest to the start of the storm or predetermined volume, such as the first inch of runoff). pH 0 1 2 3 4 5 6 7 8 4/20/2008 GR 4/20/2008 CR 8/29/2008 GR 8/29/2008 CR 9/6/2008 GR 9/6/2008 CR Date/Roof Typep H N-NO3 (mg/L) 0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.46/ 3/20 0 7 GR 6/ 3/20 07 CR 4/ 2 0/20 08 GR 4 /20 / 2 0 08 CR 8/ 2 9/20 08 GR 8 /29 / 2 0 08 CR 9/ 6/20 0 8 GR 9/ 6/20 08 CRDate/ Roof TypeNO3 mg/ L TDS 0 50 100 150 200 2506 /3/2 0 07 GR 6/3/2007 CR 4/20/2008 GR 4/20/2008 CR 8/29/2008 GR 8/29/2008 CR 9 / 6 /2 0 0 8 G R 9 / 6/ 2 008 CRDate/ Roof TypeTDS (mg/L) TSS0 2 4 6 8 10 12 14 166/ 3/2 007 G R 6/ 3/2 007 CR 4/2 0 /200 8 G R 4 /20 /20 08 C R 8/ 2 9/20 0 8 GR 8/29 / 2008 CR 9 /6/2 00 8 GR 9/ 6/2 008 CRDate/ Roof TypeTSS (mg/L)Figure 4-17. Mean concentratioins and SD for nutrients and metals measured in YSC green roof (gr) and conventional (cr) runo ff in Merrifield, VA, 2007 and 2008. 196

PAGE 197

TP 0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.86 / 3/ 2 007 G R 6 /3/200 7 CR 4 /20 / 20 0 8 GR 4/2 0 /2 0 08 CR 8/29 / 2008 GR 8/2 9 / 20 08 CR 9/6/2008 G R 9 /6 / 20 08 CRDate/ Roof TypeTP (mg/L) Fe 0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4 0.45 0.56 / 3 / 200 7 GR 6/3 / 200 7 CR 4 / 2 0/ 20 08 G R 4/20/2008 CR 8/29/2008 G R 8 /29 /2 00 8 C R 9/6/ 20 08 G R 9 / 6/2008 C RDate/ Roof TypeFe (mg/L) Figure 4-17. Continued Table 4-19. Total load of nutrien ts measured in 4 storms in 2007-2008 from YSC green roof in Virginia. Date Vol (L) N-NO3 (mg) TP (mg) SRP (mg) TSS (mg) TDS (mg) 6/3/2007 GR 340 38 450 410 340 63500 6/3/2007 CR 560 < 56 110 80 560 13400 4/20/2008 GR 1200 120 820 750 5150 67200 4/20/2008 CR 2730 640 < 270 < 270 12300 26800 8/29/2008 GR 420 < 42 596 494 60700 8/29/2008 CR 840 140 < 84 < 14 11400 9/6/2008 GR 2970 < 297 1896 1655 8770 87000 9/6/2008 CR 8020 < 802 < 802 < 80 8020 37300 Table 4-20. Total load of metals measured in 4 storms in 2007-2008 from YSC green roof in Virginia. Date Vol (L) Cu (mg) Fe (mg) Pb (mg) Cd (mg) 6/3/07 GR 340 <34.4 38 0.0 6/3/07 CR 560 <56.1 117 4/20/2008 GR 1200 <247 <121 <2.8 4/20/2008 CR 2730 <273 <273 64.0 8/29/2008 GR 420 124 138 <2.0 0.232 8/29/2008 CR 840 <383 <144 <13.0 0.000 9/6/2008 GR 2970 <297 <377 <5.9 9/6/2008 CR 8020 <802 <802 58.8 Comparisons in Concentrations of Nutrients and Metals between roof types in FL and VA and between locations Nutrients: The mean total phosphorus concentrations from the YSC green roof runoff in Virginia were significantly higher than conven tional roof runoff concen trations at the p < 0.001 (ANOVA and pairwise t-te st, SAS 9.2). SRP concentrations we re also significantly greater in 197

PAGE 198

green roof runoff than conventional roof runo ff for both the CRP green roof in Florida and theYSC green roof in Virginia, at the p < 0.001. When comparing the co nventional roof runoff from the two locations, Florida and Virginia, there were no significant differences in OP concentrations, nor any significant differences in green roof runoff from the two locations (p = 0.34 and 0.36, respectively) (Table 4-21). The results of the ANOVA and pairwise t-tests for nitrate concentrations, show that the NO3 concentrations are significan tly lower in the green roof r unoff than the conventional roof runoff in both Florida and Virginia (Table 4-21 ). There are no significant differences between NO3-N levels in conventional roof runoff between Florida versus Virginia, nor for green roof runoff for Florida compared to Virginia (Table 4-21). Ammonium was m easured in Florida only, and there were no significant diffe rences in the concentrations found in green roof runoff versus conventional roof runoff (Table 4-21). There was a significant difference in pH between conventional roofs and green roofs; the green roof buffered the pH raising it from a mean value of 4.78 0.14 in conventional roof runoff to a mean pH of 7.31 0.14 in green roof runoff. This trend of buffering acidic rain was documented in Berghage, et al. (2007) They found that the growing media they tested in their field study woul d have at least 10 years of acid rain buffering capacity with no need of lime until 10 to 30 years (Berghage, et al. (2007). There were no significant di fferences in TSS between the green roof runoff and the conventional roof runoff in either Florida (p = 0.07) or Virginia (p = 0.66), nor any significant differences in TSS concentrations in green roof from Virginia comp ared to that of Florida, nor for conventional roof runoff compared by location. However, there were significantly higher concentrations of TDS measured in green roof runoff in Virginia (the only place this parameter was measured) at a p < 0.0001 level (Table 4-21). 198

PAGE 199

Metals : With respect to metals measured in the two sites, Aluminum concentrations were found to be significantly higher in green roof runoff in Florida, th an in conventional roof runoff (Table 4-21). This may relate to the constituents used in the roof, for example the drain cover and drain itself on the green roof is fabricated of Aluminum, wh ile the down pipes are cast iron. Possibly as a consequence of the newly installed iron downpipes, iron levels in the green roof runoff from the CRP roof were si gnificantly higher than those meas ured in the conventional roof collected from the red ceramic tile roof atop Newell Hall (Table 4-21). More investigation would be necessary to pinpoint th e source of the iron in the CRP green roof runoff, it is however apparent that the concentrations from the CRP roof are about 10-fold higher than Feconcentrations measured in the Newell Hall runoff and nearly 2 times more than concentrations of Fe found in the Virginia green roof runoff. Th e difference in Fe concen trations in Florida is significantly different than the Virginia green roof concentrati ons to the p < 0.0001 level (Table 4-21); while the Fe concentrations in conven tional roof runoff measured in Florida were significantly lower than the Virgin ia conventional roof runoff at a p = 0.0129 level (Table 4-21). No significant differences were found in between conventional roof runoff and green roof runoff for iron in Virginia (Table 4-21). Copper concentrations from Virginia were higher in both conventi onal and green roof runoff than in Florida (Table 4-21), and in ne ither location were the Cu levels significantly different in the green roof runoff as compared to conventional roof runoff. Again, the presence of metals is likely due to the degradation and tran sport of building material s. Lead (Pb) was only measured in Virginia and was found to be signifi cantly lower in the green roof runoff than the conventional roof runoff at a p = 0.0023 level. Cadmium was near 0 in all samples in both locations and not significantly different in green roof runoff as compared to conventional roof 199

PAGE 200

runoff in either location. Zinc was measured in both locations, though only 2 samples were taken in Virginia. In Florida, there were no signifi cant differences in concen trations found in green roof runoff as compared to conventional roof runoff. In summary, green roofs did mitigate for NO3-N in both locations, it increased pH levels relative to rainfall (measured in VA) and lead wa s significantly lower in green roof runoff (in Virginia) compared to conventi onal roof runoff. Metals beha ved differently by location, for example Aluminum and Iron were found to be significantly higher in green roof runoff in Florida, but not in Virginia, and concentrations of metals varied more between locations than did nutrients. In the case of TP, OP, and TDS; levels were significan tly higher in green roof runoff than conventional roof runoff in Florida and Vi rginia. TSS was surprisingly not significantly different in green roof runoff as compared to conventional roof runoff, however concentrations from Florida were significantly higher than those found in the green roof runoff in Virginia. The green roof does seem to have an impact on nutrient and metal concentrations and metal concentrations vary more by loca tion than other parameters. Metal concentrations in the CRP green roof runoff varied: Aluminum levels ranged from 0.1 mg L-1 to 0.8 mg L-1, Iron ranged from 0.15 mg L-1 to 0.40 mg L -1; Zinc ranged from 0.01 mg L-1 to 0.24 mg L-1 for all samples, except one durin g the storm event on 8/21/08. Iron was lower in conventional roof runoff (0 0.05 mg L-1) at p<0.05, using SAS 9.2, than in green roof runoff. There were no significant differences between conve ntional roof and green roof runoff for Fe The mean concentration of Pb in YSC conventional roof runoff (0.034 0.006 mg P L-1) across all samples was significantly higher than th e mean concentration of Pb in green roof runoff (0.002 0.006 mg Pb L-1) at the p <0.0023 level using an ANOVA and pooled t-test in 200

PAGE 201

SAS 9.2. Compared to lead levels measured on othe r roofs, these levels are comparable to Pb levels in green runoff from both green roofs sa mpled by Berndstonn et al. in Sweden and lower than conventional roof runoff in Paris measured by Gromaire-Mertz et al. in 1999 (Table 4-22). Table 4-21. Results of the ANOVAs pairwise t-tests for nutrients and TSS and TDS between conventional roof runoff and green roof runoff in Florida and Vi rginia; and pooled ttests between locations within c onventional or green roof type. Conventional Roof Green Roof Parameter Location (Mean S.E.) (n) (Mean S.E.) (n) Prob > F (units) 0.164 0.127 9 0.755 0.085 20 p = 0.0007 Florida 0.076 0.068 20 0.87 0.068 20 Virginia p < 0.001 OP (mg P L-1) FL vs VA CRFL vs CRVA, p = 0.34 GRFL vs GRVA, p=0.36 TP (mg P L-1) Virginia 0.098 0.06 20 1.12 0.062 20 p = 0.0007 pH Virginia 4.78 0.14 12 7.31 0.14 12 p < 0.0001 0.146 0.029 9 0.055 0.019 20 p = 0.014 Florida 0.122 0.018 20 0.05 0.02 14 p = 0.0154 NO3 (mg N L-1) Virginia FL vs VA CRFL vs CRVA, p = 0.57 GRFL vs GRVA, p = 0.80 NH3 (mg N L-1) Florida 0.276 0.12 9 0.276 0.08 20 p = 0.99 1.143 1.27 9 3.95 0.85 20 p = 0.078 Florida 2.96 0.96 14 2.35 1.00 13 p = 0.66 Virginia TSS (mg L-1) FL vs VA CRFL vs CRVA, p = 0.27 GRFL vs GRVA, p = 0.23 TDS (mg L-1) Virginia 13.47 11.8 17 117. 2 11.8 17 p < 0.0001 201

PAGE 202

202 Table 4-22. Comparative values of green roofs, conventional roofs and urban lawns from various locations. Collection Site Location pH TSS (mg L-1) TP (mg L-1) PO4-P (mg L-1) Tot-N (mg L-1) NO3-N (mg L-1) NH4-N (mg L-1) NH3-N (mg L-1) Source/Reference Green Roofs Japan 5-7.5 0.01-0.02 0.01 0.03-1.0 0.01-1.2 0.1-1.0 Berndtsson, J., Lars B. and K. Jinno. 2009. Green Roofs Sweden 6-6.2 0.04-0.32 0.02-0.27 2.1-2.7 0.1-1.05 0.1-1.1 Berndtsson, J., Lars B. and K. Jinno. 2009. Green Roof North Carolina, US 0.61-1.4 0.9-6.9 Moran, A. 2005 Controls:Rain North Carolina, US 0.05 0.05-2.3 Moran, A. 2005 Conventional Roof North Carolina, US 0.7-7.1 Moran, A. 2005 Green roof Gainesville, FL 0.06-9.50 0.59-1.46 0.49-1 .37 0.01-0.19 0.05-1.04 Current study (Lang 2009) Conventional Roof (Tile) Gainesvillle, FL 0.06-0.83 0.40-0.75 0.03 0.13-0.21 0.11-0.16 Current study (Lang 2009) Green Roof Merrifield, VA 1-5 0.68-1.43 0.63-1.19 <0.10 Current study (Lang 2009) Conventional Roof (Asphalt) Merrifield, VA 1-7.8 <0.1-0.2 <0.1-0.15 <0.1-0.23 Current study (Lang 2009) Urban Roofs Paris, France 29-33 Moilleron et al. (2002) Tile, Polyester, Gravel roof Switzerland 0.52, 1.4, 0.2 Zobrist et al. (2000) in Berndstonn et al. (2009) Green Roofs Berlin, Germany 7.3 (Rain 6.1) 0.3-2.3 Kohler and Schmidt (2003) Urban Roof France 29 (3-304) Gromaire-Mertz et al. 2005 Urban Roof China 121.3 0.71 11 Ren and Wang (2006) Controls:Rain China 6.49 0.06 3.99 Ren and Wang (2006) Road China 176-567 0.71 7.40-13.62 Ren and Wang (2006) Lawn China 549 0.74 6.8 Ren and Wang (2006) Roof runoff Connecticut,US 0.019 1.2 0.5 Dietz and Claussen (2005) Mixed natives Ft. Lauderdale, FL 0.44.12** 0.16+/0.01** Erickson et al. (2005) St. Augustine Grass Ft. Lauderdale, FL 0.050.01** 0.17+/0.01g/L** Erickson et al. (2005) *Range of median values are reported unless indicated otherwise by **

PAGE 203

Comparison of Areal Loading Rate s between Florida and Virginia The areal loading rates (Table 4-23) were determined by divi ding the total load for each storm by the surface area of the roof sampled, in the case of Florida, the roof was 214.5 m2 and in Virginia, the roof section sampled was only 48.8 m2, as a result, while Virginia often had lower concentrations, the maximum areal loading rates fr om Virginia often exceeded the loading rates from Florida for many constituents. Table 4-23. Areal loading rate (mg m-2) for nutrients sampled in gr een roof (gr) and conventional roof (cr) runoff from CRP roof in Gainesville, FL, 2008. Date L m-2 NOx-N (mg m-2) NH3-N (mg m-2) TSS (mg m-2) SRP (mg m-2) 6/23/08 gr 4.1 0.13 0.2 10.0 2.9 6/25/08 gr 0.5 0.05 0.0 0.5 0.3 6/26/08 gr 1.0 0.08 0.1 3.7 0.6 6/30/08 gr 0.9 0.03 0.2 7.0 0.5 7/8/08 cr 77 3.66 4 98 0.3 7/8/08 gr 51.6 3.6 66.4 382 66.7 8/21/08 cr 109 12 17 6 3 8/21/08 gr 71.8 1.0 7.0 4.5 33.9 Table 4-24. Areal loading rate (mg/m2) for nutrients sampled in green roof (gr) and conventional roof (cr) runoff from YSC r oofs in Merrifield, Virginia, 2007 and 2008. L m-2 NOx-N (mg m-2) TSS (mg m-2) TDS (mg m-2) TP (mg m-2) SRP(mg m-2) 4/11/07 gr 1.7 0.73 0.00 0 0.00 1.02 4/11/07 cr 2.1 0.92 0.00 0 0.00 0.06 6/3/07 gr 1.6 0.77 7.01 1296 9.09 8.32 6/3/07 cr 2.6 1.14 11.44 273 2.23 1.72 4/20/2008 gr 5.6 2.47 105 1372 16.8 15.33 4/20/2008 cr 12.7 13.09 252 548 5.57 5.57 8/29/2008 gr 1.9 0.85 -1240 12.2 10.08 8/29/2008 cr 3.9 2.84 -233 1.72 0.28 9/6/2008 gr 13.8 6.06 179 1777 38.7 33.78 Conclusions The first hypothesis tested, that the overall benefit for green r oofs for water retention, peak runoff attenuation and increases in lag time to peak runoff would be less in green roofs in Florida than in Virginia, was not reje cted, though the original reason given, that it would be due to 203

PAGE 204

204 higher peak precipitation rates a nd greater recurrence of convectiv e storms, was not isolated as the specific reason for the difference in retenti on between the sites. Stor m size distributions in the two extremes of the subtropical climate were fairly similar for the study periods. Infact, the median depth of rain for the 82 rain events sa mpled in Virginia (0.63 cm) was slightly higher than the median depth of rain for the 91 rain events sampled in Florida (0.44 cm); and the maximum rain depth was nearly twice as high in Virginia (17.5 cm) as Florida (9 cm), which was unexpected. Duration of storms in Virginia however, were on average at least 3 times longer than Florida storms. Explanations for the higher retention s een in the Virginia roof may relate to differences in the underlayment between the two roofs, the lower mean intensity of storms and greater duration of storms in Virginia differences in organic matter due to the age of the roof or relate to the effect of irrigation on antecedent moisture conditions. These specific topics need to be studied furt her in greater detail to isolate why the Virginia roof had higher retention than the Florida roof, when in fact it was shallower. The green roofs in both Florida and Virginia did detain signi ficant portions of stormwater and significantly increase the lag time to peak and extend the runoff period from the roof. In Florida, small rain events (< 0.254 cm, n =31), had a significantly highe r mean retention (79%) than either medium rain events (0.254 cm 0.83 cm) or large events (>0.83 cm). There were no significant differences in retenti on between medium (43% retention) and large (26% retention) rain events in Florida, and no si gnificant differences in extension of runoff past the end of a rain event (s = 7.8 hr, m = 11 hr, l = 8.7 hr) in Florid a. In Virginia, the mean retention by size class differed significantly between small, medium an d large rain events, w ith respective retention rates of (98%, 84% and 72%). The retention rate s of the large events (72%) compared well to studies in neighboring and near by states (North Carolina and Pennsylvania) (Moran 2005 and DeNardo 2003), though the small and medium events had retention rates that were unusually

PAGE 205

high in comparison. In both Florida and Virgin ia, there were significant differences in the reduction of the peak intensity among all the diffe rent size classes of the rain events, 94% for small events, 79% for medium events and 60% for large events for Florida, and 100%, 93% and 79% for small, medium and large events, respectively in Virginia. The extension in end of runoff was not as grea t in Virginia as provided by the 15 cm thick green roof in Floridathe median extens ion time varied between 0.68 hours and 3.9 hours through the seasons for Virginia. In Florida, th e median extension in runoff varied between 2.5 hours and 14 hours by season. These differences are attributed to the th ickness of the Florida green roof and the pore-size dist ribution of the growing medium in the green roof, as well as differences in the underlayment (for example, Hydrogel packs do not re-release water until ET occurs to remove the water from the gel packs.) The major objective for the hydrological analys es was to determine if the green roofs behaved similarly hydrologically in both extremes of the subtropical cl imate. This study found that for the two roofs studied, they did behave in the same manner, though at different intensities for water retention, peak reduction and increase in lag time, with water retention and peak reduction being greater in Virginia and with a gr eater difference in the effect on extension of runoff and increase in lag to peak in Florida than Virginia. In terms of water quality analys es, the hypothesis tested was that the green roofs influence on water quality (nutrients, metals TSS and TDS) would be similar in both Virginia and Florida and that they would act as sinks for nitr ogen and sources for phosphorus and suspended sediment; and would act similarly for metals in both regions (whether as a source or sink). It was found that the green roofs did in fact act as sources for phosphorus (both TP and OP), which was similar to finding in other studies from nearby st ates and other regions of the world. The green roofs did also act as a sink for nitrogen as nitrate as hypothesized, however no differences in 205

PAGE 206

concentrations were noticed between th e green roofs and control roofs for NH3-N. The green roofs did buffer pH, which was also similar to findings in other studies (Berghage, 2005, Kohler and Schmidt, 2003; DeNardo 2003). Metals behaved differently by location, for ex ample aluminum and iron were found to be significantly higher in green roof runoff in Florida, but not in Virginia; and lead concentrations were also significantly lower in green roof runoff than conventional roof runoff in Virginia. Concentrations of metals varied more between locations than did nutrients. For both metals and nutrients, variability between st orms was greater than within storms, in both Virginia and Florida. Total Dissolved Solids (TDS) were signifi cantly higher in green roof runoff as compared to conventional roof runoff, however total susp ended solids (TSS) surprisingly was not found to be significantly different in green roof runoff as compared to conventional roof runoff. However, concentrations of TSS from the Florida green ro of were significantly high er than those found in the green roof runoff in Virginia. The load reductions attributable to the green roof, as compared to a conventional roof of the same size, varied with the size and intensity of the storms, by and large, the green roof seemed to lower nitrate and ammonium concentrations from 12-90% depending on storm size and intensity, but the presence of the green roof increased TSS, TP and SRP loads in most cases. The SRP load was often increased over 10-fold. It is recommended that the cost-savings implied by a reduced peak outflow rate and reduction in storm water volume and possible reduction in nitrate and ammonium loads be we ighed against the externalities related to increased SRP loads leaving a green roof in a future study. The imp lied cost of treating runoff for SRP or TP should be taken into account when considering implementi ng green roofs in South Fl orida, especially in new developments in close proximity to th e Everglades or surrounding areas, since the 206

PAGE 207

Everglades ecosystem is a fragile, naturally low P ecosystem, and receiving waterbodies surrounding the area have a P limit of 10 ppb, which all green roof samples exceeded. Furthermore, it is highly recommended that cist erns be used in conjunction with the green roof, as this will keep the green roof runoff out of the stormwater system and allow nutrients to be recycled to the roof and re-utilized by the pl ants. Additionally, the cisterns potentially can be used to gravity irrigate the landscape at the surface level (as is done in VA), implying a costsavings related to not having to rely on potable or even reclaimed water. Based on the current findings, it may be worthwhile to study the pos sible benefit of green roofs in taking up atmospheric nitrogen, the assumed source of N from the conventional roof, whether deposited by wet or dry deposition. Additionally, investigating the feasibility and usefulness of treating grey water on a green roof with specific plants a nd re-using the water on-site, is another possible study emanating from this research. Finally, the last recommendation is to use a low P growing medium for green roofs and to make guidelines for the State of Florida now, so th at larger green roof companies can tailor their mixes to the specific needs of our state, especially with regard to the issu es we have with P in streams and fragile ecosystems and water usage issues. Guidelines should specify that green roofs in certain areas, such as South Florida, mu st be accompanied by a cistern for on-site re-use; and reclaimed water should be used as make-up wa ter for times when the cistern is deficient in providing irrigation water. Otherw ise green roofs, if mismanaged, for example high nutrient content growing media are used, with over-irriga tion using potable water and ornamental plants, could lead to a potential ecological disaster in Fl orida. If properly managed, green roofs have the potential of even treating grey water for N, and having that water be reused on site for irrigation of the ground landscape. In no cas e is it recommendable to have green roof runoff directly conveyed to a natural waterbody with concerns about P limitations. 207

PAGE 208

CHAPTER 5 CONCLUSIONS Green roofs have the potential to reduce stor m water quantity in Florida and Virginia, however water retention and nutrient leaching de pend largely on the growing media. Choice of plant material also affected performance of green roofs in terms of retention and nutrient leaching. Perennials performed well in water retention and evapotranspiration. Succulents had the best survivability through the subtropical climate (drought and frost), though had the least amount of nutrient uptake and wate r retention of the th ree plant types. Findings suggest that a growing media rating systembased on initial nutrient content and ability to support plants with little/no irrigationshould be developed for Florid a green roofs, in order to avoid undesirable effects associated with runoff. Future research includes 12 green roof plat forms to measure plant performance and nutrient leaching under drought conditions and elevated N-deposition in the optimal soil media-plant combination determin ed from the final results of this study. Implications of Water Retention for Florida Green Roofs The results of the hydrological dynamics study re vealed that when de signing a green roof for stormwater reduction control in Florida, the most important design factor influencing retention is the initial selection of growing me dium type. When selecting from several growing media types, if interested in the greatest wa ter retention for stormwater control, it is recommended to select the medium with the finest pore size-distr ibution of the available growing media. Also the greater the abunda nce of certain constituents such as organic matter, perlite and vermiculite also play a large role in increasing re tention. Total porosity is actually less influential in determining overall retention of the medium. In the bin study retention rate s were comparable to those found in a studies in both temperate climates (DeNardo 2005) and in the subtropics eg. Texas (Timm and Rasmussen 208

PAGE 209

2008). The retention rates ranged from a low of 24% for Building Logics medium with no vegetation to a maximum of 83% for UCF grow ing medium planted wi th perennial plants. Within each growing media typeretention ra nged from 24-53% BL, 30-72% H and 31.2% -85% depending on the plant types present. Plants were found to be secondary to growing media characteristics in influencing water retention rate s in the sub-tropics, in the plant-growing media used in this study. The influence of the plants increased retention by 11-22% depending on the growing medium. For growing media with large por es and large particles, the positive influence of plants increasing water retenti on directly after a storm is less. For example in Building Logics growing medium, water uptake by plant type was increased only by 11-14%, but in Hydrotech and UCF growing media, water up take often increased by 15-22% due to the presence of plants. Evapotranspiration rates ranged from 1.5 mm day-1 for unvegetated Building Logics medium to a maximum of 3.6 mm day-1 for UCF bins planted with perennials. There was a significant difference in the amount of water rele ased by transpiration ve rsus that of just evaporation within a growing media. Having plants present in th e northern Florida climate green roof increased the amount of void pores availabl e for water storage by two-fold. It was however noticed that the plant type most capable of transpiring the greate st amount of water, perennials, also suffered the most from heat stress during droughts and were consequently more susceptible to plant diseases and succumbed more easily to death than succulents. The succulents, while surviving the largest range of climatic variation during the study, did little to contribute to increased retention of water directly after storms Bins containing succulent s did have the highest water content and healthiest overall plants duri ng the entire study, but contributed the least to nutrient retention or water retenti on. It is important to note that once a plant senesces, then which plant is present is irrelevant. For this reason the re sults of this study i ndicate that certain 209

PAGE 210

succulents do infact work well in the sub-tropics despite anecdotal information contrary to this finding. In terms of the water release curves, water co ntent was generally the lowest and retention directly after irrigating was the lowest in Building Logics growing medium and highest in UCFs Black and Gold mix. The ability of the perennial s to evapotranspire out more water over the same period of time as the other plants resulted in a larger differential between the water content in those bins containing perenni als and those containing no vegeta tion or succulents. Essentially, the water content in those planted with perennia ls and runners became successively less, to the point that the plants themselves which were resp onsible for transpiring the water out and creating the decreased water content in the bin, were in gr eater peril of suffering from plant diseases as a result; a case of a negati ve feed-back loop. It is recommended th at for a roof with no irrigation in the north central Florida climate succulents be used, however in a situation where irrigation can be used, then it is recommended that a mix of succu lents, perennials and runners be used together as each plant structure has a different adaptation which may make them compatible to be planted together. Furthermore, it is suggested that grasse s be tested in vegetate d rooftops in Florida, especially those turf grasses such as Bahia and tropical crab grass, which colonized abandoned bins on their own during a drought after the original species planted in the bins senesced from lack of water and heat. These grasses can thrive on low water inputs and do not require mowing. It was hypothesized that the pe rennial peanuts performed poorly in the bin study and only mediocrely on the Charles R. Perry Construction Yard Roof because of issues with inoculation with Rhizobium In the bin study, all excess soil was shaken off of the plants, while in the CRPCY study, plants were planted with soil around the roots, whic h may have allowed for more inoculation of the soil on the CRPCY roof; regardless Phyla nodi flora outperformed Rhizobium both in the bin study and on the CRPCY roof, in terms of survival and plant health. 210

PAGE 211

Implications of Nutrient Loads for Florida Green Roofs Similar to the case with water retention, the growing medium type is influencing the TP loads in the runoff from the green roof bins and green roofs themselves, more so than plant type. The three growing media started out with TP loads that were significantly different, with Hydrotech having the highest TP load in its leac hate and UCF with the lowest. Within 18 weeks all growing media types leached out most of thei r excess TP from their soils and eventually had comparable levels of TP in the leachate (no significant differences among soil types by week 18). Despite the fact that the TP lo ad in leachate from Hydrotech growing medium decreased from 325 mg to 25 mg, from Week 1 to Week 24, it was the overall highest TP load contributor to leachate among the three growing media ty pes, irrespective of plant type. ). It was found that the bulk of the nutrients (TP, TNTKN, NO3) were leached out of all plant-growing medium combination in the first six week period, as compared to any other time period in the 24 week study period. For example, for TN loads from the first 6 weeks of establishment comprised 89% of th e total load of TN for the en tire 24 week study period for all growing media types. The differences in load among growing media types was more significant than among plant types. For example, the TN load in the establishment period for H was 1694 mg N m-2 which was 10 times higher than that of U (169 mg N m-2) and 5 times higher than in B (384 mg N m-2); for the subsequent time periods th e difference in load among the growing medium types was less, and over time the absolu te values dropped ten fold for each growing medium to a range of 145 mg N m-2 to 31 mg N m-2 for weeks 12 to 24 respectively for H, 33 mg N m-2 to 7 mg N m-2 for B, and 1403 mg N m-2 for U, for weeks 12 to 24 respectively. The null hypothesis that there would be no signi ficant differences in to tal load of TN and TP among growing media for the establishment period or for the entire 24-week study period was rejected and the alternative hypothesis wa s accepted. Nutrient loads varied the most 211

PAGE 212

significantly among growing medium types and secondarily among pl ant types. For all nutrients there was an effect of time, this was more slight in TKN loads (p= 0.056), and seen most strongly in TP loads (p<0.001). TP concentrations from this study for leachate from B and U (ranging from 0 mg/L to 1 mg/L over all time periods), were in a similar range to TP concentrations found in other green roof studies, slightly higher than those from Berndstonn, et al. (2009) and (2006) and slightly lower than TP concentrations in Moran (2005). TP concentrations from H were hi gher than values reported in the literature for green roofs (Berndstonn (2006) and (2009), Ki m (2006) and Moran (2005). Th e hypothesis that there would be no significant differences in TP load among plant-growing medium combinations was rejected and the alternative hypothesis was accep ted. Mean TP loads for each plant-growing medium combination were analyzed using th e least Squares means method for comparing interactions among plantgrowi ng medium combinations on TP load over all time periods combined show that Um, Up, Ur, Us and Bp are not significantly different from the overall mean TP load, while Bm, Br, Bs and Hm, Hp, Hr and Hs were significantly higher than the other plantgrowing media combinations. Mean TP loads fo r plant-growing medium combinations for the entire study period ranged from 18-22 mg P m-2 for U (with Ur being the lowest and Us the highest within U), 30-42 mg P m-2 for B (with Bp being the lowe st and Bs the highest) and 90140 mg P m-2 for Hydrotech (with Hr bei ng the lowest and Hm being the highest. Plant effect on cumulative total TP load over the whole study period was that perennials had the lowest load and succulents had the highest load within each growing media type. Total cumulative TP load for the various combinations range d from a low of 150 mg P m-2 for Up to a high of 3200 mg P m-2 for Hs. Cumulative TP loads from leachate from Building Logics growin g medium was in the middle with a range from 310 mg P m-2 (Bm, Bp, Br) to 450 mg P m-2 (Bs). 212

PAGE 213

Plant effect on TN was that irregardless of growing medium type, TN loads from runners in each time period were significantly lower than TN loads from succulents, which always had the highest TN load in every time step (though no t always significantly hi gher than bare media). With regards to a concentration based first flush effect, it was found that for TP there was no CBFF within the first six hour s (concentrations were measured 20 minutes post-irrigation, 1 hour post-irrigation and 6-hours post-irrigation. There were no significant differences in concentrations in these intervals, however there was a mass based first flush effect for the runoff in the first 20 minutes, simply because the ma jority of the leachate percolated through the medium in the first 20 minutes generating the most volume of any of the time increments and hence the most mass of nutrients of any of the time increments was generated closest to the beginning of runoff. The hypothesis that concentrations in leachate fo r all nutrients will diminish over time (the whole study period) was accepted, as concentrations of TP, TN, NO3, TKN did decrease in the lysimeter samples over the whole study period. Implications of the findings are that since th e nutrient loads in leachate vary significantly among the growing medium types, but plant heal th did not vary signi ficantly between media with even ten-fold differences in nutrient levels in leachate (for example TN in H versus U), and the nutrient levels in all the media approach the same levels between weeks 12 and 18, it is recommendable to not pre-fertilize or pre-mix excessive fertilizers into the growing medium before planting. The findings indicate that all exce ss nutrients are not plan t available and are not taken up by the plants, but rather enter runoff, posing an ecological threat to receiving water bodies. (Excessive nutrients are being defined as any nutrient that washes out and is not taken up by the plants. It seems that aimi ng to have the minimum levels of nutrients necessary for plant growth and testing that those levels are not in excess is worthwhile, levels in U resulted in less 213

PAGE 214

leaching for all nutrient types and plant health in bins containing U were not significantly worse than plant health in bins cont aining H growing medium, which ha d the highest amount of excess nutrients in the leachate. Effect of Plant Type on Nutrient Loads Regarding plant types, it is interesting to not e that two species of succulents, Sedum spp and Delosperma, which are known to do well in North Carolina (Moran, 2004), but anecdotally have been said to do poorly in Florida (pers. comm. Hardin 2006), made up two of three species that survived all time periods (d rought, frost and heavy rains). However, plant effect-wise these succulents did not improve water quality, infact thei r leachate had P-loads that were significantly higher than other plant types and even surpassed th e P-loads of leachate from bare media in 3 of the 4 time periods. While runners overall had the lowest TP load s among plant types, th ey were adversely affected by frost and perished in the cold. Th erefore for future studies, instead of testing leguminous runners again, instead the plants th at eventually colonized the abandoned bins on their own and survived without ir rigation and survived frost should be tested, such as subtropical grasseswhich also have the same morphologica l tendencies as the runners from this study. Since 1) all the total phosphor us that leached out of the during the study would have entered directly to the stormwater system had th ey been on a roof, and 2) all soils regardless of initial levels, eventually reached the same comparable levels by 18 weeks and 3) there were no noticeable differences in plant health between soil types despite differences in soil P levels, it would be recommendable to begin with a low nutr ient growing media from the start or use an extremely slow release fertilizer in the mixture. Another implication of green roofs for Since S. Florida houses 40% of Floridas residents in a land mass of 28% of the states area, and us es 50% of its waterand will be growing at a 214

PAGE 215

rate of 78 million more people ove r 20 years, 85% of S. Floridas people reside in urban areas, majority of growth in SE Florida will be con centrated in the urban cen ters. South Florida, a region that naturally is a system of complex inte rconnected freshwater lakes, rivers marshes, sloughs, ponds, prairies, forest, and estuarie s extending over 18,000 squa re miles from the northern Kissimmee river to the Florida Keys (Kranzer, 2000). Urban areas when urbanized reach impervious areas of 72%, 40% of which is due to roofs. If applied to the land mass, water detention wise, gr could help w ith reducing water entering streams. However, water quality wise, we see that even in periods of low-flow and the return flow portion of the green roof runoff hydrograph, P-loading exceeds that of natural areas measured in re gions of S. Florida such as running off of agricultural fields or even STAs, which were crea ted to treat water coming off of agricultural fields. Since the American lawn and irrigation accounts for 70% of water use in an urban area and nutrient e xport increases when agricultural ar eas are converted from dairy and ranch farms to divided parcels of urban hous ing, housing developments that could require cisterns combined with a green roof could see a reduction in peak dema nd of energy for cooling, less infrastructure for reclaimed water and le ss use of potable water for irrigation. Without hooking up a cistern or reuse system with a green roof in the tropics or sub-tropics, we could see an increase in water demands that are already hi gh and further degradation of receiving water bodies. It is highly recommended that in Florida, green roofs be fi ner tuned to grow plants that need no irrigation, or absolutely require the use of some sort of recollection re-use of runoff on site. A person interested in inst alling a green roof because of inte rest in sustainability would no doubt gladly give the follow through of taking the appropriate measures to reuse the water onsite. Notethat roofs in general are most often direct ed directly to the stormwater conveyance system, therefore, even if reduction of water quantity is met, this P-laden runoff will go directly to the 215

PAGE 216

stormwater system, therefore either a rain gard en on-site or a cister n is recommendable in conjunction with a green roof in Florida. Again this dissertation is not a an endorsement of certain brands of grow ing media, rather it aims to highlight the characteristics of th e growing medium that make it a good growing medium, that manufacturers can use when cr eating mediums. For example, the pore-size distribution, the amount of organic matter and th e form/textural analysis of the OM, and the overall porosity and one engineeri ng aspect that was not analyzed in this dissertation is the compactibility of the media. For example, Hydrot ech and BL do tests to see much pressure the medium can withstand and not compress, so that it is indeed longlastingit will take several years of monitoring UCFs mix to see how it perfor ms over time with regard to this engineering aspect of the soil. However, from the stormwat er perspective, we are not concerned that the amount of medium shrinks from 15 cm to 10 cm due to oxidation of OM, as senescing plants will accrete OM, (inhibiting seedling establishment at worst). In a case where only 5 to 7 cm of growing medium is being utilized and the roof to last 40 years one may be concerned with the medium wasting away, but even here plant sene scence and regrowth can increase the medium thickness. No problems with oxidation of OM has b een seen in VA in the 6 years its been in place, nor problems with diminishing substrate or decrease in plant growth/health due to exhaustive use of nutrients and OM. If green roofs are used as a BMP in Florida in new developments, a provision in the law must be created that will allow for positive credits for using of green roof as a BMP only if it is implemented in conjunction with a cistern or re use of water on-site su ch as exfiltration, raingarden, wet or dry detention. Further points if irrigation is due to recycling roofs runoff (or collecting a smooth roofs runoff from a higher elevation and tem porarily detaining it and using that for irrigation of the roof versus reclaimed or potable wate r) and lesser points if irrigation 216

PAGE 217

water is from reclaimed water and a ban on gr een roofs installed to use potable water for irrigation (or negative points). Additionally, plant choiceif succulents are chosenwe can assume not much is being done for water qualit y, but on the other hand irrigation may be reduced or possibly avoided during some seasons. If perenn ials are used, while st ormwater detention can be expected to be slightly greater based on the results of this study, irrigation will ultimately be necessary and is recommended in short duration/high frequency applications, rather than long duration-infrequent applications, because it was found that runoff occurs before soil moisture storage capacity is 100% attained due to macropores, channels and the high permability of the growing medium, therefore it would be difficult, costly and take a long time to irrigate until soil moisture capacity is fulfilled. Role of Green Roofs in Stormwater Hydrology and Water Quality in Florida as compared to Virginia The goals of the research included 1) charac terizing the effect of green roofs on urban stormwater hydrology for Florida a nd Virginia, and 2) characterizing the effect of green roofs on the water quality of runoff for Florida and Virginia. The effect of the green roofs in both Florid a and Virginia was that they did retain significant portions of stormwater and significantly increase the lag time to peak and extend the runoff period from the roof. In Florida, small ra in events (< 0.4 cm, n =31), had a significantly higher mean retention (79%) than either medium rain events (0.42 cm 0.83 cm) or large events (>0.83 cm). There were significant differences in retention between medium (43% retention) and large (26% retention) rain events in Florida, but no significant di fferences in extension of runoff past the end of a rain event (s = 7.8 hr, m = 11 hr l = 8.7 hr) in Florida. In Virginia, the mean retention by size class differed significantly between small, medium and large rain events, with respective retention rates of (98%, 84% and 72%). The retention rates of the large events (72%) 217

PAGE 218

compared well to studies in neighboring and n earby states (North Caro lina and Pennsylvania) (Moran, 2005 and DeNardo, 2003), though the small a nd medium events had retention rates that were unusually high in comparison. In both Flor ida and Virginia, there were significant differences in the reduction of the peak intensity among all the different size classes of the rain events, 94% for small events, 79% for medium even ts and 60% for large ev ents for Florida) and 100%, 93% and 79% for small, medium and large ev ents, respectively in Virginia. The extension in end of runoff was not as great in Virginia as provided by the 15 cm thick green roof in Floridathe median extension time vari ed between 0.68 hours and 3.9 hours through the seasons. In Florida, the median extension in runoff varied between 2.5 hours and 14 hours by season. This difference is attributed to the thickn ess of the Florida green roof and the pore-size distribution of the green roof, as well as differences in the und erlayment (for example, Hydrogel packs do not re-release water until ET occurs to remove the water from the gel packs.) In summary of the major objective for hydrolog ical analyses, it was to determine if the green roofs behaved similarly hydrologically in both extremes of the subtropical climate. This study found that for the two roofs studied, they did behave in the same manner, though at different intensities for water retention, peak reduction and increase in lag time, with water retention and peak reduction being greater in Virginia and a great er difference in the effect on extension of runoff and increase in lag to peak in Florida than Virginia. In terms of water quality analys es, the hypothesis tested was that the green roofs influence on water quality (nutrients, metals, TSS and TDS) will be similar in both Virginia and Florida and that they will act as sinks for nitrogen a nd sources for phosphorus and suspended sediment; and will act similarly for metals in both regions (whether as a source or sink). It was found that the green roof did infact act as a source for phos phorus, (both TP and OP), which is similar to other studies in nearby states a nd other regions of the world. The green roofs did also act as a 218

PAGE 219

sink for nitrogen as nitrate as hypothesized, how ever no difference in concentrations were noticed between the green roof and control roof for NH3-N. The green roofs did buffer pH, which is similar to findings in other studies (Berghage, 2005; Kohler and Schmidt, 2003; DeNardo, 2003). Metals behaved differently by location, for example aluminum and iron were found to be significantly higher in green roof runoff in Florida, but not in Virginia, and lead concentrations were also signifi cantly lower in green roof runoff than conventional roof runoff in Virginia. Concentrations of metals varied more between locations than did nutrients. For both metals and nutrients, variability between storms was greater than within storms, for each location. Total Dissolved Solids (TDS) were sign ificantly higher in green roof runoff as compared to conventional roof runoff, however total suspended solids (TSS) was surprisingly not found to be significantly different in green roof runoff as compared to conventional roof runoff. However concentrations from Florida were significantly higher than those found in the green roof runoff in Virginia. The load reductions attributable to the green roof, as compared to a conventional roof of the same size, varied with the size and intensity of the storms, by and large, the green roof seemed to lower nitrate and ammonium concentrations from 12-90% depending on storm size and intensity, but the presence of the green roof increased TSS, TP and SRP loads in most cases. The SRP load was often increased over 10-fold. Summary In summary, the green roofs did behave simila rly in the two extremes of the climate and provided benefits of stormwater volume control by detaining and retainin g stormwater. In terms of water quality, in both regions and in the Florida bin study, the gree n roofs did contribute nutrients to the runoff, TP in all three cases, though the total load of nitrat e was noted to decrease in the in situ roof studi es. It is recommended that green roofs in both Florida and Virginia receive 219

PAGE 220

positive points for stormwater detention, but for storm water quality follow a design similar to those used in this studywhere green roof runoff was used on the landscape in Virginia, or held in a cistern and re-used in Florid a, as was also done at UCF. Us ing a cistern in conjunction with the green roofs in Florida is highly recommendable and will take full advantage of nutrients in the runoff, keep excess out of stormwater sy stems and receiving water bodies and reduce irrigation needs. It is not recommendable in Flor ida to have green roof runoff go directly into impaired water bodies (or conveyance systems that le ad directly into impaired water bodies.) The green roofs in Florida and Virginia did behave similarly in terms of being a P source and N sink, but differently with regards to metals. Future recommendations incl ude investigating the possibility of using green roof s as a flow-through treatment wetland for treating high N grey water on-site, and testing plants that colonized abandoned bins and needed no irrigationsuch as Tropical crab grass, that could potentially be used on green roof s that are out of sight and a model should be created weighing the economic va lue of the positive bene fits of stormwater detention/retention and being an N-sink against th e costs of TP entering the stormwater system and costs of maintenance. Provisi ons should be made in the Flor ida code governing the use of potable water for irrigationand should promote the reuse of green roof runoff to the roof for irrigation, or secondarily reclaimed water and dete r green roof users from directing green roof runoff directly into receiving water bodies. 220

PAGE 221

APPENDIX A IRRIGATION REGIME FOR GREEN ROOF BIN STUDY Table A-1. Irrigation regimen of the green roof bins for the hydrologic and nutrient green roof study in Gainesville, FL, 2007. Days of Establishment Period (6 week s) Amount and Frequency of Irrigation Day 1 to 3 1.27 cm 2xs a day Days 4 to 7 1.27 cm 1x per day Day 7 to 14 1.27 cm every other day Day 14 to 21 1.27 cm every other day Day 21 to 28 1.27 cm every third day Day 28 to 35 1.27 cm every third day Day 35 to 42 Every third day 221

PAGE 222

APPENDIX B SPECIFICATIONS OF LITE TOP GROWING MEDIUMHYDROTECH USA Property Intensive Grain Size Distribution clay fraction 0-2 % passing #200 sieve 5-15 % passing #60 sieve 10-25 % passing #18 sieve 20-50 % passing 1/8-inch sieve 55-95 % passing 3/8-inch sieve 90-100 % Density Application Density 0.7 1.1 g/cm3 (44 lbs 68 lbs/cf) Saturated Density 1.0 1.5 g/cm3 (62 lbs 93 lbs/cf) 0.6 -1.1 g/cm3 Dry Density (38 lbs 68 lbs/cf) Water & Air Management (% vol.) saturated water capacity >40 % saturated air content >10 % Saturated Hydraulic Conductivity >0.5 mm/min (>1.0 in/hr) pH, Lime, and Salt Content pH (saturated paste) 5.5 7.5 carbonate content <25 g/l salts content (water extract) <3.0 g/l (2.0 mmhos/cm) Organics OM content 6 12 mass % C/N ratio <20 Nutrients** (plant available) nitrogen (NO3) 3 15 phosphorus 1 7 potassium 6 15 calcium 19 65 magnesium 3 15 CEC Capacity >6 cmol/kg Compost Fraction 1) Meet or exceed USEPA Class A standard, 40 CFR 503.13, Tables 1 & 3 (chemical contaminants) and 40 CFR 503.32(a) (pathogens) and/or be permitted in the state of origin to produce Class A material. 2) Meet US Compost Council STA/TMECC criteria or equal for Class I or II stable, mature product. Values shall be adjusted due to availability of local materials or special project conditions related to plant Values shall be adjusted due to availability of local materials or special project conditions related to plant selection and/or environmental conditions. ** Nutrients shall be adjusted with appropriate slow-release fertilizer with micronutrient additions if below lower target range. Source: Hydrotech USA, 2008. 222

PAGE 223

APPENDIX C PHYSICAL PROPERTIES OF STALITE Source: Stalite, 2008. Figure C-1. Image of STALITE Lightweight Aggregate specificat ions sheet for properties and gradations of the material for structural applications. 223

PAGE 224

Source: Stalite, 2008. Figure C-2. Image of STALITE Lightweight Aggregate specifications sheet for physical characteristics of the material for structural applications. 224

PAGE 225

APPENDIX D MODEL FIT FOR WATER RETENTION ANOVAS Table D-1. Details of the PROC GLM model us ed for the ANOVA analyses for water retention for Weeks 1-6. Source DF Sum of Squares Mean Square F Value Pr > F Model 11 0.17095556 0.01554141 35.19 <.0001 Error 24 0.01060000 0.00044167 35 0.18155556 Corrected Total R-Square Coeff Var Root MSE Ret Mean0.941616 5.239413 0.021016 0.401111 Source DF Type I SS Mean Square F Value Pr > F P_type 3 0.04328889 0.01442963 32.67 <.0001 G_Media 2 0.12053889 0.06026944 136.46 <.0001 6 0.00712778 0.00118796 2.69G_Media*P_type 0.0385 Table D-2. Effect of plant type on water retention within grow ing medium types. Results of individual ANOVAS within each grow ing medium type and time period. Growing Medium Plant Type Sampling Period Weeks 1-6 Weeks 7-12 Weeks 13-18 Weeks 19-24 p 41% a Overflowed 58% a 30% a s 31% b Overflowed 53% a 29% a r 32% b Overflowed 53% a 28% a BL m 29% b Overflowed 36% b 24% b p 46% a Overflowed 73% a 40% a s 40% b Overflowed 75% a 46% b r 46% a Overflowed 76% a 47% b H m 39% b Overflowed 63% ab 39% a p 53% a Overflowed 89% a 51% a s 42% b Overflowed 82% b 52% a r 44% b Overflowed 86% ab 50% a UCF m 42% b Overflowed 74% c 49% a 225

PAGE 226

APPENDIX E WATER CONTENT CURVES FOR WEEKS 12, 18 AND 24 Lysimeter Week 12 Water Content (vol water/vol bin)-BL0.00 0.02 0.04 0.06 0.08 0.10 0.12 0.14 0.16 0.18 0.20 01020304050607080 Hours Post-Irrigation% water (vol water/ vol bin) Perennials Avg Succulents Runners Avg Bare Media Lysimeter Week 12 Water Content (vol water/vol bin)-H0.00 0.05 0.10 0.15 0.20 0.25 0.30 0.35 0.40 01020304050607080 Hours Post-Irrigation% water (vol water/ vol bin) Perennials Avg Succulents Runners Avg Bare Media A) B) Figure E-1. Changes in water cont ent for growing various plant types (p, r, s, m) in growing medium types B and H in Week 12, up to 72 hours post-irrigation with 1.27 cm water. Note 1.27 cm of rain fell between hours 48 and 60, causing an increase in water content. 226

PAGE 227

Lysimeter Week 12 Water Content (vol water/vol bin)-UCF0.00 0.05 0.10 0.15 0.20 0.25 0.30 0.35 0.40 0.45 01020304050607080 Hours Post-Irrigation% water (vol water/ vol bin) Perennials Avg Succulents Runners Avg Bare Media Figure E-2. Changes in water cont ent for growing various plant types (p, r, s, m) in growing medium U in Week 12, up to 72 hours post-ir rigation with 1.27 cm water. Note 1.27 cm of rain fell between hours 48 and 60, causing an increase in water content. 227

PAGE 228

Lys Wk 18 Water Content BL0.18 0.16 0.14 % Water (vol water/vol bin) 0.12 0.10 0.08 0.06 0.04 Perennials Avg Succulents Avg Runners Avg 0.02 Bare Media Avg 0.00 0 10 20 30 40 50 60 70 80 Hours post-irrigation Lys Wk 18 Water Content H0.00 0.05 0.10 0.15 0.20 0.25 0.30 0.35 0.40 01 02 03 04 05 06 07 08 0 Hours post-irrigation%water (vol water/vol bin) Perennials Avg Succulents Avg Runners Avg Bare Media Avg Lys Wk 18 Water Content UCF0.00 0.05 0.10 0.15 0.20 0.25 0.30 0.35 01 02 03 04 05 06 07 08 0 Hours post-irrigation%water (vol water/vol bin) Perennials Avg Succulents Avg Runners Avg Bare Media Avg % Water (vol water/vol bin) % Water (vol water/vol bin) Figure E-3. Changes in water co ntent in A) Building Logics, B) Hydrotech and C) UCFs Black and Gold growing media over 72 hours postirrigation of 1.27 cm water in Week 18. 228

PAGE 229

Lys Week 24--Water Content (vol water/vol of bin) 0.00 0.02 0.04 0.06 0.08 0.10 0.12 0.14 0.16 0.18 0.20 01020304050607080 Hours post-irrigation% Water Content (vol water/vol bin) B-perennials B-succulents B-runners B-bare media Lys Week 24--Water Content (vol water/vol of bin)-H0.00 0.05 0.10 0.15 0.20 0.25 0.30 0.35 0.40 01020304050607080 Hours post-irrigation% Water Content (vol water/vol bin) H-perennials H-succulents H-runners H-bare media Lys Week 24--Water Content (vol water/vol of bin)-UCF0.00 0.05 0.10 0.15 0.20 0.25 0.30 0.35 0.40 01020304050607080 Hours post-irrigation% W ater Content (vol water/vol bin) U-perennials U-succulents U-runners U-bare media Figure E-4. Changes in water co ntent in A) Building Logics, B) Hydrotech and C) UCFs Black and Gold growing media over 72 hours postirrigation of 1.27 cm water in Week 24. 229

PAGE 230

APPENDIX F WATER RETENTION CURVES FOR WEEKS 6, 12, 18 AND 24 -150% -100% -50% 0% 50% 100% 01020304050607080 Hours post-irrigationVol Water/ Vol Bi n B-perennials B-succulents B-runners B-bare media A) B) C) -150% -100% -50% 0% 50% 100% 01020304050607080 Hours post-irrigationVol Water/ Vol Bin H-Perennial H-Succulents H-Runners H-Bare Media -150% -100% -50% 0% 50% 100% 01020304050607080 Hours post-irrigationVol Water/ Vol Bin U-Succulents U-Runners U-Bare Media U-Perennials Figure F-1. Changes in water in retention in A) Building Logi cs, B) Hydrotech and C) UCF Black & Gold growing media over 72 hours post-irrigation of 1.27 cm water in Week 6 (9/5-9/8/2007). 230

PAGE 231

Lys Wk 12 Water Retention (% of input) for BL 0.00 0.05 0.10 0.15 0.20 0.25 0.30 0102030405060 Hours post-irrigation% retention (water kg/ input kg) B-perennials B-succulents B-runners B-bare media Filter Fabric Empty Lys Wk 12 Water Retention (% of input) for H0.00 0.10 0.20 0.30 0.40 0.50 0.60 0.70 0.80 0.90 01 02 03 04 05 06 0 Hours post-irrigation% retention (water kg/ input kg ) H-perennials H-succulents H-runners H-bare media Filter Fabric Empty Lys Wk 12 Water Retention (% of input) for UCF0.00 0.10 0.20 0.30 0.40 0.50 0.60 0.70 0.80 0.90 1.00 01 02 03 04 05 06 0 Hours post-irrigation% retention (water kg/ input kg ) U-perennials U-succulents U-runners U-bare media Filter Fabric Empty Figure F-2. Changes in water retention in A) Bu ilding Logics, B) Hydrotech and C) UCFs Black and Gold growing media over 72 hours post-irrigation of 1.27 cm water in Week 12. 231

PAGE 232

Lys Wk 18Water Retention (%)--BL 0.00 0.05 0.10 0.15 0.20 0.25 0102030405060708 Hours Post-Irrigation% Water Retention (kg water retained/ kg water in0 ) B-perennials B-succulents B-runners B-bare media Lys Wk 18Water Retention (%)--H0.00 0.10 0.20 0.30 0.40 0.50 0.60 0.70 0102030405060708 Hours Post-Irrigation% Water Retention (kg water retained/ kg water in)0 H-perennials H-succulents H-runners H-bare media Lys Wk 18Water Retention (%)--UCF0.00 0.10 0.20 0.30 0.40 0.50 0.60 0.70 0.80 0102030405060708 Hours Post-Irrigation% Water Retention (kg water retained/ kg water in)0 U-perennials U-succulents U-runners U-bare media Figure F-3. Changes is water retention in A) Building Logi cs, B) Hydrotech and C) UCF growing media over 72 hours post-irrigation as a percent of water applied in response to a 1.27 cm irrigation event in Week 18 (December 5th, 2007) of the study period. 232

PAGE 233

Lys Wk 24--Water Retention (% water in growing media/ water input)-BL 0% 5% 10% 15% 20% 25% 30% 35% 40% 01020304050607080 Hours post-irrigationWater Retention (% of input) B-perennials B-succulents B-runners B-bare media Lys Wk 24--Water Retention (% water in growing media/ water input)-H0% 10% 20% 30% 40% 50% 60% 70% 80% 01020304050607080 Hours post-irrigationWater Retention (% of input ) H-perennials H-succulents H-runners H-bare media Lys Wk 24--Water Retention (% water in growing media/ water input)-UCF0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 01020304050607080 Hours post-irrigationWater Retention (% of input) U-perennials U-succulents U-runners U-bare media E/F avg Figure F-4. Water retention in A) Building Logi cs, B) Hydrotech and C) U growing media over 72 hours post-irrigation (1.27 cm) as a % of water applied, in week 24 of the study. 233

PAGE 234

APPENDIX G MEAN WATER UPTAKE/RELEASE RATES FOR GROWING MEDIA AND PLANTGROWING MEDIUM COMBINATIONS Table G-1. Mean rates of wate r uptake (% volumetric water cont ent/hr) for different growing medium types (B, H, U), based on slope of water uptake from water content curves for Weeks 6, 12, 18 and 24 found in Appendix E. Significant differences between slopes based on the results of the paired t-test at an of 0.05 are denoted by different letters within a time period. Week 6 Week 12 Week 18 Week 24 Mean Std Dev Mean Std Dev Mean Std Dev Mean Std Dev B 0.023 0.007 a 0.021 0.004 a 0.012 0.006 A 0.027 0.004 a H 0.043 0.013 b 0.042 0.014 b 0.034 0.009 B 0.052 0.004 b U 0.049 0.014 b 0.059 0.007 c 0.042 0.008 C 0.061 0.005 c Table G-2. Mean rates of wate r uptake (% volumetric water conten t/hr) for different plant types from water retention curves for Weeks 6, 12, 18 and 24 for B, H and U found in Appendix C. Significant differences between sl opes based on the results of the paired t-test at an alpha =0.05 level are denote d by different letters within a time period. Week 6 Week 12 Week 18 Week 24 Level Mean Std Dev Mean Std Dev Mean Std Dev Mean Std Dev m 0.017 0.006 a 0.021 0.001 a 0.013 0.003 a 0.030 0.002 a p 0.031 0.003 a 0.024 0.007 a 0.015 0.006 a 0.026 0.004 ab r 0.018 0.006 b 0.019 0.004 a 0.007 0.006 a 0.024 0.002 b B s 0.026 0.004 ab 0.020 0.002 a 0.014 0.007 a 0.029 0.005 ab m 0.030 0.004 a 0.055 0.001 a 0.043 0.009 a 0.051 0.005 a p 0.052 0.005 b 0.041 0.017 ab 0.030 0.003 ab 0.056 0.003 a r 0.056 0.008 b 0.026 0.004 b 0.027 0.009 b 0.050 0.002 a H s 0.032 0.001 a 0.047 0.012 a 0.036 0.008 a 0.051 0.006 a m 0.038 0.008 b 0.063 0.002 a 0.050 0.007 a 0.065 0.003 a p 0.070 0.004 a 0.060 0.008 a 0.038 0.011 a 0.058 0.004 a r 0.046 0.009 b 0.061 0.010 a 0.040 0.005 a 0.060 0.007 a U s 0.041 0.003 b 0.053 0.007 a 0.040 0.007 a 0.061 0.004 a 234

PAGE 235

235 Table H-3. Evapotranspiration Ra tes in mm/day for Week 18Early Winter, based on change in mass over daylight hours on 12/5/07 and 12/6/07. APPENDIX H ET RATES FOR PLANT-GROWING MEDIUM COMBINATIONS BY TIME PERIOD Table H-1. Evapotranspiration Ra tes in mm/day for Week 6Late Summer, based on change in mass over daylight hours on 9/6/07 and 9/7/07. ET Rate (mm day-1) Plant Type B (mean SD) H (mean SD) U (mean SD) E/F (mean SD) Perennials 3.0 1 5.3 1.8 6.1 2.1 Succulents 2.2 0.5 3.6 0.9 4.2 0.3 Runners 2.3 0.3 5.6 2.0 5.6 0.6 Bare Media 1.5 0.7 3.6 0.3 3.4 0.6 E/F 0.2 0.1 Table H-2. Evapotranspiration Rates in mm/day for Week 12Fall, based on change in mass over daylight hours on 10/18/07 and 10/19/07. ET Rate (mm day-1) B (mean SD) H (mean SD) U (mean SD) Perennials 1.40 0.4 2.9 1.2 4.2 1.4 Succulents 0.70 0.07 1.9 0.09 1.8 0.13 Runners 1.10 0.4 1.5 0.4 2.5 0.4 Bare Media 0.40 0.1 1.1 0.1 1.4 0.15 ET Rate (mm/day) B U H (mean SD) (mean SD) (mean SD) Perennials 1.07 0.31 1.13 0.22 1.55 0.88 Succulents 0.67 0.19 1.05 0.26 1.64 Runners 0.59 0.07 0.75 0.00 1.68 Bare Media 0.50 0.25 1.05 0.15 1.43 0.25 0.73 0.07

PAGE 236

236 APPENDIX I CHANGES IN ORGANIC MATTER CONTEN T AND GRAIN-SIZE DEISTRIBUTION IN THE GROWING MEDIA, BEFORE AND AFTER THE GREEN ROOF STUDY Organic matter before 0.0% 1.0% 2.0% 3.0% 4.0% 5.0% 6.0% 7.0% U beforeH beforeBL before Growing Medium Type% by dry weight OM OM in smaller fractions (<2mm) Organic Matter after 1 yr (unplanted--weeded regularly) 0.0% 1.0% 2.0% 3.0% 4.0% 5.0% 6.0% 7.0% U 1yr avgH 1yr avgBL 1yr avg Growing Medium% dry weight OM OM in smaller fractions (<2mm) H after 2yrs and BL after 4yrs on roof 0.0% 1.0% 2.0% 3.0% 4.0% 5.0% 6.0% 7.0% H 2yr avgBL 4yr avg Growing Medium Type% mass (dry weight) OM OM in smaller fractions (<2mm) 0.0% 10.0% 20.0% 30.0% 40.0% 50.0% 60.0% 70.0% 80.0% 90.0% 100.0% GravelV. coarse sand Coarse sand Medium Sand Fine Sand BL before average BL 1yr avg BL 4yr avg 0.0% 10.0% 20.0% 30.0% 40.0% 50.0% 60.0% 70.0% 80.0% 90.0% 100.0% (<2mm)(1-2mm)(0.5mm1mm) (0.25mm0.5mm) (<0.25mm) H before average H 1yr avg H 2yr avg 0.0% 10.0% 20.0% 30.0% 40.0% 50.0% 60.0% 70.0% 80.0% 90.0% 100.0% (<2mm)(1-2mm)(0.5mm1mm) (0.25mm0.5mm) (<0.25mm) U before average U 1yr avg Figure I-2. Changes in grain size distribution of B, H and Us growing medium afte r 1 year in green roof bins and 4 years and 2 years on a roof top, for B and H, respectively. Figure I-1. Changes in total OM content befo re and after green roof study and in the top particle size fr action (> 2 mm). % Mass (dry weight of soil)

PAGE 237

APPENDIX J PHYSICAL PROPERTIES AND MACRO AND MICRONUTRIENTS FOR HYDROTECH GROWING MEDIUM Table J-1. Physical properties reported for Hydrotech growing medium. Property Value Dry bulk density 0.6-1.1 g/cm3 Saturated bulk density 1.0-1.5 g/cm3 Saturated water capacity >40% Saturated air content >10% Organic matter content (mass %) 6-12% C/N ratio <20 CEC >6 cmol/kg (Source: Hydrotech spec ifications sheet, 2007). Table J-2. Macro and micronutrients re ported for Hydrotech growing medium. Min Max Min Max Min Max g/m3 g/m3 g/kg g/kg g/m2 g/m2 Nitrogen (NO3 +NH4) 50 252 0.047 0.236 7.7 38.3 Phosphorus 17 118 0.016 0.110 2.6 17.9 Potassium 101 252 0.094 0.236 15.3 38.3 Calcium 319 1092 0.298 1.021 48.5 166 Magnesium 50 252 0.047 0.236 7.7 38.3 Sulfur 17 59 0.016 0.055 2.6 8.9 Iron 17 50 0.016 0.047 2.6 7.7 Manganese 17 50 0.016 0.047 2.6 7.7 Copper 4.2 8.4 0.004 0.008 0.6 1.3 Boron 4.2 8.4 0.004 0.008 0.6 1.3 Zinc 0.2 0.4 0.0002 0.0004 0.0 0.1 (Source: Adapted from Hydrotech specifications sheet, 2007). 237

PAGE 238

APPENDIX K TN LOAD ANOVA MODELS USED FOR ANALYZING EACH 6-WEEK PERIOD Table K-1. Details of the PROC GLM model used for the ANOVA analyses for TN load data for Weeks 1-6. Source DF Sum of Squares Mean Square F Value Pr > F Model 11 108.3731002 9.8521000 1027.20 <.0001 Error 24 0.2301891 0.0095912 Corrected Total 35 108.6032893 R-Square Coeff. .Var Root MSE Lntn Mean 0.997880 1.592610 0.097935 6.149322 Table K-2. Details of the PROC GLM model used for the ANOVA analyses for TN load data for Weeks 7-12. Source DF Sum of Squares Mean Square F Value Pr > F Model 11 26.27858881 2.38896262 21.43 <.0001 Error 24 2.67488556 0.11145357 Corrected Total 35 28.95347437 R-Square Coeff Var Root MSE lntn Mean 0.907614 8.774279 0.333847 3.804832 Table K-3. Details of the PROC GLM model used for the ANOVA analyses for TN load data for Weeks 13-18. Source DF Sum of Squares Mean Square F Value Pr > F Model 11 27.23662857 2.47605714 3.56 0.0045 Error 24 16.70343325 0.69597639 Corrected Total 35 43.94006182 R-Square Coeff Var Root MSE lntn Mean 0.619859 36.41717 0.834252 2.290821 238

PAGE 239

Table K-4. Details of the PROC GLM model used for the ANOVA analyses for TN load data for Weeks 19-24. Source DF Sum of Squares Mean Square F Value Pr > F Model 11 16.36982721 1.48816611 8.80 <.0001 Error 24 4.05981259 0.16915886 Corrected Total 35 20.42963980 Table K-5. Results of the ANOVA (degrees of freedom, sum of square, mean square and Fvalues) for TN load for weeks 1-6. Source DF Type I SS Mean Square F Value Pr > F P_type 3 0.4320634 0.1440211 15.02 <.0001 G_Media 2 107.806018 53.9030091 5620.04 <.0001 G_Media*P_type 6 0.1350186 0.0225031 2.35 0.0633 Table K-6. Results of the ANOVA (degrees of freedom, sum of square, mean square and Fvalues) for TN load for weeks 7-12. Source DF Type I SS Mean Square F Value Pr > F P_type 3 0.21146358 0.07048786 0.63 0.6013 G_Media 2 24.8529728 12.42648642 111.49 <.0001 G_Media*P_type 6 1.21415240 0.20235873 1.82 0.1384 Table K-7. Results of the ANOVA (degrees of freedom, sum of square, mean square and Fvalues) for TN load for weeks 13-18. Source DF Type I SS Mean Square F Value Pr > F P_type 3 3.88950127 1.29650042 1.86 0.1629 G_Media 2 20.9816477 10.49082385 15.07 <.0001 G_Media*P_type 6 2.36547959 0.39424660 0.57 0.7527 Table K-8. Results of the ANOVA (degrees of freedom, sum of square, mean square and Fvalues) for TN load for weeks 19-24. Source DF Type I SS Mean Square F Value Pr > F P_type 3 0.50119655 0.16706552 0.99 0.4153 G_Media 2 13.83974155 6.91987078 40.91 <.0001 G_Media*P_type 6 2.02888911 0.33814818 2.00 0.1055 239

PAGE 240

240 APPENDIX L TKN LOAD MODEL DETAILS (F-VALUES AND R-SQUARED VALUES) AND RESULTS FOR ANOVA ANALYSES Table L-1. Details of the PROC GLM model used for the ANOVA analyses for TKN load for Weeks 1-6. Source DF Sum of Squares Mean Square F Value Pr > F Model 11 16.81834862 1.52894078 115.15 <.0001 Error 24 0.31866454 0.01327769 Corrected Total 35 17.13701316 R-Square Coeff. .Var Root MSE Lntn Mean 0.981405 2.174536 0.115229 5.299010 Table L-2. Results of the ANOVA (degrees of freedom, sum of square, mean square and Fvalues) for TKN load for weeks 1-6. Source DF Type I SS Mean Square F Value Pr > F P_type 3 0.48541934 0.16180645 12.19 <.0001 G_Media 2 16.24423534 8.12211767 611.71 <.0001 G_Media*P_type 6 0.08869394 0.01478232 1.11 0.3839

PAGE 241

APPENDIX M NITRATE LOAD DATA FOR THE ESTABLISHMENT PERIOD Table M-1. Nitrate loads per bin (mg per m2) for each week of the establishment peri od (Weeks 1-6). Levels not connected by the same letter within a time period and growing medium type have a signif icant difference due to plant type. Growing Medium Plant Type NO3-N load in mg m-2 Week 1 Week 2 Week 3 Week 4 Week 5 Week 6 mean S.E. mean S.E. mean S.E. Mean S.E. mean S.E. mean S.E. p 30.6 12.0 a 3.5 2.1 a 9.1 0.6 a 3.2 1.1 a 5.3 1.8 a 16.2 7.6 a s 89.3 14.0 b 33.0 28.0 a 7.9 0.4 a 15.5 4.7 b 20.9 7.1 a 39.7 7.4 a r 26.0 10.3 b 7.1 1.6 a 8.0 1.4 a 11.7 4.6 ab 15.0 5.2 a 37.6 4.4 a B m 7.61 2.41 b 4.14 0.39 a 9.10 2.02 a 5.44 0.96 ab 8.12 1.49 a 16.07 7.62 a p 17100 350 ab 6540 927 a 8390 730 a 122 52 a 266 181 a 427 130 a s 17200 950 ab 7390 1390 a 9920 200 ab 184 49 a 272 58 a 212 40 ab r 18800 255 b 6400 792 a 9370 790 ab 88 7 a 304 33 a 124 11 b H m 16000 480 a 6710 1390 a 10400 150 b 381 15 b 581 16 b 702 35 c p 27.2 13.3 ab 3.7 0.6 a 3.0 2.4 a 0.2 0.1 a 0.7 0.0 a 1.1 0.0 a s 47.0 0.4 a 13.2 9.3 a 31.8 13.5 b 8.4 2.7 b 14.7 4.9 b 31.0 9.5 b r 5.56 2.61 b 6.59 2.45 a 7.29 3.82 a 2.10 0.66 a 3.69 1.20 a 9.78 6.19 a U m 12.2 4.5 b 4.8 0.3 a 7.1 1.8 a 1.6 0.2 a 2.5 0.2 a 4.3 1.9 a 241

PAGE 242

242 APPENDIX N NITRATE LOAD DATA FOR TH E 24-WEEK STUDY PERIOD Table N-1. NO3 load (mg m-2) for all plant-growing medium combinations and all time periods over the 24 week study. Significant differences between plant types within a time step and growing medium type are indi cated by different letters. Growing Medium Plant Type N-NO3 Load (mg m-2) Weeks 1-6 Weeks 7-12 Weeks 13-18 Weeks 19-24 Mean S.E. mean S.E. mean S.E. mean S.E. P 67.8 19.5 a n.d. a 0.47 0.044 a 9.16 6.46 a s 206 45 b n.d. a 0.66 0.034 ab 30.1 20.1 a r 105 13 a n.d. a 0.27 0.154 a 1.95 0.53 a B m 50.5 6.4 a n.d. a 0.22 0.129 b 2.86 2.13 a P 32700 2200 a 1150 818 a 0.56 0.028 a 1.29 0.50 a s 35200 2600 a 252 14.3 a 0.54 0.187 a 0.37 0.04 a r 35100 1280 a 155 77.7 a 0.65 0.217 a 21.9 17.4 a H m 34700 1630 a n.d. a 2.47 0.782 b 0.47 0.19 a P 35.9 16.2 a n.d. a 0.11 0.007 a 0.02 0.00 a s 146 36 b n.d. a 0.18 0.009 b 0.55 0.12 a r 35.0 14.0 a n.d. a 0.18 0.018 ab 0.42 0.30 a U m 32.5 6.6 a n.d. a 0.27 0.035 c 0.06 0.02 a n.d. non-detectable

PAGE 243

APPENDIX O DRAWINGS OF CISTERN AND PUMP SET-UP FOR CHARLES R. PERRY GREEN ROOF 243

PAGE 244

244

PAGE 245

APPENDIX P CHARLES R. PERRY GREEN ROOR HYDR OGLOGY DATA BY INDIVIDUAL STORM Table P-1. Summary of amount, duration, time of maximum intensity (peak) for rain events and runoff measured between 7/9/07 and 7/2/08 for the Charles R. Perry Construction Yard. Rain Event Runoff Date Duration Amt(cm) Max int (cm/hr) Time of Max. Int Duration RO Amt RO (cm) Max RO (cm/hr) Time of Max RO % Retention 7/9/2007 0:09 0.08 8.53 11:35 0.00 100% 7/12/2007 0:40 0.30 1.52 12:19 3:55 0.02 2.99 n/a 93% 7/14/2007 1:15 0.23 0.61 22:30 9:35 0.21 1.09 22:39 8% 7/22/2007 0:20 0.20 0.91 13:50 0:05 0.00 1.83 100% 7/24/2007 1:35 0.15 0.30 n/a 0.00 0.61 100% 7/28/2007 4:00 4.27 6.71 16:25 9:00 3.87 8.70 16:44 9% 7/31/2007 0:09 0.23 1.52 18:15 0.00 3.05 100% 8/3/2007 6:20 3.02 5.18 10:50 13:15 2.36 6.28 10:59 22% 8/9/2007 1:30 7.34 15.85 21:05 2:49 0.00 31.70 43% 8/12/2007 1:30 1.12 3.35 20:00 4:25 0.00 5.07 20:05 21% 8/13/2007 1:15 0.71 0.61 18:35 10:00 0.00 1.16 18:59 85% 8/14/2007 0:35 1.47 5.79 21:55 3:20 0.00 8.94 22:04 19% 8/25/2007 0:15 0.08 0.25 16:19 0:04 0.00 0.44 17:20 93% 8/27/2007 0:35 0.08 0.30 n/a 0.00 n/a n/a 100% 8/30/2007 0:45 0.18 0.30 n/a 0.00 n/a n/a 100% 8/31/2007 0:15 0.25 1.22 3:10 5:55 0.09 2.38 n/a 65% 9/2/2007 0:16 0.61 2.44 2:45 5:40 0.00 4.56 n/a 50% 9/2/2007 2:30 3.23 12.80 15:05 7:35 3.06 20.51 15:09 5% 9/13/2007 0:15 0.69 1.56 20:30 13:15 0.00 n/a n/a 90% 9/19/2007 0:30 0.08 n/a n/a 0.00 n/a n/a 100% 9/19/2007 3:20 1.04 0.61 22:55 7:45 0.22 0.65 23:44 75% 9/20/2007 3:45 1.73 2.74 0:00 14:30 1.17 3.79 23:59 32% 9/21/2007 1:10 2.46 6.40 21:20 9:50 1.69 9.28 21:24 31% 10/2/2007 3:45 2.01 2.32 12:25 9:30 1.80 3.58 12:54 10% 245

PAGE 246

Table P-1. Continued Rain Event Runoff Date Duration Amt(cm) Max int (cm/hr) Time of Max. Int Duration RO Amt RO (cm) Max RO (cm/hr) Time of Max RO % Retention 10/3/2007 0:55 0.13 0.30 n/a 6:20 0.07 n/a n/a 43% 10/4/2007 6:35 4.78 11.28 16:35 16:55 3.98 21.99 18:19 17% 10/5/2007 0:50 0.71 3.96 0:20 15:10 0.75 6.86 0:24 0% 10/5/2007 4:35 3.81 6.71 14:35 8:35 4.00 8.76 14:35 0% 10/6/2007 0:40 0.28 1.52 15:25 6:10 0.25 2.86 15:34 12% 10/7/2007 0:55 0.33 1.83 12:45 7:20 0.41 3.28 n/a 0% 10/19/2007 0:35 0.33 1.83 15:00 9:00 0.30 3.59 n/a 13% 10/19/2007 0:35 0.3302 19:53 15:00 9:00 0.30 3.59 n/a 13% 10/19/2007 0:40 0.56 3.66 12:45 1:00 0.01 7.25 n/a 97% 10/21/2007 4:50 0.41 0.30 n/a 7:25 0.20 0.48 22:24 50% 10/21/2007 4:50 0.4064 7:18 n/a 7:25 0.20 0.48 22:24 50% 12/21/2007 1:50 0.46 0.61 2:40 4:20 0.149 0.0628 9:59 67% 12/30/2007 0:10 0.18 1.52 14:50 13:55 0.07 0.063 15:54 61% 1/12/2008 1:51 0.5588 0.61 15:50 15:40 0.366 0.126 17:09 35% 1/16/2008 0:40 0.127 0.30 8:30 8:00 0.042 0.063 67% 1/19/2008 14:05 4.0894 2.44 3:40 2:25 2.179 1.383 3:59 1/26/2008 0:55 0.1778 0.30 9:09 12:50 0.06 0.063 0:29 66% 1/28/2008 1:46 0.0762 0.30 n/a 0 0.000 100% 2/7/2008 5:50 0.5588 2.13 18:10 20:30 0.45 0.126 18:43 19% 2/12/2008 0:40 0.254 1.22 23:35 14:43 12.4 0.251 23:43 2/12/2008 2:56 1.2192 1.83 19:50 0:00 0 1.005 100% 2/12/2008 7:30 1.5748 0.61 16:40 11:20 1.142 0.063 17:08 27% 2/21/2008 0:34 1.27 2.16 5:15 14:35 1.065 1.760 5:30 16% 2/23/2008 2:40 2.9464 6.10 2:35 12:05 1.39 0.063 9:59 2/26/2008 6:25 1.3716 3.96 13:23 1:45 1.167 0.691 15% 3/4/2008 1:30 1.0668 5.18 14:25 10:40 0.897 1.508 14:25 16% 3/8/2008 0:15 0.4318 0.30 13:25 22:00 0.228 0.063 19:47 47% 246

PAGE 247

Table P-1. Continued Rain Event Runoff Duration Amt(cm) Max int (cm/hr) Time of Max. Int Duration RO Amt RO (cm) Max RO (cm/hr) Time of Max RO % Retention Date 6/23/2008 1:20 0.56 0.61 n/a 17:55 0.21 0.1257 62% 6/25/2008 0:15 0.30 1.22 14:20 19:15 0.06 0.0628 80% 6/26/2008 2:20 0.38 0.30 n/a 18:00 0.02 n/a 95% 6/27/2008 1:00 0.30 0.61 17:20 16:20 0.27 0.0628 11% 6/28/2008 0:20 0.0508 n/a n/a 100% 6/30/2008 0:35 0.36 0.63 15:05 19:50 0.1 0.0628 72% 7/2/2008 0:50 0.66 0.91 19:00 9:25 0.27 0.063 59% 7/8/2008 6:05 4.75 6.71 17:10 11:35 4.33 8.6093 9% 7/12/2008 1:25 0.71 2.13 12:30 2:35 0.32 0.7541 55% 7/12/2008 0:55 0.08 2.13 n/a 100% 7/13/2008 0:05 0.03 n/a 100% 7/15/2008 0:35 0.15 0.30 n/a 7:30 0.03 0.06284 80% 7/15/2008 1:15 5.41 13.72 13:45 19:00 1.35 4.52459 75% 2/18/2008 5:08 1.77 1.83 10:00 21:08 1.226 0.817 10:13 31% 7/2/2008 10:47 1.07 0.91 0:50 0.78 0.0200 27% 247

PAGE 248

Table P-2. Charles R. Perry Construction Ya rd Green Roof--Summary of rain events and runoff volume and dur ation, and percent o f stormwater retention for 2007-2008 by season. July 7, 2007 to October 23, 2007 Type Antecedent Rain Duration (hr) Rain Duration (hr) Rain (cm) Duration RO (hr) Runoff (cm) % retention Total 45.2 26.7 41% Mean 17.0 1 hr 50 min 1.2 7 hr 50 min 0.7 52% Stdev 12.9 2 hr 1.6 4 hr 20 min 1.2 40% Median 14.3 1 hr 0.4 7 hr 30 min 0.2 50% Min 0.7 10 min 0.1 5 min 0.0 0% Max 48.3 6 hr 30 min 7.3 17 hr 4.2 100% December 20, 2007 to March 18, 2008 Antecedent Moist (duration [hrs] since last storm) RainDuration (hrs) RainDuration (hrs) Rain (cm) Duration RO (hr) Runoff (cm) % retention Total Rain/Runoff 22.3 14.8 34% Mean 36.0 2 hr 20 min 1.0 10 hr 50 min 1.2 33% StdDev 26.0 2 hr 5 min 1.0 7 hr 35 min 2.2 52% Median 31.8 1 hr 40 min 0.6 11 hr 40 min 0.5 33% Min 3.3 5 min 0.0 0 hr 0.0 0% Max 96.0 7 hr 30 min 4.1 22 hr 30 min 10.5 100% April 11 to June 7, 2008 Type Antecedent Moist (duration [hrs] since last storm) RainDuration (hr) RainDuration (hr) Rain (cm) Duration RO (hr) Runoff (cm) %retention TOTAL RAIN 13.60 89% Mean 25 min 0.76 2hr 20min 90% Stdev 12 min 0.30 50 min 11% median 23 min 0.87 2hr 30 min 94% Min 5 min 0.03 30 min 61% Max 1hr 1.09 3 hr 45 min 100% 248

PAGE 249

249 Table P-2. Continued. Jun 13 to Jul 26, 2008 Type Antecedent Moist (duration [hrs] since last storm) RainDuration (hr) RainDuration (hr) Rain (cm) Duration RO (hr) Runoff (cm) % retention TOTAL RAIN 23.8 10.6 55% Mean 2 hr 1.1 13 hr 20 min 0.7 67% Stdev 3 hr 1.8 7 hr 20 min 1.3 33% median 1 hr 0.4 17 hr 10 min 0.2 75% Min 5 min 0.0 45 min 0.0 9% Max 10 hr 50 min 5.4 19 hr 45 min 4.3 100% July 26 to September 15, 2008 Type Antecedent Moist (duration [hrs] since last storm) RainDuration (hr) RainDuration (hr) Rain (cm) Duration RO (hr) Runoff (cm) %retention TOTAL RAIN 20.71 15.58 25% Mean 2 hr 1.22 11 hr 0.92 53% Stdev 2 hr 45 min 2.18 8hr 45 min 1.84 38% median 35 min 0.58 6 hr 50 min 0.17 64% Min 5 min 0.03 0 hr 0.00 0% Max 9 hr 15 min 9.09 32 hr 15 min 7.44 100%

PAGE 250

Table P-3. Delay in start of runoff after rain began, extension of runoff after rain stopped, increase in lag time to peak runoff and % reduction in peak runoff for rain events July 2007 to July 2008 for the Charles R. Pe rry Construction yard Green Roof. Date Delay Start RO Extension End RO Increase Lag to Peak % Reduction of Peak 7/12/2007 2:20 5:35 n/a 96% 7/14/2007 0:15 8:35 0:09 79% 7/22/2007 0:50 0:35 n/a 100% 7/24/2007 No runoff 100% 7/28/2007 0:10 5:10 0:19 30% 7/31/2007 No runoff 100% 8/3/2007 0:00 6:55 0:09 21% 8/9/2007 0:10 4:32 n/a 100% 8/12/2007 0:14 3:09 0:05 51% 8/13/2007 0:05 8:50 0:24 90% 8/14/2007 0:00 2:45 0:09 54% 8/25/2007 1:01 0:50 1:01 75% 8/27/2007 No runoff 8/30/2007 No runoff 8/31/2007 0:15 5:55 n/a 95% 9/2/2007 0:10 5:34 n/a 87% 9/2/2007 0:05 5:10 0:04 60% 9/13/2007 0:15 13:15 n/a 9/19/2007 No runoff 9/19/2007 1:30 5:55 0:49 7% 9/20/2007 1:00 11:45 0:00 38% 9/21/2007 0:15 8:55 0:04 45% 10/2/2007 0:05 5:50 0:29 54% 10/3/2007 0:35 6:00 n/a 10/4/2007 0:05 10:25 1:44 95% 10/5/2007 0:00 14:20 0:04 73% 10/5/2007 0:00 4:00 0:00 31% 10/6/2007 0:10 5:40 0:09 88% 10/7/2007 0:05 6:30 n/a 79% 10/19/2007 0:05 8:30 n/a 97% 10/19/2007 0:05 8:30 n/a 97% 10/19/2007 0:35 0:55 n/a 98% 10/21/2007 3:25 6:00 n/a 59% 10/21/2007 3:25 6:00 n/a 59% 12/21/2007 5:59 8:29 7:19 90% 12/30/2007 0:15 14:00 1:04 96% 1/12/2008 0:50 14:39 1:19 79% 1/16/2008 1:10 8:30 79% 1/19/2008 0:19 43% 1/26/2008 0:45 12:40 15:20 79% 1/28/2008 100% 250

PAGE 251

251 Table P-3. Continued 2/7/2008 0:10 14:50 0:33 94% 2/12/2008 0:08 79% 2/12/2008 0:05 3:55 0:28 90% 2/21/2008 0:13 14:14 0:15 19% 2/23/2008 7:24 99% 2/26/2008 0:05 19:25 83% 3/4/2008 1:05 10:15 0:00 71% 3/8/2008 0:00 100% 3/8/2008 1:55 23:40 6:22 79% 6/23/2008 0:45 17:20 79% 6/25/2008 0:30 19:30 100% 6/26/2008 1:00 16:40 0% 6/27/2008 0:20 15:40 90% 6/28/2008 100% 6/30/2008 0:20 19:35 90% 7/2/2008 0:20 8:55 93% 7/8/2008 0:10 6:50 7/12/2008 0:00 1:10 65% 7/12/2008 100% 7/13/2008 100% 7/15/2008 0:15 7:10 100% 7/15/2008 0:15 18:00 67% 2/18/2008 0:13 55% 7/2/2008 0:20 14:23 98% Extension End RO = Time b/w end of storm and end of RO Increase Lag Time to Peak =time of max raintime of max ro % Reduction of Peak (Max Intensity) ([peak rain-peak r.o]./peak rain

PAGE 252

252 Table P-4. Summary table of mean, standard deviation, median and range for all rain events and r unoff from Charles R. Perry Construction Yard Green roof between July 2007 and July 2008. Runoff Date Day Duration Amt(cm) Max int (cm/hr) Duration RO Amt RO (cm) Max RO (cm/hr) % Retention Total 76.58 55.55 27% Mean 2 hr 1.14 2.77 10hr 0.88 3.54 54% Stdev 2hr 20m 1.53 3.45 6hr 15m 1.84 5.91 36% Median 1 hr 0.46 1.52 9 hr 0.21 1.12 53% Min 5 min 0.03 0.25 0min 0.00 0.00 0% Max 10hr45m 7.34 15.85 22hr50m 12.40 31.70 100%

PAGE 253

APPENDIX Q CUMULATIVE HYDROGRAPH AND SUMMA RY TABLES FOR WINTER 2007-2008, CHARLES R. PERRY GREEN ROOF IN FLORIDA Figure Q-1. Cumulative hydrograph for 12/20/07 to 3/18/08 for CRP green roof in FL. Pink line shows cumulative irrigation and rainfall, green line shows cumulative green roof runoff, and blue line show s individual rain events. Table Q-1. Mean, median and range of rain even t size, hours since prev ious storm, and green roof runoff response for rain events betw een 12/20/07 and 3/18/08 for CRP roof, FL. Hours since previous storm Rain Duration (hrs) Rain (cm) GR Runoff Duration (hr) Runoff (cm) % Retention Total 22.3 14.8 34% Mean 36.0 2 hr 20 min 1.0 10 hr 50 min 1.2 33% StdDev 26.0 2 hr 5 min 1.0 7 hr 35 min 2.2 52% Median 31.8 1 hr 40 min 0.6 11 hr 40 min 0.5 33% Min 3.3 5 min 0.0 0 hr 0.0 0% Max 96.0 7 hr 30 min 4.1 22 hr 30 min 10.5 100% Table Q-2. Mean, median and rang e of delay in start of runoff, extension of runoff, and % reduction of peak intensity for rain events between 12/ 20/07 and 3/18/08 for CRP roof in FL. Delay in Start Run Off (hr) Extension of Run Off (hr) Max. Rain Intensity (cm/hr) Max. Run Off (cm/hr) % Reduction of Peak Intensity Mean 1 hr 10 min 12 hr 1.7 0.5 74% StdDev 1hr 40 min 6 hr 1.6 0.6 24% Median 40 min 14 hr 1.2 0.2 79% Min 0 hr 1 hr 10 min 0.3 0.0 19% Max 6 hr 23 hr 45 min 6.1 1.8 100% 253

PAGE 254

APPENDIX R HYDROLOGIC SUMMARY TABLES FOR SP RING 2008, CHARLES R. PERRY GREEN ROOF IN FLORIDA Table R-1. Characterization of 14 irrigation events and 4 rain events captured between April 11, 2008 to June 7, 2008, for rainfall or irrigation duration and volume, green roof runoff duration and volume, and % volume ra infall or irrigation retention. Type Rainfall or Irrigation. Duration (hr) Rainfall or Irrigation Depth (cm) Duration RO (hr) RO (cm) Rainfall or Irrigation Retention (%) Total/Overall 13.60 1.53 89% Mean 25 min 0.76 2 hr 20 min 0.08 90% Stdev 12 min 0.30 50 min 0.10 11% Median 23 min 0.87 2 hr 30 min 0.05 94% Min 5 min 0.03 30 min 0.00 61% Max 1hr 1.09 3 hr 45 min 0.34 100% Table R-2. Results of analysis of 14 irrigation events and 4 rain events between April 11, 2008 and June 7, 2008 for % decrease in peak inte nsity, increase in lag to peak, delay to start of runoff and extension of runoff duration past the end of storm event. Type Delay in Start (min) Extension (hr) Max Rain/Irrig. Int. (cm/hr) Max RO (cm/hr) % Reduction of Peak Mean 35 min 2 hr 15 min 0.08 0.31 0.84 Stdev 14 min 50 min 0.10 0.36 0.18 Median 25 min 2 hr 30 min 0.05 0.12 0.94 Min 5 min 35 min 0.00 0.00 0.37 Max 50 min 3 hr 35 min 0.34 1.32 1.00 254

PAGE 255

APPENDIX S CUMULATIVE HYDROGRAPH AND HYDROLOGIC SUMMARY TABLES FOR SUMMER 2008, CRP GREEN ROOF IN FLORIDA Figure S-1. Cumulative green roof runoff compared to cumulative rainfall and irrigation for wet season months (June 7th to September 15th) in 2008, hyetograph shown in blue. Table S-1. Characterization of 17 rain events captured between July 26, 2008 to September 15, 2008, in terms of rainfall duration, rainfall depth (cm), green roof runoff duration and depth and percent rainfall retention. Type Rain Duration (hr) Rain (cm) Duration RO (hr) Runoff (cm) % Retention Total Rain/RO 20.71 15.58 25% Mean 2 hr 1.22 11 hr 0.92 53% StdDev 2 hr 45 min 2.18 8 hr 45 min 1.84 38% Median 35 min 0.58 6 hr 50 min 0.17 64% Min 5 min 0.03 0 hr 0.00 0% Max 9 hr 15 min 9.09 32hr15m 7.44 100% Table S-2. Mean, median and rang e of delay in start and extension in end of runoff, reduction in peak intensity for 17 rain events between July 26, 2008 and September 15, 2008. Delay in Start (min) Extension (hr) Max Rain Int (cm/hr) Max RO (cm/hr) % Reduction of Peak Mean 40 min 12 hr 1.26 0.18 65% StdDev 40 min 12 hr 0.83 0.18 87% Median 60 min 7 hr 25 min 0.75 0.45 21% Min 20 min 14 hr 30 min 0.62 0.06 96% Max 0 min 1 hr 10 min 0.00 0.00 31% 255

PAGE 256

APPENDIX T CUMULATIVE HYDROGRAPH AND HYDROLOGIC SUMMARY TABLES FOR FALL/WINTER 2006-2007, YSC GREEN ROOF IN VIRGINIA Figure T-1. Cumulative hydrograph of green roof runoff versus cumulative precipitation for YSC, Merrifield, VA for fall/early winter 2006 2007 (October 24, 2006 to February 7, 2007). Table T-1. Characterization of 36 rain events captured betw een November 1, 2006 and March 8, 2007 (Late Fall/Winter) in Merr ifield, Virginia from the YSC green roof, for rain duration, rain amount (cm), maximum inte nsity, green roof runoff duration, and maximum runoff intensity. Type Duration (hr) Rain (cm) Max int (cm hr-1) Duration RO (hr) Amt RO(cm ) Max RO (cm hr-1) Total 242.98 31.41 27.03 177.06 8.90 Mean 6.75 0.90 0.77 9.32 0.24 0.08 Stdev 6.42 1.09 1.10 5.65 0.45 0.11 Median 4.96 0.33 0.30 8.92 0.00 0.02 Min 0.05 0.05 0.00 0.60 0.00 0.00 Max 23.16 4.47 4.27 20.25 2.14 0.41 Table T-2. Mean, median and range of stormwater retention (%), delay in start and extension in end of runoff, reduction in peak intensity for 36 rain events captured between November 1, 2006 and March 8, 2007. Type %Retention Delay in Start RO (hr) Ext end RO(hr) Increase in Lag to Peak (hr) % Reduction of Peak Mean 83% 3.41 5.08 3.78 96% Stdev 23% 3.19 4.77 3.25 6% Median 98% 3.09 3.36 2.70 100% Min 19% 0.00 0.00 0.00 73% 256

PAGE 257

Max 100% 14.16 18.50 9.58 100% 257

PAGE 258

APPENDIX U CUMULATIVE HYDROGRAPH AND HYDROLOGIC SUMMARY TABLES FOR SUMMER 2007, YSC GREEN ROOF IN VIRGINIA Figure U-1. Cumulative hydrogra ph of rainfall and runoff from the YSC green roof in Merrifield, VA for June and August 2007 (Jul y data was unavailable due to technical difficulties). Table U-1. Characterization of 12 rain events captured in June 2007 and August 2007 (summer) in Merrifield, Virginia from the YSC green roof, for rain duration (hr), rain amount (cm), maximum intensity (cm hr-1), green roof runoff duration (hr), and maximum runoff intensity (cm hr-1). Type Duration (hr) Rain (cm) Max int (cm hr-1) Duration RO (hr) Amt RO(cm) Max RO (cm hr-1) Total 67.20 11.14 49.02 2.04 Mean 6.11 0.93 1.81 6.13 0.19 0.18 Stdev 7.66 0.85 2.57 3.46 0.42 0.07 Median 2.58 0.94 0.61 5.00 0.05 0.18 Min 0.08 0.05 0.04 3.42 0.00 0.04 Max 23.16 2.59 9.14 13.70 1.43 0.24 Table U-2. Mean, median and range of stormwater retention (%), delay in start and extension in end of runoff, reduction in peak intensity fo r 12 rain events collected in June and August 2007. Type %Retention Delay in Start RO (hr) Ext end RO(hr) Increase in Lag to Peak (hr) % Reduction of Peak Mean 90% 3.55 3.94 2.14 89% Stdev 16% 3.51 2.71 1.97 21% Median 94% 4.00 3.60 1.50 100% Min 45% 0.33 0.10 0.50 33% Max 100% 10.00 8.50 6.25 100% 258

PAGE 259

APPENDIX V CUMULATIVE HYDROGRAPH AND HYDROL OGIC SUMMARY TABLES FOR SPRING 2008, YSC GREEN ROOF IN VIRGINIA Figure V-1. Cumulative hydrograph of green roof runoff and rainfa ll for April and May 2008 for YSC green roof in Merrifield, VA. Table V-1. Characterization of 10 rain events cap tured in April and May 2008 (Spring) in Merrifield, Virginia from the YSC green roof, for rain durati on (hr), rain amount (cm), maximum intensity (cm hr-1), green roof runoff duration (hr), and maximum runoff intensity (cm hr-1). Type Duration (hr) Rain (cm) Max int (cm hr-1) Duration RO (hr) Amt RO (cm) Max RO (cm hr-1) Amt retained(cm) Total 223.0 46.33 16.01 30.13 Mean 7.44 1.49 1.05 9.34 0.44 0.17 0.84 Stdev 6.34 1.78 1.17 5.53 0.78 0.38 0.99 Median 5.40 0.89 0.91 8.92 0.00 0.00 0.43 Min 0.17 0.03 0.00 0.60 0.00 0.00 0.00 Max 23.25 7.65 4.87 21.17 3.48 2.08 4.16 Table V-2. Mean, median and range of stormwater retention (%), delay in start and extension in end of runoff, reduction in peak intensity for 10 rain events collected in April and May of 2008. Type %retention Delay in Start RO (hr) Ext end RO(hr) Increase in Lag to Peak (hr) % Reduction of Peak Mean 85% 3.85 2.78 2.79 92% Stdev 19% 2.37 1.66 2.93 13% Median 99% 3.30 2.90 2.04 100% Min 39% 0.15 0.90 0.25 51% Max 100% 10.00 7.30 8.90 100% 259

PAGE 260

APPENDIX W CUMULATIVE HYDROGRAPH AND HYDR OLOGIC SUMMARY TABLES FOR LATE SUMMER 2008, YSC GREEN ROOF, IN VIRGINIA Figure W-1. Cumulative hydrograph of rainfall and green roof runoff for August-September 2008 for YSC greenroof in Merrifield, VA. Table W-1. Characterization of 10 rain even ts captured in August and September 2008 (Lat Summer) in Merrifield, Virginia from the YS C green roof, for rain duration (hr), rain amount (cm), maximum intensity (cm hr-1), green roof runoff duration (hr), and maximum runoff intensity (cm hr-1). Duration (hr) Rain (cm) Max int (cm/hr) Duration RO (hr) Amt RO (cm) Max RO Type (cm hr-1) Total 42.39 20.85 26.15 6.72 Mean 8.48 4.17 0.08 4.57 8.72 1.34 Stdev 8.13 7.48 0.04 2.64 4.79 2.71 Median 10.90 1.47 0.05 3.05 6.00 0.05 Min 0.08 0.05 0.05 3.05 5.90 0.00 Max 18.33 17.47 0.15 9.14 14.25 6.19 Table W-2. Mean, median and rang e of stormwater retent ion (%), delay in start and extension in end of runoff, reduction in p eak intensity for 12 rain even ts collected in August and September of 2008. Type %Retention Delay in Start RO (hr) Ext end RO(hr) Increase in Lag to Peak (hr) % Reduction of Peak Mean 87% 6.28 0.60 0.18 91% Stdev 17% 5.83 0.89 0.11 12% Median 97% 6.00 0.00 0.18 99% Min 65% 0.60 0.00 0.10 74% Max 100% 12.25 2.00 0.25 100% 260

PAGE 261

APPENDIX X COMPARISON OF STORM DURATION/ SIZE, INTENSITY, STORM WATER RETENTION, AND REDUCTION IN RUNOFF INTENSITY FOR VA AND FL Table X-1. Comparison of length of Virginia and Florida stor ms sampled during study period. Virginia Storm Length (hr) Florida Storm Length (hr) Range 0.05 to 27.5 hr 0.08 to 10.7 hr 1st Quartile 1.2 hr Median 4.9 hr 1.0 hr 3rd Quartile 12.1 hr Mean 7.04 6.9 hr 1.9 2.8 hr Table X-2. Comparison of storm size of Virginia and Florida storms sampled. Virginia Storm Size (cm) Florida Storm Size (cm) Range 0.0254 to 17.47 cm 0.0254 to 9.09 cm 1st Quartile 0.102 cm 0.178 cm Median 0.635 cm 0.445 cm 3rd Quartile 1.6 cm 1.295 cm Mean 1.4 cm 2.5 cm 1.09 1.59 cm Table X-3. Comparison of average rain intens ity of Virginia and Fl orida storms sampled. Virginia Rain Intensity (cm hr-1) Florida Rain Intensity (cm hr-1) Range 0.0032 to 4.1 cm hr-1 0.022 to 7.3 cm hr-1 1st Quartile 0.070 cm hr-1 0.23 cm hr-1 Median 0.17 cm hr-1 0.47 cm hr-1 3rd Quartile 0.58 cm hr-1 1.03 cm hr-1 Mean 0.59 1.4 cm hr-1 0.86 1.45 cm hr-1 Table X-4. Comparison of maximu m rain intensity of Virginia and Florida storms sampled. Virginia Max. Intensity (cm hr-1) Florida Max. Intensity (cm hr-1) Range 0 to 9.14 cm hr-1 0 to 15.8 cm hr-1 1st Quartile 0.13 cm hr-1 0.38 cm hr-1 Median 0.61 cm hr-1 1.2 cm hr-1 3rd Quartile 1.6 cm hr-1 2.43 cm hr-1 Mean 1.39 1.9 cm hr-1 2.32 3.1 cm hr-1 Table X-5. Comparison of storm water retenti on of Virginia and Florida storms sampled. Virginia Florida Range 0 to 100% 0 to 100% 1st Quartile 74% 15% Median 96% 55% 3rd Quartile 100% 94% Mean 86% 19 % 49% 43% Table X-6. Comparison of max. stormwater runoff intensity fr om green roofs in VA and FL. Virginia R.O. Max. Intensity (cm hr-1) Florida R.O. Max. Intensity (cm hr-1) Range 0 to 2.4 cm hr-1 0 to 8.6 cm hr-1 1st Quartile 0.0 cm hr-1 0.06 cm hr-1 Median 0.07 cm hr-1 0.13 cm hr-1 3rd Quartile 0.23 cm hr-1 0.69 cm hr-1 Mean 0.17 0.33 cm hr-1 0.74 1.4 cm hr-1 261

PAGE 262

APPENDIX Y SUMMARY OF HYDROLOGIC CHARACTE RISTICS OF RAIN EVENTS AND CORRESPONDING GREEN ROOF RUNO FF FOR WATER QUALITY SAMPLING EVENTS FOR CRP GREEN ROOF IN GAINESVILLE, FLORIDA, 2008 Table Y-1. Characteristics of ra in events sampledstorm start and finish times, duration, total rain (cm), mean intensity, maximum intensity and lag to peak (min). Date Rain Start Rain End Duration (hr) Total Rain (cm) Rain Mean Intensity (cm hr-1) Rain Max Int (cm hr-1) Lag to Peak (min) 6/23/2008 10:40 AM 12:45 PM 2.0 0.58 0.39 0.61 1 hr 6/25/2008 2:05 PM 2:25 PM 0.3 0.30 0.91 1.22 15 min 6/26/2008 5:05 PM 7:35 PM 2.5 0.38 0.16 0.30 20 min 6/30/2008 2:50 PM 3:30 PM 0.7 0.36 0.53 1.83 15 min 7/8/2008 4:40 PM 7:50 PM 3.2 4.67 1.48 6.71 30 min 8/21-8/22/08 12:40 PM 7:00p8/22 30.0 10.20 0.34 2.13 4hr 25m Table Y-2. Characteristics of gr een roof runoff for rain events sampled for water quality in 2008Start/Finish times of green roof r unoff, runoff durati on, runoff volume (cm depth across roof surface), mean runoff rate (cm hr-1), maximum runoff rate (cm hr-1) and lag to peak (min). Date Runoff Start Time Runoff End Time Runoff Duration Runoff Volume (cm) Mean R.O. rate (cm hr-1) Max RO (cm hr1) Lag to Peak (min) 6/23/2008 11:30 AM 5:40 PM 5.2 0.194 0.038 0.06 1 hr 20m 6/25/2008 2:40 PM 5:10 PM 2.5 0.042 0.0168 0.06 1 hr 30m 6/26/2008 5:40 PM 11:20 PM 6.1 0.089 0.0146 0.12 2 hr 40m 6/30/2008 3:15 PM 11:05 PM 7.25 0.1 0.0138 0.12 n/a 7/8/2008 4:55 PM 11:30 PM 6.6 4.33 0.6561 6.3 40 min 8/21-8/22/08 1:00 PM 1:40 P 8/23 48 7.43 0.1548 1.908 4 hr 40m Table Y-3. Summary of re duction in maximum runoff rate, increas e in lag to peak, delay in start of runoff, extension of runoff past the e nd of storm and % storm water retention by the green roof for storms used for water quality sampling in 2008. Date Reduction in Max. R.O. Rate (%) Increase in Lag to Peak (min) Delay to Start of Runoff (min) Increase in duration of Return Flow (hr) Retention (%) 6/23/2008 90% 20 min 50 min 4 hr 55min 67% 6/25/2008 80% 1hr 15 min 1 hr 30 min 2 hr 45min 86% 6/26/2008 69% 1 hr 20 min 35 min 5 hr 45min 77% 6/30/2008 93% No peak 25 min 7 hr 35 min 72% 7/8/2008 6% 10 min 15 min 3 hr 40 min 9% 8/21-8/22/08 10% 15 min 20 min 18 hr 27% 262

PAGE 263

APPENDIX Z INDIVIDUAL STORM HYDROGRAPHS FO R WATER QUALITY SAMPLING EVENTS, CRP GREEN ROOF, FL, 2008 6/23/080 0.02 0.04 0.06 0.08 0.1 0.12 0.14 0.16 0.18 9:36 AM10:40 AM11:45 AM12:50 PM1:55 PM3:00 PM4:04 PM5:09 PM Time(cm) Rainfall subtot (20 min) Runoff (20 min. int) Runoff (cm) Figure Z-1. Storm hydrograph for rain event on 6/ 23/08. Green dots indicate when grab samples were taken from the green roof runoff. 6/25/080.000 0.050 0.100 0.150 0.200 0.250 12:57 PM2:09 PM3:21 PM4:33 PM5:45 PM6:57 PM Green Roof Runoff (cm) Rain (cm) Runoff (cm) Figure Z-2. Storm hydrograph for rain event sa mpled on 6/25/08. Pink dots indicate when grab samples were taken from the green roof runoff. 263

PAGE 264

6/26/080 0.01 0.02 0.03 0.04 0.05 0.06 0.07 0.08 0.09 2:24 PM 3:36 PM 4:48 PM 6:00 PM 7:12 PM 8:24 PM 9:36 PM 10:48 PM 12:00 AM 1:12 AM TimeRunoff (cm) Rain (cm) Figure Z-3. Storm hydrograph for rain event sa mpled on 6/26/08. Pink dots indicate when grab samples were taken from the green roof runoff. 6/30/08 0 0.02 0.04 0.06 0.08 0.1 0.12 0.14 0.16 0.18 0.2 1:12 PM 2:24 PM 3:36 PM 4:48 PM 6:00 PM 7:12 PM 8:24 PM 9:36 PM 10:48 PM 12:00 AM 1:12 AM TimeRunoff (cm) Green roof runoff (cm) Rain (cm) Figure Z-4. Storm hydrograph for rain event sa mpled on 6/30/08. Pink dots indicate when grab samples were taken from the green roof runoff. 264

PAGE 265

7/8/080 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 3:36 PM4:48 PM6:00 PM7:12 PM8:24 PM9:36 PM10:48 PM12:00 AM TimeRunoff (cm) Green Roof Runoff (cm) Rain as Runoff (Runoff coefficient of 1) Figure Z-5. Storm hydrograph for rain event sa mpled on 7/8/08. Pink dots indicate when grab samples were taken from the green roof runoff; blue dots indica te timing of grab samples from conventional roof runoff. 8/21-8/22/080 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.88/21/08 9:36 AM 8/21/08 2:24 PM 8/21/08 7:12 PM 8/22/08 12:00 AM 8/22/08 4:48 AM 8/22/08 9:36 AM 8/22/08 2:24 PMRunoff (cm) Rainfall as runoff (cm) Green roof runoff (cm) Figure Z-6. Storm hydrograph of r unoff from the green roof and a conventional roof of the same size based on rainfall and a runoff coefficient of 1. Pink dots indicate when grab samples were taken from the green roof r unoff, blue dots refer to the moments of conventional roof runoff sampling. 265

PAGE 266

APPENDIX AA WATER QUALITY DATA BY TIME INTERVAL FOR RAIN EVENTS SAMPLED IN FLORIDA AND VIRGINIA Table AA-1. Concentrations of nitrate, ammoni um, TSS and SRP from si x rain events in 2008. Sampling Date Time NOx-N (mg/L) NH3-N (mg/l) TSS (mg/L) SRP (mg/L) 6/23/2008 11:10 AM 11:20 AM 0.04 0.04 2.3 0.85 11:43 AM 0.03 0.05 2.6 0.59 12:46 PM 0.03 0.05 2.2 0.84 4:00 PM 0.03 0.06 2.5 0.88 6/25/2008 2:20 PM 0.33 0.10 1.50 0.40 4:00 PM 0.04 0.07 1.00 0.72 6/26/2008 6:56 PM 0.06 0.09 5.50 0.51 9:12 PM 0.03 0.07 1.00 0.59 10:30 PM 0.18 0.23 4.50 0.57 6/30/2008 4:20 PM 0.03 0.19 10.50 0.47 6:00 PM 0.04 0.26 5.00 0.65 7/8/2008 cr 5:15 PM 0.03 0.06 2.50 0.00 6:25 PM 0.30 0.14 0.00 0.04 9:20 PM 0.29 0.13 0.00 0.04 7/8/2008 gr 5:21 PM 0.08 1.36 7.50 1.26 6:35 PM 0.04 0.96 5.00 1.43 7:35 PM 0.05 1.20 11.50 1.29 9:15 PM 0.05 0.92 14.00 1.44 11:30 PM 0.06 0.78 9.50 1.43 8/21/2008 cr 4:25 PM 0.03 0.12 0.11 0.03 8:56 PM 0.05 0.09 0.00 0.02 10:45 PM 0.10 0.26 0.05 0.03 11:56 AM 0.19 0.14 0.07 0.03 4:00 PM 0.25 0.19 0.03 8/21/2008 gr 4:35 PM 0.02 0.15 0.10 0.46 8:40 PM 0.01 0.09 0.14 0.25 10:48 PM 0.01 0.09 0.11 0.81 12:00 PM 0.01 0.06 0.07 0.46 4:10 PM 0.01 0.06 0.46 266

PAGE 267

Table AA-2. Concentrations of nutrients, pH, TSS and TDS from rain events in 2007-08 in VA. Date Time N-NO3 (mg/L) N-NO2 TP OP pH TSS (mg/L) TDS (mg/L) 4/11/2007 gr 11:00 AM <0.10 <0.10 0.12 4/11/2007 gr 3:00 PM <0.10 <0.10 0.15 4/11/2007 cr 11:00 AM <0.10 <0.10 0.00 4/11/2007 cr 3:00 PM <0.10 <0.10 0.01 6/3/2007 gr 9:45 AM <0.10 <0.1 1.29 1.20 <1 181 6/3/2007 gr 1:00 PM <0.10 <0.10 1.33 1.21 <1 183 6/3/2007 gr 3:10 PM <0.10 <0.10 1.31 1.20 <1 184 6/3/2007 gr 5:20 PM <0.10 <0.10 1.30 1.23 <1 184 6/3/2007 gr 9:14 PM <0.10 <0.10 1.28 1.10 1 191 6/3/2007 cr 9:45 AM <0.10 <0.10 0.2 0.15 <1 25 6/3/2007 cr 1:00 PM <0.10 <0.10 0.2 0.16 <1 23 6/3/2007 cr 3:10 PM <0.10 <0.10 0.2 0.16 <1 26 6/3/2007 cr 5:20 PM <0.10 <0.10 0.19 0.14 <1 24 6/3/2007 cr 9:14 PM <0.10 <0.10 0.19 0.15 <1 23 4/20/2008 gr 9:30 AM <0.10 0.87 0.81 6.3 8 52 4/20/2008 gr 11:45 AM <0.10 0.78 0.72 6.8 7 55 4/20/2008 gr 13:35 PM <0.10 0.61 0.55 7.1 2 57 4/20/2008 gr 16:15 PM <0.10 0.47 0.42 7.3 <1 59 4/20/2008 cr 9:30 AM 0.34 <0.10 <0.10 4.9 <1 8 4/20/2008 cr 11:45 AM 0.16 <0.10 <0.10 4.6 9 9 4/20/2008 cr 13:35 PM <0.10 <0.10 <0.10 4.8 6 14 4/20/2008 cr 16:15 PM 0.33 <0.10 <0.10 4.6 15 <1 8/29/2008 7:27 <0.11 1.26 1.01 7.7 Insuff 84 8/29/2008 9:40 <0.10 1.49 1.24 7.8 Insuff 174 8/29/2008 10:30 <0.10 1.39 1.24 7.7 Insuff 167 8/29/2008 11:44 <0.10 1.51 1.27 7.8 1 174 8/29/2008 14:29 <0.10 1.51 1.21 7.9 Insuff 167 8/29/2008 7:37 0.13 <0.10 <0.10 3.4 Insuff 14 8/29/2008 9:28 0.13 <0.10 <0.10 5.1 Insuff 9 8/29/2008 10:36 0.21 <0.10 <0.10 4.7 2 14 8/29/2008 11:38 0.24 <0.10 <0.10 4.6 5 15 8/29/2008 14:29 0.31 <0.10 <0.10 4.8 Insuff 11 9/6/2008 9:34 <0.10 1.3 1.11 7.2 4 28 9/6/2008 11:34 <0.10 0.57 0.5 7 3 31 9/6/2008 15:29 <0.10 0.52 0.46 7.1 2 22 9/6/2008 9:30 <0.10 <0.10 <0.10 5.3 <1 3 9/6/2008 11:40 <0.10 <0.10 <0.10 5.3 <1 5 9/6/2008 15:24 <0.10 <0.10 <0.10 5.3 <1 5 267

PAGE 268

Table AA-3. Concentrations of metals sampled in 5 storms in Virginia from YSC green roof and conventional roof runoff in 2007 and 2008. Date Roof Type Cu (mg/L) Fe (mg/L) Pb (mg/L) Cd (mg/L) Zn (mg/L) Al (mg/L) 4/11/2007 gr 11:00 AM 0.086 0 0.041 0 4/11/2007 gr 3:00 PM 0.095 0 0.045 0 4/11/2007 cr 11:00 AM 0.005 0 0.040 0 4/11/2007 cr 3:00 PM 0.023 0 0.044 0 6/3/2007 gr 9:45 AM <0.10 0.12 0 6/3/2007 gr 1:00 PM <0.10 0.11 0 6/3/2007 gr 3:10 PM <0.10 0.11 0 6/3/2007 gr 5:20 PM <0.10 0.10 0 6/3/2007 gr 9:14 PM <0.10 0.11 0 6/3/2007 cr 9:45 AM <0.10 0.20 0 6/3/2007 cr 1:00 PM <0.10 0.19 0 6/3/2007 cr 3:10 PM <0.10 0.21 0 6/3/2007 cr 5:20 PM <0.10 0.23 0 6/3/2007 cr 9:14 PM <0.10 0.20 0 4/20/2008 9:30 AM 0.44 <0.10 0.003 4/20/2008 11:45 AM <0.10 <0.10 <0.002 4/20/2008 13:35 PM <0.10 <0.10 <0.002 4/20/2008 16:15 PM <0.10 <0.10 <0.002 4/20/2008 9:30 AM <0.10 <0.10 0.012 4/20/2008 11:45 AM <0.10 <0.10 0.011 4/20/2008 13:35 PM <0.10 <0.10 0.037 4/20/2008 16:15 PM <0.10 <0.10 0.042 8/29/2008 7:27 0.64 0.36 0.006 insuff 8/29/2008 9:40 0.18 0.34 <0.002 0.001 8/29/2008 10:30 0.12 0.32 <0.002 insuff 8/29/2008 11:44 0.15 0.32 0.005 0.001 8/29/2008 14:29 0.11 0.29 0.002 Insuff 8/29/2008 7:37 0.61 0.14 insuff. Insuff 8/29/2008 9:28 <0.10 0.14 insuff. Insuff 8/29/2008 10:36 <0.10 0.29 0.041 0 8/29/2008 11:38 <0.10 0.26 0.088 0.001 8/29/2008 14:29 <0.10 0.22 0.088 0 9/6/2008 9:34 <0.10 0.36 <0.002 0 9/6/2008 11:34 <0.10 insuff 0.002 Insuff 9/6/2008 15:29 <0.10 <0.10 <0.002 0 9/6/2008 9:30 <0.10 <0.10 0.014 0 9/6/2008 11:40 <0.10 <0.10 0.003 Insuff 9/6/2008 15:24 <0.10 <0.10 0.005 0 268

PAGE 269

LIST OF REFERENCES Andoh, R.Y.G and Declerck. 1997. A cost effectiv e approach to stormwater management? Source control and distributed stor age. Wat. Sci. Tech.36:8:307-311. Berndtsson, J.C, T. Emilsson, and L. Bengtss on. 2006. The influence of extensive vegetated roofs on runoff water quality. Sci. of the Tot. Env. 355:48-63. Berghage, R., A. Jarrett, D. Beattie, K. Kelley, S. Husain, F. Rezai, B. Long, A. Regassi, R. Cameron, and W. Hunt. 2007. Quantifying evapor ation and Transpirational Water Losses from green roofs and green roof media capac ity for neutralizing acid rain. NDWRCDP, EPA Agreement No. X-830851. Britton, N.L., and A. Brown. 1913. Illustrated flora of the northern states and Canada. Vol. 2.41. Courtesy of Kentucky Native Plant Society. Brown, S. 1981. A comparison of the structure, primary productivity, an d transpiratioin of cypress ecosystems in Florida. Ecological Monographs. 51:4:403-427. Butson, K. 1959. Some aspects of seasonal distribu tion of rainfall in Florida. Florida State Horticultural Society, p.171-176. Calderon, S., N.D. Poor, and S. Campbell. 2007. Es timation of the particle and gas scavenging contributions to wet deposition of or ganic nitrogen. Atmosph.Env. 41:4281-4290. Carter, T.L. and T.C. Rassmussen. 2005. Use of green roofs for ultra-urban stream restoration in the Georgia Piedmont (USA). In Proc. of the 2005 Georgia Wat. Res. Conf., April 25-27, 2005, University of Georgia, Athens GA 30602. Caruso, B. 2000. Integrated assessment of phosphorus in the Lake Hayes catchment, South Island, New Zealand. Journ. Of Hydrology. 229:168-189 Deutsch, B., H. Whitlow, M. Sullivan, A. Savi neau, and B. Busiek. 2007. The Green Build-out Model: Quantifying the Stormwater Management Benefits of Trees and Green Roofs in Washington, DC. Casey Trees and Lim noTech. EPA Cooperative Agreement CP83282101-0. April 19, 2007. 111p. Clark, S., R. Field and R. Pitt. 2001. Wet-weat her pollution prevention by product substitution. In Proc. of ASCE, UEF, EWRI Conf.: Linking Stormwater BMP Designs and Performance to Receiving Water Impact Mitigation, Aspen, Colorado. 2001. De Nardo, J.C., A.R. Jarrett, H.B. Manbeck, D. J. Beattie, and R.D.Berghage. 2005. Green Roof mitigation of stormwater and energy usage. In Transactions of the ASAE. 48:4:14911496. Del Barrio, E.P. 1998. Analysis of the green roof s cooling potential in buildings. Energy and Build. 27:2:179-193. 269

PAGE 270

Dietz, M. and J. Clausen. 2005. A field evaluation of rain garden flow and pollutant treatment. Wat., Air, and Soil Poll. 167:123-138. Douglass, J.E. and W.T. Swank. 1972. Streamflow Modification through Management of Eastern Forests. USDA For. Ser. Res. Paper. SE-94. Dunne, T. and L.B. Leopold. 1978. Wate r in environmental planning. 14th printing 1996. France, R.L. 2002. Handbook of Water Sens itive Planning and Design.CRC Press. French, E.C., G.M. Prine, and A.R. Blount. 2006 Perennial Peanut: An Alternative Forage of Growing Importance. Univ. Florida, IFAS, Coop. Ext. Ser. SS-AGR-39. Freire, M.J., C.A. Kelly Begazo, and K.H. Quesenberry. 2000. Establishment, yield and competitveness of rhizome perennial peanut germplasm on a flatwoods soil. Soil and Crop Sci. Soc. of Flor. Proc. 59:68-72. Graham, P, L. Maclean, D. Medin, P. Avinash and G. Vasarhelyi. 2004. The role of water balance modeling in the transition to Low Im pact Development. Wat. Qual. Research Journ. 39:4:331-342. Happe, D. 2005. Green roofs are sprouting up. Journ. of Soil and Wat. Conserv. 60:110. Irmak, S., D. Z. Haman and J.W. Jones. 2002. Eval uation of class A pan ev aporation coefficients for estimating reference evapotranspiration in a humid location. Journal of Irrig. and Drain. Eng. May/June 2002:153-159. Keeley, J.E. and P.W. Runde l. 2003. Evolution of CAM and C4 Carbon-Concentrating Mechanisms. Int. J. Plant. Sci. 164:S55-S77. Kurvits, A. and E. A. Kirkby. 1979. The growth and mineral composition of sunflower plants, Helianthus annus utiliazing nitrateor ammoniumn itrogen when grown in a continuous growing culture system. Department of Plant Sciences, The University of Leeds, Leeds, Yokshire. Journal of Plant Nutrition and Soil Sciences. Liptan, T. and E. Strecker. 2003. EcoRoofsA more sustainable infrastructure. p. 113-120. In Proc. of 1st North American Green Roof Conferen ce: Greening Rooftops for Sustainable Communities. Chicago. 29-30 May 2003. The Cardinal Group, Toronto. Lundholm, J.T. and S. Peck. 2008. Introduction: Front iers of green roof ecology. Urban Ecosyst. 11:335-337. Lundholm, J.T. and D. Wolf. 2008. Water uptake in green roof microcos ms: Effects of plant species and water availa bility. Ecol. Eng. 33:179-186. Lttge, U. 2004. Ecophysiology of Crassulacean Acid Metabolism (CAM). Annals of Botany 93:629-652. 270

PAGE 271

Luxmore, D., M. Jayasinghe and M. Mahendra n. 2005. Mitigating temperat ure increases in high lot density sub-tropical residential developments. Energy and Build. 37:1212-1224. Mentens, J., D. Raes and N. Hermy. 2003. Effect of orientation on the water balance of green roofs. p. 363-371. In Proc. of 1st North American Green Roof Conf.: Greening Rooftops for Sustainable Communities.29-30 May 2003. Chicago, IL. The Cardinal Group, Toronto. Mentens, J., D. Raes and N. Hermy. 2006. Green roofs as a tool for solving the rainwater runoff problem in the urbanized 21st century? Landscape and Urb. Plan. 77:217-226. Miller, C. 2003. Moisture management in green roofs. p177-182. In Proc. of 1st North American Green Roof Conf.: Greening Rooftops for Sustainable Communities.29-30 May 2003. Chicago, IL. The Cardinal Group, Toronto. Moran, A.C., W.F. Hunt and J.T. Smith. 2005. Green roof hydrologic and water quality performance from two field si tes in North Carolina. p1175-1186. In Proc. of the 2005 Watershed Management Conf.Managing Wate rsheds for Human and Natural Impacts. ASCE, Reston, VA. NOAA. 2007. 30-year rainfall record ( 1973-2000) for Gainesville, Florida. www.ncdc.noaa.gov/oa/climate/online/ccd/nrmpcp.txt Date last accessed (May 2009). Norcini, J.G. and J.H. Aldrich. 2007. Native Wildflowers: Mimosa strigillosa Torr.&A.Gray. Univ. Florida, IFAS, Coop. Ext. Ser. ENH1075. Peel, M. C., B.L. Finlayson and T.A. McMahon. 2007. Updated world map of the KppenGeiger climate classification Hydrol. Earth Syst. Sci. 11:1633-1644. Pickett, J. P. 2000. The American Heritage Dictionary of the English Language: Fourth Edition. Boston: Houghton Miffl in Company. 2,074 p Poor, N., C. Pollman, P. Tate, M. Begum, M. Evans and S. Campbell. 2006. Nature and magnitude of atmospheric fluxes of total i norganic nitrogen and ot her inorganic species to the Tampa Bay Watershed, FL, USA. Wat., Air, and Soil Poll. 170: 267. Rao, D. V. 1988. Rainfall analysies for northeast Florida. Part VI: 24hour to 96-hour maximum rainfall for return periods of 10 years, 25 years and 100 years. In Tech. Publ. SJ 88-3. Div. of Eng. Department of Wat. Res. St Johns River Water Management District. Palatka, Florida. May 1988. Proj. No. 15 200 02/20 200 02. Theodosiou, T.G. 2003. Summer pe riod analysis of the performa nce of a planted roof as a passive cooling technique. Energy and Build. 35:9:909-917. Trimble, S.W. 2007. Encyclopedia of Water Science. Ed. 2. CRC Press. 1370p. Schueler, T. and D. Caraco. 2001. Prospects fo r Low Impact Land Development at Watershed Level. In Proc. of ASCE, UEF, EWRI Conf.: Linking Stormwater BMP Designs and Performance to Receiving Water Impact Mitigation, Aspen, Colorado. 2001. 271

PAGE 272

Simmons, M.T., B. Gardiner, S. Windhager and J. Tinsley. 2008. Urban Ecosys. 11:339-348. Snodgrass, E.C. and L.L. Snodgrass. 2006. Green roof plants: a resource and planting guide. Timber Press. 203p. Stribling, E. and S. Leppo. 2001. Re lating instream biological condi tion to BMPs in Watersheds. In Proc. of ASCE, UEF, EWRI Conf.: Linking Stormwater BMP Designs and Performance to Receiving Water Impact Mitigation, Aspen, Colorado. 2001. Van Woert, N.D., D.B. Rowe, J.A. Andrese n, C.L. Rugh, R.T. Fernandez and L, Xiao. 2005. Green roof stormwater retention: effects of roof surface, slope, and media depth. Journ. of Env. Qual.34:3:1036-1044. Villareal, E.L. and L. Bengtsson. 2005. Response of Sedum green-roof to individual rain events. Eco. Eng. 25:1-7. Villareal, E.L., A. Semadeni-Davies and L. Be ngtsson. 2004. Inner city stormwater control using a combination of best manageme nt practices. Ecol. Eng. 22:279-298. Wong, N.H., D.K.W. Cheong, C.L. Ong and A. Sia. 2003. Inve stigation of ther mal benefits of rooftop garden in the tropical envi ronment. Build. and Environ. 38:261-270. Wong, N.H., S.F. Tay, R. Wong, C.L. Ong, and A. Si a. 2003. Life cycle cost analysis of rooftop gardens in Singapore. Build. And Env. 38:499-509. Zimmer, U. and W.F. Geiger. 1997. Multilayered infiltration system. Wat. Sci. Tech. 36:301306. 272

PAGE 273

BIOGRAPHICAL SKETCH Sylvia Lang was born in Madison, Wisconsin in June. She moved with her family to Michigan when she was 2 and then to Virginia when she was 5, where she lived until she was 23. She entered Jefferson High School for Science and Technology in Northern Virginia the year after it opened, then went to the College of William and Mary and studied environmental geology and Spanish literature. Sylvia worked in an outdoor school in ru ral Virginia directly after college and then moved to Costa Rica for se veral years to perfect her Spanish and work in environmental sciences. She returned to the US to pursue a Masters degree at Colorado State University in watershed scienc es. Her research project for he r masters was The Effect of Logging on Erosion in a Wet Tropi cal Forest, which took her to a remote area of Costa Rica the Osa Peninsula. Sylvia enjoyed the tropics very much and worked as a consultant for a year in El Salvador for the Center of Tropical Agriculture (CATIE) on a Strategic Watershed Management Plan for the Trinational Watershed Rio Lempa after her masters. She spent some time teaching students about watersheds in Northern Virginia and outdoors in Costa Rica again on the Osa Peninsula with American high school students. After spending much time quantifying and identifying problems in Watershed Management she realized increasing urbanization is one factor that is impacting watersheds everywhere in the world, and when she happened upon green roofs as a BMP to help increasing water volumes leaving impervious areas, she decided to make it her dissertation topic to i nvestigate the usefulness of th is BMP in the sub-tropics. 273