<%BANNER%>

Evaluation of Accuracy and Longevity of Expanding-Disk Rain Sensors

Permanent Link: http://ufdc.ufl.edu/UFE0041854/00001

Material Information

Title: Evaluation of Accuracy and Longevity of Expanding-Disk Rain Sensors
Physical Description: 1 online resource (105 p.)
Language: english
Creator: Meeks, Leah
Publisher: University of Florida
Place of Publication: Gainesville, Fla.
Publication Date: 2010

Subjects

Subjects / Keywords: irrigation, landscape, rain, residential, sensors, water
Agricultural and Biological Engineering -- Dissertations, Academic -- UF
Genre: Agricultural and Biological Engineering thesis, M.E.
bibliography   ( marcgt )
theses   ( marcgt )
government publication (state, provincial, terriorial, dependent)   ( marcgt )
born-digital   ( sobekcm )
Electronic Thesis or Dissertation

Notes

Abstract: Rain sensors are devices that connect to automatic irrigation systems to interrupt scheduled irrigations with sufficient rainfall. The goal of this research was to evaluate the performance of expanding-disk rain sensors. The primary objectives of this study were to A) evaluate rain sensor accuracy with time with respect to the selected rainfall setting, B) evaluate the amount of time rain sensors remained in interruption mode (open-switch mode) after a rainfall event, C) quantify potential irrigation savings for different rainfall settings compared with a time-based schedule, and D) determine if the hygroscopic disks in the rain sensors change length with time. Ten treatments were established at the University of Florida Agricultural and Biological Engineering Department campus turfgrass plots, Gainesville, Florida. Mini-Clik rain sensors with rainfall settings of 3, 6, and 13 mm (3MC, 6MC, and 13MC) and Wireless Rain-Clik (WL) rain sensors had four replicates for each treatment. Treatments Hunter, Irritrol, and Toro had rainfall settings of 6 mm with eight replicates each. Three other Mini-Clikregistered trademark rain sensor treatments (3R, 6R, and 13R had rainfall settings of 3, 6, and 13 mm, respectively) each had three replicates. This experiment was carried out during a relatively dry period with rainfall on 28% of the days and 15% less rainfall than average. WL, 3MC, 6MC, 13MC, Hunter, Irritrol, and Toro required 3.2, 1.9, 1.6, 6.6, 3.8, 4.3, and 5.8 mm for open-switch mode, respectively. Accuracy ranged from 27% to 97%. The rain sensor accuracy had percentile point change from -36% to 59% with time, where a negative value indicated a decrease in accuracy. Dry-out is the amount of time a rain sensor stays in open-switch mode. Rain sensors dried-out within 24 hours 79% of the time. Changing the dry-out vent settings from fully open to fully had no effect on potential irrigation savings. Dry-out occurred with decreasing relative humidity and increasing temperature and solar radiation. The hygroscopic disks in expanding-disk rain sensors increased in length after continuous rainfall exposure. Rain sensors with higher rainfall settings had the most increase. The disk length change did not influence accuracy. The potential water savings for a 2 d/wk and 1 d/wk irrigation schedule 13MC were 14% and 13% and the average for all other treatments was 24% and 21%, respectively. Potential irrigation savings should be considered in relation to the accuracy of rain sensors. Rainfall settings of 3, 6, and 13 mm are adequate for rain sensors in central Florida. If the rainfall setting needs to be changed after more than 3 months of use, a new rain sensor or new expanding disks be installed. For the best accuracy, Hunter Mini-Clik rain sensors should be replaced after 1 year while Irritrol RSF 1000 and Toro TWRS rain sensors do not need to be replaced for at least 3 years. Rain sensors could increase water savings to homeowners and have environmental benefits but should not be used in applications requiring high accuracy.
General Note: In the series University of Florida Digital Collections.
General Note: Includes vita.
Bibliography: Includes bibliographical references.
Source of Description: Description based on online resource; title from PDF title page.
Source of Description: This bibliographic record is available under the Creative Commons CC0 public domain dedication. The University of Florida Libraries, as creator of this bibliographic record, has waived all rights to it worldwide under copyright law, including all related and neighboring rights, to the extent allowed by law.
Statement of Responsibility: by Leah Meeks.
Thesis: Thesis (M.E.)--University of Florida, 2010.
Local: Adviser: Dukes, Michael D.
Electronic Access: RESTRICTED TO UF STUDENTS, STAFF, FACULTY, AND ON-CAMPUS USE UNTIL 2011-08-31

Record Information

Source Institution: UFRGP
Rights Management: Applicable rights reserved.
Classification: lcc - LD1780 2010
System ID: UFE0041854:00001

Permanent Link: http://ufdc.ufl.edu/UFE0041854/00001

Material Information

Title: Evaluation of Accuracy and Longevity of Expanding-Disk Rain Sensors
Physical Description: 1 online resource (105 p.)
Language: english
Creator: Meeks, Leah
Publisher: University of Florida
Place of Publication: Gainesville, Fla.
Publication Date: 2010

Subjects

Subjects / Keywords: irrigation, landscape, rain, residential, sensors, water
Agricultural and Biological Engineering -- Dissertations, Academic -- UF
Genre: Agricultural and Biological Engineering thesis, M.E.
bibliography   ( marcgt )
theses   ( marcgt )
government publication (state, provincial, terriorial, dependent)   ( marcgt )
born-digital   ( sobekcm )
Electronic Thesis or Dissertation

Notes

Abstract: Rain sensors are devices that connect to automatic irrigation systems to interrupt scheduled irrigations with sufficient rainfall. The goal of this research was to evaluate the performance of expanding-disk rain sensors. The primary objectives of this study were to A) evaluate rain sensor accuracy with time with respect to the selected rainfall setting, B) evaluate the amount of time rain sensors remained in interruption mode (open-switch mode) after a rainfall event, C) quantify potential irrigation savings for different rainfall settings compared with a time-based schedule, and D) determine if the hygroscopic disks in the rain sensors change length with time. Ten treatments were established at the University of Florida Agricultural and Biological Engineering Department campus turfgrass plots, Gainesville, Florida. Mini-Clik rain sensors with rainfall settings of 3, 6, and 13 mm (3MC, 6MC, and 13MC) and Wireless Rain-Clik (WL) rain sensors had four replicates for each treatment. Treatments Hunter, Irritrol, and Toro had rainfall settings of 6 mm with eight replicates each. Three other Mini-Clikregistered trademark rain sensor treatments (3R, 6R, and 13R had rainfall settings of 3, 6, and 13 mm, respectively) each had three replicates. This experiment was carried out during a relatively dry period with rainfall on 28% of the days and 15% less rainfall than average. WL, 3MC, 6MC, 13MC, Hunter, Irritrol, and Toro required 3.2, 1.9, 1.6, 6.6, 3.8, 4.3, and 5.8 mm for open-switch mode, respectively. Accuracy ranged from 27% to 97%. The rain sensor accuracy had percentile point change from -36% to 59% with time, where a negative value indicated a decrease in accuracy. Dry-out is the amount of time a rain sensor stays in open-switch mode. Rain sensors dried-out within 24 hours 79% of the time. Changing the dry-out vent settings from fully open to fully had no effect on potential irrigation savings. Dry-out occurred with decreasing relative humidity and increasing temperature and solar radiation. The hygroscopic disks in expanding-disk rain sensors increased in length after continuous rainfall exposure. Rain sensors with higher rainfall settings had the most increase. The disk length change did not influence accuracy. The potential water savings for a 2 d/wk and 1 d/wk irrigation schedule 13MC were 14% and 13% and the average for all other treatments was 24% and 21%, respectively. Potential irrigation savings should be considered in relation to the accuracy of rain sensors. Rainfall settings of 3, 6, and 13 mm are adequate for rain sensors in central Florida. If the rainfall setting needs to be changed after more than 3 months of use, a new rain sensor or new expanding disks be installed. For the best accuracy, Hunter Mini-Clik rain sensors should be replaced after 1 year while Irritrol RSF 1000 and Toro TWRS rain sensors do not need to be replaced for at least 3 years. Rain sensors could increase water savings to homeowners and have environmental benefits but should not be used in applications requiring high accuracy.
General Note: In the series University of Florida Digital Collections.
General Note: Includes vita.
Bibliography: Includes bibliographical references.
Source of Description: Description based on online resource; title from PDF title page.
Source of Description: This bibliographic record is available under the Creative Commons CC0 public domain dedication. The University of Florida Libraries, as creator of this bibliographic record, has waived all rights to it worldwide under copyright law, including all related and neighboring rights, to the extent allowed by law.
Statement of Responsibility: by Leah Meeks.
Thesis: Thesis (M.E.)--University of Florida, 2010.
Local: Adviser: Dukes, Michael D.
Electronic Access: RESTRICTED TO UF STUDENTS, STAFF, FACULTY, AND ON-CAMPUS USE UNTIL 2011-08-31

Record Information

Source Institution: UFRGP
Rights Management: Applicable rights reserved.
Classification: lcc - LD1780 2010
System ID: UFE0041854:00001


This item has the following downloads:


Full Text





EVALUATION OF ACCURACY AND LONGEVITY OF EXPANDING-DISK RAIN
SENSORS




















By

LEAH MEEKS


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

UNIVERSITY OF FLORIDA

2010

































2010 Leah Meeks
































To my grandmothers, Geraldine Rosalee Meeks and Clara Lena Davis









ACKNOWLEDGMENTS

I thank my mother, Sharon Lea Meeks, for her unconditional love and

encouragement, my fiance and best friend, James Anthony Hernandez, for his support,

and my aunt, the late Dr. Lynn Langer Meeks, for her inspiration.

I would like to thank the members of my graduate committee, Dr. Kati White

Migliaccio and Dr. Thomas Obreza, for their assistance on my research. A big thank you

goes to my advisor Dr. Michael D. Dukes for his guidance and the chance to experience

a new side of irrigation. I would also like thank Stacia Davis, Bernard Cardenas-

Lailhacar, and Mary Shedd McCready for their help on this research project.









TABLE OF CONTENTS

page

A C K N O W L E D G M E N T S ......... ........... ... ............................................................... 4

L IS T O F T A B L E S ......................................... ............................. ..................... ................. 7

L IS T O F F IG U R E S .......................... ................ .......................................... 9

L IS T O F A B B R E V IA T IO N S ........................................................ .......................................... 14

A B S T R A C T ........... ........ .. ............ .. ......................................... ........... ..... 15

CHAPTER

1 IN T R O D U C T IO N .......... ................................................... ............ .... .... ... ....... 17

2 EXPANDING-DISK RAIN SENSOR ACCURACY ..................................................31

In tro d u c tio n .................. ...................................... ...................................... 3 1
M materials and M methods ...... .......... ................................................. 34
Treatments .......... ... ............................... 34
M o n ito ri n g .................................................................................................... ........ 3 5
S tatistica l A na lysis ........ .... ..... ... ......... ... ............... ............................... 36
R results a nd D discussion ......... .... ........ ...... ................... ............. .. ......... ...... 36
C lim a ctic C o nd itio ns ........... ............... .............................. ...... ...... .... .... .. 36
Number of Times in Open Switch Mode .............. .. ............. .............. 37
A ccu racy o f R a in S e nso rs ......... ........ ........ ............... .... .. .......................................40
Change in Accuracy of Rain Sensors over Time.............. ...... ................ 40
S um m ary a nd C onclusions.............................. ............... ...................... ............... 4 1

3 EXPANDING-DISK RAIN SENSOR DRY-OUT AND POTENTIAL IRRIGATION
S A V IN G S ......... .. ..................... ......... ........ ....................................... 5 9

In tro d u ctio n ...................... .. .. .......... ... .. .............................................. 5 9
M ate ria ls a nd M methods ................. ...... .. .................................................................... 6 1
T re atm e nts ......... ............. ................................................................ .. ...... .. ........ .. ... 6 1
M o n ito ri n g .................................................................................................... ........ 6 2
S tatistica l A na lys is ......... ..................................... ................. ... ... 6 4
R results a nd D discussion ......... .... ........ ...... ................... ............. .. ......... ...... 64
C lim actic Conditions ......... ............ .. ... .......... .... ...................... .. 64
Time in Open Switch Mode (Dry-Out) .................. .........................................65
D ry-o ut T ra cking ................................................................................... ....................... 6 7
P potential Irrigation S avings....................... ..... ............................... ... ............ 68
S um m ary a nd C onclusions.......................... .. ....................................... ............... 69









4 RELATIONSHIP BETWEEN EXPANDING-DISK RAIN SENSOR DISK
LENGTH AND PERFORMANCE .................... ............................................ 83

In tro d u ctio n .................. ..................................... .................................. .... ....... .... 8 3
M materials and M methods ...... .......... ................................................. 84
Treatments .................... ......... .......... ................. 85
M o n ito ring ....................................................... 8 5
Statistical A analysis ........ ...... ........ .... ......... ................... 86
Results and Discussion ........... ...... ................ ........ .. ......... .... ................. 86
Length by Installation date and setting................. .......... ............................. 86
Disk Length and Traveling Distance ................... ................... 87
Effect on Interruption Performance ............................... ................ 88
S um m ary and C conclusions ........................... ......... ...... .................... ................. 88

5 CONCLUSIONS AND FUTURE WORK... ............................................................96

C o n c l u s i o n s ........................ .. ............... .. ......................... ..................... ....................... 9 6
F u tu re W o rk ...................... .. ............. .. ................................................. 9 9

L IS T O F R E F E R E N C E S ......... ........... .. ............................................... ...................... .... 10 0

B IO G R A P H IC A L S K E T C H ........................ ............................................. .................. ... ..... 10 5









LIST OF TABLES


Table page

2-1 Rain sensor treatment description ......... ........... .. ..................... 43

2-2 Summary of functionality problems for treatments and replicates ......................43

2-3 Average depth of rainfall before rain sensors switched to Open Switch Mode....44

2-4 Summary of changes in accuracy for change in rainfall required for Open
Sw itch M ode................................................................. ... ...... .......... .. 44

3-1 Treatm e nt description .............................................. ........................ .............. 7 1

3-2 Monthly irrigation depth to replace historical evapotranspiration values based
on Dukes and Haman (2002a). Run times are based on an irrigation
application rate of 38 mm/hr assuming system efficiency of 60% and
considering effective rainfall. The Reduced UF IFAS irrigation schedule is
60% of the UF IFAS irrigation schedule. ........................................ ....... .......... 71

3-4 Total potential water savings per treatment for all treatments compared with
a 2 d/wk and a 1 d/wk irrigation schedule for the study period (Oct/Nov 2006
to 1 Dec 2009). ..... ....................................... .................. 72

3-5 Variation in total potential water savings per replicate for the WL and MC
treatments compared with UF IFAS 2 d/wk irrigation recommendations..............72

3-6 Variation in total potential water savings per replicate for the Hunter, Irritrol,
and Toro treatments compared with UF IFAS 2 d/wk irrigation
re co m m e nd atio ns ....................... .......................................... .............. ... 7 3

3-7 Variation in total potential water savings per replicate for the WL and MC
treatments compared with UF IFAS 1 d/ wk irrigation recommendations ............ 73

3-8 Variation in total potential water savings per replicate for the Hunter, Irritrol,
and Toro treatments compared with UF IFAS 1 d/ wk irrigation
re co m m e nd atio ns ....................... .......................................... .............. ... 7 3

4-1 Description of rain sensors details for each treatment.......... ........................ 90

4-2 Average disk length for each treatment at two intervals: initial and final (276
days of installation) .................................................................................. 90

4-3 Average disk length for the treatments installed 13 February 2009 at three
intervals: initial, 81 days of installation, and final (276 days of installation).......... 91









4-4 Disk length for replicates of treatments installed 25 March 2005 at three
intervals: initial (February), 81 days of installation (May), and final
(November, 276 days of installation) ................................ ... ................. 91

4-5 Disk length for replicates of treatments installed 13 February 2009 at three
intervals: initial (February), 81 days of installation (May), and final
(November, 276 days of installation) ................................ ... ................. 91

4-6 Comparison of average length change and travel distance from closed-switch
mode to open-switch mode of treatments installed in 13 February 2009. The
February travel distance was measured on a rain sensor before installation...... 92

4-7 Comparison of average length change of each treatment from 13 February
2009 to 16 November 2009 and the travel distance each treatment from
closed-switch mode to open-switch mode. Travel distance was measured at
the end of the study. ......... .... .. ...... .............. .. ............... ...... 92









LIST OF FIGURES


Figure page

2-1 WL (model Wireless Rain-Clik, Hunter Industries, Inc., San Marcos, CA) rain
sensor. A) Expanding disks inside ventilation window, B) quick-response
expanding disks, C) Ventilation window adjustment knob, D) antenna................. 45

2-2 MC (model Mini-Clik, Hunter Industries, Inc., San Marcos, CA) rain sensor.
A) Rainfall threshold setting slots, B) expanding disks, C) dry-out adjustment
ring and vents. .................................... ............................... ......... 45

2-3 Irritrol (model RFS 1000, Irritrol Systems, Inc., Riverside, CA.) rain sensor. A)
Rainfall threshold setting slots, B) dry-out adjustment ring, C) antenna...............46

2-4 Toro (model TWRS, Toro Company, Inc., Riverside, CA) rain sensor. A)
Rainfall threshold setting slots, B) dry-out vent, C) antenna ................................ 46

2-5 Detail of expanding disk material and threshold adjustment of Mini-Clik
(Hunter Industries, Inc.) rain sensor. A) Rainfall threshold setting slots, B)
hygroscopic expanding disk m material. ...................................................................... 47

2-6 Research site located at the University of Florida Agricultural and Biological
Engineering facilities. Shown: weather station on left, WL, 3MC, 3MC, and
13MC treatments installed on left board, and Hunter, Irritrol, and Toro on
rig h t b o a rd ................ ... ...... .... ............................................................... 4 7

2-7 Installed Hunter Wireless Rain-Clik, three on left and one on right, and Mini-
Clik rain sensors with four wireless receivers for WL and data logger..................48

2-8 Installed Hunter, Irritrol, and Toro rain sensors, left to right, with wireless
receivers (Irritrol on left and Toro on right), and data logger.................................. 48

2-9 Relationship between manual rain gauge and weather station tipping bucket
rain gauge with the calibration factor applied to the tipping bucket data with
more than 15 mm of rainfall .... ...... ........... ............................ .................49

2-10 Relationship of rain events greater than 15mm between manual rain gauge
and weather station tipping bucket rain gauge with the calibration factor
applied to the tipping bucket data. ......... ............. ......... ................. 49

2-11 Relationship of rain events less than 15 mm between manual rain gauge and
weather station tipping bucket rain gauge without the calibration factor
applied to the tipping bucket data. ......... ............. ......... ................. 50

2-12 Comparison of monthly and cumulative rainfall during the study period for WL
and MC treatments and average historical rainfall for north central Florida.........50









2-13 Comparison of monthly and cumulative rainfall during the study period for
Hunter, Irritrol, and Toro treatments and average historical rainfall for north
central Florida. ............. ...... .......... ................ ........... 51

2-14 Cumulative and daily rainfall during the WL and MC treatments study period
with the rainfall setting and the respective theoretical number of times each
should have gone into O SM ................................... ....................... 52

2-15 Cumulative and daily rainfall during the Hunter, Irritrol, and Toro treatments
study period with the rainfall setting and the respective theoretical number of
times each should have gone into OSM ......... .... ............... ....... ......................53

2-16 Cumulative number of times into OSM for WL and MC treatments. Data from
28 January to 9 June 2008 are not included due to all WL replicates not
functioning. Erratic replicates within treatments are not included after their
respective improper functioning dates (WL-B 21 September 2007 and 13MC-
C 8 July 2008). Numbers with different letters indicate a statistical difference
using Tukey-Kramer adjusted p-values of p<0.05. .......... .. ........ ................. 54

2-17 Cumulative number of times the WL replicates went into OSM. WL stopped
functioning on 21 Septem ber 2007. ....................................... ........ .. .............. .....54

2-18 Cumulative number of times the 3MC replicates went into OSM. The
Theoretical value is the number of times the replicates should have gone into
O S M based o n rainfa ll ........... ..... ....... ...... .............. ............ .. ........ .... 55

2-19 Cumulative number of times the 6MC replicates went into OSM. The
Theoretical value is the number of times the replicates should have gone into
O S M based o n rainfa ll ........... ..... ....... ...... .............. ............ .. ........ .... 55

2-20 Cumulative number of times the 13MC replicates went into OSM. The
Theoretical value is the number of times the replicates should have gone into
O S M based on rainfa ll ........... ..... ....... ...... .............. ............ .. ........ .... 56

2-21 Cumulative number of times into OSM for Hunter, Irritrol, and Toro
treatments. The Theoretical value is the number of times the replicates
should have gone into OSM based on rainfall. Numbers with different letters
indicate a statistical difference using Tukey-Kramer adjusted p-values of
p <0 .0 5 ............... ..... .. ........ ............................................ 56

2-22 Cumulative number of times Hunter into OSM. The Theoretical value is the
number of times the replicates should have gone into OSM based on rainfall.... 57

2-23 Cumulative number of times Irritrol went into OSM. The Theoretical value is
the number of times the replicates should have gone into OSM based on
ra i nfa ll ................ ..... .... ......... ........................................... 5 7









2-24 Cumulative number of times Toro went into OSM. The Theoretical value is
the number of times the replicates should have gone into OSM based on
ra i nfa ll ................ ..... .... ......... ........................................... 5 8

2-25 Accuracy of each treatment with a set point over the study period with an
average (solid line) and 95% confidence bands (dashed lines). .......................... 58

3-1 Histogram and cumulative frequency of dry-out time (time from the end of the
rain event until the sensors returns to closed-switch mode) for the WL
treatment average. ......... .... ........ ....... .. ........... .. .. .... 74

3-2 Histogram and cumulative frequency of dry-out time (time from the end of the
rain event until the sensors returns to closed-switch mode) for the 3MC
treatm e nt a average ................................................................. ............. ...... 74

3-3 Histogram and cumulative frequency of dry-out time (time from the end of the
rain event until the sensors returns to closed-switch mode) for the 6MC
treatm e nt a average ................................................................. ............. ...... 75

3-4 Histogram and cumulative frequency of dry-out time (time from the end of the
rain event until the sensors returns to closed-switch mode) for the 13MC
treatm e nt a average ................................................................. ............. ...... 75

3-5 Histogram and cumulative frequency of dry-out time (time from the end of the
rain event until the sensors returns to closed-switch mode) for the Hunter
treatm e nt a average ................................................................. ............. ...... 76

3-6 Histogram and cumulative frequency of dry-out time (time from the end of the
rain event until the sensors returns to closed-switch mode) for the Irritrol
treatm e nt a average ................................................................. ............. ...... 76

3-7 Histogram and cumulative frequency of dry-out time (time from the end of the
rain event until the sensors returns to closed-switch mode) for the Toro
treatm e nt a average ................................................................. ............. ...... 77

3-8 Histogram and cumulative frequency of dry-out time (time from the end of the
rain event until the sensors returns to closed-switch mode) for the Hunter
treatment average with the dry-out vents fully open (8 November 2008 to 2
July 2009). ............. ..... .. .............. .. ....... ........... 77

3-9 Histogram and cumulative frequency of dry-out time (time from the end of the
rain event until the sensors returns to closed-switch mode) for the Hunter
treatment average with the dry-out vents fully closed (2 July2009 to 31
D ecem ber 2009) ......... ........ ........ ... ................... .............. 78

3-10 Histogram and cumulative frequency of dry-out time (time from the end of the
rain event until the sensors returns to closed-switch mode) for the Irritrol









treatment average with the dry-out vents fully open (8 November 2008 to 2
J u ly 2 0 0 9 ). ....................................................................................... 7 8

3-11 Histogram and cumulative frequency of dry-out time (time from the end of the
rain event until the sensors returns to closed-switch mode) for the Irritrol
treatment average with the dry-out vents fully closed (2 July2009 to 31
D ecem ber 2009) ......... .... ............ .................. .............. .... 79

3-12 Histogram and cumulative frequency of dry-out time (time from the end of the
rain event until the sensors returns to closed-switch mode) for the one day
water delay setting for four the Toro replicates. ................ ... ....... ................. 79

3-13 Histogram and cumulative frequency of dry-out time (time from the end of the
rain event until the sensors returns to closed-switch mode) for the three day
water delay setting for four the Toro replicates. ................ ... ....... ................. 80

3-14 Dry-out tracking of average disk length for each treatment for the natural rain
eve nt o n 10 J u ly 2 0 0 9 ....... .... ........ ............ .................. .. .................. ... 8 0

3-15 Dry-out tracking of average disk length for each treatment for the manual rain
event on 18 September 2009. ................ ..................81

3-16 Dry-out tracking of average disk length for each treatment and temperature
for the rain event on 10 July2009. ........................ ........................................ 81

3-17 Dry-out tracking of average disk length for each treatment and solar radiation
for the rain event on 10 July2009. ........................ ........................................ 82

3-18 Dry-out tracking of average disk length for each treatment and relative
humidity for the rain event on 10 July 2009 ..................... ................. 82

4-1 Mini-Clik (Hunter Industries, Inc.) rain sensor expanding disks installed in
March 2005 (left) and February 2009 (right) set at 13 mm measured in
August 2009 (1600 and 179 days of installation, respectively) ............................. 92

4-2 Mini-Clik (Hunter Industries, Inc.) rain sensors expanding disks installed in
2005 with settings (left to right) of 3 mm, 6 mm, and 13 mm and lengths 19.2,
19.8, and 20.1 mm, respectively, after 1,600 days of installation. ....................... 93

4-3 Cumulative and daily rainfall during the study period with number of rainfall
events greater than the different rain sensor rainfall settings. ................................94

4-4 Average hygroscopic disk length for Mini-Clik rain sensors installed in 2005
(MC) and 2009 (R) over installation time. The 2009 rain sensors were newly
installed on day zero. ..................................... ................ ....................... 95









4-5 Accuracy for each setting during the first 300 days of rain sensor installation
compared to change in disk length. Accuracy data are the average amount of
rainfall for open-switch mode for a given rainfall event for a treatment................ 95









LIST OF ABBREVIATIONS


Avg Average

CSM Closed Switch Mode

CV Coefficient of Variance

d/wk Day per week or days per week

ET Evapotranspiration

NOAA National Oceanic and Atmospheric Administraion

OSM Open Switch Mode

RS Rain sensor

SMS Soil moisture sensor

UF IFAS University of Flroida Institute of Food and Agriculutal Sciences

USCB United State Census Bureau









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

EVALUATION OF ACCURACY AND LONGEVITY OF EXPANDING-DISK RAIN
SENSORS

By

Leah Meeks

August 2010

Chair: Michael D. Dukes
Major: Agricultural and Biological Engineering

Rain sensors are devices that connect to automatic irrigation systems to interrupt

scheduled irrigations with sufficient rainfall. The goal of this research was to evaluate

the performance of expanding-disk rain sensors. The primary objectives of this study

were to A) evaluate rain sensor accuracy with time with respect to the selected rainfall

setting, B) evaluate the amount of time rain sensors remained in interruption mode

(open-switch mode) after a rainfall event, C) quantify potential irrigation savings for

different rainfall settings compared with a time-based schedule, and D) determine if the

hygroscopic disks in the rain sensors change length with time.

Ten treatments were established at the University of Florida Agricultural and

Biological Engineering Department campus turfgrass plots, Gainesville, Florida. Mini-

Clik rain sensors with rainfall settings of 3, 6, and 13 mm (3MC, 6MC, and 13MC) and

Wireless Rain-Clik (WL) rain sensors had four replicates for each treatment. Treatments

Hunter, Irritrol, and Toro had rainfall settings of 6 mm with eight replicates each. Three

other Mini-Clik rain sensor treatments (3R, 6R, and 13R had rainfall settings of 3, 6,

and 13 mm, respectively) each had three replicates.









This experiment was carried out during a relatively dry period with rainfall on 28%

of the days and 15% less rainfall than average. WL, 3MC, 6MC, 13MC, Hunter, Irritrol,

and Toro required 3.2, 1.9, 1.6, 6.6, 3.8, 4.3, and 5.8 mm for open-switch mode,

respectively. Accuracy ranged from 27% to 97%. The rain sensor accuracy had

percentile point change from -36% to 59% with time, where a negative value indicated a

decrease in accuracy.

Dry-out is the amount of time a rain sensor stays in open-switch mode. Rain

sensors dried-out within 24 hours 79% of the time. Changing the dry-out vent settings

from fully open to fully had no effect on potential irrigation savings. Dry-out occurred

with decreasing relative humidity and increasing temperature and solar radiation. The

hygroscopic disks in expanding-disk rain sensors increased in length after continuous

rainfall exposure. Rain sensors with higher rainfall settings had the most increase. The

disk length change did not influence accuracy. The potential water savings for a 2 d/wk

and 1 d/wk irrigation schedule 13MC were 14% and 13% and the average for all other

treatments was 24% and 21%, respectively. Potential irrigation savings should be

considered in relation to the accuracy of rain sensors.

Rainfall settings of 3, 6, and 13 mm are adequate for rain sensors in central

Florida. If the rainfall setting needs to be changed after more than 3 months of use, a

new rain sensor or new expanding disks be installed. For the best accuracy, Hunter

Mini-Clik rain sensors should be replaced after 1 year while Irritrol RSF 1000 and Toro

TWRS rain sensors do not need to be replaced for at least 3 years. Rain sensors could

increase water savings to homeowners and have environmental benefits but should not

be used in applications requiring high accuracy.









CHAPTER 1
INTRODUCTION

Introduction to Water in Florida

Florida has an increasing need for water conservation measures. Between 1950

and 2000, the population of Florida increased 475% and total public supply withdrawals

increased 1,330% (Marella, 2004). In 2000, Florida was the largest user of groundwater

east of the Mississippi River (Hutson et al., 2004). Water withdrawals for public supply

in Florida in 2000 totaled 9.2 million cubic meters per day, of which 90% was obtained

from groundwater and 10% from surface water (Marella, 2004). Florida, Texas,

Nebraska, Arkansas, and California account for more than half of the fresh groundwater

use nationwide (Hutson et al., 2004).

The five categories of factors affecting water demands are population, climate,

socioeconomic conditions, water pricing, water conservation, and alternative supply

sources (Marella, 1992). Florida receives an average of 1350 mm of rainfall per year

(National Oceanic and Atmospheric Administration, 2003). Even with Florida's

significant rainfall, the combination of relatively well drained soils and dry periods mean

that irrigation is required to maintain landscape quality. Fifty-four percent of the

freshwater withdrawn in 2000 was between February and June (Marella, 2004). Non-

municipal irrigation withdrawals in 2000 were greatest in February through June during

drier conditions and the lowest in July through September when summer rain occurred

(Marella, 2004). These withdrawals coincide with an increase in public supply use

(Marella, 1992). The biggest stresses on water supply from agricultural and municipal

sectors occur during the same time of year making water conservation even more

critical.









Residential Irrigation Practices in Florida

Between 1970 and 2000, total freshwater withdrawals for public use increased by

176% (Marella, 2004). Florida ranks fourth in overall state population with an estimated

2009 population of 18.5 million (United States Census Bureau [USCB], 2009).

Residential irrigation has been reported to account for 64% of residential water use

(Haley et al., 2007). The volume of water required for residential irrigation continues to

increase with the increasing Florida population and the years of less-than-average

rainfall. From 2000 to 2005, Florida had a net population gain of approximately 1000

people per day (USCB, 2009). Extreme dry conditions occurred between February and

June of 2000 and the result was higher water demands from public supply primarily for

lawn irrigation during these months. A study of water use in Pinellas County Florida

found that the highest water use occurred in spring due to high evaporation and low

precipitation (Dukes and Haley, 2009). Lawn irrigation in central and south Florida

occurs throughout the entire year (Marella, 2004).

Florida, which receives more rainfall than all states other than Louisiana, requires

irrigation to meet the aesthetic demands of homeowners due to the wet /dry seasons

and well drained soils. It is estimated that 70% of single family homes in southwest

Florida have automatic irrigation systems (Tampa Bay Water, 2005). In a national study,

homes that only hand watered used 33% less water than those with in-ground systems

(Mayer, et al., 1999). In-ground systems generally run off of automatic timers instead of

homeowners turning on the irrigation system. Automatic timer controls on irrigation

systems in Florida have been reported to lead to a 47% increase in water use (Mayer,

et al., 1999). Automatic systems result in more irrigation than manually controlled









systems because people tend to "set-and-forget" and do not take climatic conditions into

account. Research in Florida found that homeowners irrigate in the late fall and winter

when turfgrass is dormant because it is inconvenient to change the settings of the timer

or there is a misunderstanding of the actual amount of water that should be applied

during the year (Haley et al., 2007).

Rain Sensors

Rain sensors are devices designed to interrupt the cycle of an automatic irrigation

system controller when a specific amount of rainfall has occurred (Dukes and Haman,

2002b). The rain sensor or its receiver is wired into an automatic irrigation controller.

When rain beyond a threshold has fallen, the rain sensor will interrupt the irrigation

controller circuit to potentially bypass an irrigation event depending on the irrigation

schedule. Evaporation removes the water from the rain sensor so that irrigation will be

allowed. The water-savings potential, simple design, reliability, low cost, and ease of

installation have made them popular (Dewey, 2003). Until the addition of soil moisture

sensors in recent years, rain sensors were the only technology available commercially

for residential irrigation reduction. Rain sensors have the potential to improve irrigation

efficiency, reduce wear on the irrigation system, and reduce runoff and deep percolation

(Dukes and Cardenas-Lailhacar, 2007). States and municipalities throughout the

country have mandated the use of rain sensors to conserve water. It has been

estimated that half of single-family homes in Florida have in-ground irrigation systems

with automatic timers, of which 25% report having rain sensor shutoff devices

(Whitcomb, 2005).









Types

Several types of rain sensors are on the market. The sensors can be adjusted to

interrupt at different depths of rainfall, generally between 3 and 25 mm. One type

collects the rain water in a cup and interrupts the irrigation based on a preset weight of

water. A disadvantage of the water weight devices is that debris can get into the

collection cup and cause the system to interrupt irrigation without sufficient rainfall.

Another type has a set of electrodes that detect the water level in a small

collection dish (Dukes and Haman, 2002b). Debris is also a problem with the electrical

conductivity devices. The Rain Check TM (Rain Bird Corporation, Glendora, CA) is a rain

sensor that measures the amount of rainfall with two electrodes in a collection cup. The

stainless steel probes can be adjusted to interrupt irrigation between 3 and 13 mm of

rainfall.

The most commonly used rain sensors in Florida are expanding-disk rain sensors.

Hygroscopic disks in the sensor expand proportionally to the amount of rainfall. The

swelling and contracting of the disks opens and closes a switch. Expanding-disk

sensors require less maintenance and are cheaper than other sensors.

There are different models of rain sensors that can be used depending on the

location characteristics. Sensors can operate as "normally closed" (normally allow

irrigation) or "normally open" (normally does not allow irrigation). Most systems run on

"normally closed" rain sensors. For more versatility, expanding-disk rain sensors are

available in wired or wireless models. Some wireless models allow for the sensor to be

placed up to 300 feet from the irrigation controller (Hunter Industries, 2005).









Evaporation for Dry-out

Rain sensors rely on evaporation to allow irrigation. In the case of "normally

closed" rain sensors, the switch becomes closed after the dry-out period and the

irrigation system circuit is complete. Dry-out settings can be adjusted for most sensors.

A longer dry-out time has the potential to interrupt more scheduled irrigation cycles by

the irrigation controller. The dry-out setting should be set so that it matches the drying

rate of the site's soil (Dewey, 2003). Dry-out is determined by weather conditions such

as temperature, wind, solar radiation, and relative humidity.

Installation

Proper installation is critical to achieve water savings. In a study of single-family

homes in Florida, anecdotal evidence suggests that rain sensors are often improperly

installed (Whitcomb, 2005). The sensors need to be exposed to normal rainfall (Dewey,

2003). Inappropriate installation locations include in the spray path of sprinklers, under a

tree canopy, under leaky roof gutters, and in places easily vandalized. The effects of

sun and shade on dry-out should be considered when choosing sensor location. Unlike

many other irrigation sensors, once it is properly installed and set the rain sensor

settings do not have to be adjusted to achieve water savings (Dewey, 2003).

Water Savings

Water and cost savings vary among rain sensor model and setting. Substantial

savings can be obtained during a year of average rainfall in Florida (Dukes and Haman,

2002b). In a study evaluating rain sensors at different rainfall settings when compared

with a treatment irrigating 2 days/week without a rain sensor, a 3 mm set point reduced

irrigation 30% while a 25 mm set point reduced irrigation only 3% (Cardenas-Lailhacar









and Dukes, 2008). Other variables affecting savings include rain frequency, whether or

not the controller is left on for automatic operation, and the amount of water applied by

the system per cycle (Dukes and Haman, 2002b). Marella (1992) suggests that rain

sensors are part of long-term water conservation measures for reducing residential

irrigation.

Previous Studies Involving Rain Sensors and Smart Controllers

While research on rain sensors alone is very limited, there have been residential

irrigation studies involving rain sensors. Many of these studies have been conducted in

the Southeastern United States due to the relatively high amount of rainfall compared to

the rest of the country.

Rain Sensor Accuracy Testing

A study at the University of Florida Agricultural and Biological Engineering

Department turfgrass plots in Gainesville, Florida by Cardenas-Lailhacar and Dukes

assessed the accuracy of rain sensors at different set points. The data for the rain

sensors (Hunter Industries, Inc., San Marcos, CA) were collected 25 March 2005

through 31 December 2005. The four treatments included three Mini-Clik rain sensors

with different settings (3 mm, 13 mm, and 25 mm) and one Wireless Rain-Clik. These

rain sensors were not connected to an irrigation system. The accuracies for the 3, 13,

and 25 mm settings were 88%, 77%, and 98%, respectively. A longer study should also

be conducted on rain sensors to evaluate the Hunter Mini-Clik's 5-year warranty

(Cardenas-Lailhacar and Dukes, 2008).









Potential Rain Sensor Irrigation Savings

The rain sensor accuracy study also investigated potential irrigation savings with

the addition of a rain sensor. The average depth of potential water savings for the

wireless, 3 mm, 13 mm, and 25 mm sensors were 558, 337, 468, and 38 mm,

respectively. This study indicated that irrigation water savings with rain sensors was

dependent on the rain sensor settings (Cardenas-Lailhacar and Dukes, 2008). This

same conclusion was reached in a study in Citra, Florida with rain sensors set at 3 mm

and 6 mm during a relatively dry period. The savings for the 3 mm and 6 mm set points

were 25% and 17%, respectively, when compared with a time-based schedule with no

water conservation devices (McCready et al., 2009).

Multiple studies involving soil moisture sensors (SMS) and evapotranspiration (ET)

controllers conducted at the University of Florida have had a rain sensor included in a

treatment for comparison. One study in Gainesville, Florida found that the addition of a

rain sensor set to interrupt irrigation with 6 mm of rainfall to a time clock set to irrigation

2 days/week could reduce irrigation by 34% (Cardenas-Lailhacar et al., 2008). In a

study with ET controllers and rain sensors in southwestern Florida, an automatic

irrigation system with a rain sensor conserved 21% of water compared with an irrigation

controller operating on a time-based schedule (Davis et al., 2009). Haley and Dukes

(2007) conducted a study that included rain sensors and educational material about

irrigation controller scheduling. The addition of a rain sensor to a controller conserved

19%; the combination of a rain sensor and educational materials increased savings

58%. These studies give insight into the effects of including a rain sensor on an active

automatic irrigation system.









State Statutes


State governments are adding rain sensor devices to water conservation statute

measures. The fact that more states are considering or including rain sensors in water

conservation statutes indicates that more research into the effectiveness of rain sensors

is necessary.

Connecticut. The Connecticut statute regarding rain sensors applies to automatic lawn

irrigation systems installed by state agencies or commercial enterprises. As of 1

October 2003, all installations must be equipped with a rain sensor that interrupts the

irrigation cycle after "adequate rainfall" occurs. The statute also allows municipalities to

pass ordinances requiring rain sensors on irrigation systems installed after 1 October

2003 within their respective jurisdictions. (Connecticut Statutes, Section 29-265)

Florida. Florida Statue 373.62 required that all automatic irrigation systems installed

after 1 May 1991 must have a maintained rain sensor or switch that overrides the

irrigation cycle after "adequate rainfall."

Recently, Florida passed a new water conservation statute effective 1 July2009.

The new bill requires all new automatic irrigation system installations be equipped with

rain sensors and old installations, predating 1 May 1991, be retrofitted with rain sensors.

Licensed contractors who install the irrigation systems must properly install or check for

proper installation. A licensed contractor who does not comply can be fined $50 for a

first offense, $100 for a second offense, and $250 for a third or subsequent offense.

Funds from the penalties will be used for water-conservation programs by local

government. (Florida Senate Bill 494)









Massachusetts. In January 2009, a bill was introduced in Massachusetts requiring an

interruption device on newly installed or renovated outdoor landscapes. The device

should override irrigation during periods of "sufficient moisture." All new irrigation

systems must be inspected every 3 years by a certified irrigation contractor, a certified

landscape irrigation auditor, or a certified irrigation designer. The bill would not apply to

golf courses. It was not passed as of 30 June 2009 and was pocket vetoed by Governor

Patrick (Moriarty, 2009). (Massachusetts Senate 186th General Court)

Minnesota. Minnesota's rain sensor statute effective 1 July 2003 requires that all

automatic irrigation systems have technology that interrupts operation in the event of

"sufficient rainfall." The device must be adjustable by the irrigation system user or

installer. (Minnesota Statutes, Chapter 44-F No. 335)

New Jersey. The New Jersey rain sensor statute requires that all automatic irrigation

systems installed after 8 December 2008 have a rain sensor. The device will override

the irrigation cycle with "adequate rainfall." (New Jersey Statutes, 52:24D-123.13)

Texas. The Texas ET controller statute applies to automatic irrigation systems owned

by the state or political subdivisions of the state greater than 0.25 or 0.50 hectares (0.1

or 0.2 acres) if using nonpotable water. New or existing irrigation systems must have an

on-site ET controller as of 1 September 2007. A remote ET controller can be used if the

weather station is less than 5 miles away from the site and has a rain/freeze sensor.

Both remote and on-site systems must have an independent rain/freeze sensor. The

statute encourages the passage of local ordinances on ET controllers. (Texas House

Bill 2299)









City and Area Ordinances

There is a growing trend for local governments to require rain sensors. Many of

these rain sensor requirements are included in irrigation restriction ordinances.

Compared with state rain sensor statutes, the city and area ordinances are more likely

to have fines for noncompliance.

Metropolitan North Georgia Water Planning District, Georgia. The Metropolitan

North Georgia Water Planning District, Georgia, regulation applies all systems receiving

water from the public water system, not including golf courses. After 1 January 2005, all

automatic irrigation systems must be equipped with a rain sensor. Persons in violation

by installing a system without a rain sensor can be fined up to $100 for each violation.

(Georgia Metropolitan North Georgia Water Planning District, Water conservation action

no. 4)

Water Authority of Great Neck North, New York. The Water Authority of Great Neck

North, New York ordinance effective 15 April 1994 includes irrigation times, rain

sensors, and soil moisture sensors. The ordinance applies to persons usi ng water that

is directly from the district or beneath the district if the source is underground. Users

must irrigate no more than 3 days/week depending on the address and all irrigations

must take place between 4:00 p.m. and 10:00 a.m. from 15 April to 1 November. The

rain sensor used must be able to detect a minimum of 1/8-inch (3 mm) of rain. Rain

sensors must be set to interrupt irrigation with a 1-inch (6 mm) or less of rainfall. (Water

Authority of Great Neck North New York)

Derby, Kansas. The ordinance for Derby, Kansas applies to all persons owning

property with an automatic irrigation system, whether or not the irrigation water is









supplied by the public water system. All automatic irrigation systems installed or

"substantially replaced" after 24 May2008 must have rain sensors. After 1 July2009, all

automatic irrigation systems must be either installed with rain sensors or retrofitted with

rain sensors. Rain sensors must be set to interrupt irrigation with at least 1/-inch (13

mm) of rainfall. City code enforcement officers can inspect systems with notice if there

is doubt of compliance. Fines to property owners can range from $25 to $500.

Cessation of public water supply service may be an additional penalty if a property

ownerfails or refuses to install and maintain a rain sensor. (Derby, Kansas Ordinance

No. 1932)

Cary, North Carolina. The Cary, North Carolina rain sensor ordinance defines an

irrigation system and rain sensors for residents. This ordinance applies to all systems

that receive water from the town of Cary. Effective 14 August 1997, new automatic

irrigation systems must be equipped with rain sensors. Existing systems must be

retrofitted with a rain sensor on or before 1 May 1998. Rain sensors must be set to

interrupt the irrigation cycle after 1-inch (6 mm) of rainfall and be located in an area of

full exposure. A rain sensor set to bypass irrigation at a setting greater than 1-inch (6

mm) is considered in non-compliance. On the second notice of non-compliance, the

property owner will be fined $100 each subsequent day not in compliance (i.e. $100

first day, $200 second day, etc). Termination of service can be a consequence of

continued noncompliance. (Cary, North Carolina, Ordinance section 36-84)

Harrisburg, South Dakota. The Harrisburg, South Dakota ordinance requires rain

sensors on all automatic irrigation systems installed after the effective day of 4 April

2006 that receive water from the city public supply. Rain sensors must be set to









interrupt the irrigation cycle after 1-inch (6 mm) of rainfall. A rain sensor set to bypass

irrigation at a setting greater than 1-inch (6 mm) is considered in noncompliance.

(Harrisburg, South Dakota Ordinances, Ordinance 2006-04, Chapter 8.01)

Arlington, Texas. The Arlington, Texas Lawn and Landscape Irrigation Conservation

ordinance calls for irrigation time restrictions and rain/freeze sensor. Automatic irrigation

systems cannot operate between 10:00 a.m. and 6:00 p.m. from 1 June to 30

September unless during periods of grass establishment, dust control, maintenance,

repair, or testing.

The rain/freeze sensor requirement does not apply to a single family residential or

duplex property or an individually metered townhome or condominium unity. The city

council created a list of approved rain/freeze sensors to be used. All new automatic

irrigation systems installed after 4 March 2005 within the city limits must be equipped

with a rain/freeze sensor. As of 4 March 2007, all existing systems must be retrofitted

with a rain/freeze sensor. Those in noncompliance will by guilty of a misdemeanor with

a possible fine of up to $500 for each violation. (Arlington, Texas, Lawn and landscape

irrigation conservation ordinance section 4.27)

Colleyville, Texas. The Water Conservation Ordinance of Colleyville, Texas was

created to conserve water by preventing automatic irrigation systems from running

during wet periods. All new automatic irrigation systems installed after 31 August 2006

must be equipped with a rain sensor. As of 31 August 2008, all existing systems must

be retrofitted with a rain sensor. Termination of service can be a consequence of

continued noncompliance. (Colleyville, Texas, Water conservation ordinance 06-1579)









Dallas, Texas. The purpose of the Dallas, Texas landscape irrigation ordinance is to

promote irrigation practices that prevent waste, conserve water resources for the most

beneficial and vital use, and protect the public health. Automatic irrigation systems

cannot operate between 10:00 a.m. and 6:00 p.m. from 1 April to 31 October. Irrigation

restrictions also include not significantly irrigating on impervious surfaces, not irrigating

with a broken or missing sprinkler head, and not properly maintaining the system.

Effective 1 January 2002, new automatic irrigation systems must be equipped with

rain/freeze sensors. Existing systems must be retrofitted with a rain/freeze sensor on or

before 1 January 2005. Violators are those who own, lease, or manage property with a

system not equipped with a rain/freeze sensor or operate and/or permit operation of an

irrigation system not in compliance with the sensor or irrigation time restrictions.

Variances or exceptions can be made in cases of extreme hardship or when

approved by the city attorney. All exceptions to the ordinance must not adversely affect

the health, safety, or welfare of other people and must not cause an immediate negative

impact on the city's water supply. Violators can be fined up to $250 for the first offense,

doubled for the second offense, and continued for each subsequent offense within a 12-

month period. The total fine for a 12-month period cannot exceed $2,000. (Dallas,

Texas Ordinances, Section 49-21.1)

Lucas, Texas. The Lucas, Texas Code of Ordinances includes requirements for

rain/freeze sensors. This ordinance applies to automatic irrigation systems in the city.

The party responsible is the owner, leasee, occupier, or manager of the property on

which the irrigation system is located. All new automatic irrigation systems installed after

1 January 2006 must be equipped with a rain/freeze sensor. As of 1 July2007, all









existing systems must be retrofitted with a rain/freeze sensor. Fines to property owners

can be up to $500. (Lucas, Texas Ordinances, Article 8, Irrigation system regulations,

Section 3-220)

San Antonio, Texas. San Antonio, Texas has a permanent year-round water

conservation ordinance to reduce per capital use of water. The ordinance defines many

terms that play into conservation such as automatic irrigation controller, impervious

surface, rain sensor, recycled water, and water flow restrictor. Among the many

regulations such as ice machines and xeriscapes on new home developments is a rain

sensor regulation. Effective 1 January 2006, all automatic irrigation controllers must

have rain sensors installed and maintained (San Antonio Water Systems, Ordinance

100332 34.274.2).

Study Objectives

The goal of this research was to determine the performance of three brands of

expanding-disk rain sensors. The first objective was to evaluate the number of times in

open-switch mode and the accuracy over time with respect to the selected set point by

comparing when the rain sensors interrupted irrigation with rainfall recorded from an on-

site weather station (Chapter 2). The second objective was to evaluate the dry-out time

(amount of time in open-switch mode) and potential irrigation savings by comparing rain

sensor irrigation interruptions to a University of Florida Institute of Food and Agriculture

Science (UF IFAS) recommended irrigation schedules (Chapter 3). The third objective

was to determine if the length of the hygroscopic disks in expanding-disk rain sensors

change size based on the amount of time installed and the rainfall setting (Chapter 4).









CHAPTER
EXPANDING-DISK RAIN SENSOR ACCURACY

Introduction

Although Florida ranks second in annual state precipitation, irrigation is required to

meet the aesthetic landscape requirements. Between 1950 and 2000, the population of

Florida increased five times and total public supply withdrawals increased 13 times

(Marella, 2004). In 2005, Florida ranked first in single family home construction with

209,162 homes built (United States Census Bureau [USCB], 2007) and it has been

estimated that 70% of single family homes have automatic irrigation systems (Tampa

Bay Water, 2005). Water conservation measures are needed to reduce water volumes

applied. Haley et al. (2007) reported that residential irrigation accounted for 64% of total

residential water volumes in central Florida.

One water conservation measure is adding a rain sensor (RS) to an automatic

irrigation system. It is thought that automatic systems irrigate more than manual

irrigation because of the set-and-forget mentality of an irrigation timer with no

consideration of climatic conditions. RSs are devices designed to interrupt the cycle of

an automatic irrigation system controller when a specific amount of rainfall has occurred

(Dukes and Haman, 2002b). Unlike many other irrigation sensors, once it is properly

installed and set, the RS settings do not have to be adjusted to achieve water savings

(Dewey, 2003). Water and cost savings vary among rain sensor models and settings.

Variables affecting savings include rain frequency, whether or not the controller is left on

for automatic operation, and the amount of water applied by the system per cycle

(Dukes and Haman, 2002b).









States and municipalities throughout the country have mandated the use of rain

sensors to conserve water. States that have had bills introduced requiring RSs on

certain landscape irrigation applications include Connecticut, Florida, Massachusetts,

Minnesota, New Jersey, and Texas (see Chapter 1). Florida, Minnesota, and New

Jersey require homeowners to install and maintain a RS on all automatic irrigation

system ms.

Several types of RSs are on the market. All of the sensors can be adjusted to

interrupt at different depths of rainfall, generally between 3 and 25 mm. Research by

Cardenas-Lailhacar and Dukes (2008) determined that the 25 mm setting is too high in

Florida to net practical savings. One type of RS collects the rain water in a cup and

interrupts the irrigation based on a preset weight of water. Another type has a set of

electrodes that detect the water level in a small collection dish that measures the

amount of rainfall with two electrodes in a collection cup (Dukes and Haman, 2002b). A

disadvantage of both of these water weight devices is that debris can get into the

collection cup and cause the system to interrupt irrigation without sufficient rainfall.

The most commonly used RSs are expanding-disk rain sensors (Figures 2-1

through 2-4). Hygroscopic disks in the sensor expand proportionally to the amount of

rainfall (Figure 2-5). The swelling of the disks typically causes a switch to interrupt the

signal to open an irrigation valve. Expanding-disk sensors require less maintenance and

are less expensive than other sensors.

There are different models of RSs that can be used depending on the application.

Sensors can operate as "normally closed" (normally allow irrigation) or "normally open"

(normally does not allow irrigation). Most systems run on "normally closed" rain sensors.









For a normally closed RS, the RS is in closed-switch mode until sufficient rainfall

changes it to open-switch mode. Open-switch mode means that the irrigation circuit is

incomplete such that a scheduled irrigation event will be interrupted.

Research by Cardenas-Lailhacar and Dukes (2008) investigated the performance

of expanding disk rain sensors in a 2005 study at the University of Florida. The four

treatments were three Hunter Mini-Clik (Hunter Industries, Inc, San Marcos, CA) rain

sensors with different settings (3 mm, 13 mm, and 25 mm) and one Hunter Wireless

Rain-Clik. As would be expected, the lower set points corresponded to a greater

number of times in open-switch mode occurrences than a higher setting. All treatments

had replicate variability in the number of times in open-switch mode and depth of rainfall

required for open-switch mode. The average depth of rainfall triggering the Wireless 3

mm, 13 mm, and 25 mm settings was 1.4, 3.4, 10.0, and 24.5 mm, with resulting

accuracy of 88%, 77%, and 98% for 3 mm, 13 mm, and 25 mm, respectively. This

research provides a base-line for rain sensor research. A longer study over a variety of

precipitation conditions would offer more insights into rain sensor performance.

The average potential water savings in depth of irrigation for the Wireless Rain

Clik and Mini-Cliks@ with 3 mm, 13 mm, and 25 mm settings was 588, 337, 468, and 38

mm, respectively. For the Mini-Clik treatments, the lower setting showed more water

saving potential. This study concluded that a setting of 25 mm was too high and not

applicable in north central Florida.

The objective of this study was to evaluate the accuracy with time of three brands

of expanding-disk rain sensors with respect to the selected set point by comparing when









the rain sensors interrupted irrigation with rainfall recorded from an on-site weather

station.

Materials and Methods

This study was conducted at the University of Florida Agricultural and Biological

Engineering Department campus turfgrass plots, Gainesville, Florida. There were a total

of 40 rain sensors installed at a height of 2 m (Figure 2-6).

Treatments

Seven treatments composed of different rain sensor brands and set points were

established (Table 2-1). The Wireless Rain-Clik (WL) and Mini-Clik (MC) rain sensors

were from Hunter Industries, Inc., San Marcos, CA. The WL did not have a rainfall

setting and was designed to interrupt irrigation immediately after rain begins. The three

MC treatments had rainfall settings of 3 mm, 6 mm, and 13 mm (3MC, 6MC, and 13MC)

(Figure 2-7). Data collection for treatments WL, 3MC, 6MC, and 13MC was started 2

October 2006 and was completed 31 December 2009 (1,186 days). These four

treatments were installed on 25 March 2005 with four replications each. The 6MC

treatment was originally set to 25 mm and was changed to 6 mm on 2 October 2006

since previous work indicated that minimal savings occurred at a 25 mm setting

(Cardenas-Lailhacar and Dukes, 2008). The three remaining treatments were installed

at a later date with 6 mm settings for three brands. The brands and respective treatment

codes were Hunter Industries, Inc., model Mini-Clik (Hunter), Irritrol Systems, Inc.,

model RFS 1000, Riverside, CA (Irritrol), and Toro Company, Inc., model TWRS,

Riverside, CA, (Toro) (Figure 2-8) with eight replicates for each treatment. Data

collection for Hunter, Irritrol, and Toro treatments was started on 8 November 2006 and

was completed 31 December 2009 (1,150 days). Problems with rain sensor function are









summarized in Table 2-2. The dry out vents for Hunter and Irritrol were fully open from

installation until 2 July 2009 and were then changed to fully closed. The Toro rain

sensor receivers were set to 0.0 day dry out; the water delay feature was set to 1.0 day

for four replicates and 3.0 days for the remaining four replicates.

Monitoring

Each time a rain sensor changed mode between open switch mode (OSM) and

closed switch mode (CSM), the date and time were recorded at a one-second sampling

interval using AM16/32 multiplexers (Campbell Scientific, Inc., Logan, UT) attached to a

CR1OX model data logger (Campbell Scientific, Inc., Logan, UT).

An onsite automated weather station (Campbell Scientific, Logan, UT) located

within 15 m of the experimental site recorded weather conditions using a CR10X model

data logger (Campbell Scientific, Logan, UT). Data such as relative humidity,

temperature (model HMP45C, Vaisala, Inc., Woburn MA), solar radiation (model

LI200X, Li-Cor, Inc., Lincoln, NE), and wind speed and direction (model WAS425,

Vaisala, Inc., Sunnyvale, CA) were recorded at 15 minute intervals. Precipitation was

measured by a tipping bucket rain gauge (model TE525WS, Texas Electronics, Inc.,

Dallas, TX) with a 1-second sampling interval time stamp for each 0.25 mm of rain.

A manual rain gauge located within 5 m of the rain sensors was used to verify the

accuracy of the tipping bucket rain gauge measurements (Figure 2-9). The weather

station tipping bucket rain gauge was calibrated using the Texas Electronics Calibration

Kit (Texas Electronics, Inc., Dallas, TX). Calibration testing was conducted during

November 2009. The first test indicated that the tipping bucket was out of calibration; it

recorded 23.4 mm for every 25.4 mm that actually fell. After two calibration adjustments,

the tipping bucket rain gauge was recording 24.9 mm for every 25.4 mm of rain that fell









which is within the Texas Electronics acceptable range of 2% error (Texas Electronics,

Inc. (a)). Measured rainfall was multiplied by and adjustment factor of 1.07 so that

measured rainfall would be accurate and equal to actual rainfall. Since the tipping

bucket had not been calibrated since 2003, a calculation adjustment was applied to

recorded rain events during the duration of the study. Based on a linear regression of

the data with and without the calibration, the calibration was applied to rainfall events

greater than 15 mm and not applied to rainfall events less than 15 mm (Figures 2-10

and 2-11).

Rainfall event depths were collected and analyzed from the study period. Monthly

rainfall from a 30-year historical period from 1970 to 2000 from the National Oceanic

and Atmospheric Administration (NOAA) was used as a comparison to data collected

during the study. Establishment of current and historical weather patterns will affect

setting recommendations.

Statistical Analysis

SAS statistical software (SAS Institute, Inc., Cary, NC) was used for all statistical

analysis. A general mixed model with an auto regressive error structure was used to

model the continuous responses (PROC MIXED). Tukey-Kramer adjusted p-values

(p<0.05) were used for pairwise comparisons of mean.

Results and Discussion

Climactic Conditions

During the 1,186 days of the WL and MC study period and the 1,150 days of the

Hunter, Irritrol, and Toro study period, 28% of the days received rain. For the WL and

MC experiment, the cumulative rainfall was 3,551 mm, 14% less than the historical

average of 4,121 mm (Figure 2-12). For the Hunter, Irritrol, and Toro experiment, the









cumulative rainfall was 3,410 mm, 16% less than the historical average of 4,055 mm

(Figure 2-13). If the amount of rainfall during study would have been closer to historical

values, the rain sensors would have gone into OSM more times.

Number of Times in Open Switch Mode

Figures 2-14 and 2-15 show the daily and cumulative rainfall during each period

and the theoretical count for the number of OSM occurrences for each treatment. The

theoretical number of OSM occurrences for the 3MC, 6MC, and 13MC treatments

should was 192, 139, and 82, respectively. The percentage of rain events greater than

3, 6, and 13 mm during the WL and MC study period was 57%, 42%, and 25%. The

theoretical number of OSM occurrences for the Hunter Irritrol, and Toro treatments were

136 times and 41% of rain events being greater than 6 mm.

Figure 2-16 shows the average cumulative number of OSM events for the WL and

MC treatments. The average number of OSM events for WL, 3MC, 6MC, and 13MC

were 146, 154, 160, and 109, respectively. The 13MC treatment had fewer OSM events

than the other three treatments. The 6MC treatment went into OSM more times than

expected since it was not statistically different from 3MC (p<0.05). This result could be

due to the change of these RSs from a setting of 25 mm to 6 mm in October 2006. The

possible effects of disk size change with time are discussed in Chapter 4.

All treatments showed replicate variability and not all replicates were functioning

during the entire experiment period. One WL replicate stopped functioning on 21

September 2007, after 910 days of operation. This replicate was not considered in any

analysis for means comparisons. The remaining three WL replicates were not

functioning 28 January to 9 June 2008 due to an electrical problem such as data logger

or batteries in the receivers. One of the four 13-MC replicates displayed some erratic









behavior such as going into OSM without rainfall and not going into OSM with sufficient

rainfall starting 8 July 2008, after 1201 days of continuous operation but remained

functioning throughout the study. The cumulative numbers of OSM events for the

functioning WL replicates were 131, 145, and 162 (Figure 2-17). For 3MC with a

theoretical OSM value of 192, the replicates went into OSM 149, 162, 173, and 182

times (Figure 2-18). The 6MC replicates went into OSM 162, 173, 173, and 188 times,

which were all more than the theoretical value of 139 (Figure 2-19). The theoretical

OSM value for 13MC was 82 while the replicates went into OSM 59, 111, 116, and 131

times (Figure 2-20). The coefficient of variance (CV) for WL, 3MC, 6MC, and 13MC

were 11%, 8%, 6%, and 30%, respectively. One 13MC replicate increased the variability

from 8% to 30% because it displayed somewhat erratic behavior but did continue

functioning. The CV values for the WL, 3MC, 13MC, and 25MC after 282 days of

installation were 3%, 28%, 24%, and 8%, respectively (Cardenas-Lailhacar and Dukes,

2008). The variability of the 25MC rain sensors was 8% and after changing the setting

to 6MC the variability was 6% indicating that changing the setting of these replicates did

not influence the variability. The weighted average CV for WL, 3MC, 25/6MC, and

13MC for both studies was 8%, 12%, 7%, and 25%. The WL and 25/6MC rain sensors

had the least amount of variability with each treatment. From the initial study to this

study, the WL treatment became more variable, 13MC variability remained the same,

and 3MC became more stable. Tipping bucket rain gauges have a range of accuracies

with 0.5% to 4% variability (Omega (1995), Sutron Corporation, Spectrum

Technologies, Inc, and Texas Electronics (b)). The variability for rain sensors is









relatively high when compared to tipping buckets, which are both for a measurement

instrument.

Figure 2-21 shows the cumulative number of OSM events for the Hunter, Irritrol,

and Toro during their study period and the theoretical value based on rainfall. The

average number of OSM events for Hunter, Irritrol, and Toro were 144, 190, and 114

with a theoretical value of 136. Not all replicates were functioning during the entire

experiment period. A Hunter replicate started showing erratic behavior by not

responding to high amounts of rainfall and not reacting to manual triggering the same as

the other replicates; this replicate was not considered in evaluation analysis after 21

December 2007, corresponding to 408 days of operation. This replicate was not

considered in any analysis for means comparisons. Toro had some issues with the right

wireless receiver receiving OSM and CSM signals from the right rain sensor. Six of the

eight replicates received the same information from one rain sensor from 20 April 2009

to 22 September 2009.

Figures 2-22 to 2-24 show the theoretical number OSM occurrences for each

treatment and variability within treatments. The number OSM occurrences for the seven

functioning Hunter replicates were 106, 135, 143, 144, 144, 144, and 151 times (Figure

2-22). The number of OSM occurrences Irritrol replicates were 175, 182, 182, 190, 191,

196, 196, and 204 (Figure 2-23); they all went into OSM more than the theoretical value

due to the number of OSM occurrences without rain. The number of OSM occurrences

for the Toro treatment were 83, 87, 87 92, 96, 96, 104, and 114 with data from 20 April

2009 to 22 September 2009 excluded due to receiver problems (Figure 2-24). The CV

values for Hunter, Irritrol, and Toro were 11%, 6%, and 10%, respectively.









Accuracy of Rain Sensors

The accuracy of the instrument is its ability to indicate an exact true value (Figliola

and Beasley, 2000). Accuracy is related to the difference between the true value and

the indicated value of a measurement called the absolute error (E). The percent

accuracy (A) is calculated by:


A = 1-u *100 (3-1)
True value
Table 2-3 shows the average depth of rainfall before the RSs went into OSM.

Because of acceptable error in the tipping bucket rain gauge, there was a plus or minus

2% in the calculated rain sensor accuracy. The WL does not have a set point, so

accuracy cannot be determined. The WL had an average rainfall depth of 3.2 mm

required to go into OSM. The 3MC, 6MC, and 13MC went into OSM after 1.9, 1.6, and

6.6 mm with accuracies of 64%, 27%, and 51%, respectively. The low accuracy from

6MC could be attributed to the setting change on 6 October 2006 on the treatment from

25 mm to 6 mm because of disk size changes discussed in Chapter 4. The Hunter,

Irritrol, and Toro went into OSM after 3.8, 4.3, and 5.8 mm with accuracies of 64%,

71%, and 97%, respectively. The 3MC, 6MC, 13MC, Hunter, and Irritrol treatments

required a different depth of water for OSM than their respective rainfall setting. The CV

values for depth of rainfall required for OSM for WL, 3MC, 6MC, 13MC, Hunter, Irritrol,

and Toro were 67%, 51%, 51%, 51%, 37%, 36%, and 33%, respectively.

Change in Accuracy of Rain Sensors over Time

The accuracy of the treatments varied over the study period as summarized in

Table 2-4. The 13MC treatment became less accurate with time while 6MC, Hunter,

and Irritrol showed an increase in accuracy, and 3MC and Toro had no change in









accuracy. The amount of rainfall required before OSM for WL, 3MC, 6MC, 13MC,

Hunter, Irritrol, and Toro at the beginning and end of the study was 2.6 and 4.0 mm, 1.9

and 1.9 mm, 1.3 and 2.1 mm, 8.3 and 5.3 mm, 3.3 and 4.5 mm, 3.6 and 5.1, and 5.7

and 6.1 mm, respectively. Figure 2-25 shows the progression in the change of rainfall

required for OSM for all treatments with set points over the study period. The percentile

point change in accuracy during the study period for 3MC, 6MC, 13MC, Hunter, Irritrol,

and Toro was -1%, 59%, -36%, 36%, 42%, 7%, respectively, where a negative value

indicated a decrease in accuracy. Other than 3MC, there was a trending relationship

between rainfall depth for OSM CV value and accuracy. A treatment with low accuracy

also had a high CV value.

The initial 282-day study by Cardenas-Lailhacar and Dukes (2008) found that the

WL required 1.4 mm of rainfall for OSM. WL rainfall requirement increased 2.5 times

after an additional 904 days of installation. The 3MC and 13MC went into OSM after 3.4

and 10.0 mm with accuracies of 88% and 77%, respectively. The 3MC and 13MC were

more accurate during their first 282 days of installation. When newly installed, the rain

sensors had an average error of 15% (Cardenas-Lailhacar and Dukes, 2008). The

weighted average error of the sensors at the end of this study was 46%, a tripling in the

error of the sensors over time.

Summary and Conclusions

This experiment occurred during a relatively dry period with rainfall on 28% of the

days. The percentages of rain events greater than 3, 6, and 13 mm during the WL and

MC study period were 57%, 42%, and 25% The Hunter, Irritrol, and Toro treatments

theoretically had 136 opportunities for OSM with 41% of rain events being greater than

6 mm.









Most treatments showed variability and erratic behavior of some replicates during

the study. The coefficient of variance for depth of rainfall required for open-switch mode

varied between 33% and 67%. Some replicates showed erratic behavior such as not

detecting rainfall events much higher than their setting, going into closed-switch mode in

the middle of a rain event and returning to open-switch mode a few minutes later with

0.5 mm of rainfall, or going into open-switch mode with little or no rainfall. Two of the

replicates completely stopped functioning during the study period for unknown reasons.

The accuracy of the rain sensors changed with time. The percentile point change

in accuracy during the study period ranged from an increase of 59% to a decrease of

36%. The 13MC treatment accuracy decreased during the study while other treatments

had small change in or improved accuracy. Cardenas-Lailhacar and Dukes (2008)

showed an average weighted accuracy of 85% in the first282 days of installation while

this study had an average weighted accuracy of 47% with the same sensors and 60%

overall.

There was no single trend for all rain sensors or all rainfall settings with respect to

accuracy with time. For the best accuracy, there is evidence based on historical studies

and results from this study that Hunter Mini-Clik rain sensors should be replaced after

1 year of installation. The accuracy of a Hunter Mini-Clik set to 3 mm stabilized after 2

years of installation. The higher rainfall setting corresponded to lower accuracy with the

same brand of rain sensor. Irritrol RSF 1000 rain sensor accuracy increased as the

study progressed. Toro TWRS rain sensors retained their relatively good accuracy

during the 3 years of this study. Further research is needed to verify these results.

Changing the setting of a rain sensor after it has been installed more than 3 months is









not recommended (see Chapter 4 for more details). The change of the 6MC treatment

from a 25 mm to a 6 mm setting reduced the accuracy of the same sensors from 98% to

27%. During the same time, the 3MC and 13MC treatments average weighted accuracy

declined from 84% to 59%

Overall, the rain sensors showed that they have high variability with time.

However, due to low cost and low maintenance requirements, rain sensors can be a

useful device for potential water savings. The variability, erratic behavior, and low

accuracy of some replicates showed that rain sensors should not be used in

applications requiring high accuracy and precision.

Table 2-1. Rain sensor treatment description.
Treatment Model Replicates Set Installation Study Start
Point Date Date
WL Wireless Rai n-Clik 4a 25 Mar 2005 02 Oct 2006
3MC Mini-Clikx 4 3 mm 25 Mar 2005 02 Oct 2006
6MC Mini-Clikx 4 6mm 25 Mar 2005* 02 Oct 2006
13MC Mini-Clikx 4 13 mm 25 Mar 2005 02 Oct 2006
Hunter Mini-Clikx 8b 6 mm 02 Oct 2006 08 Nov2006
Irritrol Irritrol RFS 1000Y 8 6mm 02 Oct 2006 08 Nov2006
Toro Toro TWRSz 8 6 mm 02 Oct 2006 08 Nov 2006
a 3 replicates were included in means separation analysis due to one failed replicate
b 7 replicates were included in means separation analysis due to one failed replicate
x Hunter Industries, San Marcos, CA
Y Irritrol Systems Inc., Riverside, CA
z Toro Company, Inc., Riverside CA
*changed setting from 25-mm to 6-mm on 2 October 2006

Table 2-2. Summary of functionality problems for treatments and replicates.
Treatment Model Problems with Rain Sensors
WL Wireless Rain-Clik Not operational from 28 Jan 2008 to 9 June 2009
WL-B stopped functioning 21 Sept 2007a
3MC Mini-Clik None
6MC Mini-Clik None
13MC Mini-Clik 13-C showing somewhat erratic behavior 8 July 2008
Hunter Mini-Clik H-D showed erratic behavior after 21 Dec 2007a
Irritrol Irritrol RFS 1000 None
Toro Toro TWRS Sixwireless receivers were incorrectly connected to
one rain sensor from 20 April 2009 to 22 Sept 2009.
a These replicates were not included in means separation analysis










depth of rainfall before rain sensors switched to Open Switch Mode.


Treatment Model Set point Rainfall for OSM Accuracy
(mm) Depth (mm) Standard CV (%) (%)a
Deviation (mm)
Wireless Rain-Clik 3.2 2.1 67
Mini-Clik 3 1.9 1.0 51 64
Mini-Clik 6 1.6 0.7 51 27
Mini-Clik 13 6.6 3.3 51 51
Mini-Clik 6 3.8 1.4 37 64
Irritrol RFS 1000 6 4.3 1.5 36 71
Toro TWRS 6 5.8 1.9 33 97
a Accuracy is +/- 2% due acceptable error in the tipping bucket rain gauge


Table 2-4. Summary of changes in accuracy for change in rainfall required for Open
Switch Mode.
Treatment Model Set Days in Rainfall depth for OSM Change in accuracy
point study (mm) (percentile points)
(mm) Beginning End
Wireless Rain-Clik 1,186 2.6 4.0 a
Mini-Clik 3 1,186 1.9 1.9 -1
Mini-Clik 6 1,186 1.3 2.1 a 59
Mini-Clik 13 1,186 8.3 5.3 a -36
Mini-Clik 6 1,050 3.3 4.5 a 36
Irritrol RFS 1000 6 1,050 3.6 5.1 a 42
Toro TWRS 6 1,050 5.7 6.1 7
The depth of rainfall required for open switch mode changed over the study period.


Table 2-3. Average

























D -\
Figure 2-1. WL (model Wireless Rain-Clik, Hunter Industries, Inc., San Marcos, CA) rain
sensor. A) Expanding disks inside ventilation window, B) quick-response
expanding disks, C) Ventilation window adjustment knob, D) antenna.




A
















Figure 2-2. MC (model Mini-Clik, Hunter Industries, Inc., San Marcos, CA) rain sensor.
A) Rainfall threshold setting slots, B) expanding disks, C) dry-out adjustment
ring and vents.




























Figure 2-3. Irritrol (model RFS 1000, Irritrol Systems, Inc., Riverside, CA.) rain sensor.
A) Rainfall threshold setting slots, B) dry-out adjustment ring, C) antenna.





















Figure 2-4. Toro (model TWRS, Toro Company, Inc., Riverside, CA) rain sensor. A)
Rainfall threshold setting slots, B) dry-out vent, C) antenna.






















Figure 2-5. Detail of expanding disk material and threshold adjustment of Mini-Clik
(Hunter Industries, Inc.) rain sensor. A) Rainfall threshold setting slots, B)
hygroscopic expanding disk material.


Figure 2-6. Research site located at the University of Florida Agricultural and Biological
Engineering facilities. Shown: weather station on left, WL, 3MC, 3MC, and
13MC treatments installed on left board, and Hunter, Irritrol, and Toro on right
board.


A- -


B -










t i t IL t- t tI I I I i J I


Figure 2-7. Installed Hunter Wireless Rain-Clik, three on left and one on right, and Mini-
Clik rain sensors with four wireless receivers for WL and data logger.


S' it IL 1 0


Figure 2-8. Installed Hunter, Irritrol, and Toro rain sensors, left to right, with wireless
receivers (Irritrol on left and Toro on right), and data logger.












y= 1.0453x-0.2591


0 10 20 30 40 50 60 70 80
Rain Gauge Data (mm)

Figure 2-9. Relationship between manual rain gauge and weather station tipping bucket
rain gauge with the calibration factor applied to the tipping bucket data with
more than 15 mm of rainfall.


y = 0.9589x+ 0.681
R2 = 0.9914 ..


15 25 35 45 55 65 75
Rain Gauge Data (mm)

Figure 2-10. Relationship of rain events greater than 15mm between manual rain gauge
and weather station tipping bucket rain gauge with the calibration factor
applied to the tipping bucket data.












'a.


y= 0.9938x-0.1462

R2 = 0.9825


3 6 9 12


Rain Gauge Data (mm)


Figure 2-11. Relationship of rain events less than 15 mm between manual rain gauge
and weather station tipping bucket rain gauge without the calibration factor
applied to the tipping bucket data.


E 200
E

j 150


> 100
.c
4 50
o 50


----------- ----^ r





0 > c Cv 0) a. at > u cW > C W- ba > U 3

Date, 2006 2009


4500
4121
4000
3551E
3500 E
3000 =
2500
2000 .)

1500 "
1000 E
500 U
0


- Study Cumulative


- Historical Cumulative


Figure 2-12. Comparison of monthly and cumulative rainfall during the study period for
WL and MC treatments and average historical rainfall for north central Florida.


IStudy Period


I Historical













250


E 200
E

_ 150

I-
> 100

o 5
0 50
2


- Study Cumulative


- Historical Cumulative


Figure 2-13. Comparison of monthly and cumulative rainfall during the study period for
Hunter, Irritrol, and Toro treatments and average historical rainfall for north
central Florida.


4500
4055
4000

3500
3425
3- 000

2500

2000

I1500

1000

j500

0
Nov Jan Mar May Jul Sep Nov Jan Mar May Jul Sep Nov Jan Mar May Jul Sep Nov

Date, 2006 2009


Study Period


Historical







































u III l ll I 1111 111 1 1 I 111111 1111iilliill ii li i 11 111 1i11 1i0 1ii111 111 I i ii111 111111 Ii I I 111 111111 ili 11 i1 i iloil i 1110i* ii lM iillIl III I I 111111


43 eie rts 'ihr lore tha1 25 mm of -anr -ngingfiom: 6 to 11. nm

i' -



P1
_~~~, -_ __



_L mjmt c nt
I e en s I vith < i I > 13 rm -n





1r imssit pi t
1-- .L _- lif-3tk U1 k A I 111 111 11 11,1

A III .-.I d:1:. [ILt3 4 1ffI F.,1-111..-.1-1HIIIIIII ,[,.-....[


Date, 2006 2009
Figure 2-14. Cumulative and daily rainfall during the WL and MC treatments study period with the rainfall setting and the
respective theoretical number of times each should have gone into OSM.


4000

1551
3500


3000
E
2500
M-
C

2000

CO
1500 ro

E
1000 M


500


0









4000
C ev n swith mre than 25mm of rail ranging frorr 26 tt 77 nm
3500
S3410

3000

E
2500




'1 II 1500 "g
m n-000
E
1500







0
M^^^ ^^^^^y^^^ ^^

.l: ']I _J i Ju j i l.__ ..J il !_._.,..


?4t A


#4ttt~ep~zbO


Date, 2006 2009
Figure 2-15. Cumulative and daily rainfall during the Hunter, Irritrol, and Toro treatments study period with the rainfall
setting and the respective theoretical number of times each should have gone into OSM.


#4tI


(S QS\


,~d~$P4~6~,~











160a
154a
146a

109b


o -WW
Oct06 Jan07 Apr07 Jul07 Oct07 Dec07 Mar08 Jun08 Sep08 Dec08 Mar09 Jun09 Sep09 Dec09
Date, 2006-2009
W- L ---3MC 6MC -13MC

Figure 2-16. Cumulative number of times into OSM for WL and MC treatments. Data
from 28 January to 9 June 2008 are not included due to all WL replicates not
functioning. Erratic replicates within treatments are not included after their
respective improper functioning dates (WL-B 21 September 2007 and 13MC-
C 8 July 2008). Numbers with different letters indicate a statistical difference
using Tukey-Kramer adjusted p-values of p<0.05.


180


o --W
Oct 06


162
145
131






47


Jan07 Apr07 Jul07 Oct07 Dec07 Mar08 Jun08 Sep08 Dec08 Mar09 Jun09 Sep09 Dec09
Date, 2006-2009
--WL-A '-- WL-B WL-C -->'WL-D


Figure 2-17. Cumulative number of times the WL replicates went into OSM. WL stopped
functioning on 21 September 2007.











200


180
Theoretical= 192
160

140

120

100

80

60

40

20

0
Oct06 Jan07 Apr07 Ju107 Oct07 Dec07 Mar08 JunO8 Sep08 Dec08 Mar09 JunO9 Sep09 Dec09


-4-3MC-A -3MC-B


Date, 2006-2009
S3MC-C 3MC-D Theoretical


Figure 2-18. Cumulative number of times the 3MC replicates went into OSM. The
Theoretical value is the number of times the replicates should have gone into
OSM based on rainfall.


200
180
160
140
120
100
80 -
-80Theoretic
60
40
20
0
Oct06 Jan07 Apr07 Jul07 Oct07 Dec07 Mar08 Jun08 Sep08 Dec08 Mar09
Date, 2006-2009
--4 6MC-A -- 6MC-B 6MC-C 6MC-D -


188
173
173
162


Jun09 Sep09 Dec09

- Theoretical


Figure 2-19. Cumulative number of times the 6MC replicates went into OSM. The
Theoretical value is the number of times the replicates should have gone into
OSM based on rainfall.


182
173
162
149











131
116
111





59


Oct06 Jan07 Apr07 Jul07 Oct07 Dec07 Mar08 Jun8 Sep08 Dec08 Mar09 Jun9 Sep09 Dec09
Oct06 Jan07 Apr07 Jul07 Oct07 Dec07 Mar08 Jun08 Sep08 Dec08 Mar09 Jun09 Sep09 Dec09


---13MC-A 13MC-B


Date, 2006-2009
S-13MC-C 13MC-D Theoretical


Figure 2-20. Cumulative number of times the 13MC replicates went into OSM. The
Theoretical value is the number of times the replicates should have gone into
OSM based on rainfall.


200
200 190a
2 180
0 160
S140 -1 144b
4-
c 120
114c
100
80
S60
40 Theretical = 136
E 40
20

Nov 06Jan 07Apr 07 Jul07 Sep 07Dec 07Mar 08May 08Aug080ct08Jan 09Apr 09Jun 09Sep 09Dec09
Date, 2006 2009


-Hunter -- Irritrol


Toro -6 mm


Figure 2-21. Cumulative number of times into OSM for Hunter, Irritrol, and Toro
treatments. The Theoretical value is the number of times the replicates should
have gone into OSM based on rainfall. Numbers with different letters indicate
a statistical difference using Tukey-Kramer adjusted p-values of p<0.05.












160

140

120

100

80

60

40
20 Theoretical= 136
20


Nov06 Jan07 Apr07 Jul07 Sep07 Dec07 Mar08 May08 Aug08 Oct08 Jan09 Apr09 Jun09 Sep09 Dec09


151
144 144 144
143
135

106



57


Date, 2006 2009

A MB C D E F -F G H Theoretical



Figure 2-22. Cumulative number of times Hunter into OSM. The Theoretical value is the
number of times the replicates should have gone into OSM based on rainfall.


200
180
160
140
120
100
80
60
40 j ,-Z ~ -' '-
20 -
Nov06 Jan07 Apr07 Jul07 Sep07 Dec
Nov06 Jan07 Apr07 Jul07 Sep07 DecO


204
196 196
191
190
182 182
175


Ineoretical = 13o


7 Mar08 May08 Aug08 Oct08 Jan09 Apr09 Jun09 Sep09 Dec09


Date, 2006- 2009


-A B C -XD --E F G -H Theoretical

Figure 2-23. Cumulative number of times Irritrol went into OSM. The Theoretical value is
the number of times the replicates should have gone into OSM based on
rainfall.


L
C


r


. V












114


7 892
8787
83


60

40


Theoretial = 101


20 --


0 now-, I
Nov 06Jan 07Apr07 Jul07 Sep07Dec07Mar08May 08Aug08 Oct08 Jan 09 Apr09 Jun09 Sep09Dec09

Date, 2006- 2009

--A B -IB C D -WE F G -H Theoretical

Figure 2-24. Cumulative number of times Toro went into OSM. The Theoretical value is
the number of times the replicates should have gone into OSM based on
rainfall.




10


----M -


- -


o
E 2



Oct06 Jan07 Apr07 Jul07 Oct07 Dec07 Mar08 Jun08 Sep08 Dec08 Mar09 Jun09 Sep09 Dec09

Date, 2006 2009

3MC 6MC 13MC Hunter Irritrol Toro

Figure 2-25. Accuracy of each treatment with a set point over the study period with an
average (solid line) and 95% confidence bands (dashed lines).


..........


-- ----


- ---- ---- ----


-- i


-...-.- -- .









CHAPTER 3
EXPANDING-D ISK RAIN SENSOR DRY-OUT AND POTENTIAL IRRIGATION
SAVINGS

Introduction

Water conservation measures are becoming more critical in Florida due to

increased resource demand. Florida receives an average of 1,350 mm of rainfall a year

(NOAA, 2003). From 2000 to 2005, Florida had a net population gain of approximately

1,000 people per day and ranks fourth in population (USCB, 2009). Seventy percent of

single family homes have automatic irrigation systems (Tampa Bay Water, 2005).

Florida is second wettest state in the nation but irrigation is required to meet the

aesthetic landscape requirements. Residential irrigation has been reported to account

for 64% of residential water use in one area of the state (Haley et al., 2007).

Rain sensors are a conservation device used with automatic irrigation systems to

reduce applied irrigation. Rain sensors have adjustment settings allowing users to

choose the amount of rainfall required for the irrigation cycle to be interrupted, generally

between 3 and 25 mm. States such as Florida, Minnesota, and New Jersey and many

municipalities throughout the country have mandated the use of rain sensors to

conserve water. Variables affecting water and cost savings with rain sensors include

rain frequency, whether or not the controller is left on for automatic operation, and the

amount of water applied by the system per cycle (Dukes and Haman, 2002b).

The most common rain sensors used in Florida are expanding-disk rain sensors

(see Chapter 2, Figures 2-1 to 2-4). The swelling of the hygroscopic disks in the sensor

expand proportionally to the amount of rainfall (see Chapter 2, Figure 2-5). The

expanding-disk rain sensor is in closed-switch mode until sufficient rainfall changes it to

open-switch mode. The sensors rely on evaporation to go return to open-switch mode









(OSM). Dry-out settings can vary between 2 hours and 3 days (Hunter Industries, Inc.,

2005). Dry-out time is the amount of time the sensors stay in OSM. A longer dry-out

time has the potential to interrupt more scheduled irrigation cycles. Dry-out time is

meant to represent the time it would take for the rainfall to leave the root profile via

evaporation, transpiration, and other pathways. Dry-out settings can be adjusted for

most sensors such that it matches the drying rate of the site's soil (Dewey, 2003).

Cardenas-Lailhacar and Dukes (2008) investigated the performance of expanding

disk rain sensors in a 2005 study at the University of Florida campus in Gainesville,

Florida. The performance of Hunter Mini-Clik rain sensors with three different settings (3

mm, 13 mm, and 25 mm) and one Hunter Wireless Rain-Clik were monitored and

compared with rainfall depth. The frequency of disk dry-out within 24 hours for the

Wireless, 3 mm, and 13 mm treatments was 80%, 51%, and 47%, respectively. The

average percentage of potential water savings for the WL, 3MC, 13MC, and 25MC was

44%, 30%, 17%, and 3%, respectively, based on a 2 d/wk irrigation schedule.

Researchers at the University of Florida have conducted irrigation studies with

smart controllers that included expanding-disk rain sensors set on a 2 d/wk irrigation

schedule to represent homeowner irrigation under watering restrictions. In a study with

ET controllers and rain sensors in southwestern Florida, the addition of a rain sensor set

at 6 mm reduced irrigation 21% compared with a time-based schedule (Davis et al.,

2009). A study in central Florida compared applied irrigation among a controller only, a

controller with a rain sensor set at 3 mm, and a controller with a rain sensor set at 6

mm. The savings for the 3 mm and 6 mm set points were 25% and 17%, respectively

(McCready et al., 2009).









The objective of this study was to evaluate the dry-out time of three brands of

expanding-disk rain sensors and potential irrigation savings by comparing rain sensor

irrigation interruptions to University of Florida Institute of Food and Agriculture Science

(UF IFAS) recommended irrigation schedules irrigating 2 d/wk and 1 d/wk to represent a

homeowner schedule.

Materials and Methods

This study was conducted at the University of Florida Agricultural and Biological

Engineering Department campus turfgrass plots, Gainesville, Florida. There were a total

of 40 rain sensors installed at a height of 2 m (see Chapter 2, Figure 2-6).

Treatments

Seven treatments were established at the site (Table 3-1). The Wireless Rain-Clik

(WL) and Mini-Clik (MC) rain sensors were from Hunter Industries, Inc., San Marcos,

CA. The WL did not have a rainfall setting. The three MC treatments had rainfall

settings of 3 mm, 6 mm, and 13 mm (3MC, 6MC, and 13MC) (see Chapter 2, Figure 2-

7). Analysis for treatments WL, 3MC, 6MC, and 13MC included data collected between

2 October 2006 and 31 December 2009 (1,186 days). These treatments were installed

on 25 March 2005 with four replications each. The 6MC treatment was originally set to

25 mm and was changed to 6 mm on 2 October 2006 to have a setting better fit for

north central Florida. The remaining treatments were installed at a later date with 6 mm

settings for three brands. The brands and respective treatment codes were Hunter

Industries, Inc. (Hunter), Irritrol Systems, Inc., Riverside, CA (Irritrol), and Toro

Company, Inc., Riverside, CA, (Toro) (see Chapter 2, Figure 2-8). Analysis for

treatments Hunter, Irritrol, and Toro included data collected between 8 November 2006

and 31 December 2009 (1,150 days).









Each rain sensor brand had adjustments for the dry-out time. WL dry-out vents

were set at half open. The dry-out vents for 3MC, 6MC, and 13MC were fully open

during the study. The dry out vents for Hunter and Irritrol were fully open from

installation until 2 July2009 and were then changed to fully closed. The Toro rain

sensor receivers were set to 0.0 day dry-out; the water delay feature was set to 1.0 day

for four replicates and 3.0 days for the remaining four replicates. The dry-out setting

was changed from 0.0 to 4.0 days to validate that electrical connections were correctly

established for the Toro rain sensors and wireless receivers. The average amount of

time required for dry-out for 0.0 day setting and 4.0 days setting was 15 hours and 99

hours (4 days), respectively. The difference between dry-out times for different settings

confirmed that the Toro installation was done correctly.

To estimate the potential water savings, theoretical irrigation schedules were

compared with each treatment. The two schedules used were a 1 d/wk schedule

(Tuesdays) and a 2 d/wk schedule (Tuesdays and Saturdays) set to irrigate at 6 a.m. A

scheduled irrigation was considered interrupted if the rain sensors were in open-switch

mode due to rainfall. Potential savings was the number of irrigations interrupted by each

rain sensor multiplied by the depth of scheduled irrigation. Weekly irrigation depths were

calculated to satisfy historical net irrigation required to replace water lost to

evaoptranspiration based on Dukes and Haman (2002a) recommendations (Table 3-2).

Monitoring

Each time a rain sensor changed mode between open-switch mode (OSM) and

closed-switch mode (CSM), the date and time was recorded at a 1-second sampling

interval using AM16/32 multiplexers (Campbell Scientific, Inc., Logan, UT) attached to a

CR1 OX model data logger (Campbell Scientific, Inc., Logan, UT).









An onsite automated weather station (Campbell Scientific, Logan, UT) located

within 15 meters of the experimental site recorded weather conditions using a CR10X

model data logger (Campbell Scientific, Logan, UT). Data such as relative humidity,

temperature (model HMP45C, Vaisala, Inc., Woburn MA), solar radiation (model

LI200X, Li-Cor, Inc., Lincoln, NE), and wind speed and direction (model WAS425,

Vaisala, Inc., Sunnyvale, CA) were recorded at 15 minute intervals. Precipitation was

measured by a tipping bucket rain gauge (model TE525WS, Texas Electronics, Inc.,

Dallas, TX) with a 1-second sampling interval time stamp for each 0.25 mm of rain.

A manual rain gauge located within 5 meters of the rain sensors was used to verify

the accuracy of the tipping bucket rain gauge measurements (See Chapter 2 Figure 2-

9). The weather station tipping bucket rain gauge was calibrated using the Texas

Electronics Calibration Kit (Texas Electronics, Inc., Dallas, TX). Calibration testing was

conducted November 2009 on the tipping bucket rain gauge and is explained in detail in

Chapter 2.

Rainfall event depths from the study period were collected and analyzed. A rain

event was considered started when the tipping bucket made the first tip. A 5-hour or

longer period between tips of the tipping bucket rain gauge was defined as a new

rainfall event. Establishment of current and historical weather patterns affected rainfall

setting recommendations. Monthly rainfall from a 30-year 1970 to 2000 from the

National Oceanic and Atmospheric Administration (NOAA) was used to compare study

weather data with historical normals.

Hourly disk length measurements with a dial caliper were conducted twice to

better understand how the hygroscopic disks dry-out after a rain event. The disk length









measurements during dry-out tracking were compared with temperature, solar radiation,

and relative humidity to relate the physical parameters to rain sensor function.

One tracking period was conducted on 11 July 2009 after a 60-mm rain event on

10 July 2009. The disks were measured every hour for 20 hours after the rain stopped.

To ensure that the disks had time to expand, there were 3 hours between when the rain

stopped and measurements started. The rain sensors were inspected for interruption

mode visually before disk measurements were taken. The second tracking period on 18

September 2009 was performed by manually watering the disks. The rain sensors were

drenched and were given 2 hours to expand before measuring. Measurements were not

taken throughout the night because the first dry-out tracking showed relatively small

disk length changes during the night.

Statistical Analysis

SAS statistical software (SAS Institute, Inc., Cary, NC) was used for all statistical

analysis. A general mixed model with an auto regressive error structure was used to

model the continuous responses (PROC MIXED). Tukey-Kramer adjusted p-values

(p<0.05) were used for pairwise comparisons of mean.

Results and Discussion

Climactic Conditions

During the 1,186 days of the WL and MC study period and the 1,150 days of the

Hunter, Irritrol, and Toro study period, 28% of the days received rain. For WL and MC,

the cumulative rainfall was 3,551 mm which is 14% less than the historical average of

4,121 mm (see Chapter 2, Figure 2-12). For Hunter, Irritrol, and Toro, the cumulative

rainfall was 3,410 mm which is 16% less than the historical average of 4,055 mm (see

Chapter 2, Figure 2-13). In Chapter 2, Figures 2-14 and 2-15 show the dailyand









cumulative rainfalls during each period and the theoretical count for the number of OSM

events by treatment.

Time in Open Switch Mode (Dry-Out)

Dry-out is the amount of time the rain sensors are in OSM. After the dry-out

period, irrigation would be allowed to occur. Figures 3-1 to 3-4 show frequency

distributions of time in 6-hour intervals that the RSs stayed in OSM for all treatments.

The WL dried-out within 24 hours 84% of the time with 1% requiring 49 to 53 hours

(Figure 3-1). The 3MC dried-out within 24 hours 83% of the time and with 1% requiring

50 and 66 hours (Figure 3-2). The 6MC dried-out within 24 hours 64% of the time and

with 4% requiring 49 and 69 hours (Figure 3-3). The 13MC dried-out within 24 hours

80% of the time and with 1% requiring 50 hours (Figure 3-4). Hunter dried-out within 24

hours 71 % of the time and with 3% requiring 48 and 77 hours (Figure 3-5). Of the 3% of

events with more than 48 hours of dry-out. Irritrol dried-out within 24 hours 84% of the

time and all dried-out out within 48 hours (Figure 3-16). Toro dried-out within 24 hours

83% of the time and all dried-out within 48 hours (Figure 3-7).

The frequency of dry-out within 24 hours for 6MC was 64% while the average for

all other treatments was of 81%. This reduced percentage was a remnant of the rainfall

setting change from 25 mm to 6 mm. As discussed in detail in Chapter 4, the

hygroscopic disks change length with time in use and rainfall setting. The initial 25 mm

setting caused the hygroscopic disks to change length differently than what the disks

would have done with a 6 mm setting, which affected the dry-out of the disks for 6MC.

Table 3-3 summarizes the percentage of time each treatment dried-out in 24 hours

and how many hours each treatment required for 95% dry-out. The number of hours the

WL, 3MC, 6MC, 13MC, Hunter, Irritrol, and Toro required for dry-out 95% of the time









were 36, 36, 42, 37, 44, 36, and 33 hours, respectively. These values are relatively

close for all sensors and settings and indicated that a majority of the rain sensors dried-

out within 24 hours and 95% of the time they dried-out in less than 2 days. Rainfall

needs to occur within 24 hours of a scheduled irrigation for most rain sensors to still be

in OSM to interrupt the irrigation.

Previous research (Cardenas-Lailhacar and Dukes, 2008) investigated dry-out on

the WL, 3MC, and 13MC for the first year of installation. Their results indicated that WL

dried-out within 24 hours 80% of the time with 8% requiring 54 and 78 hours, which is

consistent with this study. The 3MC dried-out within 24 hours 51% of the time with 12%

requiring 48 and 78 hours and 7% more than 78 hours. The 13MC dried-out within 24

hours 57% of the time with 6% requiring 48 and 72 hours. The 3MC and 13MC

treatments had a shorter dry-out during this study than the Cardenas-Lailhacar and

Dukes (2008) study, but treatment dry-out times did not change within this study period.

The amount of time in dry-out did not change during the study period for any of the

treatments. On 2 July 2009, the Hunter and Irritrol dry-out vents were changed from fully

open to fully closed. Hunter (Figures 3-8 and 3-9) and Irritrol (Figures 3-10 and 3-11)

had average dry-out increase times of 14% or 3 hours and 32% of 6 hours, respectively,

after closing the dry-out vents. Hunter dried-out within 24 hours 73% and 65 % of the

time, with fully open and fully closed vents, respectively. Irritrol dried-out within 24 hours

82% and 77% of the time, with fully open and fully closed vents, respectively. The

increase in dry-out time did not increase potential water savings. All of the Toro

replicates were set to 0.0 day dry-out. There was no difference in dry-out time for the









four Toro replicates set at 1.0 day water delay and 3.0 day water delay (Figures 3-12

and 3-13).

Dry-out Tracking

Figures 3-14 to 3-15 show the dry-out of the disks over time for the two dry-out

tracking events. A 60 mm rainfall event did not trigger all rainfall sensors to OSM: all

3MC, three 6MC, two 13MC, and three Hunter remained in CSM. The second tracking

period was formed by manually watering the disks. This event did not rigger all rainfall

sensors to OSM: two 3MC, one 6MC, two 13MC, and two Hunter remained in CSM. The

same sensors did not go into OSM during both dry-out tracking events likely because of

wear of the hygroscopic disks. Though these disks did not go into OSM, the disks did

expand with the rain event and contract over the dry-out period.

The dry-out patterns in the September tracking event were similar to July. Figures

3-16 to 3-18 show the relationship between the disk length and temperature, solar

radiation, and relative humidity for the dry-out tracking on 11 July 2009 since it has

more detail. Decreases in relative humidity and increases in temperature and solar

radiation were followed about 2 to 3 hours later with significant disk length decreases

(Figures 3-14 to 3-16). Disk length reduction was caused by evaporation of rainfall from

the disks. The lag time in response to the changes in the physical parameters was due

to the time and energy required to vaporize the rain water in the disks for evaporation.

The changes in climate preceding the significant disk contraction were a temperature

increase from 24 to 31 C, an increase in solar radiation from 159 to 1005 W/m2, and a

decrease in relative humidity from 93 to 54%. Each of the three parameters influenced

dry-out time.









Potential Irrigation Savings

The total potential water savings for each treatment under the different irrigation

schedules are in Table 3-4. The values of potential irrigation savings need to be

considered with reference to the respective accuracy of the rain sensors.Tables 3-5 to

3-8 show the variation of total potential water savings for each replicate. The average

percent water savings for the 2 d/wk schedule for the WL, 3MC, 6MC, 13MC, Hunter,

Irritrol, and Toro treatments were 26%, 26%, 28%, 14%, 23%, 21%, and 21%,

respectively. The average percentage water savings for the 1 d/wk schedule for the WL,

3MC, 6MC, 13MC, Hunter, Irritrol, and Toro treatments were 25%, 23%, 25%, 13%,

20%, 19%, and 14%, respectively.

Cardenas-Lailhacar et al. (2008) found that a rain sensor set at 6 mm on a timer

with the same UF IFAS schedule used in this study could reduce applied irrigation by

34% in Gainesville, Florida. A study in southwestern Florida found 21% irrigation

savings with a rain sensor set at 6 mm with a 2 d/wk irrigation schedule (Davis et al.,

2009). The 2 d/wk irrigation schedule potential savings in this study was 28% with the

6MC indicating that the 6MC treatment acted within the range of previous studies.

McCready et al. (2009) found that the 3 mm and 6 mm settings saved 25% and 17%,

respectively, under a 2 d/wk irrigation schedule. The McCready et al. (2009) findings for

the 3 mm setting match this study's savings of 26%. McCready et al. (2009) and Davis

et al. (2009) had similar savings with a 6 mm rainfall setting while Cardenas-Lailhacar et

al. (2008) had higher savings. The increased savings of Cardenas-Lailhacar et al.

(2008) was due to the higher rainfall in 2005 compared with the later studies in 2006

and 2007. The 28% savings in this study for the 6 mm setting fall in between savings of









the other studies. The average potential irrigation savings for rain sensors set at 6 mm

was 24% also in the range of 17% to 34% from previous research.

The amount of savings should have been less for the 6MC in this study. The 6CM

treatment savings should be between 26% and 14% which are the savings for 3MC and

13MC. The lack of difference in the 3MC and 6MC savings is due to the hygroscopic

disk properties of the 6MC after being changed from a 25 mm setting to 6 mm setting a

year after original installation (see Chapter 4 for hygroscopic disk details). From Chapter

2, the 3MC, 6MC, and 13MC sensors were in OSM after 2.0, 1.7, and 7.0 mm of rainfall

with accuracies of 64%, 27%, and 51%, respectively. The 6MC treatment had a very

low accuracy, and it should have had a depth required for OSM between the depths

required by 3MC and 13MC. The Hunter treatment, which was the Mini-Clik sensor set

to 6 mm from its time of installation, went into OSM after 4.1 mm as expected based on

the performance of 3MC and 13MC. The 6MC treatment did not perform as it would

have without the setting adjustment.

Summary and Conclusions

This experiment was carried out during a relatively dry period with rainfall on 28%

of the days. Rain sensor dry-out times did not change throughout the study period. The

13MC had greater variability of dry-out times than any other treatment. Averaged across

the treatments, the rain sensors dried-out within 24 hours 79% of the time and in 38

hours 95% of the time. Changing the vent settings from fully open to fully closed on

some treatments increased the dry-out time an average of 23%, but the potential

irrigation savings was unchanged. The Toro water delay feature does not have an effect

on the number of OSM occurrences or potential irrigation savings.









During the dry out process with the expanding disks, the most significant disk

contraction occurred 2 or 3 hours after changes in climatic conditions. The disks reacted

to decreasing relative humidity (from 93 to 54%) and increasing temperature (from 24 to

31C) and solar radiation (from 159 to 1005 W/m2). Apparently, 2 to 3 hours was

needed for a sufficient amount of water to vaporize from the hygroscopic disks to result

in a significant size reduction.

Potential water savings were determined by comparing the number of times the

sensors went into OSM and dry-out time to a theoretical UF IFAS irrigation schedule.

Potential irrigation savings should be considered with the accuracy of rain sensors. All

treatments, except Toro, had accuracies of less than 65%. The potential irrigation

savings presented in this research were higher than they would be with more accurate

rain sensors since most treatments went into open-switch mode with less rainfall than

their respective rainfall setting. For a 2 d/wk and 1 d/wk irrigation schedule, the

percentage water savings for the 13MC was 14% and 13% and the average for all other

treatments was 24% and 21%, respectively. As expected, the rain sensors with lower

rainfall setting had higher potential water savings. The average irrigation savings for

previous research with a 2 d/wk irrigation schedule with rainfall settings of 3 and 6 mm

was 24% and 21%, respectively. This virtual study had similar potential water savings

as previous research. There was no difference in potential irrigation savings between

the rain sensors with 3 and 6 mm rainfall settings because of the combination of the

inherent low accuracy of rain sensors and relative closeness of the settings (3 and 6

mm versus 6 and 13 mm).









Rainfall settings of 3, 6, and 13 mm are adequate for rain sensors in central

Florida because all settings conserved water. Rain sensors should be set to 3 or 6 mm,

which conserve more than a setting of 13 mm, until the user determines that landscape

quality or climatic conditions require the higher setting. The rainfall settings for a rain

sensor should not be changed after a particular setting has been established for more

than 3 months (see Chapter 4 for more details). If the rainfall setting needs to be

changed after 3 months of use, it is recommended that a new rain sensor be installed.


Table 3-1. Treatment description.
Treatment Model Replicates Set Dry-out Vent Setting
Point
WL Wireless Rain-Clik 4 -half open
3MC Mini-Clik 4 3mm fully open
6MC Mini-Clik 4 6mm fully open
13MC Mini-Clik 4 13 mm fully open
Hunter Mini-Clik 8a 6 mm fully open and fully closed
Irritrol Irritrol RFS 1000 8a 6mm fully open and fully closed
Toro Toro TWRS 8 6 mm 0 day dry-out
a Dry-out vents changed from fully open to fully closed on 02 July 2009.

Table 3-2. Monthly irrigation depth to replace historical evapotranspiration values based
on Dukes and Haman (2002a). Run times are based on an irrigation
application rate of 38 mm/hr assuming system efficiency of 60% and
considering effective rainfall. The Reduced UF IFAS irrigation schedule is
60% of the UF IFAS irrigation schedule.
Month Irrigation depth (mm)
January 0
February 0
March 0
April 81
May 160
June 143
July 131
August 120
September 154
October 103
November 56
December 0
Total 949









Table 3-3. Summary of dry-out time for all treatments.
Treatment Model Dry-out Vent Setting


Frequency of
dry-out within
24 hours


WL Wireless Rain-Clikx half open 84% 36
3MC Mini-Clikx fully open 83% 36
6MC Mini-Clikx fully open 64% 42
13MC Mini-Clikx fully open 80% 37
Hunter Mini-Clikx fully open and fully closed 71% 44
Irritrol Irritrol RFS 1000y fully open and fully closed 84% 36
Toro Toro TWRSz 0 day dry-out 83% 33
a Dry-out vents changed from fully open to fully closed on 02 July 2009.


Table 3-4. Total potential water savings per treatment for all treatments compared with
a 2 d/wk and a 1 d/wk irrigation schedule for the study period (Oct/Nov 2006


to 1


Dec 2009).


Treatment 2 d/wk Irrigation Schedule 1 d/wk Irrigation Schedule
Irrigation Water Savings Irrigation Water Savings
depth (mm) (%) Depth (mm) (%)
(mm) (mm)
WL 3,005 772 26 2,968 748 25
3MC 3,005 775 26 2,968 677 23
6MC 3,005 841 28 2,968 745 25
13MC 3,005 415 14 2,968 401 13
Hunter 2,902 684 23 2,869 574 20
Irritrol 2,902 614 21 2,869 548 19
Toro 2,902 595 21 2,869 407 14


Table 3-5. Variation in total potential water savings per replicate for the WL and MC
treatments compared with UF IFAS 2 d/wk irrigation recommendations.
Water saved by replicates (mm)
Treatment A B C D Average CV (%)
WL 702 161 827 788 772 za 8z
3MC 846 754 661 838 775 a 11
6MC 838 822 836 868 841 a 2
13MC 352 619 187 500 415b 45
Numbers with different letters indicate a statistical difference at the 95% confidence level using
Duncan's Multiple Range Test.
z Average and CV do not include WL-B


Hours
dried-out
95% of
the time









Table 3-6. Variation in total potential water savings per replicate for the Hunter, Irritrol,
and Toro treatments compared with UF IFAS 2 d/wk irrigation
recommendations.
Water saved by replicates (mm)
Treatment A B C D E F G H Average CV (%)
Hunter 675 742 756 137 375 747 747 744 684za 20z
Irritrol 647 688 664 368 499 688 672 688 614 a 19
Toro 615 581 565 574 605 627 594 599 595 a 4
Numbers with different letters indicate a statistical difference at the 95% confidence level using
Duncan's Multiple Range Test.
z Average and CV do not include Hunter-D


Table 3-7. Variation in total potential water savings per replicate for the WL and MC
treatments compared with UF IFAS 1 d/ wk irrigation recommendations.
Water saved by replicates (mm)
Treatment A B C D Average CV (%)
WL 715 142 792 735 748 za 5z
3MC 700 634 573 803 677 a 15
6MC 765 686 771 759 745 a 5
13MC 265 646 191 503 401 b 53
Numbers with different letters indicate a statistical difference using Tukey-Kramer adjusted p-values
of p<0.05.
z Average and CV do not include WL-B


Table 3-8. Variation in total potential water savings per replicate for the Hunter, Irritrol,
and Toro treatments compared with UF IFAS 1 d/ wk irrigation
recommendations.
Water saved by replicates (mm)
Treatment A B C D E F G H Average CV (%)
Hunter 580 620 585 93 269 655 655 655 574za 24z
Irritrol 573 514 503 503 544 578 537 634 548 a 8
Toro 382 517 517 324 514 324 324 356 407 a 23
Numbers with different letters indicate a statistical difference using Tukey-Kramer adjusted p-values
of p<0.05.
z Average and CV do not include Hunter-D















30

25

20

15

10

5

0


- -- 'I

100
(V
80 3
(V
60 -

40
4-A

20 E

o 3
0 U


Interval of hours for dry-out period

Figure 3-1. Histogram and cumulative frequency of dry-out time (time from the end of
the rain event until the sensors returns to closed-switch mode) for the WL
treatment average.


(A
u
C 25
Co
3 20

- 15
0 20
0 10




5
1V
L_
L 0


0
100
(3
80 3

60 -




20 :

0o
0 U


4

Interval of hours for dry-out period

Figure 3-2. Histogram and cumulative frequency of dry-out time (time from the end of
the rain event until the sensors returns to closed-switch mode) for the 3MC
treatment average.












74
















78
1 A-


7

^~7 mmmm


120
0
100

80 3
1- (A1
60 r-

40 "
30
20 E
0


o 6, 4 '

Interval of hours for dry-out period

Figure 3-3. Histogram and cumulative frequency of dry-out time (time from the end of
the rain event until the sensors returns to closed-switch mode) for the 6MC
treatment average.


22 99100


15 --


3 3 2
1I


"^ p *'p ^ o ^

Interval of hours for dry-out period
Figure 3-4. Histogram and cumulative frequency of dry-out time (time from the end of
the rain event until the sensors returns to closed-switch mode) for the 13MC
treatment average.


30

25I


20

15


30

25

20


- 100

- 80 3
0-
-60

-40 .
4-A
-20 3
E
0 U














30

2522
22


901
83


15 I


7


4-
-100
C
-80
r"
-60 '-


-40 *


-20 E
u


Interval of hours for dry-out period

Figure 3-5. Histogram and cumulative frequency of dry-out time (time from the end of
the rain event until the sensors returns to closed-switch mode) for the Hunter
treatment average.


40 ,


(A 35
3
u
C 30

u 25
u

0 .20
0
U 15

10
c
L 5
. 5


95
92


16


1


4-
- 100 0

C
-80

L_
-60 -


-40
3
E
-20 3
U


0 6l i ^p &0

Interval of hours for dry-out period

Figure 3-6. Histogram and cumulative frequency of dry-out time (time from the end of
the rain event until the sensors returns to closed-switch mode) for the Irritrol
treatment average.













w 35
u
30
C 30

( 25

O -20
0
u 15

3 10
07
L 5
LL


Un UM U


92 9/






9-


7 2


'4-
0
100 >.
u
C
80 3
r1
L_
60 '*

40 *
-3
20 I
U


0 1 1 1 1 0

K 0""6 "0

Interval of hours for dry-out period
Figure 3-7. Histogram and cumulative frequency of dry-out time (time from the end of
the rain event until the sensors returns to closed-switch mode) for the Toro
treatment average.


3


120

0
100
u

-80 |

60 r6-
L S

> O
60 4-
3 o

-20 :

-0


.^"s '4 '-_ 90 0 V '\96

Interval of hours for dry-out period
Figure 3-8. Histogram and cumulative frequency of dry-out time (time from the end of
the rain event until the sensors returns to closed-switch mode) for the Hunter
treatment average with the dry-out vents fully open (8 November 2008 to 2
July 2009).




77


30

25

20

15

10

5


-


-


Lg"


_^<" 98100












(A 35 4-
100 0
30


0 73 <
O e o 2069 60 C
w15 15
U 15
2 40
3 10 O
4 4 4 20 :
Lu. 5
0 0







40 120
10 0
Interval of hours for dry-out period











0 0
Figure 3-9. Histogram and cumulative frequency of dry-out time (time from the end of
the rain event until the sensors returns to closed-switch mode) for the Hunter
treatment average with the dry-out vents fully closed (2 J uly 2009 to 31





December 2009).20
40 120

100
o 30o
S90 -80
25
0u v


U- 151

: 10 0
C% 5 E%%

0 0



Interval of hours for dry-out period
Figure 3-10. Histogram and cumulative frequency of dry-out time (time from the end of
the rain event until the sensors returns to closed-switch mode) for the Irritrol
treatment average with the dry-out vents fully open (8 November 2008 to 2
July 2009).











60 120
(A-
48 0
u 50 100
100 r
3 40 80
8u 84 (
u 81
O
O 30 60 L _
1 33 a
StU
C 20 40
:3 13 0

_13

0 0
0 6' 1S '

Interval of hours for dry-out period
Figure 3-11. Histogram and cumulative frequency of dry-out time (time from the end of
the rain event until the sensors returns to closed-switch mode) for the Irritrol
treatment average with the dry-out vents fully closed (2 July 2009 to 31
December 2009).




40 120
S35 4-
35 0
100 >
C 30 -
26 98
25 (A
060
O ._.


io 0
09 10
4 20 E
5 5
L.2 U
0 0




Interval of hours for dry-out period



Figure 3-12. Histogram and cumulative frequency of dry-out time (time from the end of
the rain event until the sensors returns to closed-switch mode) for the one day
water delay setti ng for four the Toro rep licates.



79












40 120
36
(A 35 5
S100
0 30
24 93 80- 8
25 2 : "

0 -V
93O
Ot v20 60 -
4 >I
U 15 t
C 12 40
: 10 30
S22 E
20V
LLu 5

0 0

o 6' "PO -6-' $


Interval of hours for dry-out period
Figure 3-13. Histogram and cumulative frequency of dry-out time (time from the end of
the rain event until I the sensors returns to closed-switch mode) for the three
day water delay setting for four the Toro replicates.




25

24


E 23

b 22

S21

20

19
0:00 3:00 6:00 9:00 12:00 15:00 18:00 21:00 0:00

Time in dry-out (hour)

-4-3MC -ll-6MC -1l3MC -KHunter -Irritrol -40Toro

Figure 3-14. Dry-out tracking of average disk length for each treatment for the natural
rain event on 10 July2009.






























9:00 12:00 1
Timer in dry-out (hours)


.5:00 18:00


21:00 0:00


-3MC -- 6MC --13MC -*Hunter -K1Irritrol --Toro

Figure 3-15. Dry-out tracking of average disk length for each treatment for the manual
rain event on 18 September 2009.


U
28

2 -4
26 "
I-

24 0.
E
__ )-


0:00 3:00 6:00 9:00 12:00 15:00 18:00 21:00 0:00

Time in dry-out (hour)
3MC 1--6MC --13MC --Hunter W-Irritrol Toro -4-Temperature


Figure 3-16. Dry-out tracking of average disk length for each treatment and temperature
for the rain event on 10 July2009.



81


19


0:00


3:00


6:00


M)II


ii i -- i i \i











25

24

23

22

21

20

19


1200

1000

800

600

400

200

0

-200


0:00 3:00 6:00 9:00 12:00 15:00 18:00 21:00 0:00

Time in dry-out (hour)

43MC -6MC --*13MC --Hunter lrrritrol -Toro --Solar Radiation


Figure 3-17. Dry-out tracking of average disk length for each treatment and solar
radiation for the rain event on 10 July 2009.


0:00 3:00 6:00 9:00 12:00 15:00 18:00 21:00 0:00

Time in dry-out (hour)

43MC --6MC -13MC --Hunter $lirritrol -Toro -Relative Humidity


Figure 3-18. Dry-out tracking of average disk length for each treatment and relative
humidity for the rain event on 10 July 2009.









CHAPTER
RELATIONSHIP BETWEEN EXPANDING-DISK RAIN SENSOR DISK LENGTH AND
PERFORMANCE

Introduction

In 2000, Florida was the largest user of groundwater east of the Mississippi River

(Hutson et al., 2004). Florida's growing population continues to put a stress on the water

supply. Florida ranked first in single family home construction in 2005 with 209,162

homes built (United States Census Bureau [USCB], 2007). Seventy percent of single

family homes have automatic irrigation systems (Tampa Bay Water, 2005). Automatic

timer controls on irrigation systems in Florida have been reported to lead to a 47%

increase in water use (Mayer, et al., 1999). Water conservation measures are needed to

reduce water use.

Rain sensors interrupt the cycle of an automatic irrigation system controller when a

specific amount of rainfall has occurred (Dukes and Haman, 2002b). Rain sensors

remain in closed-switch mode that allows irrigation until sufficient rainfall changes it to

open-switch mode. Several types of rain sensors are available to homeowners. Most

rain sensor models can be adjusted to interrupt at different depths of rainfall, generally

between 3 and 25 mm. One type of rain sensor collects the rain water in a cup and

interrupts the irrigation based on a preset weight of water. Another type has a set of

electrodes that detect the water level in a small collection dish that measures the

amount of rainfall with two electrodes in a collection cup (Dukes and Haman, 2002b). A

disadvantage to both of these devices is that debris can get into the collection cup and

cause the system to interrupt irrigation without sufficient rainfall.

The most common type of rain sensor used in Florida is the expanding-disk rain

sensor. Compared with other types of rain sensors, it requires less maintenance and is









less expensive. The swelling of the hygroscopic disks in the sensor expand

proportionally to the amount of rainfall. When the disks swell to the selected rainfall

setting, the system goes into open-switch mode. The sensors rely on evaporation to

allow irrigation after a being in open-switch mode (dry-out). A longer dry-out time has

the potential to interrupt more scheduled irrigations by the irrigation controller. Dry-out

settings can be adjusted for most sensors. Settings should be chosen to match soil

conditions such that the disks dry-out at a rate similar to water leaving the root zone soil

profile. The hygroscopic disk dry-out time is influenced by weather conditions such as

temperature, wind, solar radiation, and relative humidity.

After investigating expanding-disk rain sensors since 2005, researchers noticed

that the hygroscopic disks appear to have different sizes with time (Figures 4-1 and 4-

2). The disks may lose elasticity with time due to the repeated shrinking and swelling

from rain events resulting in a longer total length because they appear to not shrink to

their original length. A longer length of disks inside the sensor may cause the RS to

interrupt the irrigation cycle with less rain. Cardenas-Lailhacar et al. (2009) determined

that the sensitivity of the RS changed during 282 days of installation. One possible

cause of the change in sensitivity is a change in size of the hygroscopic disks

themselves.

The objectives of this study were to determine if the length of the hygroscopic

disks in expanding-disk rain sensors change size based on the amount of time installed

and the set point, and if so determine the effect of disk change on switching accuracy.

Materials and Methods

This study was conducted at the University of Florida Agricultural and Biological

Engineering Department campus turfgrass plots, Gainesville, Florida. The rain sensors









were installed on a 2-m high board next to each other. Figure 2-2 (see Chapter 2)

shows the details of the expanding-disk rain sensor used for this study.

Treatments

Five treatments with varying rain sensor set points were established (Table 4-1).

The rain sensors used were Mini-Clik (MC) rain sensors from Hunter Industries, Inc.,

(San Marcos, CA). Treatments 3MC and 13MC had rainfall settings of 3 mm and 13

mm, respectively, and were installed on 25 March 2005. The 3MC and 13MC

treatments had four replicates each. Treatments 3R, 6R, and 13R with rainfall settings

of 3 mm, 6 mm, and 13 mm, respectively, were installed on 13 February2009. The 3R,

6R, and 13R treatments were replicated three times.

Monitoring

The total disk length was measured about once per month with a Mitutoyo Series

505 dial caliper (Mitutoyo Corporation, Aurora, IL) starting February 2009. Figure 2-5

(see Chapter 2) shows the hygroscopic disks inside the rain sensor. Disks were

measured when fully contracted and all rainfall had evaporated. Treatments 3R, 6R,

and 13R were measured before installation to give the length of a new device.

To determine the accuracy of the 3MC and 13MC treatments, each time a rain

sensor changed mode between open switch mode (OSM) and closed switch mode

(CSM), the date and time were recorded at a 1-second sampling interval using AM16/32

multiplexers (Campbell Scientific, Inc., Logan, UT) attached to a CR10X model data

logger (Campbell Scientific, Inc., Logan, UT). An onsite automated weather station

(Campbell Scientific, Logan, UT) located within 15 m of the experimental site recorded

climatic conditions using a CR10X model data logger (Campbell Scientific, Logan, UT).

Precipitation was measured by a tipping bucket rain gauge with a one-second sampling









interval time stamp for each 0.25 mm of rain. A manual rain gauge located within 5

meters of the rain sensors was used to verify the accuracy of the tipping bucket rain

gauge measurements (See Chapter 2, Figure 2-9).

The travel distance from CSM to OSM of each sensor was measured to determine

if possible disk length changes were significant. The travel distance was the required

increase in disk length due to rainfall for the rain sensor to go into OSM. Travel distance

was determined with dry rain sensors by measuring the post on the top of the rain

sensor in a stationary position (CSM) and in the position of full disk expansion (OSM).

Total travel distance included the difference in the rain sensor post as described above

plus the distance required to compress the trigger device underneath the hygroscopic

disks.

Statistical Analysis

SAS statistical software (SAS Institute, Inc., Cary, NC) was used for all statistical

analysis. A general mixed model with an auto regressive error structure was used to

model the continuous responses (PROC MIXED). A paired T-Test was used for means

separation with significant F values (p<0.05).

Results and Discussion

Disk measurements were performed to track size changes in the hygroscopic disk

length. Figure 4-3 shows the daily and cumulative rainfall during the study period and

the number of rainfall events greater than the three rain sensor rainfall settings

investigated (3, 6, and 13 mm).

Length by Installation date and setting

The difference between initial and final disk length for 3MC, 13MC, 3R, 6R, and

13Rwas -0.1, 0.4, 1.2, 1.8, and 2.3 mm, respectively. Table 4-2 shows the initial and









final disk measurements for each treatment. The disk length of 3MC and 13MC did not

change significantly during the measurement period. After 81 days of installation, the

disk lengths of 3R, 6R, and 13R were significantly expanded from the initial length, by

1.2, 1.8, and 2.30 mm, respectively. The length of disks after 81 days of installation and

at the end of the study (276 days) for 3R, 6R, and 13R were 18.1 and 18.7 mm, 18.7

and 19.2 mm, and 19.0 and 19.6 mm. The disk lengths of 3Rand 13R approached the

lengths of the 3MC and 13MC, respectively.

The treatments installed in 13 February 2009 started with the same average length

of 17.4 mm (Table 4-3). After 81 days of installation, the disk lengths for 3R, 6R, and

13R were all significantly different (p-value <0.05) from the initial measurements with

lengths of 18.1, 18.7, and 19.0 mm, respectively. The disk lengths by treatment

remained statistically different during the rest of the study. This result was consistent

with the older sensors, 3MC and 13MC, in which the disks had different lengths, 19.0

and 20.2 mm respectively. Figure 4-4 shows the change in length during the study

period for each treatment.

There was little variation of disk length within treatments (Tables 4-4 and 4-5). The

average Coefficient of Variance (CV) values for February, May, and November

measurements for the 3MC, 13MC, 3R, 6R, and 13R were 1.0%, 1.2%, 0.8%, 1.6, and

1.0%, respectively. The average CV for disk length was 1.1% while the average CV for

the depth of rainfall required for OSM was 47% (see Chapter 2 for details).

Disk Length and Traveling Distance

The length that the disks travel when switching between CSM and OSM was

analyzed to determine if the disk length changes were substantial. Travel distance was

dependent on rainfall setting. For the rain sensors installed in February, the average









length increased 1.7 mm and the average travel distance decreased 0.9 mm (Table 4-

6). The changes in disk length were substantial with respect to the travel distance.

Table 4-7 shows the values of final disk length, length change, and travel distance at

the end of the study. Disk length change and the total travel distance were larger in rain

sensors with higher rainfall setting because there was more space inside the rain

sensor. Treatments 3MC and 13MC had a shorter travel distance than 3R and 13R,

respectively, because the 3MC and 13MC disks were longer.

Effect on Interruption Performance

The accuracy of the 3MC and 13MC units for the first 300 days of installation was

analyzed to evaluate the effect of disk length change on accuracy. Analysis is based on

the assumption that the disks installed in 2005 and 2009 had the same properties.

Figure 4-5 compares accuracy of 3MC and 13MC to the disk size change in the 3R and

13R. Treatments 3MC and 13MC had no correlation of accuracy and disk length

change.

Summary and Conclusions

New (0 to 276 days old) and old (1,421 to 1,697 days old) rain sensors were

analyzed to determine the effect of disk length change on rain sensor accuracy. The

hygroscopic disks in expanding-disk rain se nsors change length after rainfall exposure.

Disk length was analyzed based on time installed in the field and rainfall setting.

The new rain sensors had an average disk length increase of 1.8 mm and the old rain

sensors changed 0.1 mm in size during the same time. The average length change for

the new rain sensors was more than the travel distance inside the sensor when

switching from closed-switch mode to open-switch mode (1.7 mm versus 0.9 mm). This









finding indicated that the amount of disk length increase was significant enough to

possibly influence rain sensor performance.

The lengths of the RS disks after 81 days of installation and 178 mm of rainfall

were significantly greater than the initial measurements. Higher rainfall settings had

more disk length increase due to the larger space within the rain sensor device at higher

rainfall settings and the material losing elasticity while expanding and contracting. The

amount of time the rain sensors remained in open-switch mode was not related to the

setting; the disk length difference between 3 mm and 13 mm settings did not influence

the dry-out (see Chapter 3 for details).

Increased disk length change was not related to decreased rain sensor accuracy.

The accuracies presented in Chapter 2 showed that both 3MC and 13MC were more

accurate in the first 282 days of installation than the accuracy of days 560-1,742 of

installation. It was assumed that the properties of the disks installed in 2005 were the

same as those installed in 2009. All of the disk length change occurred during the initial

Cardenas-Lailhacar and Dukes (2008) study in which the average accuracy was 83%.

In this study, the accuracy of the same sensors was 62% in which no significant disk

length change occurred. The decreased accuracy cannot be attributed to disk length

change because the time in which the disk lengths changed did not correspond to the

time of decreased accuracy. There was no relationship between the disk length and

accuracy. Decreasing accuracy of the rain sensors was due to the aging of the entire

unit. The outer casing of the rain sensor became more brittle with sun and rain

exposure. The triggering mechanism required less movement on the older rain sensors

than the new rain sensors when manually going into open-switch mode.









The reason that the disks could have a significant length change without affecting

accuracy was because of the disks properties. The rain sensors with higher rainfall

settings had more increase in disk length because they have more space inside the rain

sensor in which to lengthen. The increase in length also meant that there was more

pore space inside each disk. While the pore space shortened the travel distance for the

sensor, the amount of rainfall required to expand the full travel distance for open-switch

mode did not change because those pores then needed to be filled with water. Thus, an

increase in disk length did not correspond to the rain sensor requiring less rainfall for

open-switch mode.

Table 4-1. Description of rain sensors details for each treatment.
Treatment Model Replicates Set Installation
Point Date
3MC Mini-Clikx 4 3 mm 25 Mar 2005
13MC Mini-Clikx 4 13 mm 25 Mar 2005
3R Mini-Clikx 3 3mm 13 Feb 2009
6R Mini-Clikx 3 6mm 13 Feb 2009
13R Mini-Clikx 3 13 mm 13 Feb 2009
x Hunter Industries, San Marcos, CA


Table 4-2. Average disk length for each treatment at two intervals: initial and final (276
days of installation).
Treatment Installation Set point Disk Length Measurement (mm)
Date (mm) February November Difference
3MC 25 Mar 2005 3 19.2 19.1 NS
13MC 25 Mar 2005 13 20.1 20.4 NS
3R 13 Feb 2009 3 17.5 18.7
6R 13 Feb 2009 6 17.4 19.2
13R 13 Feb 2009 13 17.3 19.6 **
CV (%) 6.7 3.3
NS = no statistical difference between February and November measurements
= statistical difference at 0.05 p-value level between February and November measurements
** = statistical difference at 0.001 p-value level between February and November measurements









Table 4-3. Average disk length for the treatments installed 13 February 2009 at three
intervals: initial, 81 days of installation, and final (276 days of installation).
Treatment Set point Disk Length Measurement
(mm) February May November
3R 3 17.5 a 18.1 a 18.7 a
6R 6 17.4 a 18.7 b 19.2 b
13R 13 17.3 a 19.0 c 19.6 c


Numbers with different letters indicate a statistical difference
significant F values (p<0.05)


using aT-Test for means separation with


Table 4-4. Disk length for replicates of treatments installed 25 March 2005 at three
intervals: initial (February), 81 days of installation (May), and final (November,
276 days of installation).
Treatment Disk Length Measurement Treatment Disk Length Measurement
Replicate (mm) Replicate (mm)
Feb May Nov Feb May Nov
3MC-A 18.9 19.1 19.0 13MC-A 19.8 19.9 20.2
3MC-B 19.2 19.1 19.1 13MC-B 20.0 20.6 20.5
3MC-C 19.3 19.3 19.2 13MC-C 20.3 20.4 20.5
3MC-D 19.4 18.6 19.1 13MC-D 20.4 20.4 20.5
Average 19.2 19.0 19.1 Average 20.1 20.3 20.4
CV (%) 1.0 1.5 0.5 CV (%) 1.4 1.4 0.7


Table 4-5. Disk length for replicates of treatments installed 13 February 2009 at three
intervals: initial (February), 81 days of installation (May), and final (November,
276 days of installation).
Treatment 3R 6R 13R
Replicate Disk Length Disk Length Disk Length
Measurement (mm) Measurement (mm) Measurement (mm)
Feb May Nov Feb May Nov Feb May Nov
A 17.1 18.0 18.6 17.4 18.8 19.1 17.1 19.1 19.7
B 17.7 18.1 18.7 17.9 18.8 19.2 17.7 19.0 19.8
C 17.5 18.2 18.7 16.8 18.4 19.3 17.5 19.0 19.4
Avg 17.5 18.1 18.7 17.4 18.7 19.2 17.5 19.0 19.6
CV (%) 1.7 0.4 0.2 3.0 1.3 0.5 1.7 0.3 1.0









Table 4-6. Comparison of average length change and travel distance from closed-switch
mode to open-switch mode of treatments installed in 13 February 2009. The
February travel distance was measured on a rain sensor before installation.
Treatment February November Change
Disk Travel Disk Travel Disk Travel
Length Distance Length Distance Length Distance
(mm) (mm) (mm) (mm) (mm) (mm)
3R 17.5 1.7 18.7 1.7 1.2 0.0
6R 17.4 3.0 19.2 1.8 1.8 -1.2
13RC 17.5 4.6 19.6 3.1 2.3 -1.5
Avg 17.5 3.1 19.2 2.2 1.7 -0.9

Table 4-7.Comparison of average length change of each treatment from 13 February
2009 to 16 November 2009 and the travel distance each treatment from
closed-switch mode to open-switch mode. Travel distance was measured at
the end of the study.
Treatment Set Point Final Disk Length Change Travel Distance
(mm) Length (mm) (mm) (mm)
3MC 3 19.1 -0.1 1.4
13MC 13 20.4 0.4 2.6
3R 3 18.7 1.2 1.7
6R 6 19.2 1.8 1.8
13R 13 19.6 2.3 3.1
Avg 19.4 1.1 2.1


Figure 4-1.Mini-Clik (Hunter Industries, Inc.) rain sensor expanding disks installed in
March 2005 (left) and February 2009 (right) set at 13 mm measured in August
2009 (1600 and 179 days of installation, respectively).





























Figure 4-2.Mini-Clik (Hunter Industries, Inc.) rain sensors expanding disks installed in
2005 with settings (left to right) of 3 mm, 6 mm, and 13 mm and lengths 19.2,
19.8, and 20.1 mm, respectively, after 1,600 days of installation.











events with rr o e1 han 25 mm of lain ranging frorn 16 to60 mm








.3 mm se po'





Srim sel oin t


-,1 ro st s _i. t-


--LI I. I, II I 1 11 [ I I.l


40^


1200


1000
954

E
800 .

C

600
'I


400
E


200



0


c/


VO


Date, 2006 2009

Figure 4-3. Cumulative and daily rainfall during the study period with number of rainfall events greater than the different
rain sensor rainfall settings.


*40












21.5


17.0 ,
0 35 69 104 138 173 207 242 276

Day of measurement

3MCavg 13MCavg 3Ravg -- 6Ravg -*-13Ravg


Figure 4-4. Average hygroscopic disk length for Mini-Clik rain sensors installed in
2005 (MC) and 2009 (R) over installation time. The 2009 rain sensors were
newly installed on day zero.


100%



I U *
60% -- mm-I m

40% ---

20%

0%


Disk Length Change (mm)
*3MC U13MC


Figure 4-5. Accuracy for each setting during the first 300 days of rain sensor installation
compared to change in disk length. Accuracy data are the average amount of
rainfall for open-switch mode for a given rainfall event for a treatment.









CHAPTER 5
CONCLUSIONS AND FUTURE WORK

Conclusions

The goal of this research was to evaluate the performance of expanding-disk rain

sensors. The primary objectives of this study were to use three brands to A) evaluate

the accuracy with time with respect to the selected rainfall setting, B) evaluate the

amount of time rain sensors remained in interruption mode (open-switch mode) after a

rainfall event, C) quantify potential irrigation savings for different rainfall settings

compared with a time-based schedule, and D) determine if the hygroscopic disks in the

rain sensors change length with time. The secondary objectives included a)

investigating variability within brands and rainfall settings, b) evaluating the changes in

accuracy with time, c) determining climatic parameters that influence dry-out time, and

d) evaluating the effect of possible disk length change on accuracy.

This experiment was carried out during a relatively dry period with rainfall on 28%

of the days. The percentage of rain events greater than 3, 6, and 13 mm were 57%,

42%, and 25%. The WL, which had no rainfall setting, had an average rainfall depth of

3.2 mm required to go into open-switch mode. The 3MC, 6MC, 13MC, Hunter, Irritrol,

and Toro treatments had accuracies of 64%, 27%, 51%, 64%, 71%, and 97%,

respectively. The accuracy of the rain sensors changed with time. The percentile point

change in accuracy during the study period ranged from -36% to 59%, where a negative

value indicated a decrease in accuracy. The 6MC rain sensors had a 25 mm rainfall

setting when initially installed and were changed to a 6 mm rainfall setting at the

beginning of this study. This change caused the 6MC treatment to have the lowest

overall accuracy and the largest increase in accuracy.









The amount of time a rain sensor stays in open-switch mode, called dry-out,

influences water savings. Longer dry-out times means that the rain sensors are in open-

switch longer and could potentially interrupt more irrigations. On average, rain sensors

dried-out within 24 hours 79% of the time and in 38 hours 95% of the time. Changing

the vent settings from fully open to fully closed on some treatments increased the dry-

out time but the potential irrigation savings was unchanged. The Toro water delay

feature did not have an effect on the amount of time in open-switch mode or potential

irrigation savings. Disk dry-out was related to climactic parameters. Dry-out occurred

with decreasing relative humidity and increasing temperature and solar radiation. The

most significant disk contraction occurred 2 or 3 hours after changes in climatic

conditions. The changes in climate preceding the significant disk contraction were a

temperature increase from 24 to 31 C, an increase in solar radiation from 159 to 1005

W/m2, and a decrease in relative humidity from 93 to 54%.

The hygroscopic disks in expand ng-disk rain sensors increase in length after

continuous rainfall exposure. The disk length change was related to the rainfall setting:

the 3 mm had the shortest length and the 13 mm had the longest length tending to

conform to the space available in the rain sensor body. The length of the rain sensor

disks was significantly more than the initial measurement with 81 days of installation

and 178 mm of rainfall. The rain sensors with higher rainfall settings had greater

increases in disk length because they had more space inside the rain sensor in which to

lengthen. The amount of rainfall required to expand the full travel distance for open-

switch mode did not change because those pores then needed to be filled with water.









Disk length was studied to determine its possible effects on rain sensor

performance. The data did not show that disk length change was related to decreased

accuracy. The decreased accuracy of 82% in the initial study (days 0 to 282 of

installation) to 62% in this study (days 560 to 1742 of installation) was not due to disk

length change. The disks got larger during their first 271 days of installation which was

during the initial study (days 0 to 282 of installation). The disk length increase did not

influence the dry-out time. There was also no difference in dry-out time for different

rainfall settings. The decreased accuracy was attributed to the outer casing of the rain

sensor became more brittle and the triggering mechanism become more sensitive.

The potential water savings for a 2 d/wk and 1 d/wk irrigation schedule 13MC were

14% and 13% and the average for all other treatments was 24% and 21%, respectively.

Rain sensors with 3 and 6 mm rainfall settings had similar savings due the low accuracy

of some treatments of rain sensors and relative closeness of the settings (3 and 6 mm

versus 6 and 13 mm). Potential irrigation savings should be considered in relation to the

accuracy of rain sensors. Since most treatments went into open-switch mode with less

rainfall than their respective rainfall setting, the potential irrigation savings presented in

this research were higher than they would be with more accurate rain sensors.

Rainfall settings of 3, 6, and 13 mm are adequate for rain sensors in central

Florida because all settings conserved water. Rain sensors should be set to 3 or 6 mm

until the user determines that landscape quality or climatic conditions require the higher

setting. If the rainfall setting needs to be changed after more than 3 months of use, it is

recommended that a new rain sensor or at least new expanding disks be installed. For

the best accuracy, there is evidence based on historical studies and results from this









study that Hunter Mini-Clik rain sensors should be replaced after 1 year of installation.

Based on their performance, Irritrol RSF 1000 and Toro TWRS rain sensors do not

need to be replaced for at least 3 years. Further research is needed to verify these

results. A survey of Florida homeowners with automatic irrigation systems found that

only 25% reported to have a rain sensor of which not all are likely correctly installed

(Whitcomb, 2005). The inclusion of rain sensors on more automatic irrigation systems

could increase water savings to homeowners and have environmental benefits such as

urban less runoff.

Future Work

Changing the rainfall setting as the sensors age has a significant effect on the

accuracy of the rain sensor at the new setting. Research should be conducted on rain

sensors with a variety of changed rainfall settings.

Future disk measurements of new rain sensors taken every weeks of the first 3

months of installation would offer better insight into disk length change. To best

determine the influence of disk length on accuracy, the new sensors should be

connected to a data logger.

Additional experiments studying the effect of rainfall setting on turf quality should

be conducted to determine the best settings for Florida. Both 3 and 6 mm settings have

been studied in relation to turf grass quality; the 13 mm setting has only gone through

virtual testing. The potential irrigation savings are a good starting point, but plots need

to be used for more supporting data.









LIST OF REFERENCES


Arlington, Texas. Lawn and landscape irrigation conservation ordinance. Section 4.27.
Available at:
http://www.ci.arling ton.tx.us/water/waterconservation_ordinancesummary.html
(Accessed 28 May 2009).

Cardenas-Lailhacar, B., Dukes, M.D., 2008. Expanding disk rain sensor performance
and potential irrigation water savings. Journal of Irrigation and Drainage 134(1):
67-73.

Cardenas-Lailhacar, B., Dukes, M.D, Miller, G.L., 2008. Sensor-based automation of
irrigation on bermudagrass, during wet weather conditions. Journal of Irrigation
and Drainage 134(2):120-128.

Cardenas-Lailhacar, B., Dukes, M.D., Meeks,L., 2009. Irrigation rain sensor accuracy.
Proceedings of the World Environmental & Water Resources Congress (EWRI),
Kansas City, MO. Available at: http://dx.doi.org/10.1061/41036(342)431
(Accessed 1 March 2010).

Cary, North Carolina. Rain sensors on automatic irrigation systems. Ordinance section
36-84. Available at: http://www.townofcary.org/Home.htm (Accessed 9 April
2009).

Colleyville, Texas. Water conservation ordinance 06-1579. Available at:
http://www.colleyville.com/content/view/716/12/. (Accessed 28 May 2009).

Connecticut Statutes. Section 29-265. Rain sensor devices for automatic lawn sprinkler
systems. Available at: http://search.cga.state.ct.us/dtsearch_pub_statutes.html
(Accessed 25 May 2009).

Dallas, Texas Ordinances. Section 49-21.1. Conservation measures relating to lawn
and landscape irrigation. Available at:
http://www.savedallaswater.com/pdf/Conservation_Ordinance.pdf (Accessed 9
April 2009).

Davis, S.L, Dukes, M.D., and Miller, G.L., 2009. Landscape irrigation by
evapotranspiration-based irrigation controllers under dry conditions in Southwest
Florida. Agricultural Water Management 96(2): 1828-1836. Available at:
http://dx.doi.org/10.1016/j.agwat.2009.08.005 (Accessed 28 January 2010).

Derby, Kansas Ordinances, 2008 Ordinance No. 1932. Rain sensor ordinance.
Available at: http://www.derbyweb.com/pdfs/rain-sensor-ordinance.pdf (Accessed
28 May 2009).

Dewey, C., 2003. Sensors at work. Irrigation and Green Industry, September 2003.
Available at: http://www.igin.com (Accessed 28 May 2009).


100









Dukes, M. D., Cardenas-Lailhacar, B., 2007. Smart Water Application Technologies
(SWAT) Turf and Landscape Irrigation Equipment Testing Protocol for Rain
Sensors: 1st draft. Irrigation Association, Falls Church, VA. Available at:
http://www.irrigation.org/gov/pdf/Phase 1-1 st_Draft_Test_Protocols-
RainSensors.pdf (Accessed 19 April 2009).

Dukes, M.D., Haley, M.B., 2009. Evaluation of soil moisture-based on-demand irrigation
controllers, phase II, final report. Available at:
http://irrigation.ifas.ufl.edu/pdf/publications/SMS/SMS %20Phase%2011%20F inal
%20Report%2012-17-09.pdf. (Accessed 27 January 2010).

Dukes, M. D., Haman, D.Z., 2002a. Operation of residential irrigation controllers.
CIR1421, Institute of Food and Agricultural Sciences, University of Florida,
Gainesville, FL. Available at: http://edis.ifas.ufl.edulpdffiles/AE/AE22000.pdf
(Accessed 28 January 2010).

Dukes, M. D., Haman, D.Z., 2002b. Residential irrigation system rainfall shutoff devices.
ABE325, Institute of Food and Agricultural Sciences, University of Florida,
Gainesville, FL. Available at: http://edis.ifas.ufl.edulae221 (Accessed 28 January
2010).

Figliola, R.S., Beasley, D.E., 2002. Theory and design for mechanical measurements.
3rd edition John Wiley and Sons, Inc. New York.

Florida Senate Bill 494. Water conservation/automatic sprinkler systems. Proposed
revisions to Chapter 373.62.Available at: http://www.flsenate.gov/cgi-
bin/view_page.pl?Tab=session&S ubme nu=l &FT=D&File=session/2009/Senate/
bills/amendments_com/html/sb0494c2409270.html (Accessed 28 May 2009).

Florida Statutes, 2008. Chapter 373.62. Water Conservation; automatic sprinkler
systems. Available at:
http://www.leg.state.fl.us/statutes/i ndex.cfm?App_mode=Display_Statute&U RL=
Ch0373/ch0373.htm (Accessed 19 April 2009).

Georgia Metropolitan North Georgia Water Planning District, Georgia. Water
conservation action no. 4. Available at: http://www.northgeorgiawater.com
(Accessed 26 May 2009).

Haley, M. B., Dukes. M. D., 2007. Evaluation of sensor based residential irrigation water
application. Proceedings from the American Society of Agricultural and Biological
Engineers International Meeting, 17-20 June 2007, Minneapolis, MI. Available at:
http://irrigation.ifas.ufl.edu/pdf/publications/SMS/ASABE-07-2251-Pinellas-Co.pdf
(Accessed 16 April 2009).









Haley, M.B., Dukes, M.D., Miller, G.L., 2007. Residential irrigation water use in Central
Florida. Journal of Irrigation and Drainage Engineering 133(5): 427-434.
Available at:
http://scitation.aip.org/getpdf/servlet/GetPDFServlet?filetype=pdf&id=J IDEDH000
133000005000427000001 &idtype=cvips (Accessed 16 April 2009).

Harrisburg, South Dakota Ordinances, 2006. Ordinance 2006-04 Chapter 8.01.
Available at: http://www.harrisburg.govoffice.com/vertical/Sites/%7BA7E6811A-
D879-4C7C-A19F-FOA55FD49B95%7D/uploads/%7B04211876-C006-4C4D-
BCA2-57348DD77A59%7D.PDF (Accessed 28 May 2009).

Hunter Industries, 2005. Product information-sensors. Hunter Industries, Inc., San
Marcos, CA. Available at:
http://www.hunterindustries.com/Resources/PD Fs/Product_Guides/Domestic/lit2
93w.pdf (Accessed 8 July 2009).

Hutson, S. S., Barber, N. L., Kenny, J. F., Linsey, K. S., Lumia, D. S., Maupin, M. A.,
2004. Estimated Use of Water in the United States in 2000. United States
Geological Survey Circular 1268. United States Department of the Interior,
United States Geological Survey, Reston, VA. Available at:
http://pubs.usgs.gov/circ/2004/circ1268/index. html (Accessed 9 July 2009).

Lucas, Texas Ordinances, 2005. Article 8. Irrigation system regulations. Section 3-220.
Rain and freeze sensors. Available at: http://lucastexas. us/objects/2005-11-
00540.1 Rain_Freeze_Sensors.pdf (Accessed 28 May 2009).

Marella, R.L., 1992. Water withdrawals, use, and trends in Florida, 1990. Water
Resources Investigations Report 92-4140. United States Department of the
Interior, United States Geological Survey, Tallahassee, Fl. Available at:
http://fl.water.usgs.gov/Abstracts/wri91_4123_marella. html (Accessed 21 May
2009).

Marella, R.L., 2004.Water withdrawals, use, discharge, and trends in Florida, 2000.
Scientific Investigations Report 2004-5151. United States Department of the
Interior, United States Geological Survey, Tallahassee, Fl. Available at:
http://pubs.usgs.gov/sir/2004/5151/pdf/20045151 .pdf (Accessed 21 May 2009).

Massachusetts Senate 186the General Court. Senate Bill 1396. Available at:
http://www.mass.gov/legis/bills/senate/186/st01 pdf/ST01396.PDF(Accessed 28
May 2009).
Mayer, P.W., DeOreo, W.B., Opitz, E.M., Kiefer, K.C., Davis, W.Y., Dziegielewski,B.,
Nelson, J.O., 1999. Residential end uses of water. American Water Works
Association Research Foundation. Denver, CO.









McCready, M.S., Dukes, M.D., Miller, G.L., 2009. Water conservation potential of smart
irrigation controllers on St. Augustinegrass. Agricultural Water Management
96(11 ):1623-1632. Available at: http://dx.doi.org/10.1016/j.agwat.2009.06.007
(Accessed 28 January 2010).

Minnesota Statutes, 2003. Chapter 44-F. No. 335. Available at:
https://www. revisor.m n.gov (Accessed 9 April 2009).

Moriarty, T., 2009. Front page Globe article targets lawn irrigation. Irrigation Association
of New England newsletter Issue 2Q2009. Available at:
http://www.irrigationassociationne.org/newsletters/2Q2009.pdf (Accessed 9 July
2009).

NOAA (National Oceanic and Atmospheric Administration), 2003. Monthly station
normals of temperature, precipitation, and heating and cooling degree days 1971
2000; 08 Florida. Climatographyof The United States No. 81. 28p. Available
at: http://hurricane.ncdc.noaa.gov/climatenormals/clim81/FLnorm.pdf (Accessed
27 January 2010).

New Jersey Statutes, 2000. 52:27D-123.13. Device to override automatic sprinkler after
adequate rainfall required. Available at: http://lis.njleg.state.nj.us/cgi-
bin/om_isapi.dll?clientlD=178043&Depth=4&TD=WRAP&advq uery=rain%20sens
or&headingswithhits=on&i nfobase=statutes. nfo&ra nk=&record= {15B3A}&softpag
e=Doc_Frame_Pg42&wordsaround hits =2&x=47&y=15&zz= (Accessed 26 April
2009).

Omega Engineering, Inc., 1995. RG-2501 Series: Tipping bucket rain gauge operators
manual. Available at: http://www.omega.com/manuals/manualpdf/M2235.pdf
(Accessed 29 March 2010).

San Antonio Water Systems. Ordinance 100332 34.274.2. Available at:
http://www.bexarmet.org/files/SanAntonio2005ConservationOrdi nance.pdf
(Accessed 28 May 2009).

San Antonio Water System. Rain sensors frequently asked questions. San Antonio
Water System, San Antonio, TX. Available at:
http://www.saws.org/conservation/ordinance/rainsensors/faq.shtml (Accessed 15
July 2009).

Spectrum Technologies, Inc. Tipping bucket rain gauge: Catalog #3665R. Available at:
http://www.specmeters.com/pdf/3665R%20Tipping%20Rain%20Gauge.pdf
(Accessed 29 March 2010).

Sutron Corporation. Tipping bucket rain gauge: Stainless steel 5600-0425. Available at:
http://www.sutron.com/pdfs/2006_RainGauge_StainlessSteel_5600-0425.pdf
(Accessed 29 March 2010).


103









Tampa Bay Water, 2005. Tampa Bay Water: Evaluating implementation of multiple
irrigation and landscape ordinances in the Tampa Bay region. Available at:
http://www.tampabaywater.org/documents/conservation/RegionLandscapeOrdi na
nces.pdf (Accessed 27 January 2010).

Texas Electronics, Inc., (a). Manual for field calibration kit: For TR-525 rain gauge.
Available at: http://www.texaselectronics.com/pl_supp_rainfall.htm (Accessed 30
November 2009).

Texas Electronics, Inc., (b). TR-525S siphon rain gauge tipping bucket manual.
Available at: http://www.texaselectronics.com/pdf/manuals/Siphon.pdf (Accessed
29 March 2010).

Texas House Bill 2299. Introduced 09 March 2007. Standards for irrigation system
equipment. Available at:
http://www. legis.state.tx.us/tlodocs/80R/billtext/pdf/HB02299 1.pdf (Accessed 9
April 2009).

USCB (United States Census Bureau), 2009. Population estimates. USCB. Washington,
D.C. Available at: http://www.census.gov/popest/estimates.html (Accessed 27
January 2010).

USCB (United States Census Bureau), 2007. Housing unit estimates. USCB.
Washington, D.C. Available at:
http://www.census.gov/const/C40/Table2/tb2u2005.txt (Accessed 28 January
2010).

Water Authority of Great Neck North, New York. Rules and regulations: general
information. Available at: http://www.waterauthorityofgreatnecknorth.com
(Accessed 26 May 2009).

Whitcomb, J. B., 2005. Florida water rates evaluation of single-family homes. Southwest
Florida Water Management District, Brooksville, FL. Available at:
http://www.swfwmd.state.fl.us/documents/reports/water_rate_report.pdf
(Accessed 21 May 2009).









BIOGRAPHICAL SKETCH

Leah Meeks received her Bachelor of Science in BioResource and Agricultural

Engineering at Cal Poly, San Luis Obispo with a minor in Women and Gender Studies

in 2008. While at Cal Poly, Leah served as the president of Agricultural Ambassadors

and Vice President of the Agricultural Engineering Society. She was heavily involved in

community agencies and was the scrumhalf for the women's rugby team. Leah obtained

her Engineer-In-Training certification during her fourth year at Cal Poly. Leah was

named the American Society of Agricultural and Biological Engineering National Student

Engineer of the Year in 2006 and in 2007, was one of five Society of Women Engineers

Outstanding Women in Engineering and Technology for the class of 2008, and earned

the College of Agriculture 2008 Outstanding Senior with Service to the Community.

In 2008, she accepted a graduate assistantship in the Agricultural and Biological

Engineering Department at the University of Florida. While working on her graduate

studies, Leah presented research at state and local conferences and became a member

of the American Society of Civil Engineers State Water Planning Committee.


105





PAGE 1

1 EVALUATION OF ACCURACY AND LONGEVITY OF EXPANDING DISK RAIN SENSORS By LEAH MEEKS 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 2010

PAGE 2

2 2010 Leah Meeks

PAGE 3

3 To my grandmothers, Geraldine Rosalee Meeks and Clara Lena Davis

PAGE 4

4 ACKNOWLEDGMENTS I thank my mother, Sharon Lea Meeks, for her unconditional love and encouragement, my fianc and best friend, James Anthony Hernandez, for his support and my aunt, the late Dr. Lynn Langer Meeks, for her inspiration I would like to thank the members of my graduate committee, Dr. Kati White Migliaccio and Dr. Thomas Obreza, for their assistance on my research. A big thank you goes to my advisor Dr. Michael D. Dukes for his guidance and the chance to experience a new side of irrigation. I would also like thank Stacia Davis, Bernard Cardenas Lailhacar, and Mary Shedd McCready for their help on this research project.

PAGE 5

5 TABLE OF CONTENTS page ACKNOWLEDGMENTS .............................................................................................................. 4 LIST OF TABLES ......................................................................................................................... 7 LIST OF FIGURES ....................................................................................................................... 9 LIST OF ABBREVIATIONS ...................................................................................................... 14 ABSTRACT ................................................................................................................................. 15 CHAPTER 1 INTRODUCTION................................................................................................................. 17 2 EXPANDING DISK RAIN SENSOR ACCURACY ......................................................... 31 Introduction .......................................................................................................................... 31 Materia ls and Methods ....................................................................................................... 34 Treatments .................................................................................................................... 34 Monitoring ..................................................................................................................... 35 Statistical Analys is ....................................................................................................... 36 Results and Discussion ...................................................................................................... 36 Climactic Conditions .................................................................................................... 36 Number o f Times in Open Switch Mode .................................................................. 37 Accuracy of Rain Sensors .......................................................................................... 40 Change in Accuracy of Rain Sensors over Time .................................................... 40 Summary and Conclusions ................................................................................................ 41 3 EXPANDING DISK RAIN SENSOR DRY OUT AND POTENTIAL IRRIGATION SAVINGS .............................................................................................................................. 59 Introduction .......................................................................................................................... 59 Materials and Methods ....................................................................................................... 61 Treatments .................................................................................................................... 61 Monitoring ..................................................................................................................... 62 Statistical Analysis ....................................................................................................... 64 Results and Discussion ...................................................................................................... 64 Climactic Conditions .................................................................................................... 64 Time in Open Switch Mode (Dry Out) ...................................................................... 65 Dry out Tracking ........................................................................................................... 67 Potential Irrigation Savings ......................................................................................... 68 Summary and Conclusions ................................................................................................ 69

PAGE 6

6 4 RELATIONSHIP BETWEEN EXPANDING DISK RAIN SE NSOR DISK LENGTH AND PERFORMANCE ..................................................................................... 83 Introduction .......................................................................................................................... 83 Materials and Methods ....................................................................................................... 84 Treatments .................................................................................................................... 85 Monitoring ..................................................................................................................... 85 Statistical Analysis ....................................................................................................... 86 Results and Discussion ...................................................................................................... 86 Length by Installation date and setting ..................................................................... 86 Disk Length and Traveling Distance ......................................................................... 87 Effect on Interruption Performance ........................................................................... 88 Summary and Conclusions ................................................................................................ 88 5 CONCLUSIONS AND F UTURE WORK .......................................................................... 96 Conclusions .......................................................................................................................... 96 Future Work ......................................................................................................................... 99 LIST OF REFERENCES ......................................................................................................... 100 BIOGRAPHICAL SKETCH ..................................................................................................... 105

PAGE 7

7 LIST OF TABLES Table page 2 1 Rain sensor treatment descr iption. .............................................................................. 43 2 2 Summary of functionality problems for treatments and replicates. ........................ 43 2 3 Average depth of rainfall before rain sen sors switched to Open Switch Mode. ... 44 2 4 Summary of changes in accuracy for change in rainfall required for Open Switch Mode. ................................................................................................................... 44 3 1 Treatment description. ................................................................................................... 71 3 2 Monthly irrigation depth to replace historical evapotranspiration values based on Dukes and Haman (2002a). Run times are based on an irrigation application rate of 38 mm/hr assuming system efficiency of 60% and considering effective rainfall. The Reduced UF IFAS irrigation schedule is 60% of the UF IFAS irrigation schedule. .................................................................... 71 3 4 Total potential water savings per treatment for all treatments compared with a 2 d/wk and a 1 d/wk irrigation schedule for the study period (Oct/Nov 2006 to 1 Dec 2009). ............................................................................................................... 72 3 5 Variation in total potential w ater savings per replicate for the WL and MC treatments compared with UF IFAS 2 d/wk irrigation recommendations. ............. 72 3 6 Variation in total potential water savings per replicate for the Hunter, Irritrol, and Toro treatments compared with UF IFAS 2 d/wk irrigation recommendations. .......................................................................................................... 73 3 7 Variation in total potential water savings per replicate for the WL and MC treatments compared wi th UF IFAS 1 d/ wk irrigation recommendations. ............ 73 3 8 Variation in total potential water savings per replicate for the Hunter, Irritrol, and Toro treatments compared with UF IFAS 1 d/ wk irrig ation recommendations. .......................................................................................................... 73 4 1 Description of rain sensors details for each treatment. ............................................ 90 4 2 Average disk length for each treatm ent at two intervals: initial and final (276 days of installation). ....................................................................................................... 90 4 3 Average disk length for the treatments installed 13 February 2009 at three intervals: initial, 81 days of installatio n, and final (276 days of installation). ......... 91

PAGE 8

8 4 4 Disk length for replicates of treatments installed 25 March 2005 at three intervals: initial (February), 81 days of installation (May), and fina l (November, 276 days of installation). ......................................................................... 91 4 5 Disk length for replicates of treatments installed 13 February 2009 at three intervals: initial (February), 81 days of installation (May), and fi nal (November, 276 days of installation). ......................................................................... 91 4 6 Comparison of average length change and travel distance from closedswitch mode to open switch mode of treatments installed in 13 February 2009. The February travel distance was measured on a rain sensor before installation. ..... 92 4 7 Comparison of average length change of each treatment from 13 February 2009 to 16 November 2009 and the travel distance each treatment from closed switch mode to openswitch mode. Travel distance was measured at the end of the study. ...................................................................................................... 92

PAGE 9

9 LIST OF FIGURES Figure page 2 1 WL (model Wireless Rain Clik, Hunter Industries, Inc., San Marcos, CA) rain sensor. A) Expanding disks inside ventilation window, B) quick response expanding disks, C) Ventilation window adjustment knob, D) antenna. ................ 45 2 2 MC (model Mini Clik, Hunter Industries, Inc., San Marcos, CA) rain sensor. A) Rainfall threshold setting slots, B) expanding disks, C) dry out adjustment ring and vents. ................................................................................................................ 45 2 3 Irritrol (model RFS 1000, Irritrol Systems, Inc., Riverside, CA.) rain sensor. A) Rainfall threshold setting slots, B) dry out adjustment ring, C) antenna. .............. 46 2 4 Toro (model TWRS, Toro Company, Inc., Riverside, CA) rain sensor. A) Rainfall threshold setting slots, B) dry out vent, C) antenna. .................................. 46 2 5 Detail of expanding disk material and threshol d adjustment of Mini Clik (Hunter Industries, Inc.) rain sensor. A) Rainfall threshold setting slots, B) hygroscopic expanding disk material. ......................................................................... 47 2 6 Research site located at the University of Florida Agricultural and Biological Engineering facilities. Shown: weather station on left, WL, 3MC, 3MC, and 13MC treatments installed on left board, and Hunter, Irritrol, and Toro on right board. ...................................................................................................................... 47 2 7 Installed Hunter Wireless RainClik, three on left and one on right, and Mini Clik rain sensors with four wireless receivers for WL and data logger. ................. 48 2 8 Installed Hunter, Irritrol, and Toro rain sensors, left to right, with wireless receivers (Irritrol on left and Toro on right), and data logger. .................................. 48 2 9 Relationship between manual rain gauge and weather station tipping bucket rain gauge with the calibration factor applied to the tipping bucket data with more than 15 mm of rainfall. ......................................................................................... 49 2 10 Relationship of rain events greater than 15mm between manual rain gauge and weather station tipping bucket rain gauge with the calibration factor applied to the tipping bucket data. ............................................................................... 49 2 11 Relationship of rain events less than 15 mm between manual rain gau ge and weather station tipping bucket rain gauge without the calibration factor applied to the tipping bucket data. ............................................................................... 50 2 12 Comparison of monthly and cumulative rainfall during the study period for WL and MC treatments and average historical rainfall for north central Florida. ........ 50

PAGE 10

10 2 13 Comparison of monthly and cumulative rainfall during the study period for Hunter, Irritrol, and To ro treatments and average historical rainfall for north central Florida. ................................................................................................................ 51 2 14 Cumulative and daily rainfall during the WL and MC treatments study period with the rainfall setting and the respective theoretical number of times each should have gone into OSM. ........................................................................................ 52 2 15 Cumulative and daily rainfall during the Hunter, Irritrol, and Toro treatments study period with the rainfall setti ng and the respective theoretical number of times each should have gone into OSM. .................................................................... 53 2 16 Cumulative number of times into OSM for WL and MC treatments. Data from 28 January to 9 June 2008 are not included due to all WL replicates not functioning. Erratic replicates within treatments are not included after their respective improper functioning dates (WLB 21 September 2007 and 13MC C 8 July 2008). Numbers with different letters indicate a statis tical difference using Tukey Kramer adjusted p values of p<0.05. ................................................... 54 2 17 Cumulative number of times the WL replicates went into OSM. WL stopped functioning on 21 September 2007. ............................................................................ 54 2 18 Cumulative number of times the 3MC replicates went into OSM. The Theoretical value is the number of times the replicates should have gone into OSM based on rainfall. .................................................................................................. 55 2 19 Cumulative number of times the 6MC replicates went into OSM. The Theoretical value is the number of times the replicates should have gone into OSM based on rainfall. .................................................................................................. 55 2 20 Cumulative number of times the 13MC replicates went into OSM. The Theoretical value is the number of times the replicates should have gone into OSM based on rainfall. .................................................................................................. 56 2 21 Cumulative number of times into OSM for Hunter, Irritrol, and Toro treatments. The Theoretical value is the number of times the replicates should have gone into OSM based on rainfall. Numbers with different letters indicate a statistical difference us ing Tukey Kramer adjusted p values of p<0.05. ............................................................................................................................. 56 2 22 Cumulative number of times Hunter into OSM. The Theoretical value is the number of times the replicates should have gone into OSM based on rainfall. ... 57 2 23 Cumulative number of times Irritrol went into OSM. The Theoretical value is the number of times the replicates should have gone into OSM based on rainfall. .............................................................................................................................. 57

PAGE 11

11 2 24 Cumulative number of times Toro went into OSM. The Theoretical value is the number of times the replicates should have gone into OSM based on rainfall. .............................................................................................................................. 58 2 25 Accuracy of each treatment with a set point over the study period with an average (solid line) and 95% confidence bands (dashed lines). ............................ 58 3 1 Histogram and cumulative f requency of dry out time (time from the end of the rain event until the sensors returns to closedswitch mode) for the WL treatment average. ......................................................................................................... 74 3 2 Histogram and cumulative frequency of dry out time (time from the end of the rain event until the sensors returns to closedswitch mode) for the 3MC treatment average. ......................................................................................................... 74 3 3 Histogram and cumulative frequency of dry out time (time from the end of the rain event until the sensors returns to closedswitch mode) for the 6MC treatment average. ......................................................................................................... 75 3 4 Histogram and cumulative frequency of dry out time (time from the end of the r ain event until the sensors returns to closedswitch mode) for the 13MC treatment average. ......................................................................................................... 75 3 5 Histogram and cumulative frequency of dry out time (time from the end of the rain event until the sensors returns to closedswitch mode) for the Hunter treatment average. ......................................................................................................... 76 3 6 Histogram and cumulative frequency of dry out time (time from the end of the rain event until the sensors re turns to closed switch mode) for the Irritrol treatment average. ......................................................................................................... 76 3 7 Histogram and cumulative frequency of dry out time (time from the end of the rain event until the sensors returns to clo sed switch mode) for the Toro treatment average. ......................................................................................................... 77 3 8 Histogram and cumulative frequency of dry out time (time from the end of the rain event until the sensors returns to closedswitch mode) for the Hunter treatment average with the dry out vents fully open (8 November 2008 to 2 July 2009). ....................................................................................................................... 77 3 9 Histogram and cumulative frequency of dry out time (time from the end of the rain e vent until the sensors returns to closedswitch mode) for the Hunter treatment average with the dry out vents fully closed (2 July 2009 to 31 December 2009). ............................................................................................................ 78 3 10 Histogram and cumulative f requency of dry out time (time from the end of the rain event until the sensors returns to closedswitch mode) for the Irritrol

PAGE 12

12 treatment average with the dry out vents fully open (8 November 2008 to 2 July 2009). ....................................................................................................................... 78 3 11 Histogram and cumulative frequency of dry out time (time from the end of the rain event until the sensors returns to closedswitch mode) for the Irritrol treatment average with the dry out vents fully closed (2 July 2009 to 31 December 2009). ............................................................................................................ 79 3 12 Histogram and cumulative frequency of dry out time (time from the end of the rain event until the sensors returns to closedswitch mode) for the one day water delay se tting for four the Toro replicates. ........................................................ 79 3 13 Histogram and cumulative frequency of dry out time (time from the end of the rain event until the sensors returns to closedswitch mode) for the thr ee day water delay setting for four the Toro replicates. ........................................................ 8 0 3 14 Dry out tracking of average disk length for each treatment for the natural rain event on 10 July 2009. .................................................................................................. 80 3 15 Dry out tracking of average disk length for each treatment for the manual rain event on 18 September 2009. ...................................................................................... 81 3 16 Dry out tracking of average disk length for each treatment and temperature for the rain event on 10 July 2009. .............................................................................. 81 3 17 Dry out tracking of average disk length for each treatment and solar radiation for the rain event on 10 July 2009. .............................................................................. 82 3 18 Dry out tracking of average disk length for each treatment and relative humidity for the rain event on 10 July 2009. .............................................................. 82 4 1 Mini Clik (Hunter Industries, Inc.) rain sensor expanding disks installed in March 2005 (left) and February 2009 (right) set at 13 mm measured in August 2009 (1600 and 179 days of installation, respectively). ............................. 92 4 2 Mini Clik (Hunter Industries, Inc.) rain sensors expanding disks installed in 2005 with settings (left to right) of 3 mm, 6 mm, and 13 mm and lengths 19.2, 19.8, and 20.1 mm, respectively, after 1,600 days of instal lation. ......................... 93 4 3 Cumulative and daily rainfall during the study period with number of rainfall events greater than the different rain sensor rainfall settings. ................................ 94 4 4 Average hygroscopic disk length for Mini Clik rain sensors installed in 2005 (MC) and 2009 (R) over installation time. The 2009 rain sensors were newly installed on day zero. ..................................................................................................... 95

PAGE 13

13 4 5 Accuracy for each setting during the first 300 days of rain sensor installation compared to change in disk length. Accuracy data are the average amount of rainfall for openswitch mode for a given rainfall event for a treatment. ................ 95

PAGE 14

14 LIST OF ABBREVIATION S Avg Average CSM Closed Switch Mode CV Coefficient of Variance d/wk Day per week or days per week ET Evapotranspiration NOAA National Oceanic and Atmospheric Administraion OSM Open Switch Mode RS R ain sensor SMS Soil moisture sensor UF IFAS University of Flroida Institute of Food and Agriculutal Sciences USCB United State Census Bureau

PAGE 15

15 Abstract of Thesis Presented to the Graduate School of the University of Florida in Partial F ulfillment of the Requirements for the Degree of Master of Engineering EVALUATION OF ACCURACY AND LONGEVITY OF EXPANDING DISK RAIN SENSORS By Leah Meeks August 2010 Chair: Michael D. Dukes Major: Agricultural and Biological Engineering Rain sensors are devices that connect to automatic irrigation systems to interrupt scheduled irrigations with sufficient rainfall. The goal of this research was to evaluate the performance of expandingdisk rain sensors. The primary objectives of this study were to A) eva luate rain sensor accuracy with time with respect to the selected rainfall setting, B) evaluate the amount of time rain sensors remained in interruption mode (open switch mode) after a rainfall event, C) quantify potential irrigation savings for different rainfall settings compared with a timebased schedule, and D) determine if the hygroscopic disks in the rain sensors change length with time. Ten treatments were established at the University of Florida Agricultural and Biological Engineering Department c ampus turfgrass plots, Gainesville, Florida. M i ni Clik rain sensors with rainfall settings of 3, 6, and 13 mm (3MC, 6MC, and 13MC) and Wireless Rain Clik (WL) rain sensors had four replicates for each treatment Treatments Hunter, Irritrol, and Toro had r ainfall settings of 6 mm with eight replicates each. Three other Mini Clik rain sensor treatments (3R, 6R, and 13R had rainfall settings of 3, 6, and 13 mm, respectively) each had three replicates

PAGE 16

16 This experiment was carried out during a relatively dry period with rainfall on 28% of the days and 15% less rainfall than average. WL, 3MC, 6MC, 13MC, Hunter, Irritrol, and Toro required 3.2, 1.9, 1.6, 6.6, 3.8, 4.3, and 5.8 mm for openswitch mode, respectively. Accuracy ranged from 27% to 97% The rain senso r accuracy had percentile point change from 36% to 59% with time where a negative value indicated a decrease in accuracy. Dry out is the amount of time a rain sensor stays in openswitch mode. R ain sensors dried out within 24 hours 79% of the time. Chan ging the dry out vent settings from fully open to fully had no effect on potential irrigation savings. Dry out occurred with decreasing relative humidity and increasing temperature and solar radiation. The hygroscopic disks in expandingdisk rain sensors i ncrease d in length after continuous rainfall exposure. R ain sensors with higher rainfall settings had the most increase The disk length change did not influence accuracy. The potential water savings for a 2 d/wk and 1 d/ w k irrigation schedule 13MC w ere 14% and 13% and the average for all other treatments was 24% and 21%, respectively. Potential irrigation savings should be considered in relation to the accuracy of rain sensors. Rainfall settings of 3, 6, and 13 mm are adequate for rain sensors in central Florida. If the rainfall setting needs to be changed after more than 3 months of use, a new rain sensor or new expanding disks be installed. For the best accuracy, Hunter Mini Clik rain sensors should be replaced after 1 year while Irritrol RSF 1000 and T oro TWRS rain sensors do not need to be replaced for at least 3 years. Rain sensors could increase water savings to homeowners and have environmental benefits but should not be used in applications requiring high accuracy

PAGE 17

17 CHAPTER 1 INTRODUCTION Introduct ion to Water in Florida Florida has an increasing need for water conservation measures Between 1950 and 2000, the population of Florida increased 475% and total public sup ply withdrawals increased 1,330% (Marella, 2004). In 2000, Florida was the largest u ser of groundwater east of the Mississippi River (Hutson et al., 2004). Water withdrawals for public supply in Florida in 2000 totaled 9.2 million cu bic meters per day, of which 90% was obtained from groundwater and 10% from surface water (Marella, 2004). Florida, Texas, Nebraska, Arkansas, and California account for more than half of the fresh groundwater use nationwide (Hutson et al., 2004). The f ive categories of f actors affecting water demands are population, climate, socioeconomic conditions, water pri cing, water conservation, and alternative supply sources (Marella, 1992) Florida receives an average of 1350 mm of rainfall per year (National Oceanic and Atmospheric Administration, 2003). Even with Floridas significant rainfall, the combination of rel atively well drained soils and dry periods mean that irrigation is required to maintain landscape quality. Fifty four percent of the freshwater withdrawn in 2000 was between February and June (Marella, 2004). Non municipal irrigation withdrawals in 2000 were greatest in February through June during drier conditions and the lowest in July through September when summer rain occurred (Marella, 2004). These withdrawals coincide with an increase in public supply use (Marella, 1992). The biggest stresses on water supply from agricultural and municipal sectors occur during the same time of year making water conservation even more critical.

PAGE 18

18 Residential Irrigation Practices in Florida Between 1970 and 2000, total freshwater withdrawals for pub lic use increased by 176% (Marella, 2004). Florida ranks fourth in overall state population with an estimated 2009 population of 18.5 million (United States Census Bureau [USCB], 2009). Residential i rrigation has been reported to account for 64% of residential water use (Haley et al ., 2007). The volume of water required for residential irrigation continues to increase with the increasing Florida population and the years of less thanaverage rainfall. From 2000 to 2005, Florida had a net population gain of approximately 1000 people per day (USCB, 2009). Extreme dry conditions occurred between February and June of 2000 and the result was higher water demands from public supply primarily for lawn irrigation during these months. A study of water use in Pinellas County Florida found that the highest water use occurred in spring due to high evaporation and low precipitation (Dukes and Haley, 2009). Lawn irrigation in central and south Florida occurs throughout the entire year (Marella, 2004). Florida, which receives more rainfall than all states other than Louisiana, requires irrigation to meet the aesthetic demands of homeowners due to the wet /dry seasons and well drained soils. It is estimated that 70% of single family homes in southwest Florida have automatic irrigation systems (Tam pa Bay Water, 2005). In a national study, homes that only hand watered used 33% less water than those with in ground systems (Mayer, et al., 1999). Inground systems generally run off of automatic timers instead of homeowners turning on the irrigation syst em. Automatic timer controls on irrigation systems in Florida have been reported to lead to a 47% increase in water use (Mayer, et al., 1999). Automatic systems result in more irrigation than manually controlled

PAGE 19

19 systems because people tend to set and forg et and do not take climatic conditions into account. Research in Florida found that homeowners irrigate in the late fall and winter when turfgrass is dormant because it is inconvenient to change the settings of the timer or there is a misunderstanding of the actual amount of water that should be applied during the year (Haley et al., 2007). Rain Sensors Rain sensors are devices designed to interrupt the cycle of an automatic irrigation system controller when a specific amount of rainfall has occurred (Duke s and Haman, 2002b). The rain sensor or its receiver is wired into an automatic irrigation controller. When rain beyond a threshold has fallen, the rain sensor will interrupt the irrigation controller circuit to potentially bypass an irrigation event depen ding on the irrigation schedule. Evaporation removes the water from the rain sensor so that irrigation will be allowed. The water savings potential, simple design, reliability, low cost, and ease of installation have made them popular (Dewey, 2003). Until the addition of soil moisture sensors in recent years, rain sensors were the only technology available commercially for residential irrigation reduction. Rain sensors have the potential to improve irrigation efficiency, reduce wear on the irrigation system and reduce runoff and deep percolation (Dukes and Cardenas Lailhacar, 2007). States and municipalities throughout the country have mandated the use of rain sensors to conserve water. It has been estimated that half of singlefamily homes in Florida have in ground irrigation systems with automatic timers, of which 25 % report having rain sensor shutoff devices (Whitcomb, 2005).

PAGE 20

20 Types Several types of rain sensors are on the market. T he sensors can be adjusted to interrupt at different depths of rainfall, generally between 3 and 25 mm. One type collects the rain water in a cup and interrupts the irrigation based on a preset weight of water. A disadvantage of the water weight devices is that debris can get into the collection cup and cause the system to int errupt irrigation without sufficient rainfall. Another type has a set of electrodes that detect the water level in a small collection dish (Dukes and Haman, 2002b ). Debris is also a problem with the electrical conductivity devices. The Rain Check (Rain Bird Corporation, Glendora, CA) is a rain sensor that measures the amount of rainfall with two electrodes in a collection cup. The stainless steel probes can be adjusted to interrupt irrigat ion between 3 and 13 mm of rainfall The most commonly used rain sensors in Florida are expandingdisk rain sensors. Hygroscopic disks in the sensor expand proportionally to the amount of rainfall. The swelling and contracting of the disks opens and close s a switch. Expanding disk sensors require less maintenance and are cheaper than other sensors. There are different models of rain sensor s that can be used depending on the location characteristics Sensors can operate as normally closed (normally allow irrigation) or normally open (normally does not allow irrigation). Most systems run on normally closed rain sensors. For more versatility, e xpanding disk rain sensors are available in wired or wireless models. Some w ireless models allow for the sensor to be placed up to 300 feet from the irrigation controller (Hunter Industries 2005).

PAGE 21

21 Evaporation for Dry out Rain sensors rely on evaporation to allow irrigation. In the case of normally closed rain sensors, the switch becomes closed after the dry out period and the irrigation system circuit is complete. Dry out settings can be adjusted for most sensors. A longer dry out time has the potential to interrupt more scheduled irrigation cycle s by the irrigation controller. The dry out setting should be set so that it matches the drying rate of the sites soil (Dewey, 2003). Dry out is determined by weather conditions such as temperature, wind, solar radiation and relative humidity. Installation Proper installation is critical to achiev e water savings. In a study of single family homes in Florida, anecdotal evidence suggests that rain sensors are often improperly installed (Whitcomb, 2005). The sensors need to be exposed to normal rainfall (Dewey, 2003). Inappropriate installation locations include in the sp ray path of sprinklers, under a tree canopy, under leaky roof gutters, and in places easily vandalized. The effects of sun and shade on dry out should be considered when choosing sensor location. Unlike many other irrigation sensors, once it is properly in stalled and set the rain sensor settings do not have to be adjusted to achieve water savings (Dewey, 2003). Water Savings Water and cost savings vary among rain sensor model and setting. Substantial savings can be obtained during a year of average rainfal l in Florida (Dukes and Haman, 2002b). In a study evaluating rain sensors at different rainfall settings when compared with a treatment irrigating 2 days/week without a rain sensor, a 3 mm set point reduced irrigation 30% while a 25 mm set point reduced ir rigation only 3% (Cardenas Lailhacar

PAGE 22

22 and Dukes, 2008). Other variables affecting savings include rain frequency, whether or not the controller is left on for automatic operation, and the amount of water applied by the system per cycle (Dukes and Haman, 2002b). Marella ( 1992) suggests that r ain sensors are part of long term water conservation measures for reducing residential irrigation Previous Studies Involving Rain Sensors and Smart Controllers While research on rain sensors alone is very limited, there have been residential irrigation studies involving rain sensors. Many of these studies have been conducted in the Southeastern United States due to the relatively high amount of rainfall compared to the rest of the country. Rain Sensor A ccuracy T esting A study at the University of Florida Agricultural and Biological Engineering Department turfgrass plots in Gainesville, Florida by Cardenas Lailhacar and Dukes assessed the accuracy of rain sensors at different set points. The data for the rain sensors (Hun ter Industries, Inc., San Marcos, CA) were collected 25 March 2005 through 31 December 2005. The four treatments included three Mini Clik rain sensors with different set tings (3 mm, 13 mm, and 25 mm) and one Wireless Rain Clik. These rain s ensors were not connected to an irrigation system. The accuracies for the 3, 13, and 25 mm settings were 88%, 77%, and 98%, respectively. A longer study should also be conducted on rain sensors to evaluate the Hunter Mini Cliks 5 year warranty (Cardenas Lailhacar and Du kes, 2008).

PAGE 23

23 Potential Rain Sensor Irrigation S avings The rain sensor accuracy study also investigated potential irrigation savings with the addition of a rain sensor. The average depth of potential water savings for the wireless, 3 mm, 13 mm, and 25 mm sensors were 558, 337, 468, and 38 mm, respectively. This study indicated that irrigation water savings with rain sensors was dependent on the rain sensor settings (Cardenas Lailhacar and Dukes, 2008). This same conclusion was reached in a study in Citra, Florida with rain sensors set at 3 mm and 6 mm during a relatively dry period. The savings for the 3 mm and 6 mm set points were 25% and 17%, respectively, when compared with a time based schedule with no water conservation devices (McCready et al., 2009). Multiple studies involving soil moisture sensors (SMS) and evapotranspiration (ET) controllers conducted at the University of Florida have had a rain sensor included in a treatment for comparison. One study in Gainesville, Florida found that the addition of a rain sensor set to interrupt irrigation with 6 mm of rainfall to a time clock set to irrigation 2 days/week could reduce irrigation by 34% (Cardenas Lailhacar et al., 2008). In a study with ET controllers and rain sensors in southwestern Florida, an automatic irrigation system with a rain sensor conserved 21% of water compared with an irrigation controller operating on a timebased schedule (Davis et al., 2009). Haley and Dukes (2007) conducted a study that included rain sensors and educational mater ial about irrigation controller scheduling. The addition of a rain sensor to a controller conserved 19%; the combination of a rain sensor and educational materials increased savings 58%. These studies give insight into the effects of including a rain senso r on an active automatic irrigation system.

PAGE 24

24 State Statu t es State governments are adding rain sensor devices to water conservation statute measures. The fact that more states are considering or including rain sensors in water conservation statutes indicates that more research into the effectiveness of rain sensors is necessary. Connecticut. The Connecticut statute regarding rain sensors applies to automatic lawn irrigation systems installed by state agencies or commercial enterprises. As of 1 October 2003, all installations must be equipped with a rain sensor that interrupts the irrigation cycle after adequate rainfall occurs. The statute also allows municipalities to pass ordinances requiring rain sensors on irrigation systems installed after 1 October 2003 within their respective jurisdictions. (Connecticut Statutes, Section 29265) Florida. Florida Statue 373.62 required that all automatic irrigation systems installed after 1 May 1991 must have a maintained rain sensor or switch that overr i des the irrigat ion cycle after adequate rainfall. Recently, Florida passed a new water conservation statute effective 1 July 2009. The new bill requires all new automatic irrigation system installations be equipped with rain sensors and old installations, predating 1 May 1991, be retrofitted with rain sensors. Licensed contractors who install the irrigation systems must properly install or check for proper installation. A licensed contractor who does not comply can be fined $50 for a first offense, $100 for a second o ffense, and $250 for a third or subsequent offense. Funds from the penalties will be used for water conservation programs by local government. (Florida Senate Bill 494)

PAGE 25

25 Massachusetts. In January 2009, a bill was introduced in Massachusetts requiring an int erruption device on newly installed or renovated outdoor landscapes. The device should override irrigation during periods of sufficient moisture. All new irrigation systems must be inspected every 3 years by a certified irrigation contractor, a certified landscape irrigation auditor, or a certified irrigation designer. The bill would not apply to golf courses. It was not passed as of 30 June 2009 and was pocket vetoed by Governor Patrick (Moriarty, 2009). (Massachusetts Senate 186th General Court) Minneso ta Minnesotas rain sensor statute effective 1 July 2003 requires that all automatic irrigation systems have technology that interrupts operation in the event of sufficient rainfall. The device must be adjustable by the irrigation system user or instal ler. (Minnesota Statutes, Chapter 44F No. 335) New Jersey. The New Jersey rain sensor statute requires that all automatic irrigation systems installed after 8 December 2008 have a rain sensor. The device will override the irrigation cycle with adequate r ainfall. (New Jersey Statutes, 52:24D 123.13) Texas. The Texas ET controller statute applies to automatic irrigation systems owned by the state or political subdivisions of the state greater than 0.25 or 0.50 hectares ( 0.1 or 0.2 acres ) if using nonpotabl e water. New or existing irrigation systems must have an on site ET controller as of 1 September 2007. A remote ET controller can be used if the weather station is less than 5 miles away from the site and has a rain/freeze sensor. Both remote and on site s ystems must have an independent rain/freeze sensor. The statute encourages the passage of local ordinances on ET controllers. (Texas House Bill 2299)

PAGE 26

26 City and Area Ordinances There is a growing trend for local governments to require rain sensors. Many of these rain sensor requirements are included in irrigation restriction ordinances. Compared with state rain sensor statutes, the city and area ordinances are more likely to have fines for noncompliance. Metropolitan North Georgia Water Planning District, G eorgia. The Metropolitan North Georgia Water Planning District, Georgia, regulation applies all systems receiving water from the public water system, not including golf courses. After 1 January 2005, all automatic irrigation systems must be equipped with a rain sensor. Persons in violation by installing a system without a rain sensor can be fined up to $100 for each violation. (Georgia Metropolitan North Georgia Water Planning District, Water conservation action no. 4) Water Authority of Great Neck North, N ew York The Water Authority of Great Neck North, New York ordinance effective 15 April 1994 includes irrigation times, rain sensors, and soil moisture sensors. The ordinance applies to person s using water that is directly from the district or beneath the district if the source is underground. Users must irrigate no more than 3 days / week depending on the address and all irrigations must take place between 4:00 p.m. and 10:00 a.m. from 15 April to 1 November. The rain sensor used must be able to detect a min imum of 1/8 inch (3 mm) of rain. Rain sensors must be set to interrupt irrigation with a inch (6 mm) or less of rainfall. (Water Authority of Great Neck North New York) Derby, Kansas. The ordinance for Derby, Kansas applies to all persons owning property with an automatic irrigation system whether or not the irrigation water is

PAGE 27

27 supplied by the public water system. All automatic irrigation systems installed or substantially replaced after 24 May 2008 must have rain sensors. After 1 July 2009, all automa tic irrigation systems must be either installed with rain sensors or retrofitted with rain sensors. Rain sensors must be set to interrupt irrigation with at least inch (13 mm) of rainfall. City code enforcement officers can inspect systems with notice if there is doubt of compliance. Fines to property owners can range from $25 to $500. Cessation of public water supply service may be an additional penalty if a property owner fails or refuses to install and maintain a rain sensor. (Derby, Kansas Ordinance N o. 1932) Cary, North Carolina. The Cary, North Carolina rain sensor ordinance defines an irrigation system and rain sensors for residents. This ordinance applies to all systems that receive water from the town of Cary. Effective 14 August 1997, new automat ic irrigation systems must be equipped with rain sensors. Existing systems must be retrofitted with a rain sensor on or before 1 May 1998. Rain sensors must be set to interrupt the irrigation cycle after inch (6 mm) of rainfall and be located in an area of full exposure. A rain sensor set to bypass irrigation at a setting greater than inch (6 mm) is considered in non compliance. On the second notice of noncompliance, the property owner will be fined $100 each subsequent day not in compliance (i.e. $100 first day, $200 second day, etc). Termination of service can be a consequence of continued noncompliance. (Cary, North Carolina, Ordinance section 3684) Harrisburg, South Dakota. The Harrisburg, South Dakota ordinance requires rain sensors on all autom atic irrigation systems installed after the effective day of 4 April 2006 that receive water from the city pub lic supply. Rain sensors must be set to

PAGE 28

28 interrupt the irrigation cycle after inch (6 mm) of rainfall A rain sensor set to bypass irrigation at a setting greater than inch (6 mm) is considered in noncompliance. (Harrisburg, South Dakota Ordinances, Ordinance 200604, Chapter 8.01) Arlington, Texas. The Arlington, Texas Lawn and Landscape Irrigation Conservation ordinance calls for irrigation tim e restrictions and rain/freeze sensor. Automatic irrigation systems cannot operate between 10:00 a.m. and 6:00 p.m. from 1 June to 30 September unless during periods of grass establishment, dust control, maintenance, repair, or testing. The rain/freeze se nsor requirement does not apply to a single family residential or duplex property or an individually metered townhome or condominium unity. The city council created a list of approved rain/freeze sensors to be used. All new automatic irrigation systems ins talled after 4 March 2005 within the city limits must be equipped with a rain/freeze sensor. As of 4 March 2007, all existing systems must be retrofitted with a rain/freeze sensor. Those in noncompliance will by guilty of a misdemeanor with a possible fine of up to $500 for each violation. (Arlington, Texas Lawn and landscape irrigation conse rvation ordinance section 4.27) Colleyville, Texas The Water Conservation Ordinance of Colleyville, Texas was created to conserve water by preventing automatic irrigation systems from running during wet periods. All new automatic irrigation systems installed after 31 August 2006 must be equipped with a rain sensor. As of 31 August 2008, all existing systems must be retrofitted with a rain sensor. Termination of service can be a consequence of continued noncompliance. (Colleyville, Texas, Water conservation ordinance 061579)

PAGE 29

29 Dallas, Texas. The purpose of the Dallas, Texas landscape irrigation ordinance is to promote irrigation practices that prevent waste, conserve water resources for the most beneficial and vital use, and protect the public health. Automatic irrigation systems cannot operate between 10:00 a.m. and 6:00 p.m. from 1 April to 31 October. Irrigation restrictions also include not significantly irrigating on impervious surfaces, not irrigating with a broken or missing sprinkler head, and not properly maintaining the system. Effective 1 January 2002, new automatic irrigation systems must be equipped with rain/freeze sensors. Existing systems must be retrofitted with a rain/freeze sensor on or before 1 January 2005. Violators are those who own, lease, or manage property with a system not equipped with a rain/freeze sensor or operate and/or permit operation of an irrigation system not in compliance with the sensor or irrigation time restrictions. Variances or exceptions can be made in cases of extreme hardship or when approved by the city attorney. All exceptions to the ordinance must not adversely affect the health, safety, or welfare of other people and must not cause an immediate negative impact on the citys water supply. Violators can be fined up to $250 for the first offense, doubled for the second offense, and continued for each subsequent offense within a 12month period. The total fine for a 12month period cannot exceed $2,000. ( Dallas, Texas Ordinances, S ection 4921.1) Lucas, Texas. The Lucas, Texas Code of Ordinances includes requirements for rain/freeze sensors. This ordinance applies to automatic irrigation systems in the city T he party responsible is the owner, leasee, occupier, or manager of the property on which the irrigation system is located. All new automatic irrigation systems installed after 1 January 2006 must be equipped with a rain/freeze sensor. As of 1 July 2007, all

PAGE 30

30 existing systems must be retrofitted with a rain/freeze sensor. Fines to property owners can be up to $500. (Lucas, Texas Ordinances, Article 8, Irrigation sy stem regulations, Section 3220) San Antonio, Texas. San Antonio, Texas has a permanent year round water conservation ordinance to reduce per capita use of water. Th e ordinance defines many term s that play into conservation such as automatic irrigation controller, impervious surface, ra in sensor, recycled water, and water flow restrictor. Among the many regulations such as ice machines and xeriscapes on new home developments is a rain sensor regulation. Effective 1 January 2006, all automatic irrigation controllers must have rain sensors installed and maintained (San Antonio Water System s Ordinance 100332 34.274.2) Study Objectives The goal of this research was to determine the performance of three brands of expandingdisk rain sensors. The first objective was to evaluate the number of times in open switch mode and the accuracy over time with respect to the selected set point by comparing when the rain sensors interrupted irrigation with rainfall recorded from an on site weather station (Chapter 2). The second objective was to evaluate t he dry out time (amount of time in openswitch mode) and potential irrigation savings by comparing rain sensor irrigation interruptions to a University of Florida Institute of Food and Agriculture Science (UF IFAS) recommended irrigation schedules (Chapter 3). The third objective was to determine if the length of the hygroscopic disks in expandingdisk rain sensors change size based on the amount of time installed and the rainfall setting (Chapter 4).

PAGE 31

31 CHAPTER 2 EXPANDING DISK RAIN SENSOR ACC URACY Introduction Although Florida ranks second in annual state precipitation, irrigation is required to meet the aesthetic landscape requirements. Between 1950 and 2000, the population of Florida increased five times and total public supply withdrawals increased 13 t imes (Marella, 2004). In 2005, Florida ranked first in single family home construction with 209,162 homes built (United States Census Bureau [USCB], 2007) and it has been estimated that 70% of single family homes have automatic irrigation systems (Tampa Ba y Water, 2005). Water conservation measures are needed to reduce water volumes applied. Haley et al. (2007) reported that residential irrigation accounted for 64% of total residential water volumes in central Florida. One water conservation measure is addi ng a rain sensor (RS) to an automatic irrigation system. It is thought that automatic systems irrigate more than manual irrigation because of the set and forget mentality of an irrigation timer with no consideration of climatic conditions. RSs are devices designed to interrupt the cycle of an automatic irrigation system controller when a specific amount of rainfall has occurred (Dukes and Haman, 2002b). Unlike many other irrigation sensors, once it is properly installed and set, the RS settings do not have to be adjusted to achie ve water savings (Dewey, 2003). Water and cost savings vary among rain sensor models and settings. Variables affecting savings include rain frequency, whether or not the controller is left on for automatic operation, and the amount o f water applied by the system per cycle (Dukes and Haman, 2002b).

PAGE 32

32 States and municipalities throughout the country have mandated the use of rain sensors to conserve water. States that have had bills introduced requiring RSs on certain landscape irrigation applications include Connecticut, Florida, Massachusetts, Minnesota, New Jersey, and Texas (see Chapter 1). Florida, Minnesota, and New Jersey require homeowners to install and maintain a RS on all automatic irrigation systems. S everal types of RSs are on the market. All of the sensors can be adjusted to interrupt at different depths of rainfall, generally between 3 and 25 mm. Research by Cardenas Lailhacar and Dukes (2008) determined that the 25 mm setting is too high in Florida to net practical savings. O ne type of RS collects the rain water in a cup and interrupts the irrigation based on a preset weight of water. Another type has a set of electrodes that detect the water level in a small collection dish that measures the amount of rainfall with two electr odes in a collection cup (Dukes and Haman, 2002b). A disadvantage of both of these water weight devices is that debris can get into the collection cup and cause the system to interrupt irrigation without sufficient rainfall. The most commonly used RSs are expanding disk rain sensors (Figures 21 through 24). Hygroscopic disks in the sensor expand proportionally to the amount of rainfall (Figure 2 5). The swelling of the disks typically causes a switch to interrupt the signal to open an irrigation valve. E xpanding disk sensors require less maintenance and are less expensive than other sensors. There are different models of RSs that can be used depending on the application. Sensors can operate as normally closed (normally allow irrigation) or normally ope n (normally does not allow irrigation). Most systems run on normally closed rain sensors.

PAGE 33

33 For a normally closed RS, the RS is in closed switch mode until sufficient rainfall changes it to open switch mode. Open switch mode means that the irrigation circ uit is incomplete such that a scheduled irrigation event will be interrupted. Research by Cardenas Lailhacar and Dukes (2008) investigated the performance of expanding disk rain sensors in a 2005 study at the University of Florida. The four treatments wer e three Hunter Mini Clik (Hunter Industries, Inc, San Marcos, CA) rain sensors with different set tings (3 mm, 13 mm, and 25 mm) and one Hunter Wireless Rain Clik. As would be expected, t he lower set points corresponded to a greater number of times in open switch mode occurrences than a higher setting. All treatments had replicate variability in the number of times in open switch mode and depth of rainfall required for open switch mode. The average depth of rainfall triggering the W ireless 3 mm, 13 mm, and 25 mm settings was 1.4, 3.4, 10.0, and 24.5 mm with resulting accuracy of 88% 77 % and 98% for 3 mm, 13 mm, and 25 mm, respectively This research provides a base line for rain sensor research. A longer study over a variety of precipitation conditions wo uld offer more insights into rain sensor performance. The average potential water savings in depth of irrigation for the Wireless Rain Clik and MiniCliks with 3 mm, 13 mm, and 25 mm settings was 588, 337, 468, and 38 mm, respectively. For the Mini Clik treatments, the lower setting showed more water saving potential. This study concluded that a setting of 25 mm was too high and not applicable in north central Florida. The objective of this study was to evaluate the accuracy with time of three brands of expandingdisk rain sensors with respect to the selected set point by comparing when

PAGE 34

34 the rain sensors interrupted irrigation with rainfall recorded from an on site weather station. Materials and Methods This study was conducted at the University of Florid a Agricultural and Biological Engineering Department campus turfgrass plots, Gainesville, Florida. There were a total of 40 rain sensors installed at a height of 2 m (Figure 26). Treatments Seven treatments composed of different rain sensor brands and se t points were established (Table 21). The Wireless RainClik (WL) and Mini Clik (MC) rain sensors were from Hunter Industries, Inc., San Marcos, CA. The WL d id not have a rainfall setting and wa s designed to interrupt irrigation immediately after rain begins. The three MC treatments had rainfall settings of 3 mm, 6 mm, and 13 mm (3MC, 6MC, and 13MC) (Figure 2 7). Data collection for treatments WL, 3MC, 6MC, and 13MC was started 2 October 2006 and was completed 31 December 2009 (1,186 days). These four tre atments were installed on 25 March 2005 with four replications each. The 6MC treatment was originally set to 25 mm and was changed to 6 mm on 2 October 2006 since previous work indicated that minimal savings occurred at a 25 mm setting (Cardenas Lailhacar and Dukes, 2008). The three remaining treatments were installed at a later date with 6 mm settings for three brands. The brands and respective treatment codes were Hunter Industries, Inc., model Mini Clik (Hunter), Irritrol Systems, Inc., model RFS 1000, Riverside, CA (Irritrol), and Toro Company, Inc., model TWRS, Riverside, CA, (Toro) (Figure 28) with eight replicates for each treatment Data collection for Hunter, Irritrol, and Toro treatments was started on 8 November 2006 and was completed 31 December 2009 (1,150 days). Problems with rain sensor function are

PAGE 35

35 summarized in Table 22. The dry out vents for Hunter and Irritrol were fully open from installation until 2 July 2009 and were then changed to fully closed. The Toro rain sensor receivers were se t to 0.0 day dry out; the water delay feature was set to 1.0 day for four replicates and 3.0 days for the remainin g four replicates. Monitoring Each time a rain sensor changed mode between open switch mode (OSM) and closed switch mode (CSM), the date and time were recorded at a onesecond sampling interval using AM16/32 multiplexers (Campbell Scientific, Inc., Logan, UT) attached to a CR10X model data logger (Campbell Scientific, Inc., Logan, UT). An onsite automated weather station (Campbell Scientific, Logan, UT) located within 15 m of the experimental site recorded weather conditions using a CR10X model data logger (Campbell Scientific, Logan, UT). Data such as relative humidity, temperature (model HMP45C, Vaisala, Inc., Woburn MA), solar radiation (mode l LI200X, Li Cor, Inc., Lincoln, NE), and wind speed and direction (model WAS425, Vaisala, Inc., Sunnyvale, CA) were recorded at 15 minute intervals. Precipitation was measured by a tipping bucket rain gauge (model TE525WS, Texas Electronics, Inc., Dallas, TX) with a 1second sampling interval time stamp for each 0.25 mm of rain. A manual rain gauge located within 5 m of the rain sensors was used to verify the accuracy of the tipping bucket rain gauge measurements (Figure 29). The weather station tipping bucket rain gauge was calibrated using the Texas Electronics Calibration Kit (Texas Electronics, Inc., Dallas, TX). Calibration testing was conducted during November 2009. The first test indicated that the tipping bucket was out of calibration; it recorded 23.4 mm for every 25.4 mm that actually fell. After two calibration adjustments, the tipping bucket rain gauge was recording 24.9 mm for every 25.4 mm of rain that fell

PAGE 36

36 which is within the Texas Electronics acceptable range of 2% error (Texas Electronics, Inc. (a)). Measured rainfall was multiplied by and adjustment factor of 1.07 so that measured rainfall would be accurate and equal to actual rainfall. Since the tipping bucket had not been calibrated since 2003, a calculation adjustment was applied to rec orded rain events during the duration of the study. Based on a linear regression of the data with and without the calibration, the calibration was applied to rainfall events greater than 15 mm and not applied to rainfall events less than 15 mm (Figures 21 0 and 211). Rainfall event depths were collected and analyzed from the study period. Monthly rainfall from a 30 year historical period from 1970 to 2000 from the National Oceanic and Atmospheric Administration (NOAA) was used as a comparison to data coll ected during the study. Establishment of current and historical weather patterns will affect setting recommendations. Statistical Analysis SAS statistical software (SAS Institute, Inc., Cary, NC) was used for all statistical analysis A general mixed model with an auto regressive error structure was used to model the continuous responses (PROC MIXED) Tukey Kramer adjusted p values ( p<0.05) were used for pairwise comparisons of mean. Results and Discussion Climactic Conditions During the 1,186 days of the WL and MC study period and the 1,150 days of the Hunter Irritrol, and Toro study period, 28% of the days received rain. For the WL and MC experiment the cumulative rainfall was 3,551 mm, 14% less than the historical average of 4,121 mm (Figure 212). For the Hunter, Irritrol, and Toro experiment the

PAGE 37

37 cumulative rainfall was 3,410 mm, 16% less than the historical average of 4,055 mm (Figure 2 13). If the amount of rainfall during study would have been closer to historical values, the rain sensors would hav e gone into OSM more times. Number of Times in Open Switch Mode Figures 214 and 2 15 show the daily and cumulative rainfall during each period and the theoretical count for the number of OSM occurrences for each treatment. The theoretical number of OSM o ccurrences for the 3MC, 6MC, and 13MC treatments should was 192, 139, and 82, respectively. The percentage of rain events greater than 3, 6, and 13 mm during the WL and MC study period w as 57%, 42%, and 25% The theoretical number of OSM occurrences for th e Hunter Irritrol, and Toro treatments were 136 times and 41% of rain events being greater than 6 mm. Figure 216 shows the average cumulative number of OSM events for the WL and MC treatments. The average number of OSM events for WL, 3MC, 6MC, and 13MC w ere 146, 154, 160, and 109, respectively. The 13MC treatment had fewer OSM events than the other three treatments. The 6MC treatment went into OSM more times than expected since it was not statistically different from 3MC (p<0.05). This result could be due to the change of these RSs from a setting of 25 mm to 6 mm in October 2006. The possible effects of disk size change with time are discussed in Chapter 4. All treatments showed replicate variability and not all replicates were functioning during the enti re experiment period. One WL replicate stopped functioning on 21 September 2007, after 910 days of operation. This replicate was not considered in any analysis for means comparisons. The remaining three WL replicates were not functioning 28 January to 9 Ju ne 2008 due to an electrical problem such as data logger or batteries in the receivers. One of the four 13MC replicates displayed some erratic

PAGE 38

38 behavior such as going into OSM without rainfall and not going into OSM with sufficient rainfall starting 8 July 2008, after 1201 days of continuous operation but remained functioning throughout the study. The cumulative number s of OSM events for the functioning WL replicates w ere 131, 145, and 162 (Figure 217). For 3MC with a theoretical OSM value of 192, the rep licates went into OSM 149, 162, 173, and 182 times (Figure 2 18). The 6MC replicates went into OSM 162, 173, 173, and 188 times, which were all more than the theoretical value of 139 (Figure 2 19). The theoretical OSM value for 13MC was 82 while the repli cates went into OSM 59, 111, 116, and 131 times (Figure 2 20). The coefficient of variance (CV) for WL, 3MC, 6MC, and 13MC were 11%, 8%, 6%, and 30%, respectively. One 13MC replicate increased the variability from 8% to 30% because it displayed somewhat erratic behavior but did continue functioning. The CV values for the WL, 3MC, 13MC, and 25MC after 282 days of installation w ere 3%, 28%, 24%, and 8%, respectively (Cardenas Lailhacar and Dukes, 2008). The variability of the 25MC rain sensors was 8% and aft er changing the setting to 6MC the variability was 6% indicating that changing the setting of these replicates did not influence the variability. The weighted average CV for WL, 3MC, 25/6MC, and 13MC for both studies was 8%, 12%, 7%, and 25%. The WL and 25/6MC rain sensors had the least amount of variability with each treatment. From the initial study to this study, the WL treatment became more variable, 13MC variability remained the same, and 3MC became more stable. Tipping bucket rain gauges have a range of accuracies with 0.5% to 4% variability (Omega (1995), Sutron Corporation, Spectrum Technologies, Inc and Texas Electronics (b)). The variability for rain sensors is

PAGE 39

39 relatively high when compared to tipping buckets, which are both for a measurement inst rument. Figure 221 shows the cumulative number of OSM events for the Hunter, Irritrol, and Toro during their study period and the theoretical value based on rainfall. The average number of OSM events for Hunter, Irritrol, and Toro were 144, 190, and 114 with a theoretical value of 136. Not all replicates were functioning during the entire experiment period. A Hunter replicate started showing erratic behavior by not responding to high amounts of rainfall and not reacting to manual triggering the same as th e other replicates; this replicate was not considered in evaluation analysis after 21 December 2007, corresponding to 408 days of operation. This replicate was not considered in any analysis for means comparisons. Toro had some issues with the right wirele ss receiver receiving OSM and CSM signals from the right rain sensor. Six of the eight replicates received the same information from one rain sensor from 20 April 2009 to 22 September 2009. Figures 222 to 224 show the theoretical number OSM occurrences for each treatment and variability within treatments. The number OSM occurrences for the seven functioning Hunter replicates were 106, 135, 143, 144, 144, 144, and 151 times (Figure 2 22). The number of OSM occurrences Irritrol replicates w ere 175, 182, 182, 190, 191, 196, 196, and 204 (Figure 223); they all went into OSM more than the theoretical value due to the number of OSM occurrences without rain. The number of OSM occurrences for the Toro treatment were 83, 87, 87 92, 96, 96, 104, and 114 with data from 20 April 2009 to 22 September 2009 excluded due to receiver problems (Figure 224). The CV values for Hunter, Irritrol and Toro were 11%, 6%, and 10%, respectively.

PAGE 40

40 Accuracy of Rain Sensors The accuracy of the instrument is its ability to indicate a n exact true value (Figliola and Beasley, 2000). Accuracy is related to the difference between the true value and the indicated value of a measurement called the absolute error accuracy (A) is calculated by: 100 1 value true A (3 1) Table 2 3 shows the average depth of rainfall before the RSs went into OSM. Because of acceptable error in the tipping bucket rain gauge, there was a plus or minus 2% in the cal culated rain sensor accuracy. The WL does not have a set point, so accuracy cannot be determined. The WL had an average rainfall depth of 3.2 mm required to go into OSM. The 3MC, 6MC, and 13MC went into OSM after 1.9, 1.6, and 6.6 mm with accuracies of 64% 27%, and 51%, respectively. The low accuracy from 6MC could be attributed to the setting change on 6 October 2006 on the treatment from 25 mm to 6 mm because of disk size changes discussed in Chapter 4. The Hunter, Irritrol, and Toro went into OSM after 3.8, 4.3, and 5.8 mm with accuracies of 64%, 71%, and 97%, respectively. The 3MC, 6MC, 13MC, Hunter, and Irritrol treatments required a different depth of water for OSM than their respective rainfall setting. The CV values for depth of rainfall required fo r OSM for WL, 3MC, 6MC, 13MC, Hunter, Irritrol, and Toro were 67%, 51%, 51%, 51%, 37%, 36%, and 33%, respectively. Change in Accuracy of Rain Sensors over Time The accuracy of the treatments varied over the study period as summarized in Table 2 4. The 13MC treatment became less accurate with time while 6MC, Hunter, and Irritrol showed an increase in accuracy and 3MC and Toro had no change in

PAGE 41

41 accuracy. The amount of rainfall required before OSM for WL, 3MC, 6MC, 13MC, Hunter, Irritrol, and Toro at the beginning and end of the study was 2.6 and 4.0 mm, 1.9 and 1.9 mm, 1.3 and 2.1 mm, 8.3 and 5.3 mm, 3.3 and 4.5 mm, 3.6 and 5.1, and 5.7 and 6.1 mm, respectively. Figure 225 shows the progression in the change of rainfall required for OSM for all treatments w ith set points over the study period. The percentile point change in accuracy during the study period for 3MC, 6MC, 13MC, Hunter, Irritrol, and Toro was 1%, 59%, 36%, 36%, 42%, 7%, respectively where a negative value indicated a decrease in accuracy. Ot her than 3MC, there was a trending relationship between rainfall depth for OSM CV value and accuracy. A treatment with low accuracy also had a high CV value. The initial 282day study by Cardenas Lailhacar and Dukes (2008) found that the WL required 1.4 mm of rainfall for OSM. WL rainfall requirement increased 2.5 times after an additional 904 days of installation. The 3MC and 13MC went into OSM after 3.4 and 10.0 mm with accuracies of 88% and 77%, respectively. The 3MC and 13MC were more accurate during th eir first 282 days of installation. When newly installed, the rain sensors had an average error of 15% (Cardenas Lailhacar and Dukes, 2008). The weighted average error of the sensors at the end of this study was 46%, a tripling in the error of the sensors over time. Summary and Conclusions This experiment occurred during a relatively dry period with rainfall on 28% of the days. The percentages of rain events greater than 3, 6, and 13 mm during the WL and MC study period were 57%, 42%, and 25% The Hunter, I rritrol, and Toro treatments theoretically had 136 opportunities for OSM with 41% of rain events being greater than 6 mm.

PAGE 42

42 Most treatments showed variability and erratic behavior of some replicates during the study. The coefficient of variance for depth of rainfall required for open switch mode varied between 33% and 67%. Some replicates showed erratic behavior such as not detecting rainfall events much higher than their setting, going into closedswitch mode in the middle of a rain event and returning to openswitch mode a few minutes later with 0.5 mm of rainfall, or going into open switch mode with little or no rainfall. Two of the replicates completely stopped functioning during the study period for unknown reasons. The accuracy of the rain sensors chan ged with time. The percentile point change in accuracy during the study period ranged from an increase of 59% to a decrease of 36%. The 13MC treatment accuracy decreased during the study while other treatments had small change in or improved accuracy. Cardenas Lailhacar and Dukes (2008) showed an average weighted accuracy of 85% in the first 282 days of installation while this study had an average weighted accuracy of 47% with the same sensors and 60% overall. There was no single trend for all rain sensors or all rainfall settings with respect to accuracy with time. For the best accuracy, there is evidence based on historical studies and results from this study that Hunter Mini Clik rain sensors should be replaced after 1 year of installation. The accuracy of a Hunter Mini Clik set to 3 mm stabilized after 2 years of installation. The higher rainfall setting corresponded to lower accuracy with the same brand of rain sensor. Irritrol RSF 1000 rain sensor accuracy increased as the study progressed. Toro TWRS rain sensors retained their relatively good accuracy during the 3 years of this study. Further research is needed to verify these results. Changing the setting of a rain sensor after it has been installed more than 3 months is

PAGE 43

43 not recommended (see Chapter 4 for more details). The change of the 6MC treatment from a 25 mm to a 6 mm setting reduced the accuracy of the same sensors from 98% to 27%. During the same time, the 3MC and 13MC treatments average weighted accuracy declined from 84% to 59% Overall, the rain sensors showed that they have high variability with time. However, due to low cost and low maintenance requirements, rain sensors can be a useful device for potential water savings. The variability, erratic behavior, and low accuracy of some repl icates showed that rain sensors should not be used in applications requiring high accuracy and precision. Table 2 1. Rain sensor treatment description. Treatment Model Replicates Set Point Installation Date Study Start Date WL 3MC 6MC 13MC Hunter Irritrol Toro Wireless Rain Clik X Mini ClikX Mini ClikX Mini ClikX Mini ClikX Irritrol RFS 1000Y Toro TWRS Z 4 a 4 4 4 8b 8 8 3 mm 6 mm 13 mm 6 mm 6 mm 6 mm 25 Mar 2005 25 Mar 2005 25 Mar 2005* 25 Mar 2005 02 Oct 2006 02 Oct 2006 02 Oct 2006 02 Oct 2006 02 Oct 2006 02 Oct 2006 02 Oct 2006 08 Nov 2006 08 Nov 2006 08 Nov 2006 a 3 replicates were included in means separation analysis due to one failed replicate b 7 replicates were included in means separation analysis due to one failed replicate X Hunter Industries, San Marcos, CA Y Irritrol Systems Inc., Riverside, CA Z Toro Company, Inc., Riverside CA changed setting from 25 mm to 6 mm on 2 October 2006 Table 2 2. Summary of functionality problems for treatments and replicates. Treatment Model Prob lems with Rain Sensors WL 3MC 6MC 13MC Hunter Irritrol Toro Wireless Rain Clik Mini Clik Mini Clik Mini Clik Mini Clik Irritrol RFS 1000 Toro TWRS Not operational from 28 Jan 2008 to 9 June 2009 WL B stopped functioning 21 Sept 2007a None None 13C show ing somewhat erratic behavior 8 July 2008 H D show ed erratic behavior after 21 Dec 2007a None Six wireless receivers were incorrectly connected to one rain sensor from 20 April 2009 to 22 Sept 2009. a These replicates were not included in means separatio n analysis

PAGE 44

44 Table 2 3. Average depth of rainfall before rain sensors switched to Open Switch Mode. Treatment Model Set point (mm) Rainfall for OSM Accuracy (%)a Depth (mm) Standard Deviation (mm) CV (%) Wireless Rain Clik Mini Clik Mini Clik Mini Cli k Mini Clik Irritrol RFS 1000 Toro TWRS 3 6 13 6 6 6 3.2 1.9 1.6 6.6 3.8 4.3 5.8 2.1 1.0 0.7 3.3 1.4 1.5 1.9 67 51 51 51 37 36 33 64 27 51 64 71 97 a Accuracy is +/ 2% due acceptable error in the tipping bucket rain gauge Table 2 4. Summary of changes in accuracy for change in rainfall required for Open Switch Mode. Treatment Model Set point (mm) Days in study Rainfall depth for OSM (mm) Change in accuracy (percentile points) Beginning End Wireless Rain Clik Mini Clik Mini Clik Mini Clik Min i Clik Irritrol RFS 1000 Toro TWRS 3 6 13 6 6 6 1,186 1,186 1,186 1,186 1,050 1,050 1,050 2.6 1.9 1.3 8.3 3.3 3.6 5.7 4.0 a 1.9 2.1 a 5.3 a 4.5 a 5.1 a 6.1 1 59 36 36 42 7 a The depth of rainfall required for open switc h mode changed over the study period.

PAGE 45

45 D B C A Figure 21. WL (model Wireless RainClik, Hunter Industries, Inc., San Marcos, CA) rain sensor. A) Expanding disks inside ventilation window, B) quick response expanding disks, C) Ventilation window adjustment knob, D) antenna. A B C Figure 22. MC (model Mini Clik, Hunter Industries, Inc., San Marcos, CA) rain sensor. A) Rainfall threshold setting slots, B) expanding disks, C) dry out adjustment ring and vents.

PAGE 46

46 A C B Figure 23 Irritrol (model RFS 1000, Irritrol Syst ems, Inc ., Riverside, CA. ) rain sensor. A) Rainfall threshold setting slots, B) dry out adjustment ring, C) antenna A B C Figure 24 Toro (model TWRS, Toro Company, Inc. Riverside, CA ) rain sensor. A) Rainfall threshold setting slots, B) dryout vent C) antenna

PAGE 47

47 A B Figure 25 Detail of expanding disk material and threshold adjustment of Mini Clik (Hunter Industries, Inc.) rain sensor. A) Rainfall threshold setting slots, B) hygroscopic expanding disk material. Figure 26. Research site located at th e University of Florida Agricultural and Biological Engineering facilities. Shown: weather station on left, WL, 3MC, 3MC, and 13MC treatments installed on left board, and Hunter, Irritrol, and Toro on right board.

PAGE 48

48 Figure 27. Installed Hunter Wireless Rain Clik, three on left and one on right, and Mini Clik rain sensors with four wireless receivers for WL and data logger. Figure 28. Installed Hunter, Irritrol, and Toro rain sensors, left to right, with wireless receivers (Irritrol on left and Toro on right), and data logger.

PAGE 49

49 y = 1.0453x 0.2591 R = 0.9963 0 10 20 30 40 50 60 70 80 0 10 20 30 40 50 60 70 80 Tipping Bucket Data (mm) Rain Gauge Data (mm) Figure 29. Relationship between manual rain gauge and weather station tipping bucket rain gauge with the calibration factor applied to the tipping bucket data with more than 15 mm of rainfall. y = 0.9589x + 0.681 R = 0.9914 15 25 35 45 55 65 75 15 25 35 45 55 65 75 Tipping Bucket Data (mm) Rain Gauge Data (mm) Figure 210. Relationship of rain events greater than 15mm between manual rain gauge and weather station tipping bucket rain gauge with the calibration factor applied to the tipping bucket data.

PAGE 50

50 y = 0.9938x 0.1462 R = 0.9825 0 3 6 9 12 15 0 3 6 9 12 15 Tipping Bucket Data (mm) Rain Gauge Data (mm) Figure 211. Relationship of rain events less than 15 mm between manual rain gauge and weather station tipping bucket rain gauge without the calibration factor applied to the tipping bucket data. 0 500 1000 1500 2000 2500 3000 3500 4000 4500 0 50 100 150 200 250 Oct Nov Dec Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Cumulative rainfall (mm) Monthly rainfall (mm) Date, 2006 2009 Study Period Historical Study Cumulative Historical Cumulative 4121 3551 Figure 2 12. Comparison of monthly and cumulative rainfall during the study period for WL and MC treatments and average historical rainfall for north central Florida.

PAGE 51

51 0 500 1000 1500 2000 2500 3000 3500 4000 4500 0 50 100 150 200 250 Nov Jan Mar May Jul Sep Nov Jan Mar May Jul Sep Nov Jan Mar May Jul Sep Nov Cumulative rainfall (mm) Monthly rainfall (mm) Date, 2006 2009 Study Period Historical Study Cumulative Historical Cumulative 4055 3425 Figure 213. Comparison of monthly and cumulative rainfall during the study period for Hunter, Irritrol, and Toro treatments and average historical rainfall for north central Florida.

PAGE 52

52 0 500 1000 1500 2000 2500 3000 3500 4000 0 5 10 15 20 25 Cumulalative rainfall (mm) Daily rainfall (mm) Date, 2006 2009 6 mm set point 139 events with rainfall > 6 mm43 events with more than 25 mm of rain ranging from 26 to 113 mm13 mm set point 82 events with rainfall > 13 mm 3 mm set point 192 events with rainfall > 3 mm 3551 Figure 2 14. Cumulative and daily rainf all during the WL and MC treatments study period with t he rainfall setting and the respective theoretical number of times each should have gone into OSM

PAGE 53

53 0 500 1000 1500 2000 2500 3000 3500 4000 0 5 10 15 20 25 Cumulative rainfall (mm) Daily rainfall (mm) Date, 2006 2009 6 mm set point 136 events with rainfall > 6 mm42 events with more than 25mm of rain ranging from 26 to 77 mm 3410 Figure 2 15. Cumulative and daily rainfall during the Hunter, Irritrol, and Toro treatments study period with the rainfall setting and the respective theoretical number of times each should have gone into OSM

PAGE 54

54 0 20 40 60 80 100 120 140 160 180 Oct 06 Jan 07 Apr 07 Jul 07 Oct 07 Dec 07 Mar 08 Jun 08 Sep 08 Dec 08 Mar 09 Jun 09 Sep 09 Dec 09 Cummulative times in OSM Date, 2006 2009 WL 3MC 6MC 13MC 160a 109b 146a 154a Figure 2 16. Cumulative number of times into OSM for WL and MC treatments. Data from 28 January to 9 June 2008 are not included due to all WL replicates not functioning. Erratic replicates within treatments are not included after their respective improper functioning dates (WLB 21 September 2007 and 13MC C 8 July 2008). Numbers with different letters indicate a statisti cal difference using T ukey Kramer adjusted p values of p<0.05. 0 20 40 60 80 100 120 140 160 180 Oct 06 Jan 07 Apr 07 Jul 07 Oct 07 Dec 07 Mar 08 Jun 08 Sep 08 Dec 08 Mar 09 Jun 09 Sep 09 Dec 09 Cummulative times in OSM Date, 2006 2009 WL A WL B WL C WL D 162 47 131 145 Figure 2 17. Cumulative number of times the WL replicates went into OSM. WL stopped functioning on 21 September 2007.

PAGE 55

55 0 20 40 60 80 100 120 140 160 180 200 Oct 06 Jan 07 Apr 07 Jul 07 Oct 07 Dec 07 Mar 08 Jun 08 Sep 08 Dec 08 Mar 09 Jun 09 Sep 09 Dec 09 Cummulative times in OSM Date, 2006 2009 3MC A 3MC B 3MC C 3MC D Theoretical 182 149 162 173 Theoretical = 192 Figure 2 18. Cumulative number of times the 3MC replicates went into OSM. The Theoretical v alue is the number of times the replicates should have gone into OSM based on rainfall. 0 20 40 60 80 100 120 140 160 180 200 Oct 06 Jan 07 Apr 07 Jul 07 Oct 07 Dec 07 Mar 08 Jun 08 Sep 08 Dec 08 Mar 09 Jun 09 Sep 09 Dec 09 Cummulative times in OSM Date, 2006 2009 6MC A 6MC B 6MC C 6MC D Theoretical 188 162 173 173 Theoretical = 139 Figure 2 19. Cumulative number of times the 6MC replicates went into OSM. The Theoretical value is the number of times the replicates should have gone into OSM base d on rainfall.

PAGE 56

56 0 20 40 60 80 100 120 140 Oct 06 Jan 07 Apr 07 Jul 07 Oct 07 Dec 07 Mar 08 Jun 08 Sep 08 Dec 08 Mar 09 Jun 09 Sep 09 Dec 09 Cummulative times in OSM Date, 2006 2009 13MC A 13MC B 13MC C 13MC D Theoretical 131 116 59 111 Theoretical = 82 Figure 2 20. Cumulative number of times the 13MC replicates went into OSM. The Theoretical value is the number of times the replicates should have gone into OSM based on rainfall. 0 20 40 60 80 100 120 140 160 180 200 Nov 06 Jan 07 Apr 07 Jul 07 Sep 07 Dec 07 Mar 08 May 08 Aug 08 Oct 08 Jan 09 Apr 09 Jun 09 Sep 09 Dec 09 Cumulative events in OSM Date, 2006 2009 Hunter Irritrol Toro 6 mm 114c 190a 144b Theoretical = 136 Figure 2 21. Cumulative number of times into OSM for H unter, Irritrol, and Toro treatments. The Theoretical value is the number of times the replicates should have gone into OSM based on rainfall. Numbers with different letters indicate a statisti cal difference using Tukey Kramer adjusted p values of p<0.05

PAGE 57

57 0 20 40 60 80 100 120 140 160 Nov 06 Jan 07 Apr 07 Jul 07 Sep 07 Dec 07 Mar 08 May 08 Aug 08 Oct 08 Jan 09 Apr 09 Jun 09 Sep 09 Dec 09 Cumulative events in OSM Date, 2006 2009 A B C D E F G H Theoretical 57144 144 144106 135 143 151 Theoretical = 136 Figure 2 22. Cumulative number of times Hunter into OSM. The Theoretical value is the number of times the replicates should have gone into OSM based on rainfall. 0 20 40 60 80 100 120 140 160 180 200 Nov 06 Jan 07 Apr 07 Jul 07 Sep 07 Dec 07 Mar 08 May 08 Aug 08 Oct 08 Jan 09 Apr 09 Jun 09 Sep 09 Dec 09 Cumulative events in OSM Date, 2006 2009 A B C D E F G H Theoretical 204Theoretical= 136 196 196191 190 175182 182 Figure 2 23. Cumulative number of times Irritrol went into OSM. The Theoretical value i s the number of times the replicates should have gone into OSM based on rainfall.

PAGE 58

58 0 20 40 60 80 100 120 Nov 06 Jan 07 Apr 07 Jul 07 Sep 07 Dec 07 Mar 08 May 08 Aug 08 Oct 08 Jan 09 Apr 09 Jun 09 Sep 09 Dec 09 Cumulative events in OSM Date, 2006 2009 A B C D E F G H Theoretical Theoretical = 101 87 87114 83 92 10496 96 Figure 2 24. Cumulative number of times Toro went into OSM. The Theoretical value is the number of times the replicates should have gone into OSM based on rainfall. 1 2 3 4 5 6 7 8 9 10 Oct 06 Jan 07 Apr 07 Jul 07 Oct 07 Dec 07 Mar 08 Jun 08 Sep 08 Dec 08 Mar 09 Jun 09 Sep 09 Dec 09 Amount of rainfall for OSM (mm) Date, 2006 2009 3MC 6MC 13MC Hunter Irritrol Toro Figure 22 5 Accuracy of each treatment with a set point over the study period with an average (solid line) and 95% confidence bands (dashed lines).

PAGE 59

59 CHAPTER 3 EXPANDING DISK RAIN SENSOR DRY OUT AND POTENTIAL IR RIGATION SAVINGS Introduction Water conse rvation measures are becoming more critical in Florida due to increased resource demand. Florida receives an average of 1,350 mm of rainfall a year (NOAA, 2003). From 2000 to 2005, Florida had a net population gain of approximately 1,000 people per day and ranks fou rth in population (USCB, 2009). Seventy percent of single family homes have automatic irrigation systems (Tampa Bay Water, 2005). Florida is second wettest state in the nation but irrigation is required to meet the aesthetic landscape requirement s. Residential i rrigation has been reported to account for 64% of residential water use in one area of the state (Haley et al., 2007). Rain sensors are a conservation device used with automatic irrigation systems to reduce applied irrigation. Rain sensors have adjustment settings allowing users to choose the amount of rainfall required for the irrigation cycle to be interrupted, generally between 3 and 25 mm. States such as Florida, Minnesota, and New Jersey and many municipalities throughout the country ha ve mandated the use of rain sensors to conserve water. Variables affecting water and cost savings with rain sensors include rain frequency, whether or not the controller is left on for automatic operation, and the amount of water applied by the system per cycle (Dukes and Haman, 2002b). The most common rain sensors used in Florida are expandingdisk rain sensors (see Chapter 2, Figures 21 to 24). The swelling of the hygroscopic disks in the sensor expand proportionally to the amount of rainfall (see Chapt er 2, Figure 25). The expandingdisk rain sensor is in closed switch mode until sufficient rainfall changes it to open switch mode. The sensors rely on evaporation to go return to open switch mode

PAGE 60

60 (OSM) Dry out settings can vary between 2 hours and 3 day s (Hunter Industries, Inc., 2005) Dry out time is the amount of time the sensors stay in OSM. A longer dry out time has the potential to interrupt more scheduled irrigation cycles Dry out time is meant to represent the time it would take for the rainfall to leave the root profile via evaporation, transpiration, and other pathways. Dr y out settings can be adjusted for most sensors such that it matches the drying rate of the sites soil (Dewey, 2003). Cardenas Lailhacar and Dukes (2008) investigated the pe rformance of expanding disk rain sensors in a 2005 study at the University of Florida campus in Gainesville, Florida. The performance of Hunter Mini Clik rain sensors with three different set tings (3 mm, 13 mm, and 25 mm) and one Hunter Wireless Rain Clik were monitored and compared with rainfall depth. The frequency of disk dry out within 24 hours for the Wireless, 3 mm, and 13 mm treatments was 80%, 51%, and 47%, respectively. The average percentage of potential water savings for the WL, 3MC, 13MC, and 25MC was 44%, 30%, 17%, and 3%, respectively, based on a 2 d/wk irrigation schedule. Researchers at the University of Florida have conducted irrigation studies with smart controllers that included expandingdisk rain sensors set on a 2 d/wk irrigation sche dule to represent homeowner irrigation under watering restrictions In a study with ET controllers and rain sensors in southwestern Florida, the addition of a rain sensor set at 6 mm reduced irrigation 21% compared with a timebased schedule (Davis et al., 2009). A study in central Florida compared applied irrigation among a controller only, a controller with a rain sensor set at 3 mm, and a controller with a rain sensor set at 6 mm. The savings for the 3 mm and 6 mm set points were 25% and 17%, respectivel y (McCready et al., 2009).

PAGE 61

61 The objective of this study was to evaluate the dry out time of three brands of expandingdisk rain sensors and potential irrigation savings by comparing rain sensor irrigation interruptions to University of Florida Institute of Food and Agriculture Science (UF IFAS) recommended irrigation schedules irrigating 2 d/wk and 1 d/w k to represent a homeowner schedule. Materials and Methods This study was conducted at the University of Florida Agricultural and Biological Engineering Department campus turfgrass plots, Gainesville, Florida. There were a total of 40 rain sensors installed at a height of 2 m (see Chapter 2, Figure 26). Treatments Seven treatments were established at the site (Table 3 1). The Wireless RainClik (WL) and Mini Clik (MC) rain sensors were from Hunter Industries, Inc., San Marcos, CA. The WL d id not have a rainfall setting. The three MC treatments had rainfall settings of 3 mm, 6 mm, and 13 mm (3MC, 6MC, and 13MC) (see Chapter 2, Figure 27). Analysis for treatm ents WL, 3MC, 6MC, and 13MC included data collected between 2 October 2006 and 31 December 2009 (1,186 days). These treatments were installed on 25 March 2005 with four replications each. The 6MC treatment was originally set to 25 mm and was changed to 6 m m on 2 October 2006 to have a setting better fit for north central Florida. The remaining treatments were installed at a later date with 6 mm settings for three brands. The brands and respective treatment codes were Hunter Industries, Inc. (Hunter), Irritr ol Systems, Inc., Riverside, CA (Irritrol), and Toro Company, Inc., Riverside, CA, (Toro) (see Chapter 2, Figure 2 8). Analysis for treatments Hunter, Irritrol, and Toro included data collected between 8 November 2006 and 31 December 2009 (1,150 days).

PAGE 62

62 Ea ch rain sensor brand ha d adjust ments for the dry out time. WL dry out vents were set at half open. The dry out vents for 3MC, 6MC, and 13MC were fully open during the study. The dry out vents for Hunter and Irritrol were fully open from installation until 2 July 2009 and were then changed to fully closed. The Toro rain sensor receivers were set to 0.0 day dry out; the water delay feature was set to 1.0 day for four replicates and 3.0 days for the remaining four replicates. The dry out setting was changed fr om 0.0 to 4.0 days to validate that electrical connections were correctly established for the Toro rain sensors and wireless receivers. The average amount of time required for dry out for 0.0 day setting and 4.0 days setting was 15 hours and 99 hours (4 da ys), respectively. The difference between dry out times for different settings confirmed that the Toro installation was done correctly. To estimate the potential water savings, theoretical irrigation schedules were compared with each treatment. The two sch edules used were a 1 d/wk schedule (Tuesdays) and a 2 d/wk schedule (Tuesdays and Saturdays) set to irrigate at 6 a.m. A scheduled irrigation was considered interrupted if the rain sensors were in openswitch mode due to rainfall. Potential savings was the number of irrigations interrupted by each rain sensor multiplied by the depth of scheduled irrigation. Weekly irrigation depths were calculated to satisfy historical net irrigation required to replace water lost to evaoptranspiration based on Dukes and Haman (2002a) recommendations (Table 32). Monitoring Each time a rain sensor changed mode between open switch mode (OSM) and closed switch mode (CSM), the date and time was recorded at a 1 second sampling interval using AM16/32 multiplexers (Campbell Scien tific, Inc., Logan, UT) attached to a CR10X model data logger (Campbell Scientific, Inc., Logan, UT).

PAGE 63

63 An onsite automated weather station (Campbell Scientific, Logan, UT) located within 15 meters of the experimental site recorded weather conditions using a CR10X model data logger (Campbell Scientific, Logan, UT). Data such as relative humidity, temperature (model HMP45C, Vaisala, Inc., Woburn MA), solar radiation (model LI200X, Li Cor, Inc., Lincoln, NE), and wind speed and direction (model WAS425, Vaisala, Inc., Sunnyvale, CA) were recorded at 15 minute intervals. Precipitation was measured by a tipping bucket rain gauge (model TE525WS, Texas Electronics, Inc., Dallas, TX) with a 1 second sampling interval time stamp for each 0.25 mm of rain. A manual rain gauge located within 5 meters of the rain sensors was used to verify the accuracy of the tipping bucket rain gauge measurements (See Chapter 2 Figure 29). The weather station tipping bucket rain gauge was calibrated using the Texas Electronics Calibration Kit (Texas Electronics, Inc., Dallas, TX). Calibration testing was conducted November 2009 on the tipping bucket rain gauge and is explained in detail in Chapter 2. Rainfall event depths from the study period were collected and analyzed. A rain event wa s considered started when the tipping bucket made the first tip. A 5hour or longer period between tips of the tipping bucket rain gauge was defined as a new rainfall event. Establishment of current and historical weather patterns affected rainfall setting recommendations. Monthly rainfall from a 30year 1970 to 2000 from the National Oceanic and Atmospheric Administration (NOAA) was used to compare study weather data with historical normals. Hourly disk length measurements with a dial caliper were conducted twice to better understand how the hygroscopic disks dry out after a rain event. The disk length

PAGE 64

64 measurements during dry out tracking were compared with temperature, solar radiation, and relative humidity to relate the physical parameters to rain sensor function. One tracking period was conducted on 11 July 2009 after a 60mm rain event on 10 July 2009. The disks were measured every hour for 20 hours after the rain stopped. To ensure that the disks had time to expand, there were 3 hours between when the rain stopped and measurements started. The rain sensors were inspected for interruption mode visually before disk measurements were taken. The second tracking period on 18 September 2009 was performed by manually watering the disks. The rain sensors were d renched and were given 2 hours to expand before measuring. Measurements were not taken throughout the night because the first dry out tracking showed relatively small disk length changes during the night. Statistical Analysis SAS statistical software (SA S Institute, Inc., Cary, NC) was used for all statistical analysis A general mixed model with an auto regressive error structure was used to model the continuous responses (PROC MIXED) Tukey Kramer adjusted p values ( p<0.05) were used for pairwise compar isons of mean. Results and Discussion Climactic Conditions During the 1,186 days of the WL and MC study period and the 1,150 days of the Hunter, Irritrol, and Toro study period, 28% of the days received rain. For WL and MC, the cumulative rainfall was 3,5 51 mm which is 14% less than the historical average of 4,121 mm (see Chapter 2, Figure 212). For Hunter, Irritrol, and Toro, the cumulative rainfall was 3,410 mm which is 16% less than the historical average of 4,055 mm (see Chapter 2, Figure 2 13). In Ch apter 2, Figures 214 and 215 show the daily and

PAGE 65

65 cumulative rainfall s during each period and the theoretical count for the number of OSM events by treatment Time in Open Switch Mode (Dry Out) Dry out is the amount of time the rain sensors are in OSM. Af ter the dry out period, irrigation would be allowed to occur. Figures 31 to 3 4 show frequency distributions of time in 6hour intervals that the RSs stayed in OSM for all treatments. The WL driedout within 24 hours 84% of the time with 1% requiring 49 t o 53 hours (Figure 3 1). The 3MC dried out within 24 hours 83% of the time and with 1% requiring 50 and 66 hours (Figure 3 2). The 6MC dried out within 24 hours 64% of the time and with 4% requiring 49 and 69 hours (Figure 33). The 13MC driedout within 2 4 hours 80% of the time and with 1% requiring 50 hours (Figure 3 4). Hunter driedout within 24 hours 71% of the time and with 3% requiring 48 and 77 hours (Figure 35). Of the 3% of events with more than 48 hours of dry out. Irritrol dried out within 24 h ours 84% of the time and all dr iedout out within 48 hours (Figure 3 16). Toro driedout within 24 hours 83% of the time and all driedout within 48 hours (Figure 3 7). The frequency of dry out within 24 hours for 6MC was 64% while the average for all oth er treatments was of 81%. This reduced percentage was a remnant of the rainfall setting change from 25 mm to 6 mm. As discussed in detail in Chapter 4, the hygroscopic disks change length with time in use and rainfall setting. The initial 25 mm setting cau sed the hygroscopic disks to change length differently than what the disks would have done with a 6 mm setting, which affected the dry out of the disks for 6MC. Table 3 3 summarizes the percentage of time each treatment driedout in 24 hours and how many h ours each treatment required for 95% dry out. The number of hours the WL, 3MC, 6MC, 13MC, Hunter, Irritrol, and Toro required for dry out 95% of the time

PAGE 66

66 were 36, 36, 42, 37, 44, 36, and 33 hours, respectively. These values are relatively close for all sen sors and settings and indicated that a majority of the rain sensors driedout within 24 hours and 95% of the time they dried out in less than 2 days. Rainfall needs to occur within 24 hours of a scheduled irrigation for most rain sensors to still be in OSM to interrupt the irrigation. Previous research (Cardenas Lailhacar and Dukes, 2008) investigated dry out on the WL, 3MC, and 13MC for the first year of installation. Their results indicated that WL driedout within 24 hours 80% of the time with 8% requir ing 54 and 78 hours, which is consistent with this study. The 3MC dried out within 24 hours 51% of the time with 12% requiring 48 and 78 hours and 7% more than 78 hours. The 13MC driedout within 24 hours 57% of the time with 6% requiring 48 and 72 hours. The 3MC and 13MC treatments had a shorter dry out during this study than the Cardenas Lailhacar and Dukes (2008) study, but treatment dry out times did not change within this study period. The amount of time in dry out did not change during the study peri od for any of the treatments. On 2 July 2009, the Hunter and Irritrol dry out vents were changed from fully open to fully closed. Hunter (Figures 38 and 3 9) and Irritrol (Figures 310 and 311) had average dry out increase times of 14% or 3 hours and 32% of 6 hours, respectively, after closing the dry out vents. Hunter driedout within 24 hours 73% and 65 % of the time, with fully open and fully closed vents, respectively. Irritrol driedout within 24 hours 82% and 77% of the time, with fully open and ful ly closed vents, respectively. The increase in dry out time did not increase potential water savings. All of the Toro replicates were set to 0.0 day dry out. There was no difference in dry out time for the

PAGE 67

67 four Toro replicates set at 1.0 day water delay an d 3.0 day water delay (Figures 312 and 313). Dry out Tracking Figures 314 to 315 show the dry out of the disks over time for the two dry out tracking events. A 60 mm rainfall event did not trigger all rainfall sensors to OSM: all 3MC, three 6MC, two 13MC, and three Hunter remained in CSM. The second tracking period was formed by manually watering the disks. This event did not rigger all rainfall sensors to OSM: two 3MC, one 6MC, two 13MC, and two Hunter remained in CSM. The same sensors did not go into OSM during both dry out tracking events likely because of wear of the hygroscopic disks. Though these disks did not go into OSM, the disks did expand with the rain event and contract over the dry out period. The dry out patterns in the September tracking event were similar to July. Figures 3 16 to 3 18 show the relationship between the disk length and temperature, solar radiation, and relative humidity for the dry out tracking on 11 July 2009 since it has more detail. Decreases in relative humidity and i ncreases in temperature and solar radiation were followed about 2 to 3 hours later with significant disk length decreases (Figures 3 1 4 to 3 16). Disk length reduction was caused by evaporation of rainfall from the disks. The lag time in response to the ch anges in the physical parameters was due to the time and energy required to vaporize the rain water in the disks for evaporation. The changes in climate preceding the significant disk contraction were a temperature increase from 24 to 31 C, an increase in solar radiation from 159 to 1005 W/m2, and a decrease in relative humidity from 93 to 54%. Each of the three parameters influenced dry out time.

PAGE 68

68 Potential Irrigation Savings The total potential water savings for each treatment under the different irrigati on schedules are in Table 34. The values of potential irrigation savings need to be considered with reference to the respective accuracy of the rain sensors. Tables 35 to 3 8 show the variation of total potential water savings for each replicate. The aver age percent water savings for the 2 d/wk schedule for the WL, 3MC, 6MC, 13MC, Hunter, Irritrol, and Toro treatments were 26%, 26%, 28%, 14%, 23%, 21%, and 21%, respectively. The average percentage water savings for the 1 d/wk schedule for the WL, 3MC, 6MC, 13MC, Hunter, Irritrol, and Toro treatments were 25%, 23%, 25%, 13%, 20%, 19%, and 14%, respectively. Cardenas Lailhacar et al. (2008) found that a rain sensor set at 6 mm on a timer with the same UF IFAS schedule used in this study could reduce applied irrigation by 34% in Gainesville, Florida. A study in southwestern Florida found 21% irrigation savings with a rain sensor set at 6 mm with a 2 d/wk irrigation schedule (Davis et al., 2009). The 2 d/wk irrigation schedule potential savings in this study was 28% with the 6MC indicating that the 6MC treatment acted within the range of previous studies. McCready et al. (2009) found that the 3 mm and 6 mm settings saved 25% and 17%, respectively, under a 2 d/wk irrigation schedule. The McCready et al. (2009) fi ndings for the 3 mm setting match this studys savings of 26%. McCready et al. (2009) and Davis et al. (2009) had similar savings with a 6 mm rainfall setting while Cardenas Lailhacar et al. (2008) had higher savings. The increased savings of Cardenas Lail hacar et al. (2008) was due to the higher rainfall in 2005 compared with the later studies in 2006 and 2007. The 28% savings in this study for the 6 mm setting fall in between savings of

PAGE 69

69 the other studies. The average potential irrigation savings for rain sensors set at 6 mm was 24% also in the range of 17% to 34% from previous research. The amount of savings should have been less for the 6MC in this study. The 6CM treatment savings should be between 26% and 14% which are the savings for 3MC and 13MC. The lack of difference in the 3MC and 6MC savings is due to the hygroscopic disk properties of the 6MC after being changed from a 25 mm setting to 6 mm setting a year after original installation (see Chapter 4 for hygroscopic disk details). From Chapter 2, the 3MC, 6MC, and 13MC sensors were in OSM after 2.0, 1.7, and 7.0 mm of rainfall with accuracies of 64% 2 7% and 5 1%, respectively. The 6MC treatment had a very low accuracy, and it should have had a depth required for OSM between the depths required by 3MC and 13MC. The Hunter treatment, which was the Mini Clik sensor set to 6 mm from its time of installation, went into OSM after 4.1 mm as expected based on the performance of 3MC and 13MC. The 6MC treatment did not perform as it would have without the sett ing adjustment. Summary and Conclusions This experiment was carried out during a relatively dry period with rainfall on 28% of the days. Rain sensor dry out times did not change throughout the study period. The 13MC had greater variability of dry out time s than any other treatment. Averaged across the treatments, the rain sensors driedout within 24 hours 79% of the time and in 38 hours 95% of the time. Changing the vent settings from fully open to fully closed on some treatments increased the dry out time an average of 23%, but the potential irrigation savings was unchanged. The Toro water delay feature does not have an effect on the number of OSM occurrences or potential irrigation savings.

PAGE 70

70 During the dry out process with the expanding disks, the most si gnificant disk contraction occurred 2 or 3 hours after changes in climatic conditions. The disks reacted to decreasing relative humidity (from 93 to 54%) and increasing temperature (from 24 to 31 C) and solar radiation (from 159 to 1005 W/m2). Apparently, 2 to 3 hours was needed for a sufficient amount of water to vaporize from the hygroscopic disks to result in a significant size reduction. Potential water savings were determined by comparing the number of times the sensors went into OSM and dry out time to a theoretical UF IFAS irrigation schedule. Potential irrigation savings should be considered with the accuracy of rain sensors. All treatments, except Toro, had accuracies of less than 65%. The potential irrigation savings presented in this research wer e higher than they would be with more accurate rain sensors since most treatments went into open switch mode with less rainfall than thei r respective rainfall setting. For a 2 d/w k and 1 d / w k irrigation schedule, the percent age water savings for the 13MC w as 14% and 13% and the average for all other treatments was 24% and 21%, respectively. As expected, the rain sensors with lower rainfall setting had higher potential water savings. The average irrigation savings for p re vious research with a 2 d/w k irrigati on schedule with rainfall settings of 3 and 6 mm was 24% and 21%, respectively. This virtual study had similar potential water savings as previous research. There was no difference in potential irrigation savings between the rain sensors with 3 and 6 mm rainfall settings because of the combination of the inherent low accuracy of rain sensors and relative closeness of the settings (3 and 6 mm versus 6 and 13 mm).

PAGE 71

71 Rainfall settings of 3, 6, and 13 mm are adequate for rain sensors in central Florida because a ll settings conserved water. Rain sensors should be set to 3 or 6 mm, which conserve more than a setting of 13 mm, until the user determines that landscape quality or climatic conditions require the higher setting. The rainfall settings for a rain sensor s hould not be changed after a particular setting has been established for more than 3 months (see Chapter 4 for more details). If the rainfall setting needs to be changed after 3 months of use, it is recommended that a new rain sensor be installed. Table 3 1. Treatment description. Treatment Model Replicates Set Point Dry out Vent Setting WL 3MC 6MC 13MC Hunter Irritrol Toro Wireless Rain Clik Mini Clik Mini Clik Mini Clik Mini Clik Irritrol RFS 1000 Toro TWRS 4 4 4 4 8a 8a 8 3 mm 6 mm 13 mm 6 mm 6 mm 6 mm half open fully open fully open fully open fully open and fully closede fully open and fully closede 0 day dry out a Dry out vents changed from fully open to fully closed on 02 July 2009. Table 3 2. Month ly irrigation depth to replace historical evapotranspiration values based on Dukes and Haman (2002a). Run times are based on an irrigation application rate of 38 mm/hr assuming system efficiency of 60% and considering effective rainfall. The Reduced UF IFA S irrigation schedule is 60% of the UF IFAS irrigation schedule. Month Irrigation depth (mm) January February March April May June July August September October November December 0 0 0 81 160 143 131 120 154 103 56 0 Total 949

PAGE 72

72 Table 33. Summary of dr y out time for all treatments. Treatment Model Dry out Vent Setting Frequency of dry out within 24 hours Hours driedout 95% of the time WL 3MC 6MC 13MC Hunter Irritrol Toro Wireless Rain Clik x Mini Clikx Mini Clikx Mini Clikx Mini Clikx Irritrol RFS 100 0y Toro TWRS z half open fully open fully open fully open fully open and fully closeda fully open and fully closeda 0 day dry out 84% 83% 64% 80% 71% 84% 83% 36 36 42 37 44 36 33 a Dry out vents changed from fully open to fully closed on 02 July 2009. Ta ble 3 4. Total potential water savings per treatment for all treatments compared with a 2 d/wk and a 1 d/wk irrigation schedule for the study period (Oct/Nov 2006 to 1 Dec 2009). Treatment 2 d/wk Irrigation Schedule 1 d/wk Irrigation Schedule Irrigation depth (mm) Water Savings Irrigation Depth (mm) Water Savings (mm) (%) (mm) (%) WL 3MC 6MC 13MC Hunter Irritrol Toro 3,005 3,005 3,005 3,005 2,902 2,902 2,902 772 775 841 415 684 614 595 26 26 28 14 23 21 21 2,968 2,968 2,968 2,968 2,869 2,869 2,869 7 48 677 745 401 574 548 407 25 23 25 13 20 19 14 Table 3 5. Variation in total potential water savings per replicate for the WL and MC treatments compared with UF IFAS 2 d/wk irrigation recommendations. Treatment Water saved by replicates (mm) A B C D Average CV (%) WL 3MC 6MC 13MC 702 846 838 352 161 754 822 619 827 661 836 187 788 838 868 500 772 z a 775 a 841 a 415 b 8 z 11 2 45 Numbers with different letters indicate a statistical difference at the 95% confindence level using Duncans Multiple Range Test. z Average and CV do not include WLB

PAGE 73

73 Table 3 6. Variation in total potential water savings per replicate for the Hunter, Irritrol, and Toro treatments compared with UF IFAS 2 d/wk irrigation recommendations. Treatment Water saved b y re plicates (mm) A B C D E F G H Average CV (%) Hunter Irritrol Toro 675 647 615 742 688 581 756 664 565 137 368 574 375 499 605 747 688 627 747 672 594 744 688 599 684 z a 614 a 595 a 20 z 19 4 Numbers with different letters indicate a statistical diff erence at the 95% confindence level using Duncans Multiple Range Test. z Average and CV do not include Hunter D Table 3 7. Variation in total potential water savings per replicate for the WL and MC treatments compared with UF IFAS 1 d/ wk irrigation reco mmendations. Treatment Water saved by replicates (mm) A B C D Average CV (%) WL 3MC 6MC 13MC 715 700 765 265 142 634 686 646 792 573 771 191 735 803 759 503 748 z a 677 a 745 a 401 b 5 z 15 5 53 Numbers with different letters indicate a statistical difference using Tukey Kramer adjusted p values of p<0.05. z Average and CV do not include WLB Table 3 8. Variation in total potential water savings per replicate for the Hunter, Irritrol, and Toro treatments compared with UF IFAS 1 d/ wk irrigation r ecommendations. Treatment Water saved b y replicates (mm) A B C D E F G H Average CV (%) Hunter Irritrol Toro 580 573 382 620 514 517 585 503 517 93 503 324 269 544 514 655 578 324 655 537 324 655 634 356 574 z a 548 a 407 a 24 z 8 23 Numbers with di fferent letters indicate a statistical difference using Tukey Kramer adjusted p values of p<0.05. z Average and CV do not include Hunter D

PAGE 74

74 0 20 40 60 80 100 120 0 5 10 15 20 25 30 35 Cumulative frequency of occurances (%) Frequency of occurances (%) Interval of hours for dry out period 9 20 11 21 31 2 2 1 20 32 53 84 93 95 97 99 100 2 Figure 3 1 Histogram and cumulative frequency of dry out time (time from the end of the rain event until the sen sors returns to closed switch mode) for the WL treatment average. 0 20 40 60 80 100 120 0 5 10 15 20 25 30 Cumulative frequency of occurances (%) Frequency of occurances (%) Interval of hours for dry out period 19 13 25 25 9 4 1 1 19 32 57 83 92 96 98 99 100 2 Figure 3 2 Histogram and cumulative frequency of dry out time (time from the end of the rain event until the sensors returns to closedswitch mode) for the 3MC treatment average.

PAGE 75

75 0 20 40 60 80 100 120 0 5 10 15 20 25 30 Cumulative frequency of occurances (%) Frequency of occurances (%) Interval of hours for dry out period 9 7 20 28 14 9 2 4 9 16 36 64 78 87 95 96 100 7 Fig ure 3 3 Histogram and cumulative frequency of dry out time (time from the end of the rain event until the sensors returns to closedswitch mode) for the 6MC treatment average. 0 20 40 60 80 100 120 0 5 10 15 20 25 30 35 Cumulative frequency of occurances (%) Frequency of occurances (%) Interval of hours for dry out period 9 17 32 22 11 3 2 1 9 27 58 80 91 94 97 99 100 3 Figure 3 4 Histogram and cumulative frequency of dry out time (time from the end of the rain event until the sensors returns to closedswitch mode) for the 13MC treatment average.

PAGE 76

76 0 20 40 60 80 100 120 0 5 10 15 20 25 30 35 Cumulative frequency of occurances (%) Frequency of occurances (%) Interval of hours for dry out period 12 10 8 22 31 7 4 3 10 18 41 71 83 90 93 97 100 3 Figure 3 5 Histogram and cumulative frequency of dry out time (time from the end of the rain event until the sensors returns to closedswitch mo de) for the Hunter treatment average. 0 20 40 60 80 100 120 0 5 10 15 20 25 30 35 40 Cumulative frequency of occurances (%) Frequency of occurances (%) Interval of hours for dry out period 8 16 8 23 37 3 1 16 23 47 84 92 95 99 100 5 100 Figure 3 6 Histogram and cumulative frequency of dry out time (time from the end of the rain event until the sensors returns to closedswitch mode) for the Irritrol treatment average.

PAGE 77

77 0 20 40 60 80 100 120 0 5 10 15 20 25 30 35 40 Cumulative frequency of occurances (%) Frequency of occurances (%) Interval of hours for dry out period 9 9 13 26 35 4 2 9 22 48 83 92 97 98 100 2 100 Figure 3 7 Histogram and cumulative frequency of dry out time (time from the end of the rain event until the sensors returns to closedswitch mode) for the Toro treatment average. 0 20 40 60 80 100 120 0 5 10 15 20 25 30 35 Cumulative frequency of occurances (%) Frequency of occurances (%) Interval of hours for dry out period 13 9 9 26 30 3 4 9 17 43 73 86 94 94 100 2 98 Figure 3 8 Histogram and cumulative frequency of dry out time (time from the end of the rain ev ent until the sensors returns to closedswitch mode) for the Hunter treatment average with the dry out vents fully open (8 November 2008 to 2 July 2009)

PAGE 78

78 0 20 40 60 80 100 120 0 5 10 15 20 25 30 35 40 Cumulative frequency of occurances (%) Frequency of occurances (%) Interval of hours for dry out period 8 15 8 4 35 4 4 15 23 27 62 69 73 88 100 15 92 8 Figure 3 9 Histogram and cumulative frequency of dry out time (time from the end of the rain event until the sensors returns to closedswitch mode) for the Hunter treatment average with the dry out vents fully closed (2 July 2009 to 31 December 2009) 0 20 40 60 80 100 120 0 5 10 15 20 25 30 35 40 Cumulative frequency of occurances (%) Frequency of occurances (%) Interval of hours for dry out period 9 17 9 23 34 3 5 17 25 48 82 90 94 99 100 1 100 Figure 3 10. Histogram and cumulative frequency of dry out time (time from the end of the rain event until the sensors returns to closedswitch mode) for the Irritrol treatment average with the dry out vents fully open (8 November 2008 to 2 July 2009)

PAGE 79

79 0 20 40 60 80 100 120 0 10 20 30 40 50 60 Cumulative frequency of occurances (%) Frequency of occurances (%) Interval of hours for dry out period 3 13 16 48 3 3 13 13 29 77 81 84 90 100 6 94 6 Figure 3 11. Histogram and cumulative frequency of dry out time (time from the end of the rain event until the sensors returns to closedswitch mode) for the Irritrol treatment average with the dry out vents fully closed (2 July 2009 to 31 December 2009) 0 20 40 60 80 100 120 0 5 10 15 20 25 30 35 40 Cumulative frequency of occurances (%) Frequency of occurances (%) Interval of hours for dry out period 10 9 11 26 35 4 2 9 21 47 82 92 97 98 100 2 100 Figure 3 12. Histogram and cumulative frequency of dry out time (time from the end of the rain e vent until the sensors returns to closedswitch mode) for the one day water delay setting for four the Toro replicates.

PAGE 80

80 0 20 40 60 80 100 120 0 5 10 15 20 25 30 35 40 Cumulative frequency of occurances (%) Frequency of occurances (%) Interval of hours for dry out period 12 8 24 36 4 2 8 21 46 81 93 96 98 100 2 100 13 Figure 3 13. Histogram and cumulative frequency of dry out time (time from the end of the rain event until the sensors returns to closed switch mode) for the three day water delay setting for four the Toro replicates. 19 20 21 22 23 24 25 0:00 3:00 6:00 9:00 12:00 15:00 18:00 21:00 0:00 Disk length (mm) Time in dry out (hour) 3MC 6MC 13MC Hunter Irritrol Toro Figure 3 14. Dry out tracking of average disk length for each treatment for the natural rain event on 10 July 2009.

PAGE 81

81 19 20 21 22 23 24 25 0:00 3:00 6:00 9:00 12:00 15:00 18:00 21:00 0:00 Disk length (mm) Timer in dry out (hours) 3MC 6MC 13MC Hunter Irritrol Toro Figure 3 15. Dry out tracking of average disk len gth for each treatment for the manual rain event on 18 September 2009. 20 22 24 26 28 30 32 19 20 21 22 23 24 25 0:00 3:00 6:00 9:00 12:00 15:00 18:00 21:00 0:00 Temperature (deg C) Disk length (mm) Time in dry out (hour) 3MC 6MC 13MC Hunter Irritrol Toro Temperature Figure 3 16. Dry out tracking of average disk length for each treatment and temperature for the rain event on 10 July 2009.

PAGE 82

82 200 0 200 400 600 800 1000 1200 19 20 21 22 23 24 25 0:00 3:00 6:00 9:00 12:00 15:00 18:00 21:00 0:00 Solar Radiation (W/m 2 ) Disk length (mm) Time in dry out (hour) 3MC 6MC 13MC Hunter Irritrol Toro Solar Radiation Figure 3 17. Dry out tracking of average disk length for each treatment and solar radiation for the rain event on 10 July 2009. 0 10 20 30 40 50 60 70 80 90 100 19 20 21 22 23 24 25 0:00 3:00 6:00 9:00 12:00 15:00 18:00 21:00 0:00 Relative Humidity (%) Disk length (mm) Time in dry out (hour) 3MC 6MC 13MC Hunter Irritrol Toro Relative Humidity Figure 3 18. Dry out tracking of average disk length for each treatment and relative humidity for the rain event on 10 July 2009.

PAGE 83

83 CHAPTER 4 RELATIONSHIP BETWEEN EXPANDING DIS K RAIN SENSOR DISK L ENGTH AND PERFORMANCE Introduction In 2000, Florida was the largest user of groundwater east of the Mississippi River (Hutson et al., 2004). Floridas growing population continues to put a stress on the water supply. Florida ranked firs t in single family home construction in 2005 with 209,162 homes built (United States Census Bureau [USCB], 2007). Seventy percent of single family homes have automatic irrigation systems (Tampa Bay Water, 2005). Automatic timer controls on irrigation systems in Florida have been reported to lead to a 47% increase in water use (Mayer, et al., 1999). Water conservation measures are needed to reduce water use. Rain sensors interrupt the cycle of an automatic irrigation system controller when a specific amount of rainfall has occurred (Dukes and Haman, 2002b). Rain sensors remain in closedswitch mode that allows irrigation until sufficient rainfall changes it to open switch mode. S everal types of rain sensors are available to homeowners Most rain sensor model s can be adjusted to interrupt at different depths of rainfall, generally between 3 and 25 mm. One type of rain sensor collects the rain water in a cup and interrupts the irrigation based on a preset weight of water. Another type has a set of electrodes th at detect the water level in a small collection dish that measures the amount of rainfall with two electrodes in a collection cup (Dukes and H aman, 2002b) A disadvantage to both of these devices is that debris can get into the collection cup and cause the system to interrupt irrigation without sufficient rainfall. The most common type of rain sensor used in Florida is the expandingdisk rain sensor. Compared with other types of rain sensors, it requires less maintenance and is

PAGE 84

84 less expensive. The swelling of the hygroscopic disks in the sensor expand proportionally to the amount of rainfall. When the disks swell to the selected rainfall setting, the system goes into open switch mode. The sensors rely on evaporation to allow irrigation after a being in open switch mode (dry out). A longer dry out time has the potential to interrupt more scheduled irrigations by the irrigation controller. Dry out settings can be adjusted for most sensors. Settings should be chosen to match soil conditions such that the disks dry out at a rate similar to water leaving the root zone soil profile. The hygroscopic disk dry out time is influenced by weather conditions such as temperature, wind, solar radiation and relative humidity. After investigating expandingdisk rain sensors since 2005, researchers noticed that the hygroscopic disks appear to have different sizes with time (Figures 4 1 and 4 2 ). The disks may lose elasticity with time due to the repeated shrinking and swelling from rain events resulting in a longer total leng th because they appear to not shrink to their original length. A longer length of disks inside the sensor may cause the RS to interrupt the irrigation cycle with less rain. Cardenas Lailhacar et al. (2009) determined that the sensitivity of the RS changed during 282 days of installation. One possible cause of the change in sensitivity is a change in size of the hygroscopic disks themselves. The objectives of this study were to determine if the length of the hygroscopic disks in expanding disk rain sensors change size based on the amount of time installed and the set point, and if so determine the effect of disk change on switching accuracy. Materials and Methods This study was conducted at the University of Florida Agricultural and Biological Engineering Department campus turfgrass plots, Gainesville, Florida. The rain sensors

PAGE 85

85 were installed on a 2m high board next to each other. Figure 22 (see Chapter 2) shows the details of the expandingdisk rain sensor used for this study. Treatments Five treatments wi th varying rain sensor set points were established (Table 41). The rain sensors used were Mini Clik (MC) rain sensors from Hunter Industries, Inc., ( San Marcos, CA ) Treatments 3MC and 13MC had rainfall settings of 3 mm and 13 mm, respectively, and were installed on 25 March 2005. The 3MC and 13MC treatments had four replicates each. Treatments 3R, 6R, and 13R with rainfall settings of 3 mm, 6 mm, and 13 mm, respectively, were installed on 13 February 2009. The 3R, 6R, and 13R treatments were replicated t hree times Monitoring The total disk length was measured about once per month with a Mitutoyo Series 505 dial caliper ( Mitutoyo Co rporation Aurora, IL) starting February 2009. Figure 25 (see Chapter 2) shows the hygroscopic disks inside the rain sensor Disks were measured when fully contracted and all rainfall had evaporated. Treatments 3R, 6R, and 13R were measured before installation to give the length of a new device. To determine the accuracy of the 3MC and 13MC treatments, each time a rain sensor changed mode between open switch mode (OSM) and closed switch mode (CSM), the date and time were recorded at a 1second sampling interval using AM16/32 multiplexers (Campbell Scientific, Inc., Logan, UT) attached to a CR10X model data logger (Campbell Sci entific, Inc., Logan, UT). An onsite automated weather station (Campbell Scientific, Logan, UT) located within 15 m of the experimental site recorded climatic conditions using a CR10X model data logger (Campbell Scientific, Logan, UT). Precipitation was measured by a tipping bucket rain gauge with a onesecond sampling

PAGE 86

86 interval time stamp for each 0.25 mm of rain. A manual rain gauge located within 5 meters of the rain sensors was used to verify the accuracy of the tipping bucket rain gauge measurements (See Chapter 2, Figure 2 9). The travel distance from CSM to OSM of each sensor was measured to determine if possible disk length changes were significant. The travel distance was the required increase in disk length due to rainfall for the rain sensor to go into OSM. Travel distance was determined with dry rain sensors by measuring the post on the top of the rain sensor in a stationary position (CSM) and in the position of full disk expansion (OSM). Total travel distance included the difference in the rain se nsor post as described above plus the distance required to compress the trigger device underneath the hygroscopic disks. Statistical Analysis SAS statistical software (SAS Institute, Inc., Cary, NC) was used for all statistical analysis A general mixed model with an auto regressive error structure was used to model the continuous responses (PROC MIXED) A paired T Test was used for means separation with significant F values ( p<0.05). Results and Discussion Disk measurements were performed to track size changes in the hygroscopic disk length. Figure 4 3 shows the daily and cumulative rainfall during the study period and the number of rainfall events greater than the three rain sensor rainfall settings investigated (3, 6, and 13 mm). Length by I nstallation date and setting The difference between initial and final disk length for 3MC, 13MC, 3R, 6R, and 13R was 0.1, 0.4, 1.2, 1.8, and 2.3 mm, respectively. Table 42 shows the initial and

PAGE 87

87 final disk measurements for each treatment. The disk length of 3MC and 13MC did not change significantly during the measurement period. After 81 days of installation, the disk lengths of 3R, 6R, and 13R were significantly expanded from the initial length, by 1.2, 1.8, and 2.30 mm, respectively. The length of disks after 81 days of installation and at the end of the study (276 days) for 3R, 6R, and 13R were 18.1 and 18.7 mm, 18.7 and 19.2 mm, and 19.0 and 19.6 mm. The disk lengths of 3R and 13R approached the lengths of the 3MC and 13MC, respectively The treatments installed i n 13 February 2009 started with the same average length of 17.4 mm (Table 4 3). After 81 days of installation, the disk lengths for 3R, 6R, and 13R were all significantly different (pvalue <0.05) from the initial measurements with lengths of 18.1, 18.7, a nd 19.0 mm, respectively. The disk lengths by treatment remained statistically different during the rest of the study. This result was consistent with the older sensors, 3MC and 13MC, in which the disks had different lengths, 19.0 and 20.2 mm respectively. Figure 44 shows the change in length during the study period for each treatment. There was little variation of disk length within treatments (Tables 44 and 4 5). The average Coefficient of Variance (CV) values for February, May, and November measurement s for the 3MC, 13MC, 3R, 6R, and 13R w ere 1.0%, 1.2%, 0.8%, 1.6, and 1.0%, respectively. The average CV for disk length was 1.1% while the average CV for the depth of rainfall required for OSM was 47% (see Chapter 2 for details). Disk Length and Traveling D istance The length that the disks travel when switching between CSM and OSM was analyzed to determine if the disk length changes were substantial. Travel distance was dependent on rainfall setting. For the rain sensors installed in February, the average

PAGE 88

88 length increased 1.7 mm and the average travel distance decreased 0.9 mm (Table 4 6). The changes in disk length were substantial with respect to the travel distance. Table 4 7 shows the values of final disk length, length change, and travel distance at th e end of the study. Disk length change and the total travel distance were larger in rain sensors with higher rainfall setting because there was more space inside the rain sensor. Treatments 3MC and 13MC had a shorter travel distance than 3R and 13R, respec tively, because the 3MC and 13MC disks were longer. Effect on Interruption P erformance The accuracy of the 3MC and 13MC units for the first 300 days of installation was analyzed to evaluate the effect of disk length change on accuracy Analysis is based o n the assumption that the disks installed in 2005 and 2009 had the same properties. Figure 45 compares accuracy of 3MC and 13MC to the disk size change in the 3R and 13R. Treatments 3MC and 13MC had no correlation of accuracy and disk length change. Summary and Conclusions New (0 to 276 days old) and old (1,421 to 1,697 days old) rain sensors were analyzed to determine the effect of disk length change on rain sensor accuracy. The hygroscopic disks in expandingdisk rain sensors change length after rainfal l exposure. Disk length was analyzed based on time installed in the field and rainfall setting. The new rain sensors had an average disk length increase of 1.8 mm and the old rain sensors changed 0.1 mm in size during the same time. The average length change for the new rain sensors was more than the travel distance inside the sensor when switching from closed switch mode to openswitch mode (1.7 mm versus 0.9 mm). This

PAGE 89

89 finding indicated that the amount of disk length increase was significant enough to pos sibly influence rain sensor performance. The lengths of the RS disks after 81 days of installation and 178 mm of rainfall were significantly greater than the initial measurements. Higher rainfall settings had more disk length increase due to the larger space within the rain sensor device at higher rainfall settings and the material losing elasticity while expanding and contracting. The amount of time the rain sensors remained in openswitch mode was not related to the setting; the disk length difference bet ween 3 mm and 13 mm settings did not influence the dry out (see Chapter 3 for details). Increased disk length change was not related to decreased rain sensor accuracy. The accuracies presented in Chapter 2 showed that both 3MC and 13MC were more accurate i n the first 282 days of installation than the accuracy of days 5601,742 of installation. It was assumed that the properties of the disks installed in 2005 were the same as those installed in 2009. All of the disk length change occurred during the initial Cardenas Lailhacar and Dukes (2008) study in which the average accuracy was 83%. In this study, the accuracy of the same sensors was 62% in which no significant disk length change occurred. The decreased accuracy cannot be attributed to disk length change because the time in which the disk lengths changed did not correspond to the time of decreased accuracy. There was no relationship between the disk length and accuracy. Decreasing accuracy of the rain sensors was due to the aging of the entire unit. The outer casing of the rain sensor became more brittle with sun and rain exposure. The triggering mechanism required less movement on the older rain sensors than the new rain sensors when manually going into openswitch mode.

PAGE 90

90 The reason that the disks could hav e a significant length change without affecting accuracy was because of the disks properties. The rain sensors with higher rainfall settings had more increase in disk length because they have more space inside the rain sensor in which to lengthen. The incr ease in length also meant that there was more pore space inside each disk. While the pore space shortened the travel distance for the sensor, the amount of rainfall required to expand the full travel distance for open switch mode did not change because those pores then needed to be filled with water. Thus, an increase in disk length did not correspond to the rain sensor requiring less rainfall for open switch mode. Table 4 1. Description of rain sensors details for each treatment. Treatment Model Replicate s Set Point Installation Date 3MC 13MC 3R 6R 13R Mini Clik x Mini Clikx Mini Clikx Mini Clikx Mini Clik x 4 4 3 3 3 3 mm 13 mm 3 mm 6 mm 13 mm 25 Mar 2005 25 Mar 2005 13 Feb 2009 13 Feb 2009 13 Feb 2009 x Hun ter Industries, San Marcos, CA Table 4 2. Aver age disk length for each treatment at two intervals: initial and final (276 days of installation). Treatment Installation Date Set point (mm) Disk Length Measurement (mm) February November Difference 3MC 13MC 3R 6R 13R CV (%) 25 Mar 2005 25 Mar 2005 13 Feb 2009 13 Feb 2009 13 Feb 2009 3 13 3 6 13 19.2 20.1 17.5 17.4 17.3 6.7 19.1 20.4 18.7 19.2 19.6 3.3 NS NS *** NS = no statistical difference between February and November measurements = statistical difference at 0.05 p value level betw een February and November measurements *** = statistical difference at 0.001 pvalue level between February and November measurements

PAGE 91

91 Table 4 3. Average disk length for the treatments installed 13 February 2009 at three intervals: initial, 81 days of in stallation, and final (276 days of installation). Treatment Set point (mm) Disk Length Measurement February May November 3R 6R 13R 3 6 13 17.5 a 17.4 a 17.3 a 18.1 a 18.7 b 19.0 c 18.7 a 19.2 b 19.6 c Numbers with different letters indicate a statist ical difference using a T Test for means separation with significant F values (p<0.05) Table 4 4. Disk length for replicates of treatments installed 25 March 2005 at three intervals: initial (February), 81 days of installation (May), and final (November, 276 days of installation). Treatment Replicate Disk Length Measurement (mm) Treatment Replicate Disk Length Measurement (mm) Feb May Nov Feb May Nov 3MC A 3MC B 3MC C 3MC D Average CV (%) 18.9 19.2 19.3 19.4 19.2 1.0 19.1 19.1 19.3 18.6 19.0 1.5 19.0 19.1 19.2 19.1 19.1 0.5 13MC A 13MC B 13MC C 13MC D Average CV (%) 19.8 20.0 20.3 20.4 20.1 1.4 19.9 20.6 20.4 20.4 20.3 1.4 20.2 20.5 20.5 20.5 20.4 0.7 Table 4 5. Disk length for replicates of treatments installed 13 February 2009 at three intervals: initial (February), 81 days of installation (May), and final (November, 276 days of installation). Treatment 3R 6R 13R Replicate Disk Length Measurement (mm) Disk Length Measurement (mm) Disk Length Measurement (mm) Feb May Nov Feb May Nov Feb May Nov A B C Avg CV (%) 17.1 17.7 17.5 17.5 1.7 18.0 18.1 18.2 18.1 0.4 18.6 18.7 18.7 18.7 0.2 17.4 17.9 16.8 17.4 3.0 18.8 18.8 18.4 18.7 1.3 19.1 19.2 19.3 19.2 0.5 17.1 17.7 17.5 17.5 1.7 19.1 19.0 19.0 19.0 0.3 19.7 19.8 19.4 19.6 1.0

PAGE 92

92 Table 4 6. Compari son of average length change and travel distance from closedswitch mode to open switch mode of treatments installed in 13 February 2009. The February travel distance was measured on a rain sensor before installation. Treatment February November Change D isk Length (mm) Travel Distance (mm) Disk Length (mm) Travel Distance (mm) Disk Length (mm) Travel Distance (mm) 3R 6R 13RC Avg 17.5 17.4 17.5 17.5 1.7 3.0 4.6 3.1 18.7 19.2 19.6 19.2 1.7 1.8 3.1 2.2 1.2 1.8 2.3 1.7 0.0 1.2 1.5 0.9 Table 4 7.Comparis on of average length change of each treatment from 13 February 2009 to 16 November 2009 and the travel distance each treatment from closed switch mode to openswitch mode. Travel distance was measured at the end of the study. Treatment Set Point (mm) Final Disk Length (mm) Length Change (mm) Travel Distance (mm) 3MC 13MC 3R 6R 13R Avg 3 13 3 6 13 19.1 20.4 18.7 19.2 19.6 19.4 0.1 0.4 1.2 1.8 2.3 1.1 1.4 2.6 1.7 1.8 3.1 2.1 Figure 41 .Mini Clik (Hunter Industries, Inc.) rain sensor expanding disks in stalled in March 2005 (left) and February 2009 (right) set at 13 mm measured in August 2009 (1600 and 179 days of installation, respectively).

PAGE 93

93 Figure 42 Mini Clik (Hunter Industries, Inc.) rain sensors expanding disks installed in 2005 with settings (l eft to right) of 3 mm, 6 mm, and 13 mm and lengths 19.2, 19.8, and 20.1 mm, respectively, after 1,600 days of installation.

PAGE 94

94 0 200 400 600 800 1000 1200 0 5 10 15 20 25 Cumulalative rainfall (mm) Daily rainfall (mm) Date, 2006 2009 6 mm set point 44 events with rainfall > 6 mm11 events with more than 25 mm of rain ranging from 26 to 60 mm13 mm set point 30 events with rainfall > 13 mm 3 mm set point 57 events with rainfall > 3 mm 954 Figure 4 3 Cumulative and daily rainfall during the study period with number of rainfall events greater than the different rain sensor rainfall settings

PAGE 95

95 17.0 17.5 18.0 18.5 19.0 19.5 20.0 20.5 21.0 21.5 0 35 69 104 138 173 207 242 276 Average disk length (mm) Day of measurement 3MC avg 13MC avg 3R avg 6R avg 13R avg Figure 44. Average hygroscopic disk length for Mini Clik rain sensors installed in 2005 (MC) and 2009 (R) over installation time. The 2009 rain sensors were newly installed on day zero. 0% 20% 40% 60% 80% 100% 0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 Accuracy (%) Disk Length Change (mm) 3MC 13MC Figure 45 Accuracy f or each setting during the first 300 days of rain sensor installation compared to change in disk length. Accuracy data are the average amount of rainfall for openswitch mode for a given rainfall event for a treatment.

PAGE 96

96 CHAPTER 5 CONCLUSIONS AND FUTURE WORK Co nclusions The goal of this research was to evaluate the performance of expandingdisk rain sensors. The primary objectives of this study were to use three brands to A) evaluate the accuracy with time with respect to the selected rainfall setting, B) evaluate the amount of time rain sensors remained in interruption mode (open switch mode) after a rainfall event, C) quantify potential irrigation savings for different rainfall settings compared with a time based schedule, and D) determine if the hygroscopic di sks in the rain sensors change length with time. The secondary objectives included a) investigating variability within brands and rainfall settings, b) evaluating the changes in accuracy with time, c) determining climatic parameters that influence dry out time, and d) evaluating the effect of possible disk length change on accuracy. This experiment was carried out during a relatively dry period with rainfall on 28% of the days. The percentage of rain events greater than 3, 6, and 13 mm were 57%, 42%, and 25%. The WL, which had no rainfall setting, had an average rainfall depth of 3.2 mm required t o go into open switch mode. The 3MC, 6MC, 13MC Hunter, Irritrol, and Toro treatments had accuracies of 64%, 27%, 51%, 64%, 71%, and 97%, respectively The accuracy of the rain sensors changed with time. The percentile point change in accuracy during the study period ranged from 36% to 59%, where a negative value indicated a decrease in accuracy. The 6MC rain sensors had a 25 mm rainfall setting when initially insta lled and were changed to a 6 mm rainfall setting at the beginning of this study. This change caused the 6MC treatment to have the lowest overall accuracy and the largest increase in accuracy.

PAGE 97

97 The amount of time a rain sensor stays in open switch mode, cal led dry out, influences water savings. Longer dry out times means that the rain sensors are in open switch longer and could potentially interrupt more irrigations. On average, rain sensors driedout within 24 hours 79% of the time and in 38 hours 95% of th e time. Changing the vent settings from fully open to fully closed on some treatments increased the dry out time but the potential irrigation savings was unchanged. The Toro water delay feature did not have an effect on the amount of time in open switch mo de or potential irrigation savings. Disk dry out was related to climactic parameters. Dry out occurred with decreasing relative humidity and increasing temperature and solar radiation. The most significant disk contraction occurred 2 or 3 hours after changes in climatic conditions. The changes in climate preceding the significant disk contraction were a temperature increase from 24 to 31 C, an increase in solar radiation from 159 to 1005 W/m2, and a decrease in relative humidity from 93 to 54%. The hy groscopic disks in expandingdisk rain sensors increase in length after continuous rainfall exposure. The disk length change was related to the rainfall setting: the 3 mm had the shortest length and the 13 mm had the longest length tending to conform to th e space available in the rain sensor body. The length of the rain sensor disks was significantly more than the initial measurement with 81 days of installation and 178 mm of rainfall. The rain sensors with higher rainfall settings had greater increases in disk length because they had more space inside the rain sensor in which to lengthen. The amount of rainfall required to expand the full travel distance for open switch mode did not change because those pores then needed to be filled with water.

PAGE 98

98 Disk lengt h was studied to determine its possible effects on rain sensor performance. The data did not show that disk length change was related to decreased accuracy. The decreased accuracy of 82% in the initial study (days 0 to 282 of installation) to 62% in this s tudy (days 560 to 1742 of installation) was not due to disk length change. The disks got larger during their first 271 days of installation which was during the initial study (days 0 to 282 of installation). The disk length increase did not influence the d ry out time. There was also no difference in dry out time for different rainfall settings. The decreased accuracy was attributed to the outer casing of the rain sensor became more brittle and the triggering mechanism become more sensitive. The potential w ater savings for a 2 d/wk and 1 d/ w k irrigation schedule 13MC w ere 14% and 13% and the average for all other treatments was 24% and 21%, respectively. Rain sensors with 3 and 6 mm rainfall settings had similar savings due the low accuracy of some treatment s of rain sensors and relative closeness of the settings (3 and 6 mm versus 6 and 13 mm). Potential irrigation savings should be considered in relation to the accuracy of rain sensors. Since most treatments went into openswitch mode with less rainfall tha n their respective rainfall setting, the potential irrigation savings presented in this research were higher than they would be with more accurate rain sensors. Rainfall settings of 3, 6, and 13 mm are adequate for rain sensors in central Florida because all settings conserved water. Rain sensors should be set to 3 or 6 mm until the user determines that landscape quality or climatic conditions require the higher setting. If the rainfall setting needs to be changed after more than 3 months of use, it is re commended that a new rain sensor or at least new expanding disks be installed. For the best accuracy, there is evidence based on historical studies and results from this

PAGE 99

99 study that Hunter Mini Clik rain sensors should be replaced after 1 year of installat ion. Based on their performance, Irritrol RSF 1000 and Toro TWRS rain sensors do not need to be replaced for at least 3 years. Further research is needed to verify these results. A survey of Florida homeowners with automatic irrigation systems found that o nly 25% reported to have a rain sensor of which not all are likely correctly installed (Whitcomb, 2005). The inclusion of rain sensors on more automatic irrigation systems could increase water savings to homeowners and have environmental benefits such as u rban less runoff. Future Work Changing the rainfall setting as the sensors age has a significant effect on the accuracy of the rain sensor at the new setting. Research should be conducted on rain sensors with a variety of changed rainfall settings. Fut ure disk measurements of new rain sensors taken every 2 weeks of the first 3 months of installation would offer better insight into disk length change. To best determine the influence of disk length on accuracy, the new sensors should be connected to a dat a logger. Additional experiments studying the effect of rainfall setting on turf quality should be conducted to determine the best settings for Florida. Both 3 and 6 mm settings have been studied in relation to turf grass quality; the 13 mm setting has on ly gone through virtual testing. The potential irrigation savings are a good starting point, but plots need to be used for more supporting data.

PAGE 100

100 LIST OF REFERENCES Arlington, Texas. Lawn and landscape irrigation conservation ordinance. Section 4.27. Ava ilable at: http://www.ci.arlington.tx.us/water/waterconservation_ordinancesummary.html ( Accessed 28 May 2009). Cardenas Lailhacar, B., Dukes M.D 2008. Expandin g disk rain sensor performance and potential irrigation water savings. Journal of Irrigation and Drainage 134(1): 6773. Cardenas Lailhacar, B., Dukes M.D, Miller, G.L 2008. Sensor based automation of irrigation on bermudagrass, during wet weather cond itions. Journal of Irrigation and Drainage 134(2):120128. Cardenas Lailhacar, B., Dukes M.D., Meeks ,L 2009. Irrigation rain sensor accuracy. Proceedings of the World Environmental & Water Resources Congress (EWRI), Kansas City, MO. Available at: http://dx.doi.org/10.1061/41036(342)431 ( Accessed 1 March 2010). Cary, North Carolina. Rain sensors on automatic irrigation systems. Ordinance section 3684. Available at : http://www.townofcary.org/Home.htm ( Accessed 9 April 2009). Colleyville, Texas. Water conservation ordinance 061579. Available at: http://www.colleyville.com/content/view /716/12/ ( Accessed 28 May 2009) Connecticut Statutes. Section 29 265. Rain sensor devices for automatic lawn sprinkler systems. Available at: http://search.cga.state.ct.us/dtsearc h_pub_statutes.html ( Accessed 25 May 2009) Dallas, Texas Ordinances. Section 4921.1. Conservation measures relating to lawn and landscape irrigation. Available at: http://www .savedallaswater.com/pdf/Conservation_Ordinance.pdf (Accessed 9 April 2009). Davis, S.L, Dukes M.D., and Miller G.L., 2009. Landscape irrigation by evapotranspiration based irrigation controllers under dry conditions in Southwest Florida. Agricultural Water Management 96(2): 18281836. Available at: http://dx.doi.org/10.1016/j.agwat.2009.08.005 ( Accessed 28 January 2010) Derby, Kansas Ordinances, 2008 Ordinance No. 1932. Rain sensor ordinan ce. Available at: http://www.derbyweb.com/pdfs/rain sensor ordinance.pdf ( Accessed 28 May 2009) Dewey, C. 2003. Sensors at work. Irrigation and Green Industry, September 2003. Available at : http://www.igin.com ( Accessed 28 May 2009)

PAGE 101

101 Dukes, M. D., Cardenas Lailhacar, B., 2007. Smart Water Application Technologies (SWAT) Turf and Landscape Irrigation Equipment Testing Protocol for Rain Sensors: 1st dra ft Irrigation Association, Falls Church, VA. Available at: http://www.irrigation.org/gov/pdf/Phase11st_Draft_Test_Protocols RainSensors.pdf ( Accessed 19 Apri l 2009) Dukes, M.D., Haley M.B 2009. Evaluation of soil moisture based on demand irrigation controllers, phase II, final report. Available at: http://irrigation.ifas.ufl.edu/pdf/publications/SMS/SMS%20Phase%20II%20Final %20Report%2012 1709.pdf ( Accessed 27 January 2010) Dukes, M. D., Haman D.Z 2002a. Operation of residential irrigation controllers. CIR1421, Institute of Food and Agricultural Sciences, University of Florida, Gainesville, FL. Available at: http://edis.ifas.ufl.edu/pdffiles/AE/AE22000.pdf ( Accessed 28 January 2010) Dukes, M. D., Haman D.Z 2002b. Resi dential irrigation system rainfall shutoff devices. ABE325, Institute of Food and Agricultural Sciences, University of Florida, Gainesville, FL. Available at: http://edis.ifas.ufl.edu/ae221 ( Accessed 28 January 2010) Figliola, R.S., Beasley, D.E 2002. Theory and design for mechanical measurements. 3rd edition John Wiley and Sons, Inc. New York. Florida Senate Bill 494. Water conservation/automatic sprinkler systems. Proposed revisions to Chapter 373.62. Av ailable at: http://www.flsenate.gov/cgi bin/view_page.pl?Tab=session&Submenu=1&FT=D&File=session/2009/Senate/ bills/amendments_com/html/sb0494c2409270.html ( Accessed 28 May 2009) Florida Statutes 2008. Chapter 373.62. Water Conservation; automatic sprinkler systems. Available at: http://www.leg.state.fl.us/statutes/index.cfm?App_mode=Display_Statute&URL= Ch0373/ch0373.htm ( Accessed 19 April 2009) Georgia Metropolitan North Georgia Water Planning District, Georgia. Water conservation action no 4. Available at: http://www.northgeorgiawater.com ( Accessed 26 May 2009) Haley, M. B., Dukes. M. D., 2007. Evaluation of sensor based residential irrigation water application. Proceedings from the America n Society of Agricultural and Biological Engineers International Meeting, 1720 June 2007, Minneapolis, MI. Available at: http://irrigation.ifas.ufl.edu/pdf/pub lications/SMS/ASABE 072251Pinellas Co.pdf ( Accessed 16 April 2009)

PAGE 102

102 Haley, M.B., Dukes M.D., Miller G.L 2007. Residential irrigation water use in Central Florida. Journal of Irrigation and Drainage Engineering 133(5): 427434. Available at: http://scitation.aip.org/getpdf/servlet/GetPDFServlet?filetype=pdf&id=JIDEDH000 133000005000427000001&idtype=cvips ( Accessed 16 April 2009) Harrisburg, South Dakota Ordinances, 2006. Ordinance 200604 Chapter 8.01. Available at: http://www.harrisburg.govoffice.com/vertical/Sites/%7BA7E6811A D8794C7C A19F F0A55FD49B95%7D/uploads/%7B04211876C0064C4D BCA2 57348DD77A59%7D.PDF ( Accessed 28 May 2009) Hunter Industries, 2005. Product information sensors. Hunter Industries, Inc., San Marcos, CA. Available at: http://www.hunterindustries.com/Resources/PDFs/Product_Guides/Domestic/lit2 93w.pdf ( Accessed 8 July 2009) Hutson, S S., Barber, N. L., Kenny, J. F., Linsey, K. S., Lumia, D. S., Maupin, M. A., 2004. Estimated Use of Water in the United States in 2000. United States Geological Survey Circular 1268. United States Department of the Interior, United States Geological Surv ey, Reston, VA. Available at: http://pubs.usgs.gov/circ/2004/circ1268/index.html ( Accessed 9 July 2009) Lucas, Texas Ordinances, 2005. Article 8. Irrigation system regulations. Section 32 20. Rain and freeze sensors. Available at: http://lucastexas.us/objects/20051100540.1_Rain_Freeze_Sensors.pdf ( Accessed 28 May 2009) Marella, R.L. 1992. Water withdraw als, use, and trends in Florida, 1990. Water Resources Investigations Report 924140. United States Department of the Interior, United States Geological Survey, Tallahassee, Fl. Available at: http://fl.water.usgs.gov/Abstracts/wri91_4123_marella.html ( Accessed 21 May 2009) Marella, R.L. 2004.Water withdrawals, use, discharge, and trends in Florida, 2000. Scientific Investigations Report 20045151. United States Department of the In terior, United States Geological Survey, Tallahassee, Fl. Available at: http://pubs.usgs.gov/sir/2004/5151/pdf/20045151.pdf ( Accessed 21 May 2009) Massachusetts Senate 186the General Cou rt. Senate Bill 1396. Available at: http://www.mass.gov/legis/bills/senate/186/st01pdf/ST01396.PDF ( Accessed 28 May 2009) Mayer, P.W., DeOreo, W.B., Opitz, E.M., Kiefer, K.C., Davis, W.Y., Dziegielewski, B., Nelson J.O 1999. Residential end uses of water. American Water Works Association Research Foundation. Denver, CO.

PAGE 103

103 McCready, M.S., Dukes, M.D., Miller G.L 2009. Water conservation potential of smart irrigation controlle rs on St. Augustinegrass. Agricultural Water Management 96(11):16231632. Available at: http://dx.doi.org/10.1016/j.agwat.2009.06.007 ( Accessed 28 January 2010) Minnesota Statutes, 2003. Chapt er 44F. No. 335. Available at: https://www.revisor.mn.gov ( Accessed 9 April 2009) Moriarty, T. 2009. Front page Globe article targets lawn irrigation. Irrigation Association of New England newsletter Issue 2Q2 009. Available at: http://www.irrigationassociationne.org/newsletters/2Q2009.pdf (Accessed 9 July 2009). NOAA ( National Oceanic and Atmospheric Administration) 2003. Monthly st ation normals of temperature, precipitation, and heating and cooling degree days 1971 2000; 08 Florida. Climatography of The United States No. 81. 28p. Available at: http://hurricane.ncdc.noaa.gov/climatenormals/clim81/FLnorm.pdf ( Accessed 27 January 2010) New Jersey Statutes, 2000. 52:27D 123.13. Device to override automatic sprinkler after adequate rainfall required. Available at: http://lis.njleg.state.nj.us/cgi bin/om_isapi.dll?clientID=178043&Depth=4&TD=WRAP&advquery=rain%20sens or&headingswithhits=on&infobase=statu tes.nfo&rank=&record={15B3A}&softpag e=Doc_Frame_Pg42&wordsaroundhits=2&x=47&y=15&zz = ( Accessed 26 April 2009) Omega Engineering, Inc. 1995. RG 2501 Series: Tipping bucket rain gauge operators manual. Available at: http://www.omega.com/manuals/manualpdf/M2235.pdf ( Accessed 29 March 2010) San Antonio Water Systems. Ordinance 100332 34.274.2. Available at: http://www.bexarmet.org/files/SanAntonio2005ConservationOrdinance.pdf ( Accessed 28 May 2009) San Antonio Water System. Rain sensors frequently asked questions. San Antonio Water System, San Antonio, TX. Available at: http://www.saws.org/conservation/ordinance/rainsensors/faq.shtml ( Accessed 15 July 2009) Spectrum Technologies, Inc. Tipping bucket rain gauge: Catalog #3665R. Available at: http://www.specmeters.com/pdf/3665R%20Tipping%20Rain%20Gauge.pdf ( Accessed 29 March 2010) Sutron Corporation. Tipping bucket rain gauge: Stainless steel 56000425. Available at: http://www.sutron.com/pdfs/2006_RainGauge_StainlessSteel_56000425.pdf ( Accessed 29 March 2010)

PAGE 104

104 Tampa Bay Water, 2005. Tampa Bay Water: Evaluating implementation of multiple irrigation and landscape ordinances in the Tampa Bay region. Available at: http://www.tampabaywater.org/documents/conservation/RegionLandscapeOrdina nces.pdf ( Accessed 27 January 2010) Texas Electronics, Inc. (a). Manual for field calibration kit: For TR 525 rain gauge. Available at: http://www.texaselectronics.com/pl_supp_rainfall.htm ( Accessed 30 November 2009) Texas Electronics, Inc. (b). TR 525S siphon rain gauge tipping bucket manual. Available at: http://www.texaselectronics.com/pdf/manuals/Siphon.pdf ( Accessed 29 March 2010) Texas Hou se Bill 2299. Introduced 09 March 2007. Standards for irrigation system equipment. Available at: http://www.legis.state.tx.us/tlodocs/80R/billtext/pdf/HB02299I.pdf ( Accessed 9 April 2009) USCB ( United States Census Bureau), 2009. Population estimates. USCB. Washington, D.C. Available at: http://www.census.gov/popest/estimates.html ( Accessed 27 January 2010) USCB ( United States Census Bureau), 2007. Housing unit estimates. USCB. Washington, D.C. Available at: http://www.census.gov/const/C40/Table2/tb2u2005.txt ( Accessed 28 January 2010) Water Auth ority of Great Neck North, New York. Rules and regulations: general information. Available at: http://www.waterauthorityofgreatnecknorth.com ( Accessed 26 May 2009) Whitcomb, J. B. 2005. Florida water rates evaluation of singlefamily homes. Southwest Florida Water Management District, Brooksville, FL. Available at: http://www.swfwmd.state.fl.us/documents/reports /water_rate_report.pdf ( Accessed 21 May 2009)

PAGE 105

105 BIOGRAPHICAL SKETCH Leah Meeks received her Bachelor of Science in BioResource and Agricultural Engineering at Cal Poly, San Luis Obispo with a minor in Women and Gender Studies in 2008. While at Cal Poly, Leah served as the president of Agricultural Ambassadors and Vice President of the Agricultural Engineering Society. She was heavily involved in community agencies and was the scrumhalf for the womens rugby team. Leah obtained her Engineer In Training cert ification during her fourth year at Cal Poly. Leah was named the American Society of Agricultural and Biological Engineering National Student Engineer of the Y ear in 2006 and in 2007, was one of five Society of Women Engineers Outstanding Women in Engineering and Technology for the class of 2008, and earned the College of Agriculture 2008 Outstanding Senior with Service to the Community. In 2008, she accepted a graduate assistantship in the Agricultural and Biological Engineering Department at the University of Florida. While working on her graduate studies, Leah presented research at state and local conferences and became a member of the American Society of Civ il Engineers State Water Planning C ommittee.