<%BANNER%>

Spatiotemporal Throughfall Characterization of Heterogeneous Forest Communities in the Southeastern U.S.


PAGE 1

SPATIOTEMPORAL THROUGHFA LL CHARACTERIZATION OF HETEROGENEOUS FOREST COMMUNITI ES IN THE SOUTHEASTERN U.S. By MALCOLM LEWIS BRYANT A THESIS PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLOR IDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF MASTER OF ENGINEERING UNIVERSITY OF FLORIDA 2002

PAGE 2

Copyright 2002 by MALCOLM LEWIS BRYANT

PAGE 3

ACKNOWLEDGMENTS I wish to thank Dr. Jennifer Jacobs, Dr. Kirk Hatfield, and Dr. Louis Motz for serving on my committee. I extend special appreciation to Dr. Jacobs for her help and guidance throughout this project. I want to thank Shirish Bhat for helping with the data collection and processing and Sudheer Reddy Satti for his technical advice. Thanks go to Hugh Westbury at Fort Benning for his coordination efforts and D. L. Price and M. R. Kress with the U.S. Army Corp of Engineers for the meteorological data. Funding from the Strategic Environmental Research and Development program grant (DACA88-98-R-0010) is gratefully acknowledged. Also, thanks go to the rest of the Water Resources gang: Gerard Ripo, Brent Whitfield, En-Ching Hsu, Aniruddha Guha, and Erik Howard. Finally, I want to thank my wife Kim and son Trevor for their support and understanding. iii

PAGE 4

TABLE OF CONTENTS page ACKNOWLEDGMENTS.................................................................................................iii LIST OF TABLES.............................................................................................................vi LIST OF FIGURES..........................................................................................................vii ABSTRACT.......................................................................................................................ix CHAPTER 1 INTRODUCTION............................................................................................................1 2 STUDY AREA................................................................................................................4 3 INSTRUMENTATION / EXPERIMENT DESIGN.......................................................8 Study Plots......................................................................................................................8 Canopy Cover Data.......................................................................................................10 Climate Data.................................................................................................................10 4 METHODOLOGY........................................................................................................15 5 RESULTS / DISCUSSION...........................................................................................18 Precipitation and Evapotranspiration............................................................................18 Measured Throughfall...................................................................................................18 Gash Model Parameters................................................................................................19 Stemflow.......................................................................................................................20 Model Results using Annual Average Canopy Cover..................................................21 Model Results using Seasonal Canopy Cover..............................................................21 Canopy Density Comparison........................................................................................22 6 WATERSHED SCALE APPLICATION......................................................................36 7 CONCLUSION..............................................................................................................44 APPENDIX A PRECIPITATION VERSUS THROUGHFALL GRAPHS.........................................45 B PRECIPITATION VERSUS STEMFLOW GRAPHS.................................................48 iv

PAGE 5

LIST OF REFERENCES...................................................................................................51 BIOGRAPHICAL SKETCH.............................................................................................54 v

PAGE 6

LIST OF TABLES Table page 1 Land use distribution in the Bonham-1 and Bonham-2 watersheds................................6 2 Plot vegetation and distribution statistics......................................................................12 3 Measured precipitation, throughfall, and derived stemflow for 4/04/01 through 6/11/02...................................................................................................................26 4 Derived canopy specific parameters, climatic variables, and interception components............................................................................................................27 5 Model results using average annual canopy cover values.............................................28 6 Model results using seasonal canopy cover values........................................................29 7 Density comparison results for 2/01/02 to 4/29/02 using seasonal canopy cover.........30 8 Throughfall results using the distributed approach........................................................39 vi

PAGE 7

LIST OF FIGURES Figure page 1 Land use for the Bonham-1 and Bonham-2 study watersheds........................................7 2 Plot locations in the Bonham-1 and Bonham-2 study watersheds.................................14 3 Cumulative precipitation event totals for 4/04/01 through 6/11/02...............................31 4 Precipitation event durations for 4/04/01 through 6/11/02............................................32 5 Canopy cover measurements for the five forest types...................................................33 6 Measured and modeled interception results using average canopy cover.....................34 7 Measured and modeled interception results using seasonal canopy cover....................35 8 Watershed scale results for the Spring season (April 2001 through June 2001)...........40 9 Watershed scale results for the Summer season (July 2001 through September 2001) ................................................................................................................................41 10 Watershed scale results for the Fall season (October 2001 through December 2001) ................................................................................................................................42 11 Watershed scale results for the Winter season (January 2002 through March 2002)..43 A-1 Precipitation versus throughfall graph used to determine canopy storage capacity for the wetland plot......................................................................................................45 A-2 Precipitation versus throughfall graph used to determine canopy storage capacity for the pine plot............................................................................................................46 A-3 Precipitation versus throughfall graph used to determine canopy storage capacity for the pine plantation plot...........................................................................................46 A-4 Precipitation versus throughfall graph used to determine canopy storage capacity for the mixed plot........................................................................................................47 A-5 Precipitation versus throughfall graph used to determine canopy storage capacity for the hardwood plot..................................................................................................47 vii

PAGE 8

B-1 Precipitation versus stemflow graph used to determine trunk storage capacity and precipitation reaching the trunks for the wetland plot...........................................48 B-2 Precipitation versus stemflow graph used to determine trunk storage capacity and precipitation reaching the trunks for the pine plot.................................................49 B-3 Precipitation versus stemflow graph used to determine trunk storage capacity and precipitation reaching the trunks for the pine plantation plot................................49 B-4 Precipitation versus stemflow graph used to determine trunk storage capacity and precipitation reaching the trunks for the hardwood plot........................................50 viii

PAGE 9

Abstract of Thesis Presented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Master of Engineering SPATIOTEMPORAL THROUGHFALL CHARACTERIZATION OF HETEROGENEOUS FOREST COMMUNITIES IN THE SOUTHEASTERN U.S. By Malcolm Lewis Bryant December 2002 Chair: Dr. Jennifer Jacobs Major Department: Civil and Coastal Engineering The spatial and temporal influence of heterogeneous forest communities on interception loss was characterized for the southeastern United States. Throughfall was measured simultaneously in five forest types, i.e., mature pine, 13 year-old pine plantation, wetland, hardwood, and mixed hardwood/pine, that comprise the landscape in Fort Benning, Columbus, GA. The 1995 Gash interception model was determined to be valid for use in all forest types using seasonal canopy cover values and annual average canopy cover values. The model predicted interception loss with an agreement of -8.1 to 10.5% using annual average canopy cover values. Application of seasonal canopy cover values in lieu of annual averages improved accuracy in all cases with an overall range from -7.3 to 4.5%. An investigation of the models performance within forests of varying ix

PAGE 10

canopy cover showed that the model predicts interception with an agreement of -5.7 to 8.2% using seasonal canopy cover values. A watershed scale intercomparison of the influence of forest community and seasonal variation on interception showed that appropriate characterization of forests is necessary when applying the 1995 Gash model over seasonal or shorter duration time periods. Additionally, application of the model at sub-watershed spatial scales showed that significant variation among results can be expected as the extent of the spatial scale is reduced. The field experimentation and water-budget analysis in this study provide insight into the relative net water input into the heterogeneous forest communities that are typical of the southeastern United States. x

PAGE 11

CHAPTER 1 INTRODUCTION Interception losses modify the dynamics of the watershed hydrological and geochemical cycles. They lower the intensity of precipitation and wash solid particles and dissolved carbon from leaves affecting soil-water chemistry and weathering processes. Nutrient loading is impacted by interception losses because microbial activity is affected by soil moisture. Interception losses also affect rainfall-runoff by reducing the water input available for runoff. Depending on the vegetation type, interception losses can account for 10 to 40% of the total incident precipitation (Dingman, 1994). Furthermore, the distribution of vegetation types among heterogeneous landscapes can influence the spatial variability of interception losses (Crockford and Richardson, 2000). Most methods for calculating interception use a running water balance approach. In this type of approach, the total incident precipitation equals the sum of the throughfall, stemflow, evaporation from the trunk, and evaporation from the canopy. Models that use a water balance approach are affected by the spatiotemporal accuracy of the interception loss calculation. The first physically based model for calculating rainfall interception was the Rutter model (cited in Rutter et al., 1971). The Rutter model is a running water balance model in which total evaporation from a wet canopy is calculated on a per storm basis. Rutter et al. (1975) were able to predict throughfall within 10% of the measured values in a pine stand. Gash and Morton (1978) used the Rutter model to predict interception loss from a Scots pine forest within 7% of the measured values (Gash and Morton, 1978). However, the Rutter model underestimated interception by 20% to 32% 1

PAGE 12

2 in an artificially defoliated hardwood stand (Rutter et al., 1975). Gash (1979) introduced a variation of the Rutter model. The 1979 Gash model differs from the Rutter model in that it considers the arithmetic average instead of the actual rainfall and evaporation rates for individual storm events, thus simplifying the model and minimizing the data requirements. In both the Rutter and Gash models, evaporation is calculated per unit area of ground. This formulation is problematic for sparsely vegetated forests. The 1979 Gash model overestimated interception losses by as much as 29 to 44% in a sparse pine forest in Portugal (Valente et al., 1997). The Gash model was revised (Gash et al., 1995) to calculate evaporation per unit area of canopy rather than per unit area of ground. The revised 1979 Gash model, the 1995 Gash model, greatly improves the accuracy of the interception loss predictions in sparse forests. The 1995 Gash model was used by Valente et al. (1997) to predict interception losses for the same pine forest in Portugal to within 3% of the measured values. Using the 1995 Gash model, Dykes (1997) estimated interception losses for a tropical rain forest in Borneo within 5% of measured values. Carlyle-Moses and Price (1999) estimated interception losses on a hardwood stand in Ontario, Canada, to less than 1% of measured values using the 1995 Gash model. Many studies have validated the 1995 Gash interception model for use in homogeneous landscapes (van Dijk and Bruinjnzeel, 2001; Jackson, 2000; Carlyle-Moses and Price, 1999; Nvar et al., 1999; Dykes, 1997; Valente et al., 1997; Dolman, 1987); few studies have incorporated interception models over large heterogeneous landscapes. The goal of this research was to characterize the net water input into the heterogeneous forest communities that are distinctive of the southeast United States. The

PAGE 13

3 1995 Gash interception model was chosen for this study since it has been validated in many different landscapes and it accounts for canopy cover variations. The objectives of this study were to 1) simultaneously calibrate the 1995 Gash interception model for use in the five forest types considered in this study, 2) use the model to predict rainfall interception and intercompare the results across forest types and canopy densities, 3) explore the influence of seasonal canopy characteristics on the models performance, and 4) apply the model at the watershed scale to identify critical differences among available approaches to modeling interception in a heterogeneous landscape.

PAGE 14

CHAPTER 2 STUDY AREA The study was conducted at the Fort Benning military reservation, located in southwest Georgia. Long, hot summers and mild winters characterize the regions climate. Average annual precipitation is about 830 mm with a monthly average of 69 mm. Most of the precipitation occurs in the spring and summer as a result of thunderstorms. Heavy rains are typical during the summer but can occur in any month. Snow accounts for less than 1% of the annual precipitation. The soils in the area are dominated by loamy sand with some sandy loam. Two second-order watersheds, Bonham-1 and Bonham-2, were selected for this study. These watersheds were selected because they contain most of the regions predominant forest types. The Bonham-1 watershed has an area of 762 ha, a minimum elevation of 87.8 m and maximum elevation of 144.2 m, and an average slope of 9.6%. The Bonham-2 watershed has an area of 2,211 ha, a minimum elevation of 90.5 m, a maximum elevation of 159.1 m, and an average slope of 8.6%. The two watersheds have a heterogeneous land cover consisting of either open or forested areas (Figure 1). The open areas are either military, brush, or wildlife openings. The military openings are clear-cut parcels of land dominated by grass and bare soil that are used as military training grounds. The brush openings consist of tall grass and immature crateagus. The wildlife openings are natural openings in the forests that are vegetated primarily by grass. The forested areas include five forest typesmature pine, pine plantation, riparian wetland, hardwood, and mixed hardwood/pine. The mature pine, 4

PAGE 15

5 which consists of loblolly (Pinus taeda) and short leaf pine (Pinus echinata), is the dominant forest type of each watershed. The pine plantation is a 13 year-old long leaf pine (Pinus palustris) stand planted in rows. The wetland vegetation consists mostly of various hardwood trees along with a range of wetland understory vegetation. The hardwood stands are generally mature scrub oak (Quercus berberidifolia). White oak (Quercus alba), short leaf pine, and loblolly pine are the dominant species in the mixed pine/hardwood stands. Table 1 describes the total area and relative contribution of each forest type. The forests are managed using prescribed burning on a three-year cycle. A majority of the land area included in this study was burned within one year of the study period.

PAGE 16

6 Table 1. Land use distribution in the Bonham-1 and Bonham-2 watersheds. Land Cover Watershed Area (m 2 ) Watershed Area (%) Mature Pine 1,293,800 43.7 Pine Plantation 118,700 4.0 Wetland 299,000 10.1 Hardwood 542,600 18.3 Mixed 506,700 17.1 Openings (military, wildlife and brush) 202,900 6.8

PAGE 17

7 Figure 1. Land use for the Bonham-1 and Bonham-2 study watersheds.

PAGE 18

CHAPTER 3 INSTRUMENTATION/EXPERIMENT DESIGN The experiment was conducted from April 4, 2001 through June 11, 2002. Canopy parameters, climatic variables, and interception components were measured throughout this period. The measured data include precipitation, throughfall, stemflow, atmospheric conditions, and canopy cover. The following sections describe the experimental equipment and methods applied to this study. Study Plots A rectangular plot was established in each forest community. Table 2 lists the vegetation distribution and statistics for the all plots. The plots were randomly selected within areas having vegetation that is consistent with the average vegetation density and distribution for the respective forest community (Figure 2). The dimensions of each plot were determined by the size and geometry of each land-use type and ranged from 10 x 40 meters to 30 x 30 meters. Each plot was subdivided into four sampling grids of equal size. Each grid was outfitted with four throughfall collectors and one tipping bucket rain gauge. The throughfall collectors and tipping buckets were randomly placed on the ground within the confines of the grid. To ensure randomness, each instrument was relocated within the grid after data were collected. The throughfall data were collected using eight-inch diameter tipping bucket rain gauges (model RG-100a, RainWise ) and throughfall collectors. The tipping buckets were calibrated to 0.01 inches per tip. Each tipping bucket was equipped with a data logger (HOBO-8000, Onset Inc.) that recorded the number of tips and the date and time 8

PAGE 19

9 of each tip. Clogging of the tipping buckets resulted in the omission of data for brief time periods. To reduce clogging, the installed plastic screen and cotter pin at the throat of the tipping bucket collector was removed and a plastic mesh was placed over the mouth of the collector. The mesh was pushed down into the collector to create a sock-filter arrangement so that the effective collection area of the gauge was not reduced. The throughfall collectors consisted of two-liter plastic bottles with six-inch diameter funnels. The plastic bottles and funnels were supported by stands made of 3/8-inch plywood and six-inch PVC pipe. The PVC pipe was secured to the plywood base using silicon adhesive. Data were collected from the instruments on a bi-weekly basis. A parallel experiment with a shorter duration was conducted to study the effect of varying canopy cover on interception modeling. Two additional plots for both the wetland and mature pine communities were monitored from February 1 through April 29, 2002. Each additional plot had a significantly different canopy cover than the other plots in the respective forest community. The additional plots, referred to as wetland B, wetland C, mature pine B, and mature pine C, were selected based on a visual inspection and a spherical densiometer survey. The instrumentation and methods used for these plots were identical to those described above. Stemflow was measured in the wetland, mature pine, pine plantation, and hardwood plots. For these plots, the dominant tree species were further sub-divided into three classes of diameter at breast height (DBH) and projected crown radius. Stemflow gauges were installed in each plot such that the dominant tree species were sampled and that a representative tree from each DBH and projected crown radius category was included.

PAGE 20

10 The mixed plot stemflow was determined by averaging the pine and hardwood measurements. Stemflow was measured by attaching split plastic tubing to the representative trees in each plot. Nails or staples and silicon sealant were used as the method of attachment. The tubing was wound around each tree in a 360-degree spiral and routed to a tipping bucket (model RG-100a, RainWise ) equipped with a data logger (HOBO-8000, Onset Inc.). Canopy Cover Data Canopy cover was determined by direct measurement with a Model-A spherical densiometer using the method outlined by Lemmon (1956). This method provides a simple and reliable measure of canopy cover (Bunell and Vales, 1990). Readings were taken in the center of each study plot on a bi-weekly basis throughout the study period. Climate Data Precipitation data were measured at a height of 15 centimeters by two tipping bucket rain gauges (model RG-100a, RainWise ). Each tipping bucket gauge was equipped with an 8-inch diameter collector and calibrated to 0.01 inches per tip. Data loggers (HOBO-8000, Onset Inc.) were used to store the date and time of each tip. The tipping buckets were located in a 50 x 150 meter clearing on the boundary between the Bonham-1 and Bonham-2 watersheds. The distance from the precipitation gauges to the throughfall plots ranged from 315 to 900 meters. The clearing is centrally located and is within 900 meters of each land-use plot. Storm totals were calculated as the average recording of the two rain gauges. Data were collected from the instruments on a bi-weekly basis.

PAGE 21

11 Atmospheric measurements were continuously recorded by the Ecosystem Characterization and Monitoring Initiative (ECMI) meteorological monitoring station ME-04 McKenna Mount. The atmospheric data includes air temperature, barometric pressure, relative humidity, solar radiation, wind speed, and precipitation. The station is located at UTM coordinates 706387 Easting and 3583703 Northing. The station is approximately 7 kilometers south of the Bonham-2 watershed. Data were collected in 1-minute intervals and averaged over 30-minute periods. The air temperature and relative humidity data were collected with a temperature and relative humidity probe (model HMP45C, Vaisala) at an installation height of 2 meters. Barometric pressure data were collected with a pressure transmitter (model PT101B, Vaisala) at an installation height of 1.75 meters. Solar radiation data were collected with a pyranometer with a silicon photovoltaic detector (model LI200X, Li-Cor) at an installation height of 2.5 meters. Wind speed data were collected using a four-blade heilcoid propeller sensor (model 05103-5, R. M. Young) at 3 meters. Finally, precipitation data were collected with a tipping bucket rain gauge (model TE525MM, Texas Electronics) mounted at 3 meters.

PAGE 22

Table 2. Plot vegetation and distribution statistics. Plot Tree Type Number of Trees Percent of Total Minimum Height (meters) Maximum Height (meters) Avera g e Height (meters) Average Pro j ected Canopy Radius (meters) Average Diameter (meters) Trees per Hectare Distance From B2 Precipitation Gauge (meters) Wet Sweet Gum 29 60 5.0 31.0 12.88 1.62 0.14 1200 325 Bay 8 17 5.0 16.0 9.75 1.31 0.07 Long L. Pine 4 8 4.0 8.0 6.00 0.90 0.05 Water Oak 4 8 4.0 28.0 11.13 1.23 0.12 Dog Wood 3 6 6.0 14.5 10.50 2.33 0.11 Wet B Sweet Gum 21 46 6.0 28.0 17.60 1.70 0.23 1150 343 Long L. Pine 9 20 14.0 21.0 18.89 1.80 0.23 Dog Wood 6 13 4.0 14.0 7.83 1.82 0.08 Water Oak 6 13 9.0 28.0 16.58 2.33 0.19 Bay 4 9 5.0 13.0 9.13 1.60 0.07 Wet C Sweet Gum 21 54 3.0 25.0 14.98 2.18 0.20 975 395 Long L. Pine 8 21 3.8 5.0 4.25 0.78 0.05 Dog Wood 8 21 3.0 6.5 4.44 1.19 0.06 Water Oak 2 5 9.5 23.0 16.25 1.80 0.18 Pine Loblolly Pine 41 82 4.0 22.0 13.76 2.08 0.21 556 573 Oak 5 10 6.0 15.0 9.80 2.34 0.14 Crateagus 4 8 5.0 9.0 6.50 2.38 0.11 Pine B Loblolly Pine 33 100 4.0 25.5 12.86 2.14 0.20 367 675 12

PAGE 23

13 Plot Tree Type Number of Trees Percent of Total Minimum Height (meters) Maximum Height (meters) Avera g e Height (meters) Average Pro j ected Canopy Radius (meters) Average Diameter (meters) Trees per Hectare Distance From B2 Precipitation Gau g e (meters) Pine C Long Leaf Pine 13 76 2.5 19.0 11.31 1.21 0.14 189 529 Loblolly Pine 3 18 16.0 22.0 19.33 3.30 0.33 Dogwood 1 6 7.0 7.0 7.00 1.30 0.20 Plantation Long Leaf Pine 80 98 5.0 9.0 8.00 1.00 0.10 2050 900 Oak 2 2 8.0 8.0 8.00 0.50 0.07 Mixed Oak 30 47 3.0 17.0 11.77 2.30 0.16 711 480 Loblolly Pine 23 36 5.0 18.0 13.17 1.47 0.18 Cherry 3 5 3.0 8.0 5.67 1.50 0.07 Plum 3 5 3.0 4.0 3.50 1.00 0.08 Dogwood 3 5 6.0 7.0 6.33 2.17 0.07 Crateagus 1 2 5.0 5.0 5.00 1.50 0.07 Sassafras 1 2 15.0 15.0 15.00 1.00 0.13 Hardwood Oak 125 98 5.0 10.0 9.00 1.20 0.14 1411 523 Long Leaf Pine 2 2 8.0 8.0 8.00 0.70 0.60

PAGE 24

14 Figure 2. Plot locations in the Bonham-1 and Bonham-2 study watersheds.

PAGE 25

CHAPTER 4 METHODOLOGY The 1995 Gash model uses a canopy water balance approach. The precipitation reaching the canopy either evaporates, runs down the trunk, or falls to the ground as canopy drip. The model considers each precipitation event as an individual event with enough time between events to allow the trunk and canopy to completely dry. The total interception loss is the summation of the interception losses from a series of individual events over a period of time. Each precipitation event consists of a wetting up period, a saturation period, and a drying out period. During each event, intercepted precipitation is lost through evaporation. The 1995 Gash model assumes the amount of precipitation lost due to evaporation is a function of the unit area of canopy. This approach requires the determination of specific canopy and trunk parameters and the measurement of several atmospheric parameters. The necessary canopy parameters include the canopy cover expressed as a percent of canopy per unit area, the canopy storage capacity, and the trunk storage capacity. The required atmospheric parameters are the gross precipitation per storm event, the average evaporation rate from a saturated canopy, and the mean rainfall rate per storm event. 15

PAGE 26

16 Interception losses using the 1995 Gash model are calculated using the following equation: qnjjnjjnjjcmjjmnjjPGptstqGPcGPPGRcEPGcI11111')'()/( (1) where I is the interception loss, n is the number of saturation events, m is the number of non-saturation events, c is the mean canopy cover, PG is the total rainfall during the event, E c is the mean evaporation rate from a saturated canopy scaled in proportion to canopy cover, R is the mean rainfall rate, PG is the amount of rainfall necessary to fill the canopy storage capacity, q is the number of events that saturate the trunk storage capacity, st is the trunk storage capacity, and pt is the incident precipitation reaching the trunks. The experimental data were used to determine the canopy specific parameters, climatic variables, and interception components. The total rainfall during the event is the average of the two tipping bucket rain gauges. The mean rainfall rate is the total rainfall during the event divided by the storm duration. Average annual canopy cover is calculated by averaging all measured values during the experiment. An event is any period where precipitation was recorded without a break of more than three hours between successive recordings. The mean evaporation rate from a saturated canopy scaled in proportion to canopy cover is equal to the mean evaporation rate multiplied by the canopy cover. The mean evaporation rate E was determined using the REF-ET Reference Evapotranspiration program (Allen, 2000) with the Penman equation (Penman, 1948; 1963). The hourly average atmospheric measurements taken from the meteorological monitoring station were used to calculate evaporation.

PAGE 27

17 The amount of rainfall necessary to fill the canopy storage capacity PG is given by the following equation (Carlyle-Moses and Price, 1999): RESERGPccc/1ln*/' (2) where R is the mean precipitation rate falling on a saturated canopy, E c is the mean evaporation rate scaled in proportion to canopy cover where E c = E c, and S c is the canopy storage capacity per unit area of cover where S c = S / c. The canopy storage capacity S was determined by plotting precipitation versus throughfall for saturation events and drawing a regression line. The negative regression line intercept divided by the canopy cover is the canopy storage capacity. Only storm events of 2.8 mm or more were used to determine the canopy storage capacity. The trunk storage capacity and the incident precipitation reaching the trunks were determined by the method used by Gash and Morton (1978) and Carlyle-Moses and Price (1999) where st is the slope and pt is the negative regression line intercept of the stemflow versus incident precipitation graph.

PAGE 28

CHAPTER 5 RESULTS / DISCUSSION Precipitation and Evapotranspiration During the study period, 140 discrete storm events generated 752.8 mm of precipitation. The events ranged in intensity from 0.3 to 14.4 mm/hr with an average intensity of 1.8 mm/hr. Total rainfall accumulation for each event ranged from 0.3 to 73.2 mm with an average of 5.4 mm (Figure 3). Approximately 46% of all storms deposited less than 1 mm. During this period, 45% of all precipitation fell between 1900 and 0700. The duration of each event ranged from 0.5 to 34 hours with 50% of all events being one hour or less (Figure 4). The regional precipitation network showed no systematic spatial trend. Although some spatial variation of individual events is indicated, no bias is expected over the study period. The average evaporation rate during precipitation events is 0.1 mm hr -1 This value is somewhat lower than the 0.18 0.45 mm hr -1 range cited by Carlyle-Moses and Price (1999) for Ontario, Canada. However, as 45% of the events recorded in this study occurred at night, this rate is reasonable. Measured Throughfall The five forest types exhibited a range of measured throughfall. Observed throughfall and derived interception values are summarized in Table 3. The total throughfall measurements ranged from 553.8 mm in the mixed plot to 614.6 mm in the wetland plot. Throughfall plus stemflow accounted for 77.7 to 82.5% of incident precipitation for mature pine and hardwood forests, respectively. Interception losses were largest in the 18

PAGE 29

19 mature pine forest (22.3%) and smallest in the hardwood forest (17.4%). Annual interception losses were very consistent, within 2%, for all forest communities except pine. Overall, these values compare well with published results. Dolman (1987) found interception to be 18% of incident precipitation in an oak hardwood forest. Klaassen et al. (1998) found interception to be approximately 22.3% of incident precipitation for a mixed forest comprised of mostly Douglas fir, Scotch pine, and oak. Huber and Iroume (2001) report interception losses between 11 and 39% of incident precipitation for Monterey pine forests while Lankreijer et al. (1993) and Liu (2001) found interception to account for 13% and 12%, respectively, in a maritime pine forest. Gash Model Parameters The 1995 Gash model parameters were derived from the experimental measurements. The canopy specific parameters, climatic variables, and interception components are summarized in Table 4 by forest type. Despite the similarity in interception percentage, the canopy parameters exhibit a significant range of variability. The precipitation required to saturate the canopy was determined using equation 2. The PG values ranged from 1.14 mm for the wetland plot to 4.00 mm for the pine plantation plot. The measured canopy cover values are shown in Figure 5. Annual average canopy cover ranges from 43% in the pine plantation to 88% in the wetland forest. However, the wetland, hardwood, and, to some extent, the mixed plots exhibit a strong seasonal variability. These deciduous canopies drop most of their leaves at the end of the year resulting in a significantly lower canopy cover in January, February, and March. New leaf growth during April and May gradually increases the canopy cover.

PAGE 30

20 The canopy storage capacity was determined for each plot by scaling the precipitation versus throughfall regression line intercept by the annual average canopy cover. The canopy storage capacity values ranged from 0.98 mm for the wetland plot to 1.97 mm for the mature pine plot. The mature pine and pine plantation values fall within the range of 0.4 to 3 mm reported by Liu (1998) and Llorens (2000), respectively. Little experimental data exist for wetland forests. However, Lius (1998) 0.94 mm canopy storage capacity for a cypress wetland in Florida compares favorably with this studys 0.98 mm. This studys hardwood and mixed species canopy storage capacities are 1.40 and 1.58 mm, respectively. These values are slightly higher than the 1.0 mm for an oak and maple hardwood forest reported by Carlyle-Moses and Price (1999) and 1.2 mm for a Douglas fir, Scotch pine, and oak mixed forest reported by Klaassen et al. (1998). Stemflow Stemflow was measured on an individual event basis and problems with the instrumentation preclude the use of this data to quantify the total amount of stemflow over the study period. However, the collected data were sufficient to develop a linear regression model to predict stemflow on an event basis. The cumulative calculated stemflow during the study period ranged from 3.68 mm for the mixed plot to 14.23 mm for the pine plantation plot. The percent of incident precipitation for calculated stemflow ranged from 0.5% for the pine, hardwood, and mixed plots to 2.0% for the pine plantation plots. Published stemflow values for various pine species range from 0.3% (Valente et al., 1995) to 2.4% (Hanchi and Rapp, 1997). Stemflow values published for hardwood forests range from negligible amounts (Liu, 1998; Lankreijer et al., 1993) to 4.3% (Carlyle-Moses and Price, 1999) of incident precipitation. Klaassen et al. (1996) reports that stemflow accounted for 2% of gross precipitation in a mixed hardwood forest.

PAGE 31

21 Additionally, Liu (1998) states that stemflow in a cypress wetland accounts for less that 3% of total precipitation. The agreement of the stemflow values with published values also provides validation for the parameter values determined from the regression line, i.e. trunk storage capacity and incident precipitation reaching the trunks. Model Results using Annual Average Canopy Cover Interception was modeled using equation 1 and the measured canopy parameters. Table 5 summarizes the results using the average annual canopy cover values. The model performed well with little error for the mature pine, wetland, and mixed plots. The model overestimated interception by 8.1% in the pine plantation plot. This is reasonable considering the dynamic growth of the pine plantation. However, the model under-estimated interception by 10.5% in the hardwood plot. Overall, these results demonstrate that the model predicts interception with reasonable accuracy across a range of forest communities when annual average canopy cover values are used (Figure 6). Model Results using Seasonal Canopy Cover As previously noted, the wetland, mixed, and hardwood plots exhibit a distinct seasonal variation in canopy cover. The wetland canopy cover drops from an average of 93% during the spring and summer (typically mid April to late November) to 75% during the winter (typically late November to mid April). The mixed plot experiences a smaller decrease, from 77% to 70%, during the same time period. The hardwood plot experiences the most pronounced seasonal variation. Its canopy cover decreases from an average of 60% during the spring and summer to 37% during the winter. The pine and pine plantation plots do not show a distinct seasonal variation in canopy cover. The canopy cover in the pine plot is 64% throughout the year. The pine plantation canopy

PAGE 32

22 cover increases from 40% in the spring and summer to 50% in the winter. Tree growth is the most likely cause of this change. The 1995 Gash model (equation 1) was reapplied using seasonal canopy cover values. Table 6 summarizes the results using the seasonal canopy cover values. In all cases, the application of seasonal canopy cover values improved the results (Figure 7). This is most evident with the hardwood forest where predicted interception error decreases from 10.5 to 2.4%. This result strongly suggests that forests having a significant seasonal canopy cover variation will benefit from the inclusion of routine vegetation cover information. Overall, the 1995 Gash model results show that excellent interception predictions are possible using measured canopy parameters. Canopy Density Comparison Teklehaimanot and Jarvis (1991) examined the effect of tree density on canopy storage capacity for a Sitka spruce plantation. They concluded that canopy storage capacity is a property of individual trees and is unaffected by tree density. This suggests that canopy storage capacity is consistent among tree species and spatial variations of interception losses are a function of canopy cover only. To test this assumption, our study established additional wetland and mature pine plots with varying canopy density. Data for the canopy density comparison were collected from February 1, 2002 to April 29, 2002. During this period, 24 individual storm events generated 243.0 mm of precipitation. The event intensities were comparable to the yearlong study while the total rainfall accumulation for each event covered the entire range observed during the year. The net values of throughfall, stemflow, and interception loss are summarized in Table 7. The

PAGE 33

23 canopy cover averaged 64, 46, and 29% for the mature pine (MP), mature pine B (MP B ), and mature pine C (MP C ) plots, respectively. The calibrated Gash model was applied to the additional mature pine plots. The agreement between predicted and measured interception using a seasonal canopy cover value was -0.5 (MP), -4.5 (MP B ), and 1.8% (MP C ). The 1995 Gash model predicts interception with excellent agreement for the wide range of mature pine forest canopy covers, 29 to 64%, included in this study. The canopy cover averaged 80, 78, and 66% for the wetland (W), wetland B (W B ), and wetland C (W C ) plots, respectively. A seasonal variation in canopy density was noticed in all plots, the most significant being a 33% change in the W C plot. Interception in the three plots was modeled using equation 1, the plot canopy cover, and the wetland parameters. The agreement between predicted and measured interception using seasonal canopy cover values was -5.7 (W), 25.2 (W B ), and 28.5% (W C ). The agreement for W, the wetland calibration plot, is reasonable, however, plots W B and W C are in poor agreement. One possible explanation for the poor agreement is the physical difference between plot W and plots W B and W C While the mature pine plots had similar understories, plot W had dense understory vegetation and plots W B and W C had sparse understories. The plot W understory vegetation consisted mostly of immature sweet gum, water oak, and dogwood up to 4 meters tall. Additionally, plot W contains a different tree species distribution plots W B and W C (Table 2). Plot W is composed of only 5% pine while plots W B and W C are 20% pine. As pine has a high canopy storage capacity, a higher pine percentage will effectively increase the plots overall canopy storage capacity.

PAGE 34

24 Physically based corrections were used to adjust the canopy cover and canopy storage capacity to account for the variation in wetland composition. Plot statistics were used to determine the percentage of total canopy area contributed by the overstory for plot W. The canopy cover contributed by the overstory vegetation may be described as C Wadj = A O / A T (3) where C Wadj is the adjusted canopy cover for the W plot, A o represents the sum of the projected canopy area for all trees greater than 12 meters tall, and A T represents the sum of the projected canopy area for all trees in the plot scaled by the measured canopy cover. Using equation 3, C Wadj is 66%. The canopy storage capacity for plot W was adjusted to account for the difference in species composition. Weighted averaging was used to account for the difference in pine contribution. The adjusted canopy storage capacity S adj is described by the following equations: S adj = (S W R B + S P R P ) / C Wadj (4) R P = (% pine in the W B and W C plots % pine in W plot) (5) R B = (1 R P ) (6) where S W and S P are the precipitation versus throughfall graph linear regression intercept for plot W and plot MP, respectively. The above algorithm results in an adjusted wetland canopy storage capacity of 1.50 mm. This canopy storage capacity is applicable for the wetland communities with sparse understories. Using the adjusted canopy storage capacity and the seasonal canopy cover in equation 1, the difference between the measured and predicted interception loss improves to 6.3 and 8.2% for plots W B and W C respectively. The 1995 Gash model predicts interception

PAGE 35

25 within reasonable agreement for the wetland forest included in this study once adjustments are made to compensate for the physical differences among plots. Variations in tree species and understory composition among heterogeneous forests have a significant impact on model parameters and subsequent interception prediction. The present results suggest that forests that are comprised of multiple species may require species-specific corrections to model parameters. In addition, the relative composition of overstory and understory should be considered prior to applying experimentally determined parameters to other sites. The methods introduced here to correct canopy cover measurements and canopy storage capacity provide a preliminary approach to characterize canopy specific parameters on the basis of site characteristics. While the applied methods draw from a physically based approach, the corrections were based on a limited dataset and require additional study.

PAGE 36

26 Table 3. Measured precipitation, throughfall, and derived stemflow for 4/04/01 through 6/11/02. Wetland Pine Pine Plantation Hardwood Mixed Gross Measured Precipitation (mm) 752.8 752.8 724.8 724.8 684.9 Measured Throughfall (mm) 614.5 580.8 583.3 594.5 553.8 Stemflow (mm) 4.9 4.1 14.2 3.9 3.7 Actual Interception (mm) 133.4 167.9 127.3 126.4 127.4 Throughfall Percent of Total Precipitation 81.6 77.2 80.5 82.0 80.9 Stemflow Percent of Total Precipitation 0.65 0.54 1.96 0.54 0.54 Interception Percent of Total Precipitation 17.7 22.3 17.6 17.4 18.6

PAGE 37

27 Table 4. Derived canopy specific parameters, climatic variables, and interception components. Wetland Pine Pine Plantation Hardwood Mixed PG (mm) 752.8 752.8 724.8 724.8 684.9 n 71 53 45 52 51 m 69 87 90 83 76 R (mm hr -1 ) 2.03 2.03 2.02 2.02 1.95 S (mm) 0.98 1.97 1.70 1.40 1.58 E (mm hr -1 ) 0.10 0.10 0.10 0.10 0.10 c 0.88 0.64 0.43 0.52 0.74 pt (mm) 0.02 0.01 0.05 0.01 0.01 st (mm) 0.16 0.13 0.46 0.08 0.10 P't (mm) 9.41 9.29 9.20 6.82 8.20 Ec (mm hr -1 ) 0.09 0.06 0.04 0.05 0.07 P'G (mm) 1.14 3.13 4.00 2.73 2.18

PAGE 38

28 Table 5. Model results using average annual canopy cover values. Wetland Pine Pine Plantation Hardwood Mixed Gross Measured Precipitation (mm) 752.8 752.8 724.8 724.8 684.9 Measured Throughfall (mm) 614.5 580.8 583.3 594.5 553.8 Stemflow (mm) 4.9 4.1 14.2 3.9 3.7 Actual Interception (mm) 133.4 167.9 127.3 126.4 127.4 Predicted Interception Using Average Canopy Cover (mm) 126.7 161.7 137.6 113.1 135.2 Percent Difference (Actual and Predicted Interception) 5.0 3.7 -8.1 10.5 -6.1 Throughfall Percent of Total Precipitation 81.6 77.2 80.5 82.0 80.9 Stemflow Percent of Total Precipitation 0.65 0.54 1.96 0.54 0.54 Interception Percent of Total Precipitation 17.7 22.3 17.6 17.4 18.6

PAGE 39

29 Table 6. Model results using seasonal canopy cover values. Wetland Pine Pine Plantation Hardwood Mixed Gross Measured Precipitation (mm) 752.8 752.8 724.8 724.8 684.9 Measured Throughfall (mm) 614.5 580.8 583.3 594.5 553.8 Stemflow (mm) 4.9 4.1 14.2 3.9 3.7 Actual Interception (mm) 133.4 167.9 127.3 126.4 127.4 Predicted Interception Using Seasonal Canopy Cover (mm) 127.4 166.7 136.6 123.4 134.1 Percent Difference (Actual and Predicted Interception) 4.5 0.7 -7.3 2.4 -5.3 Throughfall Percent of Total Precipitation 81.6 77.2 80.5 82.0 80.9 Stemflow Percent of Total Precipitation 0.65 0.54 1.96 0.54 0.54 Interception Percent of Total Precipitation 17.7 22.3 17.6 17.4 18.6

PAGE 40

30 Table 7. Density comparison results for 2/01/02 to 4/29/02 using seasonal canopy cover. Wetland Wetland B Wetland C Pine Pine B Pine C Canopy Cover Range (%) 67 87 69 85 49 82 48 80 41 55 17 34 Average Canopy Cover (%) 88 78 66 64 44 29 Gross Measured Precipitation (mm) 243.0 243.0 243.0 243.0 243.0 243.0 Measured Throughfall (mm) 215.3 205.8 208.1 204.0 211.9 211.9 Stemflow (mm) 2.3 2.3 2.3 1.9 1.9 1.9 Actual Interception (mm) 25.4 34.9 32.6 37.1 29.2 29.2 Predicted Interception (mm) 26.8 26.1 1 32.82 23.3 1 30.02 37.3 32.2 28.7 Percent Difference (Actual and Predicted Interception) -5.7 25.2 1 6.32 28.5 1 8.22 -0.5 -10.2 1.7 Throughfall Percent of Total Precipitation 88.6 84.7 85.6 83.9 87.2 87.2 Stemflow Percent of Total Precipitation 0.96 0.96 0.96 0.80 0.80 0.80 Interception Percent of Total Precipitation 10.4 14.4 13.4 15.3 12.0 12.0 1 Calculated using unadjusted canopy cover and canopy storage capacity. 2 Calculated using adjusted canopy cover and canopy storage capacity.

PAGE 41

31 01020304050607080901510152025MorePrecipitation (mm)Frequency Figure 3. Cumulative precipitation event totals for 4/04/01 through 6/11/02.

PAGE 42

32 0102030405060708090100125101520MoreEvent Duration (hours)Frequency Figure 4. Precipitation event durations for 4/04/01 through 6/11/02.

PAGE 43

0204060801001203/10/015/19/017/28/0110/6/0112/15/012/23/025/4/027/13/02DateCanopy Cover (%) Wetland Pine Mixed Hardwood Plantation 33 Figure 5. Canopy cover measurements for the five forest types.

PAGE 44

0.020.040.060.080.0100.0120.0140.0160.0180.0WetlandPinePine PlantationHardwoodMixedInterception (mm) Actual Interception (mm) Predicted Interception Using Average Canopy Cover (mm) 5.0% Difference 3.7% Difference -8.1% Difference -6.1% Difference 10.5% Difference 34 Figure 6. Measured and modeled interception results using average canopy cover.

PAGE 45

0.020.040.060.080.0100.0120.0140.0160.0180.0WetlandPinePine PlantationHardwoodMixedLand UseInterception (mm) Actual Interception (mm) Predicted Interception Using Seasonal Canopy Cover (mm) 0.7% Difference 4.5% Difference -7.3% Difference 2.4% Difference -5.3% Difference 35 Figure 7. Measured and modeled interception results using seasonal canopy cover.

PAGE 46

CHAPTER 6 WATERSHED SCALE APPLICATION Watershed hydrology has transitioned from the prediction of rainfall-runoff and land surface-atmospheric interactions using lumped approaches (Liou et al., 1999) to the more advanced application of distributed land-use, soils, and topographic data (Bonan et al., 2002). The current studys results were used to consider the relative importance of capturing the spatiotemporal variability of the water input resulting from distributed throughfall. Three approaches to predict throughfall using the 1995 Gash interception model were compared at a watershed scale. The first approach is a lumped approach wherein the predominant vegetation type is used to predict the magnitude of water input on a seasonal basis. The second and third approaches use land use maps to distribute throughfall in the watershed spatially and temporally. The second approach assumes a constant canopy cover value while the third captures the seasonal dynamics of leaf fall and growth. The interception for the period from April 29, 2001 to April 29, 2002 was modeled using all three approaches. During this period, 836 mm of precipitation fell on the study watershed. The lumped approach was applied with the mature pine forest type that covers 47% of the watershed. Table 8 summarizes the total throughfall by forest type and aggregate watershed on a seasonal basis. The 1995 Gash model predicts 659 mm of throughfall and 79% of total gross precipitation, using the lumped approach. By taking into account the spatial variation in forest type and applying the annual average canopy cover, the model predicts 687 mm of throughfall or 36

PAGE 47

37 82% of gross precipitation. When the model is further refined to include seasonal canopy cover as well as spatially distributed forest types, the predicted throughfall is 680 mm or 81% of gross precipitation. The predicted annual throughfall varies by 4% between the lumped approach and the spatially and seasonally distributed forests. The choice of approach does not appear to be significant when the 1995 Gash model is applied over long temporal periods and when the interception by the dominant species is similar to that of the other species. Larger differences among the watershed responses are observed for smaller spatial scales and shorter temporal periods (Figures 8 through 11). An examination of the watershed results by individual forest type shows that the lumped approach under-predicts annual throughfall for all forest types. Most significantly, it under-predicts throughfall by 7% for hardwood forests and 6% for wetland forests when an annual average canopy cover is used or by 6% for hardwood and wetland forests using seasonal canopy cover. This error is of particular concern for the riparian wetland forest as the watershed storm response is most critical for areas closest to the stream in watersheds dominated by the saturation excess mechanisms of runoff generation. When shorter temporal periods are examined, i.e. seasonal instead of annual, the associated errors with the lumped approach are more pronounced. For example, the lumped approach predicts wetland throughfall within 1% of the spatially distributed approach using seasonal canopy cover during the winter. However, the difference between the approaches is 10% during the summer. A large seasonal variation is also seen in the pine plantation communities where the error ranges from a 1% over-prediction to an 11% under-prediction in throughfall. Clearly, throughfall is controlled by the plant architecture, plant physiology, and rainfall input and timing. The best model would ideally include all details. However, often the details are

PAGE 48

38 not available. These results demonstrate that there is not a significant variation among approaches when applying models over aggregated spatial scales and long temporal periods. However, when smaller scales or shorter temporal periods are of interest, an appropriate landscape characterization is necessary to capture the variability of water input.

PAGE 49

Table 8. Throughfall results using the distributed approach. 39 Precipitation (mm) Mature Pine (mm) Wetland (mm) Pine Plantation (mm) Mixed (mm) Hardwood (mm) Watershed Total (mm) Spring 250.5 201.8 (0%) 212.6 (5%) 207.6 (3%) 206.0 (2%) 215.2 (6%) 209.6 (4%) Summer 231.3 173.6 (0%) 188.3 (8%) 180.0 (4%) 178.6 (3%) 189.1 (8%) 183.0 (5%) Fall 138.2 110.3 (0%) 115.8 (5%) 112.8 (2%) 112.3 (2%) 117.9 (7%) 114.6 (4%) Annual Average Canopy Cover Winter 215.7 173.4 (0%) 182.2 (5%) 178.5 (3%) 176.6 (2%) 184.9 (6%) 180.0 (4%) Total 835.6 659.1 (0%) 699.0 (6%) 679.0 (3%) 673.4 (2%) 707.2 (7%) 687.2 (4%) Spring 250.5 196.5 (0%) 211.3 (7%) 204.4 (4%) 203.6 (4%) 210.3 (7%) 205.7 (4%) Summer 231.3 166.7 (0%) 184.2 (10%) 188.1 (11%) 173.4 (4%) 174.8 (5%) 176.3 (5%) Fall 138.2 113.0 (0%) 113.6 (1%) 115.1 (2%) 110.4 (-2%) 119.2 (5%) 115.5 (2%) Seasonal Canopy Cover Winter 215.7 174.5 (0%) 185.1 (6%) 173.2 (-1%) 181.1 (4%) 189.9 (8%) 182.3 (4%) Total 835.6 650.6 (0%) 694.3 (6%) 680.7 (4%) 668.5 (3%) 694.1 (6%) 679.9 (4%) Numbers in parenthesis represent the percent difference from the lumped prediction using mature pine.

PAGE 50

40 Figure 8. Watershed scale results for the Spring season (April 2001 through June 2001).

PAGE 51

41 Figure 9. Watershed scale results for the Summer season (July 2001 through September 2001).

PAGE 52

42 Figure 10. Watershed scale results for the Fall season (October 2001 through December 2001).

PAGE 53

43 Figure 11. Watershed scale results for the Winter season (January 2002 through March 2002).

PAGE 54

CHAPTER 7 CONCLUSION This study derived a set of parameters, coefficients, and physical properties for wetland, mature pine, pine plantation, mixed, and hardwood land uses that are appropriate for studying diverse forested communities. Application of the parameters in the 1995 Gash interception model demonstrates its ability to predict interception losses accurately provided that the model parameters are representative of the modeled region. Application of seasonal canopy cover values in lieu of annual average values improved the agreement of the modeled and the actual interception loss for all five land uses included in this study. Furthermore, the model predicts interception accurately when applied over land uses of varying canopy cover as long as the canopy cover is adjusted for the area of interest. A new approach is proposed to correct derived parameters for site-specific vegetation in riparian wetlands. A watershed scale intercomparison of the influence of forest community and seasonal variation on interception demonstrated that appropriate characterization of forests is necessary when applying the 1995 Gash model over seasonal or shorter duration time periods. Additionally, application of the model at sub-watershed spatial scales demonstrated that significant variation between results can be expected as the extent of the spatial scale is reduced. The field experimentation and water budget analysis in this study provide insight into the characterization of the net water input into the heterogeneous forest communities distinctive to the southeast United States. 44

PAGE 55

APPENDIX A PRECIPITATION VERSUS THROUGHFALL GRAPHS y = 1.0126x 0.8621R2 = 0.87520510152025303505101520253035Precipitation (mm)Throughfall (mm) Figure A-1. Precipitation versus throughfall graph used to determine canopy storage capacity for the wetland plot. 45

PAGE 56

46 y = 0.9932x 1.2575R2 = 0.91080102030405060708001020304050607080Precipitation (mm)Throughfall (mm) Figure A-2. Precipitation versus throughfall graph used to determine canopy storage capacity for the pine plot. y = 0.9682x 0.7256R2 = 0.9319010203040506070010203040506070Precipitation (mm)Throughfall (mm) Figure A-3. Precipitation versus throughfall graph used to determine canopy storage capacity for the pine plantation plot.

PAGE 57

47 y = 0.9831x 1.1695R2 = 0.96090102030405060708001020304050607080Precipitation (mm)Throughfall (mm) Figure A-4. Precipitation versus throughfall graph used to determine canopy storage capacity for the mixed plot. y = 0.9912x 0.7287R2 = 0.91760102030405060708001020304050607080Precipitation (mm)Throughfall (mm) Figure A-5. Precipitation versus throughfall graph used to determine canopy storage capacity for the hardwood plot.

PAGE 58

APPENDIX B PRECIPITATION VERSUS STEMFLOW GRAPHS y = 0.0172x 0.1577R2 = 0.44150.00.20.40.60.81.01.21.41.61.82.00.010.020.030.040.050.060.0Precipitation (mm)Stemflow (mm) Figure B-1. Precipitation versus stemflow graph used to determine trunk storage capacity and precipitation reaching the trunks for the wetland plot. 48

PAGE 59

49 y = 0.0179x 0.1958R2 = 0.59130.00.20.40.60.81.01.21.41.61.82.00.010.020.030.040.050.060.070.080.0Precipitation (mm)Stemflow (mm) Figure B-2. Precipitation versus stemflow graph used to determine trunk storage capacity and precipitation reaching the trunks for the pine plot. y = 0.0518x 0.4598R2 = 0.94870.00.51.01.52.02.53.00.010.020.030.040.050.060.0Precipitation (mm)Stemflow (mm) Figure B-3. Precipitation versus stemflow graph used to determine trunk storage capacity and precipitation reaching the trunks for the pine plantation plot.

PAGE 60

50 y = 0.0109x 0.0754R2 = 0.510700.10.20.30.40.50.60.70.80.910.010.020.030.040.050.060.070.080.0Precipitation (mm)Stemflow (mm) Figure B-4. Precipitation versus stemflow graph used to determine trunk storage capacity and precipitation reaching the trunks for the hardwood plot.

PAGE 61

LIST OF REFERENCES Allen, R. G. 2000. REF-ET, Reference Evapotranspiration Calculator Version Windows 2.0. Univ. of Idaho Res. and Ext. Center, Kimberly, ID, 82 p. Bonan, G.B., K.W. Oleson, M. Vertenstein, S. Levis, X. Zeng, Y. Dai, R.E. Dickinson, and Z.-L.Yang, 2002: The land surface climatology of the Community Land Model coupled to the NCAR Community Climate Model. Journal of Climate, submitted. Bunnell, F. L., Vales, D. J., 1990. Comparison of methods for estimating forest overstory cover: differences among techniques. Canadian Journal of Forest Research 20: 101-107. Carlyle-Moses, D. E., Price, A. G., 1999. An evaluation of the Gash interception model in a northern hardwood stand. Journal of Hydrology 214: 103. Crockford, R. H., Richardson, D. P., 2000. Partitioning of rainfall into throughfall, stemflow and interception: effect of forest type, ground cover and climate. Hydrological Processes 14: 2903. Dingman, S. L., 1994. Physical Hydrology. Upper Saddle River, New Jersey: Prentice-Hall, Inc. Dolman, A. J., 1987. Summer and winter rainfall interception in an oak forest. Predictions with an analytical and a numerical simulation model. Journal of Hydrology 90: 1. Dykes, A. P., 1997. Rainfall interception from a lowland tropical rainforest in Brunei. Journal of Hydrology 200: 260. Gash, J. H. C., 1979. An analytical model of rainfall interception by forests. Quarterly Journal of the Royal Meteorological Society 105: 43. Gash, J. H. C., Lloyd, C. R., Lachaud, G., 1995. Estimating sparse forest rainfall interception with an analytical model. Journal of Hydrology 170: 79. Gash, J. H. C., Morton, A. J., 1978. An application of the Rutter model to the estimation of the interception loss from Thetford forest. Journal of Hydrology 38: 49. Hanchi, A., Rapp, M., 1997. Stemflow determination in forest stands. Forest Ecology and Management 97: 231. 51

PAGE 62

52 Huber, A., Iroume, A., 2001. Variability of annual rainfall partitioning for different sites and forest covers in Chile. Journal of Hydrology 248: 78. Jackson, N. A., 2000. Measured and modeled rainfall interception loss from an agroforestry system in Kenya. Agricultural and Forest Meteorology 100: 323. Klaassen, W., Bosveld, F., de Water, E., 1998. Water storage and evaporation as constituents of rainfall interception. Journal of Hydrology 212: 36. Klaassen, W., Lankreijer, H. J. M., Veen, A. W. L., 1996. Rainfall interception near a forest edge. Journal of Hydrology 185: 349. Lankreijer, H., Lundberg, A., Grelle, A., Lindroth, A., Seibert, J., 1993. Evaporation and storage of intercepted rain analyzed by comparing two models applied to a boreal forest. Agricultural and Forest Meteorology 98: 595. Lemmon, P. E., 1956. A spherical densiometer for estimating forest overstory density. Forest Science 2: 314 320. Liou, Y., Galantowicz, J. F., England, A. W., 1999. A Land Surface Process/Radiobrightness Model with Coupled Heat and Moisture Transport for Prairie Grassland. IEEE Transactions on Geoscience and Remote Sensing, Volume 37, No. 4: 1848. Liu, S., 1998. Estimation of rainfall storage capacity in the canopies of cypress wetlands and slash pine uplands in North-Central Florida. Journal of Hydrology 207: 32. Liu, S., 2001. Evaluation of the Liu model for prediction rainfall interception in forests world-wide. Hydrological Processes 15: 2341. Llorens, P., Gallart, F., 2000. A simplified method for forest water storage capacity measurement. Journal of Hydrology 240: 131. Nvar, J., Charles, F., Jurando, E., 1999. Spatial variations of interception loss components by Tamaulipan thornscrub in northeastern Mexico. Forest Ecology and Management 124: 231. Penman, H.L., 1948. Natural Evaporation from open water, bare soil and grass. Proceedings of the Royal Society, Series A. 193: 120-145. Penman, H.L., 1963. Vegetation and Hydrology. Tech. Comm. No. 53, Commonwealth Bureau of Soils, Harpenden, England. 125 pp.

PAGE 63

53 Rutter, A. J., 1963. Studies in the water relations of pinus sylvestris in plantation conditions I. Measurements of rainfall and interception. Journal of Ecology 51: 191203. Rutter, A. J., Dershaw, K. A., Robins, P. C., Morton, A. J., 1971. A predictive model of rainfall interception in forests. I. Derivation of the model from observations in a plantation of Corsican pine. Agricultural Meteorology 9: 367. Rutter, A. J., Morton, A. J., 1977. A predictive model of rainfall interception in forests. III. Sensitivity of the model to stand parameters and meteorological variables. Journal of Applied Ecology, Volume 14, Issue 2: 567. Rutter, A. J., Morton, A. J., Robins, P. C., 1975. A predictive model of rainfall interception in forests. II. Generalization of the model and comparison with observations in some coniferous and hardwood stands. Journal of Applied Ecology 12: 367. Teklehaimanot, Z., Jarvis, P. G., 1991. Direct Measurement of Evaporation of Intercepted Water from Forest Canopies. Journal of Applied Ecology. Volume 28, Issue 2: 603. Valente, F., David, J. S., Gash, J. H. C., 1997. Modeling interception loss for two sparse eucalypt and pine forests in central Portugal using reformulated Rutter and Gash analytical models. Journal of Hydrology 190: 141. van Dijk, A. I. J. M., Bruijnzeel, L. A., 2001. Modeling rainfall interception by vegetation of variable density using an adapted analytical model. Part 1. Model description. Journal of Hydrology 247: 230. van Dijk, A. I. J. M., Bruijnzeel, L. A., 2001. Modeling rainfall interception by vegetation of variable density using an adapted analytical model. Part 2. Model validation for a tropical upland mixed cropping system. Journal of Hydrology 247: 239.

PAGE 64

BIOGRAPHICAL SKETCH Malcolm Lewis Bryant was born in Morgantown, West Virginia, on September 15, 1965. He is a graduate of Vanguard High School in Ocala, Florida. He attended Georgia Southern University from 1983 to 1984 and Central Florida Community College from 1985 to 1987 where he earned an Associate of Science degree in radiation protection. From 1987 to 1998 he worked in the commercial nuclear power industry as a health physics technician and radiological engineer. In 1993 he married his wife, Kim. He earned a Bachelor of Science degree in technology (nuclear) in 1997 from Regents College in Albany, New York. In 2000 his son, Trevor, was born. He received a second Bachelor of Science degree in civil engineering from the University of Florida in 2001. He enrolled in the Master of Science in water resources engineering in the summer of 2001. 54


Permanent Link: http://ufdc.ufl.edu/UFE0000510/00001

Material Information

Title: Spatiotemporal Throughfall Characterization of Heterogeneous Forest Communities in the Southeastern U.S.
Physical Description: Mixed Material
Copyright Date: 2008

Record Information

Source Institution: University of Florida
Holding Location: University of Florida
Rights Management: All rights reserved by the source institution and holding location.
System ID: UFE0000510:00001

Permanent Link: http://ufdc.ufl.edu/UFE0000510/00001

Material Information

Title: Spatiotemporal Throughfall Characterization of Heterogeneous Forest Communities in the Southeastern U.S.
Physical Description: Mixed Material
Copyright Date: 2008

Record Information

Source Institution: University of Florida
Holding Location: University of Florida
Rights Management: All rights reserved by the source institution and holding location.
System ID: UFE0000510:00001


This item has the following downloads:


Full Text











SPATIOTEMPORAL THROUGHFALL CHARACTERIZATION OF
HETEROGENEOUS FOREST COMMUNITIES IN THE SOUTHEASTERN U.S.















By

MALCOLM LEWIS BRYANT


A THESIS PRESENTED TO THE GRADUATE SCHOOL
OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT
OF THE REQUIREMENTS FOR THE DEGREE OF
MASTER OF ENGINEERING

UNIVERSITY OF FLORIDA


2002




























Copyright 2002

by

MALCOLM LEWIS BRYANT















ACKNOWLEDGMENTS

I wish to thank Dr. Jennifer Jacobs, Dr. Kirk Hatfield, and Dr. Louis Motz for serving

on my committee. I extend special appreciation to Dr. Jacobs for her help and guidance

throughout this project. I want to thank Shirish Bhat for helping with the data collection

and processing and Sudheer Reddy Satti for his technical advice. Thanks go to Hugh

Westbury at Fort Benning for his coordination efforts and D. L. Price and M. R. Kress

with the U.S. Army Corp of Engineers for the meteorological data. Funding from the

Strategic Environmental Research and Development program grant (DACA88-98-R-

0010) is gratefully acknowledged. Also, thanks go to the rest of the Water Resources

gang: Gerard Ripo, Brent Whitfield, En-Ching Hsu, Aniruddha Guha, and Erik Howard.

Finally, I want to thank my wife Kim and son Trevor for their support and understanding.















TABLE OF CONTENTS
page

A C K N O W L E D G M E N T S ......... ............... ................................................................... iii

LIST OF TABLES ................. .. ...... .. ............ ..... .. ...... ....... ....... vi

L IST O F F IG U R E S .... ...... ................................................ .. .. ..... .............. vii

ABSTRACT .............. ......................................... ix

CHAPTER

1 INTRODUCTION ..................................................... ............ .. ......... 1

2 S T U D Y A R E A ........................................ ............................................. .................... 4

3 INSTRUMENTATION / EXPERIMENT DESIGN ...................................................8

S tu d y P lo ts ...................................... ...................................... ............... 8
Canopy Cover D ata ............................................................. .............. 10
Climate Data ............................................. 10

4 M E T H O D O L O G Y ............................................................................. .....................15

5 R E SU L T S / D ISC U SSIO N .............................................. ....................................... 18

Precipitation and Evapotranspiration.................. ........ .......................... 18
M measured Throughfall ............. ................. ...................... .. .. .............. 18
G ash M odel Param eters ...... .. ............. ........................................ .............. 19
Stemflow ................................... ............................. ......... 20
Model Results using Annual Average Canopy Cover .............................................. 21
Model Results using Seasonal Canopy Cover ........................................................... 21
Canopy D ensity Com prison .......................................................... ............ 22

6 WATERSHED SCALE APPLICATION ........... ...............................................36

7 CONCLUSION ............... .............................. .......................... 44

APPENDIX

A PRECIPITATION VERSUS THROUGHFALL GRAPHS .......................................45

B PRECIPITATION VERSUS STEMFLOW GRAPHS.............. ...............48


iv










L IST O F R E F E R E N C E S ....................................................................... .......................5 1

B IO G R A PH IC A L SK E T C H ...................................................................... ..................54


























































v
















LIST OF TABLES


Table page

1 Land use distribution in the Bonham-1 and Bonham-2 watersheds. ............................6

2 Plot vegetation and distribution statistics. ........................................ ............... 12

3 Measured precipitation, throughfall, and derived stemflow for 4/04/01 through
6/11/02 ............................................................................26

4 Derived canopy specific parameters, climatic variables, and interception
com ponents. ...................................................... ................. 27

5 Model results using average annual canopy cover values. .........................................28

6 Model results using seasonal canopy cover values .............. ....... ...............29

7 Density comparison results for 2/01/02 to 4/29/02 using seasonal canopy cover.........30

8 Throughfall results using the distributed approach...................... ...................39
















LIST OF FIGURES


Figure p

1 Land use for the Bonham-1 and Bonham-2 study watersheds. ......................................7

2 Plot locations in the Bonham-1 and Bonham-2 study watersheds..............................14

3 Cumulative precipitation event totals for 4/04/01 through 6/11/02..............................31

4 Precipitation event durations for 4/04/01 through 6/11/02..................... ..............32

5 Canopy cover measurements for the five forest types ................................................33

6 Measured and modeled interception results using average canopy cover ...................34

7 Measured and modeled interception results using seasonal canopy cover ..................35

8 Watershed scale results for the Spring season (April 2001 through June 2001). ..........40

9 Watershed scale results for the Summer season (July 2001 through September 2001)
...................................................................................................... . 4 1

10 Watershed scale results for the Fall season (October 2001 through December 2001)
...................................................................................................... . 4 2

11 Watershed scale results for the Winter season (January 2002 through March 2002)..43

A-i Precipitation versus throughfall graph used to determine canopy storage capacity for
th e w etlan d p lo t ............................. .. .................. .............. ................ 4 5

A-2 Precipitation versus throughfall graph used to determine canopy storage capacity for
th e pin e plot ......... .... .. ........................................... ..................... 4 6

A-3 Precipitation versus throughfall graph used to determine canopy storage capacity for
the pine plantation plot.................................. ......... .. ... ... ............ 46

A-4 Precipitation versus throughfall graph used to determine canopy storage capacity for
the m ixed plot. .................................................... ................. 47

A-5 Precipitation versus throughfall graph used to determine canopy storage capacity for
the hard ood plot. ..................... ...... ............ ................. .... ....... 47









B-l Precipitation versus stemflow graph used to determine trunk storage capacity and
precipitation reaching the trunks for the wetland plot. ........................................48

B-2 Precipitation versus stemflow graph used to determine trunk storage capacity and
precipitation reaching the trunks for the pine plot..........................................49

B-3 Precipitation versus stemflow graph used to determine trunk storage capacity and
precipitation reaching the trunks for the pine plantation plot.............................49

B-4 Precipitation versus stemflow graph used to determine trunk storage capacity and
precipitation reaching the trunks for the hardwood plot................... ............50
















Abstract of Thesis Presented to the Graduate School
of the University of Florida in Partial Fulfillment of the
Requirements for the Degree of Master of Engineering






SPATIOTEMPORAL THROUGHFALL CHARACTERIZATION OF
HETEROGENEOUS FOREST COMMUNITIES IN THE SOUTHEASTERN U.S.


By

Malcolm Lewis Bryant

December 2002

Chair: Dr. Jennifer Jacobs
Major Department: Civil and Coastal Engineering

The spatial and temporal influence of heterogeneous forest communities on

interception loss was characterized for the southeastern United States. Throughfall was

measured simultaneously in five forest types, i.e., mature pine, 13 year-old pine

plantation, wetland, hardwood, and mixed hardwood/pine, that comprise the landscape in

Fort Benning, Columbus, GA. The 1995 Gash interception model was determined to be

valid for use in all forest types using seasonal canopy cover values and annual average

canopy cover values. The model predicted interception loss with an agreement of -8.1 to

10.5% using annual average canopy cover values. Application of seasonal canopy cover

values in lieu of annual averages improved accuracy in all cases with an overall range

from -7.3 to 4.5%. An investigation of the model's performance within forests of varying









canopy cover showed that the model predicts interception with an agreement of -5.7 to

8.2% using seasonal canopy cover values. A watershed scale intercomparison of the

influence of forest community and seasonal variation on interception showed that

appropriate characterization of forests is necessary when applying the 1995 Gash model

over seasonal or shorter duration time periods. Additionally, application of the model at

sub-watershed spatial scales showed that significant variation among results can be

expected as the extent of the spatial scale is reduced. The field experimentation and

water-budget analysis in this study provide insight into the relative net water input into

the heterogeneous forest communities that are typical of the southeastern United States.














CHAPTER 1
INTRODUCTION

Interception losses modify the dynamics of the watershed hydrological and

geochemical cycles. They lower the intensity of precipitation and wash solid particles

and dissolved carbon from leaves affecting soil-water chemistry and weathering

processes. Nutrient loading is impacted by interception losses because microbial activity

is affected by soil moisture. Interception losses also affect rainfall-runoff by reducing the

water input available for runoff Depending on the vegetation type, interception losses

can account for 10 to 40% of the total incident precipitation (Dingman, 1994).

Furthermore, the distribution of vegetation types among heterogeneous landscapes can

influence the spatial variability of interception losses (Crockford and Richardson, 2000).

Most methods for calculating interception use a running water balance approach. In

this type of approach, the total incident precipitation equals the sum of the throughfall,

stemflow, evaporation from the trunk, and evaporation from the canopy. Models that use

a water balance approach are affected by the spatiotemporal accuracy of the interception

loss calculation. The first physically based model for calculating rainfall interception

was the Rutter model (cited in Rutter et al., 1971). The Rutter model is a running water

balance model in which total evaporation from a wet canopy is calculated on a per storm

basis. Rutter et al. (1975) were able to predict throughfall within 10% of the measured

values in a pine stand. Gash and Morton (1978) used the Rutter model to predict

interception loss from a Scots pine forest within 7% of the measured values (Gash and

Morton, 1978). However, the Rutter model underestimated interception by 20% to 32%









in an artificially defoliated hardwood stand (Rutter et al., 1975). Gash (1979) introduced

a variation of the Rutter model. The 1979 Gash model differs from the Rutter model in

that it considers the arithmetic average instead of the actual rainfall and evaporation rates

for individual storm events, thus simplifying the model and minimizing the data

requirements.

In both the Rutter and Gash models, evaporation is calculated per unit area of ground.

This formulation is problematic for sparsely vegetated forests. The 1979 Gash model

overestimated interception losses by as much as 29 to 44% in a sparse pine forest in

Portugal (Valente et al., 1997). The Gash model was revised (Gash et al., 1995) to

calculate evaporation per unit area of canopy rather than per unit area of ground. The

revised 1979 Gash model, the 1995 Gash model, greatly improves the accuracy of the

interception loss predictions in sparse forests. The 1995 Gash model was used by

Valente et al. (1997) to predict interception losses for the same pine forest in Portugal to

within 3% of the measured values. Using the 1995 Gash model, Dykes (1997) estimated

interception losses for a tropical rain forest in Borneo within 5% of measured values.

Carlyle-Moses and Price (1999) estimated interception losses on a hardwood stand in

Ontario, Canada, to less than 1% of measured values using the 1995 Gash model.

Many studies have validated the 1995 Gash interception model for use in

homogeneous landscapes (van Dijk and Bruinjnzeel, 2001; Jackson, 2000; Carlyle-Moses

and Price, 1999; Navar et al., 1999; Dykes, 1997; Valente et al., 1997; Dolman, 1987);

few studies have incorporated interception models over large heterogeneous landscapes.

The goal of this research was to characterize the net water input into the

heterogeneous forest communities that are distinctive of the southeast United States. The









1995 Gash interception model was chosen for this study since it has been validated in

many different landscapes and it accounts for canopy cover variations. The objectives of

this study were to 1) simultaneously calibrate the 1995 Gash interception model for use in

the five forest types considered in this study, 2) use the model to predict rainfall

interception and intercompare the results across forest types and canopy densities, 3)

explore the influence of seasonal canopy characteristics on the model's performance, and

4) apply the model at the watershed scale to identify critical differences among available

approaches to modeling interception in a heterogeneous landscape.














CHAPTER 2
STUDY AREA

The study was conducted at the Fort Benning military reservation, located in

southwest Georgia. Long, hot summers and mild winters characterize the region's

climate. Average annual precipitation is about 830 mm with a monthly average of 69

mm. Most of the precipitation occurs in the spring and summer as a result of

thunderstorms. Heavy rains are typical during the summer but can occur in any month.

Snow accounts for less than 1% of the annual precipitation. The soils in the area are

dominated by loamy sand with some sandy loam.

Two second-order watersheds, Bonham-1 and Bonham-2, were selected for this study.

These watersheds were selected because they contain most of the region's predominant

forest types. The Bonham-1 watershed has an area of 762 ha, a minimum elevation of

87.8 m and maximum elevation of 144.2 m, and an average slope of 9.6%. The Bonham-

2 watershed has an area of 2,211 ha, a minimum elevation of 90.5 m, a maximum

elevation of 159.1 m, and an average slope of 8.6%.

The two watersheds have a heterogeneous land cover consisting of either open or

forested areas (Figure 1). The open areas are either military, brush, or wildlife openings.

The military openings are clear-cut parcels of land dominated by grass and bare soil that

are used as military training grounds. The brush openings consist of tall grass and

immature crateagus. The wildlife openings are natural openings in the forests that are

vegetated primarily by grass. The forested areas include five forest types-mature pine,

pine plantation, riparian wetland, hardwood, and mixed hardwood/pine. The mature pine,









which consists of loblolly (Pinus taeda) and short leaf pine (Pinus echinata), is the

dominant forest type of each watershed. The pine plantation is a 13 year-old long leaf

pine (Pinuspalustris) stand planted in rows. The wetland vegetation consists mostly of

various hardwood trees along with a range of wetland understory vegetation. The

hardwood stands are generally mature scrub oak (Quercus berberidifolia). White oak

(Quercus alba), short leaf pine, and loblolly pine are the dominant species in the mixed

pine/hardwood stands. Table 1 describes the total area and relative contribution of each

forest type. The forests are managed using prescribed burning on a three-year cycle. A

majority of the land area included in this study was burned within one year of the study

period.













Table 1. Land use distribution in the Bonham-1 and Bonham-2 watersheds.
Land Cover Watershed Area (m2) Watershed Area (%)
Mature Pine 1,293,800 43.7
Pine Plantation 118,700 4.0
Wetland 299,000 10.1
Hardwood 542,600 18.3
Mixed 506,700 17.1
Openings (military, 2 0
wildlife and brush)
























4 4. 4- /.


























Sil i
p ~P-
##- p -f"t.-



)M -eged

Lan de
a7 0 k;


Figure 1. Land use for the Bonham-1 and Bonham-2 study watersheds.














CHAPTER 3
INSTRUMENTATION/EXPERIMENT DESIGN

The experiment was conducted from April 4, 2001 through June 11, 2002. Canopy

parameters, climatic variables, and interception components were measured throughout

this period. The measured data include precipitation, throughfall, stemflow, atmospheric

conditions, and canopy cover. The following sections describe the experimental

equipment and methods applied to this study.

Study Plots

A rectangular plot was established in each forest community. Table 2 lists the

vegetation distribution and statistics for the all plots. The plots were randomly selected

within areas having vegetation that is consistent with the average vegetation density and

distribution for the respective forest community (Figure 2). The dimensions of each plot

were determined by the size and geometry of each land-use type and ranged from 10 x 40

meters to 30 x 30 meters. Each plot was subdivided into four sampling grids of equal

size. Each grid was outfitted with four throughfall collectors and one tipping bucket rain

gauge. The throughfall collectors and tipping buckets were randomly placed on the

ground within the confines of the grid. To ensure randomness, each instrument was

relocated within the grid after data were collected.

The throughfall data were collected using eight-inch diameter tipping bucket rain

gauges (model RG-100a, RainWise) and throughfall collectors. The tipping buckets

were calibrated to 0.01 inches per tip. Each tipping bucket was equipped with a data

logger (HOBO-8000, Onset Inc.) that recorded the number of tips and the date and time









of each tip. Clogging of the tipping buckets resulted in the omission of data for brief time

periods. To reduce clogging, the installed plastic screen and cotter pin at the throat of the

tipping bucket collector was removed and a plastic mesh was placed over the mouth of

the collector. The mesh was pushed down into the collector to create a sock-filter

arrangement so that the effective collection area of the gauge was not reduced.

The throughfall collectors consisted of two-liter plastic bottles with six-inch diameter

funnels. The plastic bottles and funnels were supported by stands made of 3/8-inch

plywood and six-inch PVC pipe. The PVC pipe was secured to the plywood base using

silicon adhesive. Data were collected from the instruments on a bi-weekly basis.

A parallel experiment with a shorter duration was conducted to study the effect of

varying canopy cover on interception modeling. Two additional plots for both the

wetland and mature pine communities were monitored from February 1 through April 29,

2002. Each additional plot had a significantly different canopy cover than the other plots

in the respective forest community. The additional plots, referred to as wetland B,

wetland C, mature pine B, and mature pine C, were selected based on a visual inspection

and a spherical densiometer survey. The instrumentation and methods used for these

plots were identical to those described above.

Stemflow was measured in the wetland, mature pine, pine plantation, and hardwood

plots. For these plots, the dominant tree species were further sub-divided into three

classes of diameter at breast height (DBH) and projected crown radius. Stemflow gauges

were installed in each plot such that the dominant tree species were sampled and that a

representative tree from each DBH and projected crown radius category was included.









The mixed plot stemflow was determined by averaging the pine and hardwood

measurements.

Stemflow was measured by attaching split plastic tubing to the representative trees in

each plot. Nails or staples and silicon sealant were used as the method of attachment.

The tubing was wound around each tree in a 360-degree spiral and routed to a tipping

bucket (model RG-100a, RainWise) equipped with a data logger (HOBO-8000, Onset

Inc.).

Canopy Cover Data

Canopy cover was determined by direct measurement with a Model-A spherical

densiometer using the method outlined by Lemmon (1956). This method provides a

simple and reliable measure of canopy cover (Bunell and Vales, 1990). Readings were

taken in the center of each study plot on a bi-weekly basis throughout the study period.

Climate Data

Precipitation data were measured at a height of 15 centimeters by two tipping bucket

rain gauges (model RG-100a, RainWise). Each tipping bucket gauge was equipped

with an 8-inch diameter collector and calibrated to 0.01 inches per tip. Data loggers

(HOBO-8000, Onset Inc.) were used to store the date and time of each tip. The tipping

buckets were located in a 50 x 150 meter clearing on the boundary between the Bonham-

1 and Bonham-2 watersheds. The distance from the precipitation gauges to the

throughfall plots ranged from 315 to 900 meters. The clearing is centrally located and is

within 900 meters of each land-use plot. Storm totals were calculated as the average

recording of the two rain gauges. Data were collected from the instruments on a bi-

weekly basis.









Atmospheric measurements were continuously recorded by the Ecosystem

Characterization and Monitoring Initiative (ECMI) meteorological monitoring station

ME-04 "McKenna Mount." The atmospheric data includes air temperature, barometric

pressure, relative humidity, solar radiation, wind speed, and precipitation. The station is

located at UTM coordinates 706387 Easting and 3583703 Northing. The station is

approximately 7 kilometers south of the Bonham-2 watershed. Data were collected in 1-

minute intervals and averaged over 30-minute periods. The air temperature and relative

humidity data were collected with a temperature and relative humidity probe (model

HMP45C, Vaisala) at an installation height of 2 meters. Barometric pressure data were

collected with a pressure transmitter (model PT101B, Vaisala) at an installation height of

1.75 meters. Solar radiation data were collected with a pyranometer with a silicon

photovoltaic detector (model LI200X, Li-Cor) at an installation height of 2.5 meters.

Wind speed data were collected using a four-blade heilcoid propeller sensor (model

05103-5, R. M. Young) at 3 meters. Finally, precipitation data were collected with a

tipping bucket rain gauge (model TE525MM, Texas Electronics) mounted at 3 meters.














Table 2. Plot vegetation and distribution statistics.
Average
Projected
Minimum Maximum Average Canopy Average Distance From B2
Number Percent Height Height Height Radius Diameter Trees per Precipitation Gauge
Plot Tree Type of Trees of Total (meters) (meters) (meters) (meters) (meters) Hectare (meters)
Wet Sweet Gum 29 60 5.0 31.0 12.88 1.62 0.14 1200 325
Bay 8 17 5.0 16.0 9.75 1.31 0.07
Long L. Pine 4 8 4.0 8.0 6.00 0.90 0.05
Water Oak 4 8 4.0 28.0 11.13 1.23 0.12
Dog Wood 3 6 6.0 14.5 10.50 2.33 0.11
Wet B Sweet Gum 21 46 6.0 28.0 17.60 1.70 0.23 1150 343
Long L. Pine 9 20 14.0 21.0 18.89 1.80 0.23
Dog Wood 6 13 4.0 14.0 7.83 1.82 0.08
Water Oak 6 13 9.0 28.0 16.58 2.33 0.19
Bay 4 9 5.0 13.0 9.13 1.60 0.07
Wet C Sweet Gum 21 54 3.0 25.0 14.98 2.18 0.20 975 395
Long L. Pine 8 21 3.8 5.0 4.25 0.78 0.05
Dog Wood 8 21 3.0 6.5 4.44 1.19 0.06
Water Oak 2 5 9.5 23.0 16.25 1.80 0.18
Pine Loblolly Pine 41 82 4.0 22.0 13.76 2.08 0.21 556 573
Oak 5 10 6.0 15.0 9.80 2.34 0.14
Crateagus 4 8 5.0 9.0 6.50 2.38 0.11
Pine B Loblolly Pine 33 100 4.0 25.5 12.86 2.14 0.20 367 675















Minimum Maximum Average


Number Percent


Plot Tree Type
Pine C Long Leaf Pine
Loblolly Pine
Dogwood
Plantation Long Leaf Pine
Oak
Mixed Oak
Loblolly Pine
Cherry
Plum
Dogwood
Crateagus
Sassafras
Hardwood Oak
Long Leaf Pine


of Trees
13
3
1
80
2
30
23
3
3
3
1
1
125
2


of Total
76
18
6
98
2
47
36
5
5
5
2
2
98
2


Height
(meters)
2.5
16.0
7.0
5.0
8.0
3.0
5.0
3.0
3.0
6.0
5.0
15.0
5.0
8.0


Height
(meters)
19.0
22.0
7.0
9.0
8.0
17.0
18.0
8.0
4.0
7.0
5.0
15.0
10.0
8.0


Height
(meters)
11.31
19.33
7.00
8.00
8.00
11.77
13.17
5.67
3.50
6.33
5.00
15.00
9.00
8.00


Average
Projected
Canopy Average Distance From B2
Radius Diameter Trees per Precipitation Gauge
(meters) (meters) Hectare (meters)
1.21 0.14 189 529
3.30 0.33
1.30 0.20
1.00 0.10 2050 900
0.50 0.07
2.30 0.16 711 480
1.47 0.18
1.50 0.07
1.00 0.08
2.17 0.07
1.50 0.07
1.00 0.13
1.20 0.14 1411 523
0.70 0.60

























































Kilometers


Figure 2. Plot locations in the Bonham-1 and Bonham-2 study watersheds.














CHAPTER 4
METHODOLOGY

The 1995 Gash model uses a canopy water balance approach. The precipitation

reaching the canopy either evaporates, runs down the trunk, or falls to the ground as

canopy drip. The model considers each precipitation event as an individual event with

enough time between events to allow the trunk and canopy to completely dry. The total

interception loss is the summation of the interception losses from a series of individual

events over a period of time.

Each precipitation event consists of a wetting up period, a saturation period, and a

drying out period. During each event, intercepted precipitation is lost through

evaporation. The 1995 Gash model assumes the amount of precipitation lost due to

evaporation is a function of the unit area of canopy. This approach requires the

determination of specific canopy and trunk parameters and the measurement of several

atmospheric parameters. The necessary canopy parameters include the canopy cover

expressed as a percent of canopy per unit area, the canopy storage capacity, and the trunk

storage capacity. The required atmospheric parameters are the gross precipitation per

storm event, the average evaporation rate from a saturated canopy, and the mean rainfall

rate per storm event.









Interception losses using the 1995 Gash model are calculated using the following

equation:

n+m m n n n-q
I =c cPG, +(cE /R) (PG P'G) +cy P'G +qxst +ptPGJ (1)
J=1 J=1 J=1 J=1 J=1

where I is the interception loss, n is the number of saturation events, m is the number of

non-saturation events, c is the mean canopy cover, PG is the total rainfall during the

event, Ec is the mean evaporation rate from a saturated canopy scaled in proportion to

canopy cover, R is the mean rainfall rate, P 'G is the amount of rainfall necessary to fill

the canopy storage capacity, q is the number of events that saturate the trunk storage

capacity, st is the trunk storage capacity, andpt is the incident precipitation reaching the

trunks.

The experimental data were used to determine the canopy specific parameters,

climatic variables, and interception components. The total rainfall during the event is the

average of the two tipping bucket rain gauges. The mean rainfall rate is the total rainfall

during the event divided by the storm duration. Average annual canopy cover is

calculated by averaging all measured values during the experiment. An event is any

period where precipitation was recorded without a break of more than three hours

between successive recordings. The mean evaporation rate from a saturated canopy

scaled in proportion to canopy cover is equal to the mean evaporation rate multiplied by

the canopy cover. The mean evaporation rate E was determined using the REF-ET

Reference Evapotranspiration program (Allen, 2000) with the Penman equation (Penman,

1948; 1963). The hourly average atmospheric measurements taken from the

meteorological monitoring station were used to calculate evaporation.









The amount of rainfall necessary to fill the canopy storage capacity P 'G is given by

the following equation (Carlyle-Moses and Price, 1999):

P'G = -(R/E,)S ln[1-(E, /R)] (2)

where R is the mean precipitation rate falling on a saturated canopy, Ec is the mean

evaporation rate scaled in proportion to canopy cover where Ec = E c, and Sc is the

canopy storage capacity per unit area of cover where Sc = S / c. The canopy storage

capacity S was determined by plotting precipitation versus throughfall for saturation

events and drawing a regression line. The negative regression line intercept divided by

the canopy cover is the canopy storage capacity. Only storm events of 2.8 mm or more

were used to determine the canopy storage capacity. The trunk storage capacity and the

incident precipitation reaching the trunks were determined by the method used by Gash

and Morton (1978) and Carlyle-Moses and Price (1999) where st is the slope andpt is the

negative regression line intercept of the stemflow versus incident precipitation graph.














CHAPTER 5
RESULTS / DISCUSSION

Precipitation and Evapotranspiration

During the study period, 140 discrete storm events generated 752.8 mm of

precipitation. The events ranged in intensity from 0.3 to 14.4 mm/hr with an average

intensity of 1.8 mm/hr. Total rainfall accumulation for each event ranged from 0.3 to

73.2 mm with an average of 5.4 mm (Figure 3). Approximately 46% of all storms

deposited less than 1 mm. During this period, 45% of all precipitation fell between 1900

and 0700. The duration of each event ranged from 0.5 to 34 hours with 50% of all events

being one hour or less (Figure 4). The regional precipitation network showed no

systematic spatial trend. Although some spatial variation of individual events is

indicated, no bias is expected over the study period.

The average evaporation rate during precipitation events is 0.1 mm hr-1. This value is

somewhat lower than the 0.18 0.45 mm hr-1 range cited by Carlyle-Moses and Price

(1999) for Ontario, Canada. However, as 45% of the events recorded in this study

occurred at night, this rate is reasonable.

Measured Throughfall

The five forest types exhibited a range of measured throughfall. Observed throughfall

and derived interception values are summarized in Table 3. The total throughfall

measurements ranged from 553.8 mm in the mixed plot to 614.6 mm in the wetland plot.

Throughfall plus stemflow accounted for 77.7 to 82.5% of incident precipitation for

mature pine and hardwood forests, respectively. Interception losses were largest in the









mature pine forest (22.3%) and smallest in the hardwood forest (17.4%). Annual

interception losses were very consistent, within 2%, for all forest communities except

pine. Overall, these values compare well with published results. Dolman (1987) found

interception to be 18% of incident precipitation in an oak hardwood forest. Klaassen et

al. (1998) found interception to be approximately 22.3% of incident precipitation for a

mixed forest comprised of mostly Douglas fir, Scotch pine, and oak. Huber and Iroume

(2001) report interception losses between 11 and 39% of incident precipitation for

Monterey pine forests while Lankreijer et al. (1993) and Liu (2001) found interception to

account for 13% and 12%, respectively, in a maritime pine forest.

Gash Model Parameters

The 1995 Gash model parameters were derived from the experimental

measurements. The canopy specific parameters, climatic variables, and interception

components are summarized in Table 4 by forest type. Despite the similarity in

interception percentage, the canopy parameters exhibit a significant range of variability.

The precipitation required to saturate the canopy was determined using equation 2. The

P 'G values ranged from 1.14 mm for the wetland plot to 4.00 mm for the pine plantation

plot.

The measured canopy cover values are shown in Figure 5. Annual average

canopy cover ranges from 43% in the pine plantation to 88% in the wetland forest.

However, the wetland, hardwood, and, to some extent, the mixed plots exhibit a strong

seasonal variability. These deciduous canopies drop most of their leaves at the end of the

year resulting in a significantly lower canopy cover in January, February, and March.

New leaf growth during April and May gradually increases the canopy cover.









The canopy storage capacity was determined for each plot by scaling the precipitation

versus throughfall regression line intercept by the annual average canopy cover. The

canopy storage capacity values ranged from 0.98 mm for the wetland plot to 1.97 mm for

the mature pine plot. The mature pine and pine plantation values fall within the range of

0.4 to 3 mm reported by Liu (1998) and Llorens (2000), respectively. Little experimental

data exist for wetland forests. However, Liu's (1998) 0.94 mm canopy storage capacity

for a cypress wetland in Florida compares favorably with this study's 0.98 mm. This

study's hardwood and mixed species canopy storage capacities are 1.40 and 1.58 mm,

respectively. These values are slightly higher than the 1.0 mm for an oak and maple

hardwood forest reported by Carlyle-Moses and Price (1999) and 1.2 mm for a Douglas

fir, Scotch pine, and oak mixed forest reported by Klaassen et al. (1998).

Stemflow

Stemflow was measured on an individual event basis and problems with the

instrumentation preclude the use of this data to quantify the total amount of stemflow

over the study period. However, the collected data were sufficient to develop a linear

regression model to predict stemflow on an event basis. The cumulative calculated

stemflow during the study period ranged from 3.68 mm for the mixed plot to 14.23 mm

for the pine plantation plot. The percent of incident precipitation for calculated stemflow

ranged from 0.5% for the pine, hardwood, and mixed plots to 2.0% for the pine plantation

plots. Published stemflow values for various pine species range from 0.3% (Valente et

al., 1995) to 2.4% (Hanchi and Rapp, 1997). Stemflow values published for hardwood

forests range from negligible amounts (Liu, 1998; Lankreijer et al., 1993) to 4.3%

(Carlyle-Moses and Price, 1999) of incident precipitation. Klaassen et al. (1996) reports

that stemflow accounted for 2% of gross precipitation in a mixed hardwood forest.









Additionally, Liu (1998) states that stemflow in a cypress wetland accounts for less that

3% of total precipitation. The agreement of the stemflow values with published values

also provides validation for the parameter values determined from the regression line, i.e.

trunk storage capacity and incident precipitation reaching the trunks.

Model Results using Annual Average Canopy Cover

Interception was modeled using equation 1 and the measured canopy parameters.

Table 5 summarizes the results using the average annual canopy cover values. The

model performed well with little error for the mature pine, wetland, and mixed plots. The

model overestimated interception by 8.1% in the pine plantation plot. This is reasonable

considering the dynamic growth of the pine plantation. However, the model under-

estimated interception by 10.5% in the hardwood plot. Overall, these results demonstrate

that the model predicts interception with reasonable accuracy across a range of forest

communities when annual average canopy cover values are used (Figure 6).

Model Results using Seasonal Canopy Cover

As previously noted, the wetland, mixed, and hardwood plots exhibit a distinct

seasonal variation in canopy cover. The wetland canopy cover drops from an average of

93% during the spring and summer (typically mid April to late November) to 75% during

the winter (typically late November to mid April). The mixed plot experiences a smaller

decrease, from 77% to 70%, during the same time period. The hardwood plot

experiences the most pronounced seasonal variation. Its canopy cover decreases from an

average of 60% during the spring and summer to 37% during the winter. The pine and

pine plantation plots do not show a distinct seasonal variation in canopy cover. The

canopy cover in the pine plot is 64% throughout the year. The pine plantation canopy









cover increases from 40% in the spring and summer to 50% in the winter. Tree growth is

the most likely cause of this change.

The 1995 Gash model (equation 1) was reapplied using seasonal canopy cover values.

Table 6 summarizes the results using the seasonal canopy cover values. In all cases, the

application of seasonal canopy cover values improved the results (Figure 7). This is most

evident with the hardwood forest where predicted interception error decreases from 10.5

to 2.4%. This result strongly suggests that forests having a significant seasonal canopy

cover variation will benefit from the inclusion of routine vegetation cover information.

Overall, the 1995 Gash model results show that excellent interception predictions are

possible using measured canopy parameters.

Canopy Density Comparison

Teklehaimanot and Jarvis (1991) examined the effect of tree density on canopy

storage capacity for a Sitka spruce plantation. They concluded that canopy storage

capacity is a property of individual trees and is unaffected by tree density. This suggests

that canopy storage capacity is consistent among tree species and spatial variations of

interception losses are a function of canopy cover only. To test this assumption, our

study established additional wetland and mature pine plots with varying canopy density.

Data for the canopy density comparison were collected from February 1, 2002 to April

29, 2002.

During this period, 24 individual storm events generated 243.0 mm of precipitation.

The event intensities were comparable to the yearlong study while the total rainfall

accumulation for each event covered the entire range observed during the year. The net

values of throughfall, stemflow, and interception loss are summarized in Table 7. The









canopy cover averaged 64, 46, and 29% for the mature pine (MP), mature pine B (MPB),

and mature pine C (MPc) plots, respectively.

The calibrated Gash model was applied to the additional mature pine plots. The

agreement between predicted and measured interception using a seasonal canopy cover

value was -0.5 (MP), -4.5 (MPB), and 1.8% (MPc). The 1995 Gash model predicts

interception with excellent agreement for the wide range of mature pine forest canopy

covers, 29 to 64%, included in this study.

The canopy cover averaged 80, 78, and 66% for the wetland (W), wetland B (WB), and

wetland C (Wc) plots, respectively. A seasonal variation in canopy density was noticed

in all plots, the most significant being a 33% change in the We plot. Interception in the

three plots was modeled using equation 1, the plot canopy cover, and the wetland

parameters. The agreement between predicted and measured interception using seasonal

canopy cover values was -5.7 (W), 25.2 (WB), and 28.5% (Wc). The agreement for W,

the wetland calibration plot, is reasonable, however, plots WB and We are in poor

agreement.

One possible explanation for the poor agreement is the physical difference between

plot W and plots WB and Wc. While the mature pine plots had similar understories, plot

W had dense understory vegetation and plots WB and We had sparse understories. The

plot W understory vegetation consisted mostly of immature sweet gum, water oak, and

dogwood up to 4 meters tall. Additionally, plot W contains a different tree species

distribution plots WB and We (Table 2). Plot W is composed of only 5% pine while plots

WB and We are 20% pine. As pine has a high canopy storage capacity, a higher pine

percentage will effectively increase the plot's overall canopy storage capacity.









Physically based corrections were used to adjust the canopy cover and canopy storage

capacity to account for the variation in wetland composition. Plot statistics were used to

determine the percentage of total canopy area contributed by the overstory for plot W.

The canopy cover contributed by the overstory vegetation may be described as

Cwadj = Ao / AT (3)

where Cwadj is the adjusted canopy cover for the W plot, Ao represents the sum of the

projected canopy area for all trees greater than 12 meters tall, and AT represents the sum

of the projected canopy area for all trees in the plot scaled by the measured canopy cover.

Using equation 3, CWadj is 66%.

The canopy storage capacity for plot W was adjusted to account for the difference in

species composition. Weighted averaging was used to account for the difference in pine

contribution. The adjusted canopy storage capacity Sadj is described by the following

equations:

Sadj = (Sw RB + Sp Rp) / CWadj (4)

Rp = (% pine in the WB and We plots % pine in W plot) (5)

RB = (1 RP) (6)

where Sw and Sp are the precipitation versus throughfall graph linear regression intercept

for plot W and plot MP, respectively. The above algorithm results in an adjusted wetland

canopy storage capacity of 1.50 mm. This canopy storage capacity is applicable for the

wetland communities with sparse understories.

Using the adjusted canopy storage capacity and the seasonal canopy cover in equation

1, the difference between the measured and predicted interception loss improves to 6.3

and 8.2% for plots WB and Wc, respectively. The 1995 Gash model predicts interception









within reasonable agreement for the wetland forest included in this study once

adjustments are made to compensate for the physical differences among plots.

Variations in tree species and understory composition among heterogeneous forests

have a significant impact on model parameters and subsequent interception prediction.

The present results suggest that forests that are comprised of multiple species may require

species-specific corrections to model parameters. In addition, the relative composition of

overstory and understory should be considered prior to applying experimentally

determined parameters to other sites. The methods introduced here to correct canopy

cover measurements and canopy storage capacity provide a preliminary approach to

characterize canopy specific parameters on the basis of site characteristics. While the

applied methods draw from a physically based approach, the corrections were based on a

limited dataset and require additional study.











Table 3. Measured precipitation, throughfall, and derived stemflow for 4/04/01 through
6/11/02.


Pine
Wetland Pine Plantation Hardwood


Gross Measured Precipitation
(mm)
Measured Throughfall (mm)
Stemflow (mm)
Actual Interception (mm)
Throughfall Percent of Total
Precipitation
Stemflow Percent of Total
Precipitation
Interception Percent of Total
Precipitation


752.8
614.5
4.9
133.4


752.8
580.8
4.1
167.9


724.8
583.3
14.2
127.3


81.6 77.2 80.5

0.65 0.54 1.96

17.7 22.3 17.6


724.8
594.5
3.9
126.4

82.0

0.54


Mixed

684.9
553.8
3.7
127.4

80.9

0.54


17.4 18.6












Table 4. Derived canopy specific parameters, climatic variables, and interception
components.


PG (mm)
n
m
R (mm hr )
S (mm)
E (mm hr 1)
C
pt (mm)
st (mm)
P't (mm)
Ec (mm hr )
P'G (mm)


Wetland
752.8
71
69
2.03
0.98
0.10
0.88
0.02
0.16
9.41
0.09
1.14


Pine
752.8
53
87
2.03
1.97
0.10
0.64
0.01
0.13
9.29
0.06
3.13


Pine Plantation Hardwood
724.8 724.8
45 52
90 83
2.02 2.02
1.70 1.40
0.10 0.10
0.43 0.52
0.05 0.01
0.46 0.08
9.20 6.82
0.04 0.05
4.00 2.73


Mixed
684.9
51
76
1.95
1.58
0.10
0.74
0.01
0.10
8.20
0.07
2.18












Table 5. Model results using average annual canopy


Pine
Wetland Pine Plantation


Gross Measured
Precipitation (mm)
Measured Throughfall
(mm)
Stemflow (mm)
Actual Interception (mm)
Predicted Interception
Using Average Canopy
Cover (mm)
Percent Difference (Actual
and Predicted
Interception)
Throughfall Percent of
Total Precipitation
Stemflow Percent of Total
Precipitation
Interception Percent of
Total Precipitation


752.8

614.5
4.9
133.4


126.7


5.0

81.6

0.65


752.8

580.8
4.1
167.9


161.7


3.7

77.2

0.54


724.8

583.3
14.2
127.3


137.6


-8.1

80.5

1.96


724.8 684.9


594.5
3.9
126.4


113.1


10.5

82.0

0.54


553.8
3.7
127.4


135.2


-6.1

80.9

0.54


17.7 22.3 17.6


Hardwood


Mixed


cover values.


17.4 18.6












Table 6. Model results using


Gross Measured
Precipitation (mm)
Measured Throughfall
(mm)
Stemflow (mm)
Actual Interception (mm)
Predicted Interception
Using Seasonal Canopy
Cover (mm)
Percent Difference (Actual
and Predicted
Interception)
Throughfall Percent of
Total Precipitation
Stemflow Percent of Total
Precipitation
Interception Percent of
Total Precipitation


seasonal canopy cover values.
Pine
Wetland Pine Plantation


752.8

614.5
4.9
133.4


752.8

580.8
4.1
167.9


127.4 166.7


4.5

81.6

0.65


0.7

77.2

0.54


724.8

583.3
14.2
127.3


136.6


-7.3

80.5

1.96


Hardwood


Mixed


724.8 684.9


594.5
3.9
126.4


123.4


2.4

82.0

0.54


553.8
3.7
127.4


134.1


-5.3

80.9

0.54


17.7 22.3 17.6


17.4 18.6











Table 7. Density comparison


results for 2/01/02 to 4/29/02 using seasonal canopy cover.
Wetland Wetland


Canopy Cover Range (%)
Average Canopy Cover
(%)
Gross Measured
Precipitation (mm)
Measured Throughfall
(mm)
Stemflow (mm)
Actual Interception (mm)
Predicted Interception
(mm)
Percent Difference (Actual
and Predicted Interception)
Throughfall Percent of
Total Precipitation
Stemflow Percent of Total
Precipitation
Interception Percent of
Total Precipitation


Wetland B
67-87 69-85


C Pine Pine B Pine C
49-82 48-80 41-55 17-34


88 78 66 64 44 29

243.0 243.0 243.0 243.0 243.0 243.0


215.3 205.8
2.3 2.3
25.4 34.9
26.11
26.8 32.82
25.21
-5.7 6.32


208.1
2.3
32.6
23.31
30.02
28.51
8.22


204.0
1.9
37.1


211.9
1.9
29.2


211.9
1.9
29.2


37.3 32.2 28.7


-10.2


88.6 84.7 85.6 83.9 87.2 87.2

0.96 0.96 0.96 0.80 0.80 0.80


10.4


14.4


13.4


15.3


12.0


12.0


1 Calculated using unadjusted canopy cover and canopy storage capacity.
2 Calculated using adjusted canopy cover and canopy storage capacity.











90

80

70

60

c 50

g40
LL
30

20

10

0
0 -- ---- ^ ---- ^ -- B

1 5 10 15 20 25 More
Precipitation (mm)

Figure 3. Cumulative precipitation event totals for 4/04/01 through 6/11/02.






























1 2


100

90

80

70

60

50

40

30

20

10

0


20 More


Figure 4. Precipitation event durations for 4/04/01 through 6/11/02.


5 10 15
Event Duration (hours)


































5/19/01


7/28/01


10/6/01


12/15/01


2/23/02


Date


-- Wetland -X- Pine


Mixed --- Hardwood -- Plantation


Figure 5. Canopy cover measurements for the five forest types.


100


80 -


60


40


20 -


0/0
3/10/01


5/4/02


7/13/02














180.0


160.0

140.0

120.0

100.0

5.0% 3.7% -8.1% 10.5% -6.1%
o 80.0 Difference Difference Difference Difference Difference


60.0

40.0

20.0

0.0
Wetland Pine Pine Plantation Hardwood Mixed

U Actual Interception (mm) O Predicted Interception Using Average Canopy Cover (mm)

Figure 6. Measured and modeled interception results using average canopy cover.
















180.0

160.0

140.0

120.0 -

a 100.0 -

4 80.0 4.5% 0.7% -7.3% 2.4% -5.3%
S8.Difference Difference Difference Difference Differenc
60.0

40.0

20.0

0.0
Wetland Pine Pine Plantation Hardwood Mixed


Actual Interception (mm) O Predicted Interception Using Seasonal Canopy Cover (mm)

Figure 7. Measured and modeled interception results using seasonal canopy cover.

















CHAPTER 6
WATERSHED SCALE APPLICATION

Watershed hydrology has transitioned from the prediction of rainfall-runoff and land surface-

atmospheric interactions using lumped approaches (Liou et al., 1999) to the more advanced

application of distributed land-use, soils, and topographic data (Bonan et al., 2002). The current

study's results were used to consider the relative importance of capturing the spatiotemporal

variability of the water input resulting from distributed throughfall. Three approaches to predict

throughfall using the 1995 Gash interception model were compared at a watershed scale. The

first approach is a lumped approach wherein the predominant vegetation type is used to predict

the magnitude of water input on a seasonal basis. The second and third approaches use land use

maps to distribute throughfall in the watershed spatially and temporally. The second approach

assumes a constant canopy cover value while the third captures the seasonal dynamics of leaf fall

and growth.

The interception for the period from April 29, 2001 to April 29, 2002 was modeled using all

three approaches. During this period, 836 mm of precipitation fell on the study watershed. The

lumped approach was applied with the mature pine forest type that covers 47% of the watershed.

Table 8 summarizes the total throughfall by forest type and aggregate watershed on a seasonal

basis. The 1995 Gash model predicts 659 mm of throughfall and 79% of total gross

precipitation, using the lumped approach. By taking into account the spatial variation in forest

type and applying the annual average canopy cover, the model predicts 687 mm of throughfall or












82% of gross precipitation. When the model is further refined to include seasonal canopy cover

as well as spatially distributed forest types, the predicted throughfall is 680 mm or 81% of gross

precipitation. The predicted annual throughfall varies by 4% between the lumped approach and

the spatially and seasonally distributed forests. The choice of approach does not appear to be

significant when the 1995 Gash model is applied over long temporal periods and when the

interception by the dominant species is similar to that of the other species.

Larger differences among the watershed responses are observed for smaller spatial scales and

shorter temporal periods (Figures 8 through 11). An examination of the watershed results by

individual forest type shows that the lumped approach under-predicts annual throughfall for all

forest types. Most significantly, it under-predicts throughfall by 7% for hardwood forests and

6% for wetland forests when an annual average canopy cover is used or by 6% for hardwood and

wetland forests using seasonal canopy cover. This error is of particular concern for the riparian

wetland forest as the watershed storm response is most critical for areas closest to the stream in

watersheds dominated by the saturation excess mechanisms of runoff generation.

When shorter temporal periods are examined, i.e. seasonal instead of annual, the associated

errors with the lumped approach are more pronounced. For example, the lumped approach

predicts wetland throughfall within 1% of the spatially distributed approach using seasonal

canopy cover during the winter. However, the difference between the approaches is 10% during

the summer. A large seasonal variation is also seen in the pine plantation communities where the

error ranges from a 1% over-prediction to an 11% under-prediction in throughfall.

Clearly, throughfall is controlled by the plant architecture, plant physiology, and rainfall input

and timing. The "best" model would ideally include all details. However, often the details are






38





not available. These results demonstrate that there is not a significant variation among

approaches when applying models over aggregated spatial scales and long temporal periods.

However, when smaller scales or shorter temporal periods are of interest, an appropriate

landscape characterization is necessary to capture the variability of water input.














Table 8. Throughfall results using the distributed approach.


Precipitation
(mm)
250.5
231.3
138.2
215.7
835.6


Mature Pine
(mm)


201.8 (0%)
173.6 (0%)
110.3 (0%)
173.4 (0%)
659.1 (0%)


Wetland Pine Plantation


(mm)


212.6
188.3
115.8
182.2
699.0


(mm)


(5%)
(8%)
(5%)
(5%)
(6%)


207.6
180.0
112.8
178.5
679.0


Mixed Hardwood


(mm)


(3%)
(4%)
(2%)
(3%)
(3%)


206.0
178.6
112.3
176.6
673.4


(mm)


(2%)
(3%)
(2%)
(2%)
(2%)


215.2
189.1
117.9
184.9
707.2


(6%)
(8%)
(7%)
(6%)
(7%)


Watershed
Total
(mm)


209.6 (4%)
183.0(5%)
114.6(4%)
180.0 (4%)
687.2 (4%)


Spring 250.5 196.5 (0%) 211.3 (7%) 204.4 (4%) 203.6 (4%) 210.3 (7%) 205.7 (4%)
Seasonal Summer 231.3 166.7 (0%) 184.2 (10%) 188.1 (11%) 173.4 (4%) 174.8 (5%) 176.3 (5%)
Canopy Fall 138.2 113.0 (0%) 113.6 (1%) 115.1 (2%) 110.4 (-2%) 119.2 (5%) 115.5 (2%)
Cover Winter 215.7 174.5 (0%) 185.1 (6%) 173.2 (-1%) 181.1 (4%) 189.9 (8%) 182.3 (4%)
Total 835.6 650.6 (0%) 694.3 (6%) 680.7 (4%) 668.5 (3%) 694.1 (6%) 679.9 (4%)
Numbers in parenthesis represent the percent difference from the lumped prediction using mature pine.


Annual
Average
Canopy
Cover


Spring
Summer
Fall
Winter
Total

















DISTRIBUTED (Average c) DISTRIBUTED (Seasonal c)


Throughfll (rrm)
195- 200
20 205
S205-210
- 210-215
- 215-220
22o-


Figure 8. Watershed scale results for the Spring season (April 2001 through June 2001).


LUMPED


















DISTRIBUTED (Average c)


Tht uughfall (mm)
| 165-170
170-175
M 12 5
- 175 1
S180-185

M 190-


Figure 9. Watershed scale results for the Summer season (July 2001 through September 2001).


DISTRIBUTED (Seasonal c)


LUMPED


















DISTRIBUTED (Average c)


Throughfall (mm)
S110 -115
S115- 120
M 120+


Figure 10. Watershed scale results for the Fall season (October 2001 through December 2001).


DISTRIBUTED (Seasonal c)


LUMPED













DISTRIBUTED (Average c)


'C





Throughfall (mm)
S170-175
S175-1180
-180-185 s
185 193
190+


Figure 11. Watershed scale results for the Winter season (January 2002 through March 2002).


DISTRIBUTED (Seasonal c)


LUMPED














CHAPTER 7
CONCLUSION

This study derived a set of parameters, coefficients, and physical properties for

wetland, mature pine, pine plantation, mixed, and hardwood land uses that are

appropriate for studying diverse forested communities. Application of the parameters in

the 1995 Gash interception model demonstrates its ability to predict interception losses

accurately provided that the model parameters are representative of the modeled region.

Application of seasonal canopy cover values in lieu of annual average values improved

the agreement of the modeled and the actual interception loss for all five land uses

included in this study. Furthermore, the model predicts interception accurately when

applied over land uses of varying canopy cover as long as the canopy cover is adjusted

for the area of interest. A new approach is proposed to correct derived parameters for

site-specific vegetation in riparian wetlands. A watershed scale intercomparison of the

influence of forest community and seasonal variation on interception demonstrated that

appropriate characterization of forests is necessary when applying the 1995 Gash model

over seasonal or shorter duration time periods. Additionally, application of the model at

sub-watershed spatial scales demonstrated that significant variation between results can

be expected as the extent of the spatial scale is reduced. The field experimentation and

water budget analysis in this study provide insight into the characterization of the net

water input into the heterogeneous forest communities distinctive to the southeast United

States.















APPENDIX A
PRECIPITATION VERSUS THROUGHFALL GRAPHS


35

30

25

20

15

10


0 5 10 15 20 25 30


Precipitation (mm)


Figure A-1. Precipitation versus throughfall graph used to determine canopy storage
capacity for the wetland plot.

























0 10 20 30 40 50 60 70 80


Precipitation (mm)


Figure A-2. Precipitation versus throughfall graph used to determine canopy storage
capacity for the pine plot.



70

60

50
4y = 0.9682x- 0.7256
40-R R= 0.9319

0 30

20 -

10 -

0
0 10 20 30 40 50 60 70
Precipitation (mm)

Figure A-3. Precipitation versus throughfall graph used to determine canopy storage
capacity for the pine plantation plot.










80
70
^ 60
50
40
o 30


i0n-
10 I I I

0 10 20 30 40 50 60 70 80
Precipitation (mm)

Figure A-4. Precipitation versus throughfall graph used to determine canopy storage
capacity for the mixed plot.


50
40
o 30
H 'on


10 ~-w'


0 10 20 30 40 50 60 70 80
Precipitation (mm)

Figure A-5. Precipitation versus throughfall graph used to determine canopy storage
capacity for the hardwood plot.















APPENDIX B
PRECIPITATION VERSUS STEMFLOW GRAPHS


10.0 20.0 30.0 40.0 50.0
Precipitation (mm)


60.0


Figure B-1. Precipitation versus stemflow graph used to determine trunk storage capacity
and precipitation reaching the trunks for the wetland plot.

























0.0 10.0 20.0 30.0 40.0 50.0 60.0 70.0 80.0
Precipitation (mm)

Figure B-2. Precipitation versus stemflow graph used to determine trunk storage capacity
and precipitation reaching the trunks for the pine plot.


3.0

2.5

S2.0

1.5

1.0


10.0 20.0 30.0 40.0 50.0


60.0


Precipitation (mm)

Figure B-3. Precipitation versus stemflow graph used to determine trunk storage capacity
and precipitation reaching the trunks for the pine plantation plot.


























0.0 10.0 20.0 30.0 40.0 50.0 60.0 70.0 80.0
Precipitation (mm)

Figure B-4. Precipitation versus stemflow graph used to determine trunk storage capacity
and precipitation reaching the trunks for the hardwood plot.
















LIST OF REFERENCES


Allen, R. G. 2000. "REF-ET, Reference Evapotranspiration Calculator Version
Windows 2.0". Univ. of Idaho Res. and Ext. Center, Kimberly, ID, 82 p.

Bonan, G.B., K.W. Oleson, M. Vertenstein, S. Levis, X. Zeng, Y. Dai, R.E. Dickinson,
and Z.-L.Yang, 2002: The land surface climatology of the Community Land Model
coupled to the NCAR Community Climate Model. Journal of Climate, submitted.

Bunnell, F. L., Vales, D. J., 1990. Comparison of methods for estimating forest overstory
cover: differences among techniques. Canadian Journal of Forest Research 20: 101-
107.

Carlyle-Moses, D. E., Price, A. G., 1999. An evaluation of the Gash interception model
in a northern hardwood stand. Journal of Hydrology 214: 103-110.

Crockford, R. H., Richardson, D. P., 2000. Partitioning of rainfall into throughfall,
stemflow and interception: effect of forest type, ground cover and climate.
Hydrological Processes 14: 2903-2920.

Dingman, S. L., 1994. Physical Hydrology. Upper Saddle River, New Jersey: Prentice-
Hall, Inc.

Dolman, A. J., 1987. Summer and winter rainfall interception in an oak forest.
Predictions with an analytical and a numerical simulation model. Journal of Hydrology
90: 1-9.

Dykes, A. P., 1997. Rainfall interception from a lowland tropical rainforest in Brunei.
Journal of Hydrology 200: 260-279.

Gash, J. H. C., 1979. An analytical model of rainfall interception by forests. Quarterly
Journal of the Royal Meteorological Society 105: 43-55.

Gash, J. H. C., Lloyd, C. R., Lachaud, G., 1995. Estimating sparse forest rainfall
interception with an analytical model. Journal of Hydrology 170: 79-86.

Gash, J. H. C., Morton, A. J., 1978. An application of the Rutter model to the estimation
of the interception loss from Thetford forest. Journal of Hydrology 38: 49-58.

Hanchi, A., Rapp, M., 1997. Stemflow determination in forest stands. Forest Ecology
and Management 97: 231-235.










Huber, A., Iroume, A., 2001. Variability of annual rainfall partitioning for different sites
and forest covers in Chile. Journal of Hydrology 248: 78-92.

Jackson, N. A., 2000. Measured and modeled rainfall interception loss from an
agroforestry system in Kenya. Agricultural and Forest Meteorology 100: 323-336.

Klaassen, W., Bosveld, F., de Water, E., 1998. Water storage and evaporation as
constituents of rainfall interception. Journal of Hydrology 212-213: 36-50.

Klaassen, W., Lankreijer, H. J. M., Veen, A. W. L., 1996. Rainfall interception near a
forest edge. Journal of Hydrology 185: 349-361.

Lankreijer, H., Lundberg, A., Grelle, A., Lindroth, A., Seibert, J., 1993. Evaporation and
storage of intercepted rain analyzed by comparing two models applied to a boreal
forest. Agricultural and Forest Meteorology 98-99: 595-604.

Lemmon, P. E., 1956. A spherical densiometer for estimating forest overstory density.
Forest Science 2: 314 320.

Liou, Y., Galantowicz, J. F., England, A. W., 1999. A Land Surface
Process/Radiobrightness Model with Coupled Heat and Moisture Transport for Prairie
Grassland. IEEE Transactions on Geoscience and Remote Sensing, Volume 37, No. 4:
1848-1859.

Liu, S., 1998. Estimation of rainfall storage capacity in the canopies of cypress wetlands
and slash pine uplands in North-Central Florida. Journal of Hydrology 207: 32-41.

Liu, S., 2001. Evaluation of the Liu model for prediction rainfall interception in forests
world-wide. Hydrological Processes 15: 2341-2360.

Llorens, P., Gallart, F., 2000. A simplified method for forest water storage capacity
measurement. Journal of Hydrology 240: 131-144.

Navar, J., Charles, F., Jurando, E., 1999. Spatial variations of interception loss
components by Tamaulipan thornscrub in northeastern Mexico. Forest Ecology and
Management 124: 231-239.

Penman, H.L., 1948. Natural Evaporation from open water, bare soil and grass.
Proceedings of the Royal Society, Series A. 193: 120-145.

Penman, H.L., 1963. Vegetation and Hydrology. Tech. Comm. No. 53, Commonwealth
Bureau of Soils, Harpenden, England. 125 pp.









Rutter, A. J., 1963. Studies in the water relations of pinus sylvestris in plantation
conditions I. Measurements of rainfall and interception. Journal of Ecology 51: 191-
203.

Rutter, A. J., Dershaw, K. A., Robins, P. C., Morton, A. J., 1971. A predictive model of
rainfall interception in forests. I. Derivation of the model from observations in a
plantation of Corsican pine. Agricultural Meteorology 9: 367-384.

Rutter, A. J., Morton, A. J., 1977. A predictive model of rainfall interception in forests.
III. Sensitivity of the model to stand parameters and meteorological variables. Journal
of Applied Ecology, Volume 14, Issue 2: 567-588.

Rutter, A. J., Morton, A. J., Robins, P. C., 1975. A predictive model of rainfall
interception in forests. II. Generalization of the model and comparison with
observations in some coniferous and hardwood stands. Journal of Applied Ecology 12:
367-380.

Teklehaimanot, Z., Jarvis, P. G., 1991. Direct Measurement of Evaporation of
Intercepted Water from Forest Canopies. Journal of Applied Ecology. Volume 28,
Issue 2: 603-618.

Valente, F., David, J. S., Gash, J. H. C., 1997. Modeling interception loss for two sparse
eucalypt and pine forests in central Portugal using reformulated Rutter and Gash
analytical models. Journal of Hydrology 190: 141-162.

van Dijk, A. I. J. M., Bruijnzeel, L. A., 2001. Modeling rainfall interception by
vegetation of variable density using an adapted analytical model. Part 1. Model
description. Journal of Hydrology 247: 230-238.

van Dijk, A. I. J. M., Bruijnzeel, L. A., 2001. Modeling rainfall interception by
vegetation of variable density using an adapted analytical model. Part 2. Model
validation for a tropical upland mixed cropping system. Journal of Hydrology 247:
239-262.















BIOGRAPHICAL SKETCH

Malcolm Lewis Bryant was born in Morgantown, West Virginia, on September 15,

1965. He is a graduate of Vanguard High School in Ocala, Florida. He attended Georgia

Southern University from 1983 to 1984 and Central Florida Community College from

1985 to 1987 where he earned an Associate of Science degree in radiation protection.

From 1987 to 1998 he worked in the commercial nuclear power industry as a health

physics technician and radiological engineer. In 1993 he married his wife, Kim. He

earned a Bachelor of Science degree in technology (nuclear) in 1997 from Regent's

College in Albany, New York. In 2000 his son, Trevor, was born. He received a second

Bachelor of Science degree in civil engineering from the University of Florida in 2001.

He enrolled in the Master of Science in water resources engineering in the summer of

2001.