<%BANNER%>

Sensor-Based Automation of Irrigation in Bermudagrass


PAGE 1

SENSOR-BASED AUTOMATION OF IRRIGATION OF BERMUDAGRASS By BERNARD CARDENAS-LAILHACAR A THESIS PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLOR IDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF MASTER OF SCIENCE UNIVERSITY OF FLORIDA 2006

PAGE 2

Copyright 2006 by Bernard Cardenas-Lailhacar

PAGE 3

To my parents and sons

PAGE 4

iv ACKNOWLEDGMENTS In the first place, I wish to thank my parents for their al ways enormous and unconditional love, support, a nd guidance; my sons for being the most adorable human beings I have ever met; my ex-wife for taki ng care of them while I was completing my studies, her understanding, patien ce, and sacrifice; and my amorcita for her enormous support, understanding, patience an d, most of all, her immens e love. Next, I would like to thank all my thesis committee members for be ing not just professors, but great human beings: Dr. Dorota Z. Haman and Dr. Gra dy L. Miller, also for their guidance and patience, and a huge thank you goes to Dr. Michael D. Dukes, for giving me the opportunity to work with him, which was always a lot of work, but also a pleasure. Also, I wish to give a special thank you to Melissa B. Haley for being alwa ys ready to help me. Lastly, I would also like to thank Engineer Larry Miller, Senior E ngineering Technician Danny Burch, and students Mary Shedd, Ste phen Hanks, Clay Coarsey, Brent Addison, Jason Frank, and Clay Breazeale for their assi stance on this research. This research was supported by the Pinellas-Anclotte Basin Bo ard of the Southwest Water Management District, the Florida Nursery and Lands cape Growers Associa tion, and the Florida Agricultural Experiment Station.

PAGE 5

v TABLE OF CONTENTS page ACKNOWLEDGMENTS.................................................................................................iv LIST OF TABLES...........................................................................................................viii LIST OF FIGURES.............................................................................................................x ABSTRACT....................................................................................................................... xv CHAPTER 1 INTRODUCTION........................................................................................................1 Water.......................................................................................................................... ...1 Water Demand..............................................................................................................2 Water Use.....................................................................................................................2 Water Use Restrictions.................................................................................................3 Landscapes in Florida...................................................................................................5 Irrigation..................................................................................................................... ..6 Irrigation Timers....................................................................................................6 Soil Moisture Content Measurement.....................................................................8 Granular matrix sensor...................................................................................8 Modern soil moisture sensors.........................................................................9 Controllers...........................................................................................................11 Automatic Control of Irrigation...........................................................................12 Rain Sensors........................................................................................................13 Irrigation and Turfgrass Quality.................................................................................15 2 SENSOR-BASED AUTOMATION OF IRRIGATION OF BERMUDAGRASS....20 Introduction.................................................................................................................20 Materials and Methods...............................................................................................25 Treatments...........................................................................................................27 Uniformity Test...................................................................................................28 Dry-Wet Analysis................................................................................................29 Plot Irrigation Management and Data Collection................................................30 Data Analysis.......................................................................................................33 Results and Discussion...............................................................................................34 Uniformity Tests..................................................................................................34

PAGE 6

vi Dry-Wet Analysis................................................................................................34 Rainfall................................................................................................................35 Irrigation Events..................................................................................................36 Irrigation Application Comparisons....................................................................42 Time-based treatments vs. SMS-based treatments.......................................42 Time-based treatments.................................................................................42 Comparisons between SMS-irrigation frequencies......................................44 Soil moisture sensor-brands comparison......................................................45 Brand comparisons within irrigation frequencies........................................46 Overall comparison......................................................................................47 Automation of Irrigation Systems.......................................................................49 Turfgrass Quality.................................................................................................50 Summary and Conclusions.........................................................................................51 3 EXPANDING DISK RAIN SENSOR PERFORMANCE AND POTENTIAL IRRIGATION WATER SAVINGS...........................................................................93 Rain Sensors...............................................................................................................93 Advantages..........................................................................................................94 Types and Methods..............................................................................................94 Installation...........................................................................................................96 Objectives............................................................................................................96 Materials and Methods...............................................................................................97 Data......................................................................................................................97 Treatments...........................................................................................................98 Statistical Analysis..............................................................................................99 Results and Discussion...............................................................................................99 Climatic Conditions.............................................................................................99 Number of Times in Bypass Mode......................................................................99 Depth of Rainfall Before Shut Off....................................................................100 Duration in Irrigation Bypa ss Mode (Dry-Out Period).....................................102 Potential Water Savings....................................................................................103 Payback Period..................................................................................................104 Summary and Conclusions.......................................................................................105 4 GRANULAR MATRIX SENSOR PERFORMANCE COMPARED TO TENSIOMETER IN A SANDY SOIL.....................................................................119 Tensiometers.............................................................................................................119 Granular Matrix Sensors...........................................................................................121 GMS – Tensiometer Comparison.............................................................................122 Objectives.................................................................................................................122 Materials and Methods.............................................................................................122 Experimental Set-Up.........................................................................................123 ECH2O Probes Calibration................................................................................123 Treatments.........................................................................................................124 Data....................................................................................................................124

PAGE 7

vii Results and Discussion.............................................................................................125 Calibration of the ECH2O Probe.......................................................................125 GMSs versus Tensiometers...............................................................................125 Conclusions...............................................................................................................126 5 CONCLUSIONS AND FUTURE WORK...............................................................140 Conclusions...............................................................................................................140 Future Work..............................................................................................................142 APPENDIX A LIST OF ABBREVIATIONS...................................................................................144 B STATISTICAL ANALYSES...................................................................................145 LIST OF REFERENCES.................................................................................................199 BIOGRAPHICAL SKETCH...........................................................................................208

PAGE 8

viii LIST OF TABLES Table page 2-1. Irrigation treatment codes and descriptions...............................................................54 2-2. Monthly irrigation depth to replace historical evapotra nspiration, assuming system efficiency of 60%, and c onsidering effective rainfall..................................54 2-3. Total number and percent of overri dden scheduled irrigation cycles; 2004 and 2005..........................................................................................................................5 5 2-4. Percent of irrigation cycles allowed by the SMS-based treatments through the experimental months of 2004 and 2005...................................................................56 2-5. Cumulative irrigation de pth applied to treatments, statistical comparisons between them, and percent of water savi ngs compared to 2-WRS, 2-DWRS, and 2-WORS; year 2004.................................................................................................57 2-6. Cumulative irrigation de pth applied to treatments, statistical comparisons between them, and percent of water savi ngs compared to 2-WRS, 2-DWRS, and 2-WORS; year 2005.................................................................................................58 2-7. Total cumulative irrigation depth applie d to treatments, statistical comparisons between them, and percent of water savi ngs compared to 2-WRS, 2-DWRS, and 2-WORS; years 2004 + 2005...................................................................................59 3-1. Treatments description............................................................................................107 3-2. Average depth of rainfall before rain sensors switched to bypass mode.................107 3-3. Large rainfall events not bypassed by treatment 3-MC...........................................107 3-4. Large rainfall events not bypassed by treatment 13-MC.........................................108 3-5. Large rainfall events not bypassed by treatment 25-MC.........................................108 3-6. Hours after rain stoppe d and sensors switched to bypa ss mode; treatment 3-MC..108 3-7. Hours after rain stoppe d and sensors switched to bypass mode; treatment 13-MC.109

PAGE 9

ix 3-8. WL replications that switched to bypa ss mode in absence of rainfall, elapsed time that they remained in bypass mode, and re lative humidity at the time when this occurred..................................................................................................................109 3-9. Total potential water savings per treatment.............................................................109 3-10. Potential payback period per treatment.................................................................110 4-1. Treatments............................................................................................................... 128 4-2. GMS-Tensiometer crossing points..........................................................................128

PAGE 10

x LIST OF FIGURES Figure page 1-1. Components of an automated irriga tion system: A) timer, B) power supply, C) soil moisture sensor-controller circuitr y, D) soil moisture sensor, and E) solenoid valve...........................................................................................................16 1-2. Granular matrix sensors (GMS)................................................................................17 1-3. Components of an automated irrigation system. 1) Timer, and 2) soil moisture sensor-controllers from different brands..................................................................18 1-4. Rain shut-off switch...................................................................................................19 1-5. The expanding material of a rain shut-off switch......................................................19 2-1. Soil water retention curve from tensiometers and calibrated ECH2O probe readings....................................................................................................................60 2-2. Soil moisture sensor br ands tested in this study........................................................61 2-3. Irrigation controls as installed for this study: soil moisture sensors-controllers brands: A) Rain Bird, B) Water Watcher, C) Acclima, and D) Irrometer, and irrigation timer E) Rain Bird....................................................................................62 2-4. Rain sensor installed for this study............................................................................63 2-5. Catch-can display for unifo rmity tests on turfgrass plots..........................................64 2-6. General view of the irrigati on controls used in this study.........................................65 2-7. Pipes, flowmeters, valves and wirings for this study...............................................65 2-8. Control board showing timers, soil mois ture sensor-controllers, solenoid valves wiring, and flowmeters-datalogger (det ails are shown in the next s).......................66 2-9. Control board detail showing the solenoid valves control box.................................67 2-10. Control board detail, flowmeter-dat alogger boxes showing A) multiplexers, B) CR 10X datalogger used for this study....................................................................67 2-11. Automated weather station ne ar turf plots for this study.........................................68

PAGE 11

xi 2-12. ECH2O probe, capacitance soil moisture probe shown with a HOBO data logger as installed for this study..........................................................................................69 2-13. Plot plan showing the low-quarter distribution uniformity testing results on each plot........................................................................................................................... .70 2-14. Plot plan showing aver age volumetric water conten t (%) on each plot during a relatively “dry” period. Plots in red were discarded, and plots in green were used for placement of SMSs.............................................................................................71 2-15. Plot plan showing aver age volumetric water conten t (%) on each plot during a relatively “wet” condition. Plots in red we re discarded, and plots in green were used for placement of SMSs.....................................................................................72 2-16. Plot plan with the modified comple tely randomized design (same color depicts treatment repetitions)................................................................................................73 2-17. Daily and cumulative rainfall in 2004. Note: rainfall for 5 Sep. (188 mm) and 6 Sep. (81 mm) is shown as a cumulative total (269 mm)..........................................74 2-18. Daily and cumulative rainfall in 2005.....................................................................74 2-19. Cumulative number of irrigation even ts per treatment in 2004; A) time-based treatments, and soil moisture sensor-based treatments at irrigation frequencies of B) 1 d/w, C) 2 d/w, and D) 7 d/w.............................................................................75 2-20 Cumulative number of irrigation even ts per treatment in 2005; A) time-based treatments, and soil moisture sensor-based treatments at irrigation frequencies of B) 1 d/w, C) 2 d/w, and D) 7 d/w.............................................................................76 2-21. Maximum weekly irriga tion water requirement (rainfa ll ETo difference); year 2004..........................................................................................................................7 7 2-22. Maximum weekly irriga tion water requirement (rainfa ll ETo difference); year 2005..........................................................................................................................7 7 2-23. Volumetric moisture content th rough time, on treatment 0-NI, year 2004.............78 2-24. Volumetric moisture content (VMC ) through time, showing results of the scheduled irrigation cycles (SIC ); treatment 1-AC, year 2004................................79 2-25. Volumetric moisture content (VMC ) through time, showing results of the scheduled irrigation cycles (SIC ); treatment 1-IM, year 2004.................................80 2-26. Volumetric moisture content (VMC ) through time, showing results of the scheduled irrigation cycles (SIC ); treatment 1-RB, year 2004................................81

PAGE 12

xii 2-27. Volumetric moisture content (VMC ) through time, showing results of the scheduled irrigation cycles (SIC ); treatment 1-WW, year 2004..............................82 2-28. Volumetric moisture content (VMC ) through time, showing results of the scheduled irrigation cycles (SIC ); treatment 2-AC, year 2004................................83 2-29. Volumetric moisture content (VMC ) through time, showing results of the scheduled irrigation cycles (SIC ); treatment 2-IM, year 2004.................................84 2-30. Volumetric moisture content (VMC ) through time, showing results of the scheduled irrigation cycles (SIC ); treatment 2-RB, year 2004................................85 2-31. Volumetric moisture content (VMC ) through time, showing results of the scheduled irrigation cycles (SIC ); treatment 2-WW, year 2004..............................86 2-32. Volumetric moisture content (VMC ) through time, showing results of the scheduled irrigation cycles (SIC ); treatment 7-AC, year 2004................................87 2-33. Volumetric moisture content (VMC ) through time, showing results of the scheduled irrigation cycles (SIC ); treatment 7-IM, year 2004.................................88 2-34. Volumetric moisture content (VMC ) through time, showing results of the scheduled irrigation cycles (SIC ); treatment 7-RB, year 2004................................89 2-35. Volumetric moisture content (VMC ) through time, showing results of the scheduled irrigation cycles (SIC ); treatment 7-WW, year 2004..............................90 2-36. Average irrigation depth applie d by brand; years 2004 and 2005 (P<0.0001)........91 2-37. View of different plots where no evid ent turfgrass quality differences could be detected; A) good quality, B) dormant.....................................................................92 3-1. Mini-Click (Hunter Industrie s, Inc.) rain sensor. A) Rain threshold set slots, B) vent ring..................................................................................................................111 3-2. Wireless Rain-Click (Hunter Industries, Inc.) rain sensor. A) Ventilation window adjustment knob, B) ventilation windows, C) antenna..........................................111 3-3. The expanding material of a rain shut-off switch....................................................112 3-4. Rain sensor experiment layout: A) Wi reless Rain-Click rain sensors, B) MiniClick rain sensors, C) Wireless Rain -Click receivers, D) multiplexers, E) CR 10X datalogger.......................................................................................................112 3-5. Manual rain gauge measurements co mpared to tipping bucket rain gauge measurements.........................................................................................................113 3-6. Daily and cumulative rainfall..................................................................................113

PAGE 13

xiii 3-7. Cumulative number of times rain sensors switched to bypass mode; average per treatment. Different letters indicat e a significant difference by Duncan’s Multiple Range Test (P<0.05)................................................................................114 3-8. Cumulative number of times rain sensors switched to bypass mode; WL treatment, with replicates indicated by A-D...........................................................115 3-9. Cumulative number of times rain sensors switched to bypass mode; 3-MC treatment, with replicates indicated by A-D...........................................................115 3-10. Cumulative number of times rain sensors switched to bypass mode; 13-MC treatment, with replicates indicated by A-D...........................................................116 3-11. Cumulative number of times rain sensors switched to bypass mode; 25-MC treatment, with replicates indicated by A-D...........................................................116 3-12. Histogram and frequency distribution for 6-hour intervals in bypass mode; WL.117 3-13. Histogram and frequency distributi on for 6-hour intervals in bypass mode; 3MC..........................................................................................................................117 3-14. Histogram and frequency distributi on for 6-hour intervals in bypass mode; 13MC..........................................................................................................................118 4-1. MLT-RSU Tensiometer...........................................................................................129 4-2. Watermark GMS......................................................................................................129 4-3. Temperature sensor..................................................................................................130 4-4. ECH2O probe...........................................................................................................130 4-5. Experimental layout (top view). A) Tens iometers, B) Granular matrix sensors, C) ECH2O probe, and D) Thermometer......................................................................131 4-6. Watermark monitor..................................................................................................132 4-7. ECH2O probe hooked up to a HOBO Mi cro Station datalogger.............................132 4-8. Volumetric moisture content (VMC) from all three ECH2O probes compared to gravimetric m easurements......................................................................................133 4-9. Soil water tension th rough time; treatment T0........................................................134 4-10. Soil water tension through time; treatment T5......................................................134 4-11. Soil water tension through time; treatment T15....................................................135 4-12. Soil water tension through time; treatment T50....................................................135

PAGE 14

xiv 4-13. Soil water tension through time; deta il showing when curves from GMS and tensiometers cross; treatment T0............................................................................136 4-14. Soil water tension through time; deta il showing when curves from GMS and tensiometers cross; treatment T5............................................................................136 4-15. Soil water tension through time; deta il showing when curves from GMS and tensiometers cross; treatment T15..........................................................................137 4-16. Soil water tension through time; deta il showing when curves from GMS and tensiometers cross; treatment T50..........................................................................137 4-17. Relation between the average soil matric potential (SMP) from tensiometers and GMS.......................................................................................................................138 4-18. Relation between the average soil matric potential (SMP) from tensiometers and GMS; excluding GMS data < 10 kPa.....................................................................139

PAGE 15

xv Abstract of Thesis Presen ted to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Master of Science SENSOR-BASED AUTOMATION OF IRRIGATION OF BERMUDAGRASS By Bernard Cardenas-Lailhacar August 2006 Chair: Michael D. Dukes Major Department: Agricultura l and Biological Engineering Turfgrass in landscapes cont ributes to substa ntial cropped area in Florida. New irrigation technologies could improve irrigati on efficiency, promoting water conservation and reducing the environm ental impacts. The objectives of this research were to quantify differences in irrigation water use and turf quality among 1) a soil moisture sensor-based irrigation system compared to a time-based sc heduling, 2) different commercial irrigation soil moisture sensor (SMSs), 3) a time-based scheduling system with or without a rain sensor (RS), and 4) the reliability of tw o commercially available expanding disk RStypes. The experimental area co nsisted of common bermudagrass ( Cynodon dactylon L. ) plots (3.66 x 3.66 m) in a completely randomized design, located in Ga inesville, Florida. The monitoring period for the irrigation treat ments took place from 20 July through 14 December of 2004 and from 25 March through 31 August of 2005. Treatments consisted of irrigating one, two, or seven days a wee k, each with four different commercial SMSs brands. A non-irrigated control and time-bas ed treatments were also implemented. In

PAGE 16

xvi addition, twelve Mini-Click (MC) and four Wi reless Rain-Click (WL) rain sensor models not connected to irrigation were monitored from 25 March through 31 December 2005. For the MCs, three different thresholds we re established: 3, 13, and 25 mm (codes 3-MC, 13-MC, and 25-MC, respectively). No significan t differences in turfgrass quality among irrigation treatments were detected. On aver age, SMS-based treatments reduced irrigation water application compared to time-based tr eatments. The treatment without-rain-sensor (2-WORS) used significantly (52%) more wa ter than the with-rain-sensor treatment (2WRS). Most brands recorded significant irrigation water savings compared to 2-WRS, which ranged from 54% to 88%, for the best performing sensors, and depending on the irrigation frequency. Therefore SMS-systems represent a promising technology, because of the water savings that they can accomplis h, while maintaining an acceptable turfgrass quality during rainy periods (944 and 732 mm of rainfall, for seasons 2004 and 2005, respectively). On average, RS treatments WL, 3-MC, 13-MC, and 25-MC responded close to their rainfall se t points (1.4, 3.4, 10.0, and 24.5 mm, respectively). However, some replications showed erratic behavior th rough time. The number of times that these sensors shut off irriga tion was inversely proportional to the magnitude of their set point (81, 43, 30, and 8 times, respectively) with potential water savings following a similar trend (363, 245, 142, and 25 mm, respectively) Under the relatively wet testing conditions typical to Florida, the payback period could be less than a year, except for 25MC (around 7 years). Consequently, RSs are strongly recommended for use by homeowners as a means to save water, but not when accuracy is required.

PAGE 17

1 CHAPTER 1 INTRODUCTION Turfgrass is the main cultivated crop in Fl orida with nearly four times the acreage as the next largest crop, citrus (Hodges et al., 1994; United States Department of Agriculture [USDA], 2005). Ir rigation of residential, i ndustrial, commercial, and recreational turf areas is necessary to ensure acceptable turf quality. As a consequence of problems related to drought, coupled with a steadily increasi ng demand for water resources, the state of Florida has imposed restrictions on irrigation water use. Water used for turfgrass irrigation, however, remain s to be publicly discussed. The development of Best Management Practices (BMPs) for ir rigation water use in landscapes has become an undeniable strategic, economic, and envir onmental issue for the state. New irrigation technologies could improve irrigation effi ciency, promoting water conservation and reducing the environmental impacts of turfgr ass culture, which is a major component of landscapes in Florida. Water Florida receives an average of around 1400 mm of rainfall a year (National Oceanic and Atmospheric Administration [NOAA], 2003). Unlike many areas dependent on irrigation, annual rainfall in Florida typical ly exceeds evapotranspiration. Nevertheless, irrigation is required because total annual rainfall for Florida t ypically varies both geographically and temporally (USDA, 1981; Carriker, 2000). Such rainfall variation has a direct impact on surface wa ter and groundwater supplies. Lack of rainfall for even a few days causes depletion of moisture in sandy soils commonly found in Florida; along

PAGE 18

2 with reduction of stream flow and groundw ater recharge (Ca rriker, 2000; National Research Council, 1996). Water Demand Florida has the second largest withdrawal of groundwater for public supply in the United States (Solley et al., 1998). Groundwater was the source of more than 88% of the water withdrawn for public supply in 1990 (Carriker, 2000). In 1995, nearly 93% of population in Florida used groundwater as a drinking water source (S olley et al., 1998). Water withdrawals for public supply in Florida have increased rapidly, from 600,000 m3/day in 1950 to 7.3 million m3/day in 1990 (Carriker, 2000). The population served by public-supply systems increased from 5.42 million in 1970 to 11.23 million in 1990 (Marella, 1992). Florida has a fast-growing population with a net inflow of more than 1100 people a day, and ranks as the second largest net ga in in the nation. The population of 17 million in 2004 is projected to exceed 21 million people by 2015, becoming the third most populous state in the nation (United States Census Bureau [USCB], 2004a). The U.S. Census Bureau estimated 156.8 thousand singl e-family housing starts and 56.7 thousand multi-family housing starts in Florida in 2003, accounting for approximately 11% of all new homes constructed in the United States, the largest amount in a ny single state in the U.S. (USCB, 2004b). As urban populations swe ll, pressures on limited supplies of clean water are increasing, and it may become a scarce resource. Water Use Indoor water use per person in the U.S. is relatively constant across all geographic and social lines. Depending on climate, resi dential outdoor water use can account for 22% to 67% of total annual water use (Mayer et al., 1999). The primary use of residential

PAGE 19

3 outdoor water is irrigation. Hi storically, Florida exhibit dry and warm spring and fall weather, as well as sporadic large rain even ts in the summer. These climatic conditions, coupled with low water holding capacity of the soil, make irrigation indispensable for the high quality landscapes desired by homeowners (Haley et al., 2006; National Research Council, 1996). Recent studies in the U.S. indicate that, on average, 58% of potable water is used for landscape irrigation (Mayer et al., 1999). In the Central Florida Ridge, this average has been show to be as high as 74% (Haley et al., 2006). Consequently, proper irrigation water use clearly represents a substantial opportunity for re sidential water savings. Furthermore, residential water use res earch, carried out by Mayer et al. (1999), found that homeowners with a standard lands cape used 77 mm per month, on average, for irrigation purposes in U.S. However, in Cent ral Florida, Haley et al. (2006) found that typical homeowners with a standard lands cape for the region, which consisted of approximately three-quarters tu rfgrass across the irrigated ar ea, used an average of 149 mm per month. Therefore, opportunities that result in better irrigation scheduling by homeowners may lead to substantia l savings in irrigation water use. Water Use Restrictions The Florida Water Resources Act of 1972 established a form of administrative water law that brought all waters of the st ate under regulatory control. Five Water Management Districts (WMDs) were forme d, encompassing the entire state (National Research Council, 1996; Burney et al., 1998). These agencies have the legal authority and financial capacity to ma nage water comprehensively, and can impose conservation and water shortage management (National Research Council, 1996).

PAGE 20

4 The Florida Department of Environmenta l Protection (FDEP, 2002) specifies some water use classifications to be employed when implementing water use restrictions, describing landscape irrigation as “the outdoor irrigation of gr ass, trees and other plants in places such as residences, businesses, golf courses, parks, recreational areas, cemeteries, and public buildings.” When a WMD declares a water shortage, it will impose water use restrictions in different phases depending upon the severity of the shortage. The phase names and their specific goals in water use reduction are: I M oderate: 15%, II Severe: 30%, III Extreme: 45%, and IV Critical: 60%. Moreover, any loca l government has the right to impose even stronger water restrictions (FDEP, 2002). Where there is a year-round watering rule it applies to every one who uses water outdoors–homes, businesses, parks, golf cour ses, etc.–regardless of the water source, whether private well, public utility or surf ace water. However, there are some exceptions to the water restrictions, such as when reclaimed or reuse water is being used (St. John’s River Water Management District [SJRWMD], 2006). Much of Florida is under Phase II water rest rictions. Basically, th is means that lawn watering is limited to two days a week (Wednesdays and Saturdays for odd-number addresses, Thursdays and Saturdays for evens), and restricted to certa in hours to reduce evaporative and wind losses (before 1000 h and after 1600 h in the Orlando area, for instance, or before 0800 h and after 1600 h in parts of South Florida). As of the end of March 2001, the densely populated southernmost part of the state–including Palm Beach, Broward, Dade and Monroe counties–wa s under even tougher regulations. Lawn watering was allowed for only three hours, one day a week (SJRWMD, 2006). Since

PAGE 21

5 1991, there have been water restrictions enforced by the St. Johns River Water Management District (SJRWMD) district where this study wa s carried out. Residential irrigation is limited to two days per week and prohibited between 1000 h and 1600 h, regardless of the water source (SJRWMD, 2006). Violating Florida's water restrictions is punishable with pe nalties of up to $500, with additional fees as app licable. South Florida is enforcing a tough zero-tolerance policy (SJRWMD, 2006). Landscapes in Florida Florida homeowners now maintain more than 1.5 million hectares of lawn with 20,000 hectares of new grass planted every year (American Water Works Association [AWWA], 2005). In an effort to meet Florida's water conservation goals, Volusia County has passed an ordinance requiring new homes to have le ss grass. The ordinance mandates that new yards at homes and businesses have la ndscapes requiring littl e or no irrigation. Homeowners can have up to 75% of the yard with grass if the rest of the landscape retains the original, na tural vegetation without irrigation. Under the ordinance, 50 percent of a new landscape can be irrigated up to 25 mm of water per week (AWWA, 2005). Likewise, in Sarasota County, according to its Ordinance #2001-081, from year 2001, new single and multi-family residences will have no more than 50% of the total irrigated landscape dedicated to high ir rigation water use zones in cluding turf, annuals and vegetable gardens (Sarasota County, 2006). Similar restrictions have been in effect in the Tampa Bay area (Tampa Bay Water, 2005). These types of ordinances that limit plan t type assume that turfgrass water needs are responsible for excessive water applica tion. However, recent research in Central

PAGE 22

6 Florida indicates that excessi ve water application is due to homeowner mis-management of irrigation (Haley et al., 2006). Simila r conclusions were found in the Tampa Bay region, where approximately 30 percent of irrigation water use is wasted due to inefficient irrigation system design, instal lation, operation, or maintenance (Tampa Bay Water, 2005). Irrigation An efficient irrigation schedule is the a pplication of water in the correct amount and only when needed. Under-irrigation a nd over-irrigation can negatively affect turfgrass quality. Over-irrigation tends to have environmentally costly effects because of wasted water and energy, leaching of nutrien ts and/or agricultural chemicals into groundwater supplies, degradation of surface wa ter supplies by sediment-laden irrigation water runoff, and erosion (Ley et al., 2000), and increased evapotranspiration (Biran et al., 1981). Increasing irrigation e fficiency, using just the appr opriate amount of water to irrigate lawns, can be achieved by a number of different methods. Irrigation Timers Irrigation time clock controllers, or timers, are an integral part of an automatic irrigation system. They are an essential tool to apply water in the necessary quantity and at the right time; however, through incorrect programming, timers can result in overirrigation. Time clock controllers have been available for many years in the form of mechanical and electromechanical irrigation timers. These devices have evolved into electronic systems that rely on so lid state and integrated circui ts, so they tend to be very flexible and provide a large number of featur es at a relatively low cost, allowing accurate control of water, while responding to envir onmental changes and plant demands (Zazueta et al., 2002; Boman et al., 2002).

PAGE 23

7 Two general types of timers are used in au tomatic irrigation systems: Open Control Loop systems and Closed Control Loop syst ems. Open Control Loop systems apply a preset action, as is done with simple mechan ical irrigation timers. In a Closed Control Loop (CCL) the system receives feedback from one or more sensors, make decisions, and apply the results of these decisions to the i rrigation system (Zazueta et al., 2002). First, it is necessary to set up a genera l strategy in the timer. Then, the control system takes over and makes decisions of whether or not to ap ply water based on data from the sensor(s). For example, soil moisture sensors can avoid irrigation when adequate soil moisture is already present, rain sensors can prevent i rrigation during or after significant rain, wind sensors can stop the system when a speed-thres hold is surpassed, sensors can be used to detect pressure and shut the system down if th e pump is not primed or to initiate flush cycles in filters, etc. (Zazueta et al., 2002; Boman et al., 2002). The simplest form of a CCL system is to set up a high-frequency irrigation in the timer, which could be interrupted by a soil mois ture sensor. The sensor is wired into the line that supplies power from the timer to the electric solenoid valve (Figure 1-1). The sensor operates as a switch th at responds to soil moisture content. When sufficient soilmoisture is available, the sensor maintains an open circuit between the timer and the solenoid valve. When soil-moisture drops be low a certain threshold, the sensing device closes the circuit. Thus, the irrigation control system can bypass a pre-programmed schedule, or maintain the soil water cont ent within a specified range. These two approaches are known as bypass and on-de mand, respectively (Dukes and MuozCarpena, 2005). Bypass configur ations skip an entire timed irrigation event based on the

PAGE 24

8 soil water status at the beginning of that ev ent or by checking the so il water status at intervals within a time-based event (Muoz-Carpena and Dukes, 2005). Soil Moisture Content Measurement The standard method of measuring soil mois ture content is the thermogravimetric method, which requires oven drying of a known volume of soil at 105 C and determining the weight loss. This method is time consuming and destructive to the sampled soil, meaning that it cannot be used for repetitive measurements at the same location. However, it is indispensable as a standard method for cal ibration and evaluation purposes (Walker et al., 2004). Among the widely used on-site soil mois ture measurement techniques are neutron scattering, gamma ray attenuation, soil elect rical conductivity (i ncluding electrical conductivity probes, el ectrical resistance blocks a nd electromagnetic induction), tensiometry, hygrometry (including electrica l resistance, capacitance, piezoelectric sorption, infra-red absorption and transmi ssion, dimensionally varying element, dew point, and psychometric), and so il dielectric constant (i ncluding capacitance and time domain reflectometry). Reviews on the adva ntages, disadvantages, and basis of these measurement techniques may be found in Sc hmugge et al., 1980; Campbell and Mulla, 1990; Charlesworth, 2000; Ley et al., 2000; Topp, 2003; Muoz-Carpena, 2004; and Walker et al., 2004. Granular matrix sensor The granular matrix sensor (GMS) is a device that measures soil electrical resistance, that can be convert ed to soil water tension (SWT ), either usin g a calibration formula provided in the literature for sandy soils (Irmak and Haman 2001) and silt loam

PAGE 25

9 soils (Eldredge et al., 1993), or calibrating th em for a specific soil type (Hanson et al., 2000b; Intrigliolo and Castel, 2004). The GMS (Figure 1-2) is made of a porous ceramic external she ll with an internal granular matrix material, which approxi mates compressed fine sand, containing two electrodes. A synthetic porous membrane for protection against dete rioration surrounds the matrix material. The GMS includes an internal gypsum cylindrical tablet, which provides buffering against salinity effects that may cause erroneous readings. A stainless steel casing, with holes dr illed in it, surrounds the s ynthetic porous membrane. The GMS operates on the electrical resist ance principle: water conditions in the unit change with corresponding variations in water conditions in the soil, and changes within the block are reflected by differences in resistance between the electrodes. The transmission matrix material was de signed to respond faster than gypsum blocks to SWT in the 0 to 100 kPa range. Some commercial GMSs exhibit good sensitivity to SWT over a range from 0 to 200 kPa. This makes them more adaptable to a wider range of soil textures and irrigation regimes than tr aditional gypsum blocks and tensiometers (Thomson et al., 1996; Charle sworth, 2000). Also, the GMSs are much more stable and have a longer life than gyps um blocks and, compared to tensiometers, require little maintenance and can be left in the soil under fr eezing conditions (Ley et al., 2000). Modern soil moisture sensors The concept of connecting to timers one or more soil moisture sensors (SMSs) to determine irrigation needs, and to automate irrigation systems, has moved forward in recent years. Over the last decade, the SM S industry has advanced dramatically. Two basic reasons can explain this advancement. The first has been the major development of

PAGE 26

10 computer technology (with more powerful, smal ler and economical inte grated circuits). The other phenomenon has been the signifi cant advances in th e application of electromagnetic methods to the measurement of soil water content. These methods make use of the high relative permittivity (dielectric constant) of the water in soil for estimating the water content The relative permittivity of wate r is about 80, whereas the other components in soil, including air, have relativ e permittivities in the ra nge of one to seven. Hence, methods that measure the relative pe rmittivity are effective for the measurement of soil water content (Topp, 2003). Combining the computer technology and th e soil dielectric co ncept has allowed manufacturers to produce a number of different types of inexpensive SMSs for irrigation scheduling. An increasing adopti on of the dielectric methods has been observed, because they are non-destructive, provide almost instantaneous measurements, do not require maintenance, and can provide continuous r eadings through automation. However, they have important differences in terms of calibration requirements, accuracy, cost, installation and maintenance requirements etc. (Muoz-Carpena and Dukes, 2005). The main techniques used by these sensor s can be classified as Time Domain Reflectometry (TDR) and Frequency Domain Reflectrometery (FDR) (Leib et al., 2003). Time domain reflectometry. The speed of an electromagnetic signal passing through a material varies with the dielectr ic of the material. Most TDR instruments operate by sending a step pulse signal down stee l rods (called wave-guides) buried in the soil. The signal reaches the end of the probe s and is reflected back to the TDR control unit where it is detected and analyzed. The time taken for the pulse to return varies with

PAGE 27

11 the soil dielectric, which is related to the water content of the so il surrounding the probe (Topp, 2003). According to Charlesworth (2000) and Ed is and George (2000), TDR instruments give the most robust soil water content data with little need fo r recalibration between different soil types. An important advantage of TDRs in turfgrass irrigation management, is that accurate measurements may be made near the surface compared to techniques such as the neutron probe (Ley et al., 2000). Frequency Domain Reflectometry. Frequency domain reflectometry (FDR) measures the soil dielectric by placing the soil (in effect) between two electrical plates to form a capacitor. Hence ‘capacitance’ is the term commonly used to describe what these instruments measure. When a voltage is applie d to the electric plates a frequency can be measured. This frequency varies with the soil dielectric (Charlesworth, 2000). In spite of the advances and advantages of these modern SMSs, when comparing the performance of different brand/types, si gnificant differences were found in respect to set-up requirements, accuracy, data interpretatio n, maintenance, and in itial cost (Ley et al., 2000) and the ability to repeat measurements accurately over time and under various moisture regimes after initial calibration (Yoder et al., 1998). Controllers Modern commercially available SMS-system s include a controller. This piece of equipment is the one that sends the signal to the buried SMS and reads the soil moisture content. The controller has an adjustable threshold (Figure 1-3), which can be set between relatively dry to wet soil moisture conditions; depending on the plant material, soil type, depth-installation of the SMS, etc. In general, manufacturers recommend setting the thresholds 24 hours af ter a significant rainfall even t or after an irrigation that

PAGE 28

12 filled the soil profile with water to field cap acity. The controller is connected in series with the residential irrigati on timer and acts as a switch depending on the pre-set soil moisture threshold. Automatic Control of Irrigation An automatic SMS-based irrigation system seeks to maintain a desired soil moisture range in the root zone that is op timal or adequate for plant growth and/or quality. This type of system adapts the amount of water applied according to plant requirements without managers having to undertake daily monitoring or make adjustments according to actual weather c onditions (Muoz-Carpena and Dukes, 2005; Pathan et al., 2003). The continuous monitoring of the soil moisture status becomes particularly important in sandy soils. A wide range of app lications to automatical ly control irrigation events has been investigated in coarse text ured soils. In Florida, switching tensiometers have been studied for agricultural produc tion (Smajstrla and Koo, 1986, Clark et al., 1994; Smajstrla and Locascio, 1994; Muoz-Car pena et al., 2003, M uoz-Carpena et al., 2005), and for maintaining bermudagrass tu rf (Augustin and Snyder, 1984). Although they found water savings, these investiga tions suggest that tensiometers require calibration and frequent maintenance, up to twice per week. Consequently, the adoption of this technology will not lead to automa tically controlled irri gation since it will not eliminate human interaction in irrigation management. Other types of sensors have been adapte d to automate irrigation based on soil moisture status. Nogueira et al. (2002) used TDR sensors to maintain soil moisture within two preset limits (upper and lower soil moistu re thresholds). Dukes and Scholberg (2005) and Dukes et al. (2003) found 11% and 50% in water savings, without diminishing yields

PAGE 29

13 on sweet corn and green bell pepper, usi ng TDR probes and a commercially available dielectric sensor, respectively. Granular matrix sensors (GMSs) have also been used to automatically irrigate agricultural products (Muoz-Carpena et al ., 2003; Shock et al., 2002) and, as with other solid-state sensors, do not require as much maintenance as tensiometers. Although TDR and GMS, as well as similar types of sensors, have been successfully used in agricultu re, they have found limited use in residential landscape irrigation (Qualls et al., 2001). Rain Sensors A rain sensor (RS), also called rain s hut-off device (Figure 14), is a piece of equipment designed to interrupt a scheduled cycle of an automatic irrigation system controller when a specific am ount of rainfall has occurred and, depending on the weather conditions, after the said rainfall (Dukes and Haman, 2002b; Hunter Industries Inc., 2006). Florida law requires a RS device on all automatic lawn sprinkler systems (Florida Statutes, Chapter 373.62, n.d.). The original text said: “Any person who purchases and installs an automatic lawn sprinkler system after May 1, 1991, shall in stall a rain sensor device or switch which will override the irriga tion cycle of the sprinkler system when adequate rainfall has occurred.” In 2001, this Chapter was amended to require the owner not only to install, but also to maintain and operate a RS device or switch (Florida Statutes, 2001). Moreover, some local laws al so require older systems to be retrofitted with rain shut-off switches (SJRWMD, 2006). Florida is the only state in the nation with an overall RS statute. However, recently, Georgia Gov. Sonny Perdue has signed into law H1277 requiring RSs on newly installed

PAGE 30

14 irrigation systems in the Atlanta metro region. The new law affects systems installed after January 1, 2005 (AWWA, 2004). As with soil moisture sensors, rain se nsors can be connected to any automatic irrigation system controller and mounted in an open area where they are exposed to rainfall. The new irrigation timers have a sp ecial connection, which allows a RS to be attached directly. If it is not available, or the sensor does not work with a given timer, the sensor can always be “hard-wired” into the co ntroller, wiring the RS in series with the common wire. When a specific amount of rain fall has occurred, the RS will interrupt the irrigation system common wire, which disables the solenoid valves until the sensor dries (Dukes and Haman, 2002b). Figure 1-4 shows a simple and low cost RS. Rain causes the hygroscopic porous disks in the device to swell and open a micro-switch (Fig ure 1-5). The switch remains open as long as the disks are swollen. When th e rain has passed and the disks dry out, the switch will close again. According to Dukes and Haman (2002b), the use of rain sensors has several advantages: they conserve water, preventing irrigation after recent rain events; reduce wear on the irrigation system, because the system runs only when necessary; reduce disease and weeds development, by elimin ating unnecessary irrigation events; help protect surface and groundwater by reducing the runoff and deep percolation that carries pollutants, such as fertilizers and pesticides ; and, finally, RSs save money, because they reduce utility bills and maintenance costs. Rain sensors should be mounted on any surface where they will be exposed to unobstructed rainfall, but should not be in the path of sprinkler sp ray. These sensors are

PAGE 31

15 typically installed near the roofline on the side of a building, but manufacturers recommend mounting it in a location that r eceives about the same amount of sun and shade as the turf (Hunter Industries Inc., 2006). Irrigation and Turfgrass Quality Under-irrigation and over-irrigation can negati vely affect turfgrass quality. It has been reported that, deeper and reduced irrigation frequency impr oves turfgrasses quality. Augustin and Snyder (1984) conc luded that this practice te nds to reduce N leaching in sandy soils, increasing N utiliz ation, resulting in a better color rating (bet ter quality). Bonos and Murphy (1999) reported an increase in a Kentucky bluegrass ( Poa pratensis L. ) cultivar root growth as drought stress was imposed. Recently, Jordan et al. (2003) found that bentgrass irrigated every 4 days produced a signif icantly denser and deeper root system, a higher shoot density, and greate r overall plant healt h, resulting in better turf quality, than grass watered every 1 or 2 days (even under putting green management conditions). McCarty (2005) su mmarizes that drier conditio ns slow shoot growth and increase root growth and leaf water content. Moreover, limitations to establishment and survival of some turfgrass weeds (Colbaugh and Elmore, 1985; Youngner et al., 1981), and reduction of some pathogens severity (Davis and Dernoeden, 1991; Kackle y et al., 1990) has been associated with deep, infrequent irrigation.

PAGE 32

16 Figure 1-1. Components of an automated irrigation system: A) timer, B) power supply, C) soil moisture sensor-controller circui try, D) soil moisture sensor, and E) solenoid valve.

PAGE 33

17 Figure 1-2. Granular matrix sensors (GMS)

PAGE 34

18 Figure 1-3. Components of an au tomated irrigation system. 1) Timer, and 2) soil moisture sensor-controllers from different brands.

PAGE 35

19 Figure 1-4. Rain shut-off switch. Figure 1-5. The expanding material of a rain shut-off switch.

PAGE 36

20 CHAPTER 2 SENSOR-BASED AUTOMATION OF IRRIGATION OF BERMUDAGRASS Introduction Turfgrass in landscape applications is the most extensively cultivated crop in Florida (Hodges et al., 1994; USDA, 2005). Irrigation of residential, industrial, commercial, and recreational turf areas is co mmonly employed to ensure acceptable turf quality. As a consequence of problems re lated to drought, coupled with a steadily increasing demand for water, the state of Flor ida has imposed restrictions on irrigation water use. The development of Best Manage ment Practices (BMPs) for irrigation water use in turf has become an undeniable strate gic, economic, and environmental issue for the state. New irrigation technologi es could improve irrigation efficiency promoting water conservation and reducing the environmental impacts of th e landscapes, which are often composed of turfgrass as a majo r portion of the irrigated area. Florida receives an average of around 1400 mm of rainfall a year, which typically exceeds evapotranspiration. Nevertheless, irri gation is required because total annual rainfall for Florida typically varies both geographically and temporally (USDA, 1981; Carriker, 2000; NOAA, 2003), and lack of rain fall for even a few days causes depletion of moisture in Florida's predominately sa ndy soils (Carriker, 2000; National Research Council, 1996). Florida has the second largest withdrawal of groundwater for public supply in the United States. In 1995, nearly 93% of popula tion in Florida used groundwater as a drinking water source (Solley et al., 1998). Florida has a fa st-growing population with a

PAGE 37

21 net inflow of more than 1100 people a day. By 2025, it is projected to be the third most populous state in the nation (Office of Econo mic and Demographic Research [ODR], 2006; USCB 2004a). The U.S. Census Bureau estimated that Florida accounted for approximately 11% of all new homes construc ted in the U.S. in 2003, the largest amount in any single state in the U.S. (USCB, 2004b), the majority of them with in-ground irrigation systems1 (Tampa Bay Water, 2005). As urban populations swell, pressures on limited supplies of clean wate r are increasing. Even saltwa ter intrusion in groundwater from the Floridan aquifer have been found in coastal Hillsborough, Manatee and Sarasota counties (Southern Water Use Caution Ar ea Recovery Strategy [SWUCA], 2006) The primary use of residential outdoor wa ter is irrigation. Recent studies in the U.S. indicate that, on average, 58% of potable water is used for landscape irrigation, that households that use automatic timers to contro l their irrigation systems used 47% more water outdoors than those w ithout timers, and that homes with in-ground sprinkler systems use 35% more water outdoors than those without in-ground systems (Mayer et al., 1999). In the Central Florida Ridge, the pot able water used for landscape irrigation is as high as 74%, with an average of 64% (Hal ey et al., 2006), and ev en when irrigation is restricted to two days a week and from 1000 h to 1600 h (SJRWMD, 2006), typically homeowners tended to over-irr igate (Haley et al., 2006). Over-irrigation or under-irrigation can nega tively affect turfgr ass quality. It has been reported that deeper and reduced irrigation frequency impr oves turfgrass quality. Augustin and Snyder (1984) conclu ded that this practice tended to reduce N leaching in 1 57% and 85% of new homes built in Pasco and Hills borough counties, respectively, have in-ground irrigation systems. Actual percentages may be higher since many homeowners install irrigation systems after moving into the home. In the Tampa region, 70% of homes are estimated to have in-ground irrigation.

PAGE 38

22 sandy soils, increasing N utiliz ation, resulting in a better color rating (bet ter quality). Bonos and Murphy (1999) reported an increase in a Kentucky bluegrass ( Poa pratensis L. ) cultivar root growth as drought stress was imposed. Recently, Jordan et al. (2003) found that bentgrass irrigated every 4 days produced a significantly larger and deeper root system, a higher shoot density, and an overa ll plant health–resulting in greater turf quality–than that watered every 1 or 2 da ys (even under golf putting green management conditions). McCarty (2005) su mmarizes that drier conditio ns slow shoot growth, and increase root growth and leaf water content. Moreover, limitations to the establishment and survival of some turfgrass weeds (Colbaugh and Elmore, 1985; Youngner et al., 1981), and reduction of some pathogens seve rity (Davis and Dernoeden, 1991; Kackley et al., 1990) have been associated with deep, infrequent irrigation. He nce, better irrigation scheduling by homeowners may lead to impr oved turfgrass quality coupled with potential savings in irrigation water use. Over the last decade, the soil moisture sensor (SMS) industry has advanced dramatically. Two basic reasons can explain this advancement. The first has been the major development of computer technology (with more powerful, smaller and more economical integrated circuits), and the other phenomenon has been the significant advances in the application of electromagnetic methods to the measurement of soil water content. These methods make use of the high relative permittivity (die lectric constant) of the water in soil for estimating the water cont ent. The relative permittivity of water is about 80, whereas the other components in soil, including air, have relative permittivities in the range of one to seven. Hence, met hods that measure the relative permittivity are effective for the measurement of the soil water content (Topp, 2003).

PAGE 39

23 Combining the computer technology and th e soil dielectric co ncept has allowed manufacturers to design and produce a number of different types of inexpensive SMSs for irrigation scheduling. However, when comparing the performance of different brand/types of sensors for measurement of soil moisture, differences were found. For example, Ley et al. (2000) found significant differences betw een sensors with respect to set-up requirements, accuracy, data interpretatio n, maintenance, and in itial cost. Yoder et al. (1998) obtained differences related w ith error, accuracy, re liability, durability, installation factors, and the ability to repe at measurements accurately over time and under various moisture regimes after initial calibration. Automation of irrigation systems, based on SMSs, has the potential to provide maximum water use efficiency, by maintaining soil moisture between a desired range that is optimal or adequate for plant growth a nd/or quality; allowing irrigation only when necessary (Muoz-Carpena and Dukes, 2005). A wide range of applications to automati cally control irrigati on events have been investigated in coarse textured soils. In Florida, switching te nsiometers have been studied for agricultural production (S majstrla and Koo, 1986, Clark et al., 1994; Smajstrla and Locascio, 1994; Muoz-Carpena et al., 2003; Muoz-Carpena et al., 2005), and for maintaining bermudagrass turf (Augustin and Snyder, 1984). Although they found water savings, these investigations suggest that tensiometers re quire calibration and frequent maintenance, up to twice per week. Conseque ntly, the adoption of this technology will not lead to an automatically controlled irrigation system, since it will not eliminate human interaction in irrigation management.

PAGE 40

24 Other types of sensors have been adapte d to automate irrigation based on soil moisture status in Florida. Nogueira et al (2002) used TDR sensors to maintain soil moisture within two preset limits (upper and lower soil moisture thresholds). Dukes and Scholberg (2005) and Dukes et al. (2003) found 11% and 50% in water savings–without diminishing yields–using TDR probes on sweet corn, and a commercially available dielectric sensor on green bell pepper, resp ectively. Granular ma trix sensors (GMSs) have also been used to automatically irriga te agricultural products (Muoz-Carpena et al., 2003; Shock et al., 2002) and, as with other so lid-state sensors, do not require as much maintenance as tensiometers. Although SMSs have been successfully used in agriculture, th ey have found limited use in residential lands cape irrigation and further invest igation is required to provide evidence of their potential us e in this area. A study using GMSs to control urban landscape irrigation in Colorado, used 533 mm of water for irrigation when compared to the theoretical requirement of 726 mm, a re duction of 27% (Qualls et al., 2001). Since 1991, Florida law requires a rain se nsor device or switch hooked up to all automatic lawn sprinkler systems (Florida Statutes, Chapter 373.62, n.d.). A rain sensor (RS) is a piece of equipment designed to in terrupt a scheduled cycle of an automatic irrigation system controller when a specifi c amount of rainfall has occurred (Dukes and Haman, 2002b; Hunter Industries Inc., 2005). Be nefits and advantages of its use are similar to those of SMSs, and have been summarized by Dukes and Haman (2002b). Even when this law has been in effect fo r a long time, and RSs have been commercially available for many years, little evidence related to their usefulness and/ or to quantify their water savings exists

PAGE 41

25 The goals of this research were to find out if different SMS-systems (sensor with a proprietary controlle r) could reduce irrigation water application–while maintaining acceptable turf quality–compared to current prac tices. The objectives of this experiment were to quantify differences in irrigation wa ter use and turf quality between: 1) a SMSbased irrigation system compared to a tim e-based scheduling, 2) different commercial irrigation SMSs, and 3) a time-based sche duling system with or without a RS. Materials and Methods The experimental area was located at the Agricultural and Biological Engineering Department facilities, Univers ity of Florida, Gainesville, Florida; on an Arredondo fine sand (loamy, siliceous, semiactive, hyperthermic Grossa renic Paleudults) (Thomas et.al, 1985; USDA, 2003). This soil has a field capac ity of 7% (Figure 2-1), as determined from repacked soil columns (see Chapter 4 for methodology details). Seventy-two 3.66 m x 3.66 m plots were established on a field covered with common bermudagrass ( Cynodon dactylon L. ). Each plot was spri nkler irrigated by four quarter-circle pop-up spray heads, with an appl ication rate of 38 mm /hr and regulated at 172 kPa (Hunter 12A, Hunter Indu stries, Inc., San Marcos, CA ). Much of the irrigation hardware was in place from a previous rese arch project; however, extensive renovations were performed to make the equipment serviceable. Plots were mowed twice weekly at a height of 5.5 cm. Chemicals were applied as needed to control weeds and pests. Nutrient applications were made using ammonium sulfate (21-0-0), at a N rate of 50 kg ha-1, on April and May of 2004, before the beginning

PAGE 42

26 of the experiment. Then, a granulated N c ontrolled-release fertilizer (Polyon, PTI, Sylacauga, AL) 2 was applied at a rate of 180 kg ha-1, on July 2004 and April 2005. Four commercially available SMSs were selected for evaluation (Figure 2-2): Acclima Digital TDT RS-500 (Acclima In c., Meridian, ID), Watermark 200SS-5 (Irrometer Company, Inc., Riverside, CA), Ra in Bird MS-100 (Rain Bird International, Inc., Glendora, CA), and Water Watcher D PS-100 (Water Watcher, Inc., Logan, UT), codified as AC, IM, RB, and WW, respectivel y. In order to find similar outcomes to those that homeowners would encounter, sensors were not calibrated, and were used directly “out of the box.” Each one of these SMSs systems includes a SMS and a controller (Figure 2-3). The controller’s thresholds can be adjusted be tween “dry” and “wet” on the RB (on a 1 to 8 scale), and between “moist” and “dry” on the WW (on a -3 to 3 scale). The IM can be set at a specific soil water tension (kPa) and th e AC can be set direct ly to a specific soil volumetric moisture content (VMC ), expressed in percent. As recommended by manufacturers, all cont roller thresholds, except for the AC, were set 24 hours after a significant rainfall event (on 20 July 2004, after four days of rain with a total of 107 mm) that filled the so il profile with water. On RB controllers, the thresholds were set by adjusting the dial until the LED turned off and on. On the WW, initially the unit could not be calibrated since the soil moisture was outside the range of the controller. After discussion with the manu facturer, a resistor was added between the solenoid valve wire and the valve common wi re. The calibration pro cedure consisted of activating the reset button, which allowed its au to-calibration. The IM controller was set 2 The mention of trade and company names is for th e benefit of the reader and does not imply an endorsement of the product.

PAGE 43

27 at number 1 (equivalent to 10 kPa, and approxi mately to field capacity, according to the manufacturer), whereas the AC controller was set on their display at a VMC of 7%, based on the measured soil water release curve of th e soil. All these contro llers were connected in series with typical residential irri gation timers (see description of timers under Treatments sub-heading). Treatments Two basic types of treatments were defi ned: SMS-based treatments, and time-based treatments (Table 2-1). In the SMS-based treat ments, all four brands were tested with three irrigation frequencies: one two, and seven days per week (1 d/w, 2 d/w and 7 d/w, respectively). The 1 d/w and 2 d/w wateri ng frequencies represent typical watering restrictions imposed in Florid a (FDEP, 2002; SJRWMD, 2006). Within the time-based treatments, a fre quency of 2 d/w was defined (the most common in Florida, and current watering restriction in the area of study). Two treatments were connected to a rain sensor (2-WRS and 2-DWRS), to simulate requirements imposed on homeowners by Florida Statutes (Chapter 373.62, n.d.). The rain sensor (Figure 2-4) (Mini-click II, Hunter Industries, Inc., San Marcos, CA) was set at 6 mm rainfall threshold. A without-ra in-sensor treatment (2-WORS) was also included, in order to simulate homeowner irrigation systems with an absent or non-f unctional rain sensor. Finally, a non-irrigated treatme nt (0-NI) was also implemente d as a control for turfgrass quality. All experimental treatments were repeat ed four times, for a total of 64 plots, in a modified completely randomized design3. 3 See Dry-Wet Analysis subheading for details

PAGE 44

28 The weekly irrigation depth was set to re place the historical ET-based irrigation schedule recommended by Dukes and Haman ( 2002a) for the area where this experiment was carried out (Table 2-2) All treatments were progr ammed to have the equal opportunity to apply the same amount of ir rigation per week, except for treatments 2DWRS (deficit-with-rain-sens or, 60% of this amount), and 0-NI (non-irrigated). The irrigation depths were adjusted monthly. The irrigation cycles were programme d on two ESP-6, and three ESP-4Si model timers (Rain Bird International, Inc., Glendora, CA) (Figure 2-3). They were programmed to start between 0100 and 0500 h, w ith the purpose of diminishing wind drift and decreasing evaporation. Uniformity Test An irrigation uniformity test measures the relative distribution application of water depth over a given area. This concept results in a numeric value to quantify the variability in depth of sprinkler irrigati on over a target area. Two met hods have been developed to quantify uniformity: distribution uniformity ( DU) and Christiansen’s coefficient of uniformity (CU). According to Merriam and Keller (1978), the low-quarter irri gation distribution uniformity (DUlq) can be calculated with the following equation: tot lq lqD D DU [2-1] where lqD is the lower quarter of the average of a group of catch-can measurements, and totD is the total average of a group of catch-can measurements. This method emphasizes the areas that receive the least irrigation by focusing on the lowest quarter. Although a

PAGE 45

29 system may have even distribution, over-irr igation can occur because of mismanagement (Burt et al., 1997). On the other hand, the CU treats over-i rrigation and under-irrigation equally as compared to the mean, and can be calculated by the Christiansen (1942) formula: n i i n i iV V V CU1 11 [2-2] where iVequals the volume in a given catch-can, and Vrefers to the mean volume. To carry out the uniformity test on th e field, 16 catch-cans on a 0.9 m x 0.9 m square grid pattern were placed on each plot. To minimize edge effects, this grid was positioned 0.4 m inside the plot boundaries (Figure 2-5). The cans had an opening diameter of 15.9 cm and a depth of 20.3 cm. Pressure at the two farthest plots was verified with a pressure gauge. The system wa s set to run for 35 min, to ensure that the average water application dept h was at least 13 mm. Wind velo city during th e test period was measured with a hand held anemometer The American Society of Agricultural Engineers (ASAE) standards (ASAE, 2000) al low uniformity testing with wind speeds up to 5 m/s. However, if wind was over 2.5 m/ s or the distribution was affected by wind gusts, the test was discontinued. Dry-Wet Analysis In accordance with manufacturer recomme ndations for the products tested, the SMS should be buried in the driest zone of a multiple-zone system. Thus, that particular zone would receive sufficient irrigation, whereas the other zo nes would be slightly overirrigated. Accordingly, to identify the driest and wettest plots in the experimental area, a

PAGE 46

30 survey was carried out on each plot, before the beginning of the experiment. In addition, because a total of 64 plots were required, this analysis was used to discard 8 plots from a pool of 72 plots available. On 12 March 2004, after 14 days without rain fall, a relatively “dry” soil moisture condition was evident. The VMC was measured in each plot by means of a hand held TDR device, which measured the moisture in the top 20 cm (Field Scout 300, Spectrum Technologies, Inc., Plainfield, IL). Measuremen ts were taken at five locations in the center 1 m X 1 m of each plot and aver aged. On 17 March 2004, 24 hr after a 23 mm rainfall filled the soil profile, the VMC on a “wet” condition was measured as well. After selecting the driest plots, the SMS that controlled a particul ar treatment was buried in the center of one of these plots, thereby controlling all four rep lications. In all cases, SMSs were installed in the top 7-10 cm of the soil, where most of the roots were present. Plot Irrigation Management and Data Collection Figures 2-6 to 2-10 show the set up of the different control features of the experiment. Each plot was managed individua lly, and data were collected independently from each plot as well. Pulse-type positive displacement flowmeters (PSMT 20mm x 190mm, Amco Water Metering Systems, Inc ., Ocala, FL) were connected to nine AM16/32 multiplexers (Campbell Scientific, Logan, UT), which were hooked up to a CR 10X model datalogger (Cam pbell Scientific, Logan, UT), to continually measure irrigation volume and frequency applied to e ach plot (Figure 2-10). In addition, meters were read manually each week. Weather data were collected by an automated weather station (Campbell Scientific, Logan, UT), located within 1 m of the experimental site (Fig ure 2-11). Measurements, made every fifteen minutes, included air te mperature, relative humidity, wind speed,

PAGE 47

31 wind direction, solar radiation, barometric pressure, and soil heat flux. Rainfall was recorded continuously by a manual rain ga uge during 2004 and by a tipping bucket rain gauge in 2005. Reference evapotranspiration (ETo) was calculated from the Penman–Monteith equation described in FAO-56 (A llen et al., 1998) as follows: ETo 2 234 0 1 273 900 408 0 u e e u T G Ra s n [2-5] 23 237 3 237 27 17 exp 6108 0 4098 T T T [2-6] Rn = Rns Rnl [2-7] Rnl = 35 0 35 1 14 0 34 0 24 min, 4 max, so s a K KR R e T T [2-8] Rns = (1)Rs [2-9] Rso = (0.75 + z(2 x 10-5))Ra [2-10] Ra = ) cos( ) cos( ) sin( ) sin( ) sin( ) 60 ( 24 s s r scd G [2-11] dr = 1 + 0.033 cos J 365 2 [2-12] = 0.409 sin 39 1 365 2 J [2-13] s = arcos [-tan( )tan( )] [2-14] es = 2 ) (T e ) (T emin o max o [2-15]

PAGE 48

32 ea = 2 100 RH ) (T e 100 RH ) (T emin max o max min o [2-16] eo(T) = 3 237 27 17 exp 6108 0T T [2-17] where: ETo = Potential evapot ranspiration, mm/day slope of the vapor pressure curve, kPa oC-1 Rn = net radiation of the turf surface, MJ m-2 day-1 Rnl = net outgoing longwave radiation, MJ m-2 day-1 Rns = net solar or shortwave radiation, MJ m-2 day-1 Rso = clear sky solar radiation, MJ m-2 day-1 Rs = measured solar radiation W/m2 x 0.0864, MJ m-2 day-1 Ra = extraterrestrial radiation, MJ m-2 day-1 G = measured soil heat flux density, MJ m-2 day-1 Gsc = solar constant, 0.0820 MJ m-2 min-1 T = measured air temperat ure at a 1.5 m height, oC u2 = measured wind speed at a 2 m height, m s-1 es = saturation vapor pressure, kPa ea = actual vapor pressure, kPa eo(T) = saturation vapour pressure at air temperature, kPa RH = relative humidity at 1.5 m height, % dr = inverse relative distance Earth-Sun s = sunset hour angle, rad = solar declination, rad = psychrometric constant, 0.067 kPa oC-1 = Stefan-Boltzmann constant, 4.903 x 10-9 MJ K-4 m-2 J = Julian day = latitude, radians The soil moisture content was monitored with a capacitance soil water probe (20 cm ECH20, Decagon Devices, Inc., Pullman, WA) inst alled in each plot (Figure 2-12). These probes were connected to HOBO microloggers (Onset Computer Corp., Bourne, MA). Each HOBO datalogger had four probe s connected to it, and readings were recorded every 15 minutes duri ng 2004 and every hour in 2005.

PAGE 49

33 Before the beginning of the expe riment, calibration of the ECH2O probes was performed at the research site using the thermogravimetric soil sampling method (Gardner, 1986). Four probes connected to a HOB O datalogger were installed in the field. Undisturbed soil samples were collected from the field at less than 20 cm from the probes, and at the same depth where the probe s were placed. Samples were taken from a saturated through a dry condition. When each sample was removed, date and time was recorded. The volumetric soil water content of each sample was then compared to the ECH2O probe readings at the same date a nd time when the samples were taken (see Chapter 4 for methodology details) and a site -specific calibration curve was developed. Turfgrass quality was visually assessed and rated using a scale of 1 to 9, where 1 represents brown, dormant or dead turf, a nd 9 represents the best quality (Skogley and Sawyer, 1992). A rating of 5 was considered the minimum acceptable turf quality for a homeowner. Ratings were carried out in July, October and December of 2004, and in April, May and July of 2005. The data were obtained from 20 July through 14 December of 2004–when the turfgrass went dormant due to cool temper atures, and irrigation was discontinued–and from 25 March through 31 August of 2005. Data Analysis Statistical data analyses were perfor med using the general linear model (GLM) procedure of the Statistical An alysis System software (SAS 2000). Analysis of Variance was used to determine treatment differences and Duncan's Multiple Range Test was used to identify mean differences. The combined data from both years were analyzed. When interactions between years a nd other parameters were dete cted, each year's data were analyzed separately and, when needed, a m onthly data analysis was made as well.

PAGE 50

34 Results and Discussion Uniformity Tests The uniformity tests resulted in a wide range of DUlq values across the plots (15% to 78%), with an average of 52% that, acco rding to the Irrigation Association (2003) overall system quality ratings, is considered “fair.” Obvious problems such as leaks and broken heads were repaired prior to testing, but in some cases problems were discovered as a result of testing (Figure 2-13, plots A 8, A12, B9, and D1) and action was taken to correct the problem. Baum et al. (2003) pe rformed uniformity tests on irrigation systems of homes in Central Florida having spray heads. That research found an average DUlq of 41%, with a range of 12% to 67%. The average CU for all the plots was 71%, w ith a range of 50% to 89%. Baum et al. (2003), found a CU average of 59%, with a range between 50% and 72%. Therefore, these experimental plots had a better distribut ion application of wate r depth, expressed as DUlq and CU, than actual spray irrigation zone s on homes sampled in Central Florida. It is interesting to mention that, consid ering all the catch-cans of the experiment responsible for the lowest read ings, 99% of them were placed on the edges of the plots, indicating that substantial e dge effects occurred in the testing. This is common for sprinkler irrigation systems a nd did not negatively impact the results, because soil moisture and turf quality ratings of each plot were always taken inside this perimeter. In addition, this situation tended to minimize th e effect of irrigation overlapping between plots. Dry-Wet Analysis The data collected during the dry-wet anal ysis are shown in Figures 2-14 and 2-15. The statistical analysis (Appendix B, Dry) re vealed that at the dry condition, plots A11

PAGE 51

35 and A12 were similar, but had significantly higher VMC than the other plots. Under the wet condition, again plots A11 and A12 were si gnificantly different from the rest, also having a higher VMC. Therefore, plots A11 and A12 were discar ded for this experiment because they were too “wet” (Figures 2-14 and 2-15). On the other hand, plots A4, A5, A7, A8, F11, and F12 were also discarded, because they appeared to have the lowest VMC values of all plots, coupled with a comparatively lowe r turfgrass quality before the beginning of the experiment. These discarded plots appear in red in the plot plans shown in Figures 2-14 and 2-15. Consequently, a total of 64 plots were left for this research. When a statistical analysis was performed (Appendix B, We t) with the 64 plots included, the only significant difference was that F1 and E1 show ed higher values at the wet condition than the rest, so they were discarded as locations for SMS placement. The green plots shown in Figures 2-14 a nd 2-15 were selected to bury the SMS for the control plots, because they appeared to be similarly dry and on the dryer end, and because of the practical convenience of bur ying the cables of the different SMS on the same trench, at the same time, and closer to the control board (Fi gure 2-8). A statistical analysis on these plots (Appendix B, Plots w ith SMS) indicated th at they were not statistically different (P>0.94). Figure 216 shows the plot plan containing all the treatments and repetitions, in a modified completely randomized design, where the same color depicts treatment repetitions (plots showing an X were the discarded ones). Rainfall Both 2004 and 2005 were rainy years (Figur es 2-17 and 2-18), with high frequency rainfall events and a large amount of cumulative precipitatio n, which is not uncommon in this region. During 2004, a tropi cal storm and two hurricane s–Frances and Jeanne–passed

PAGE 52

36 over the research area duri ng the experiment, resulting in 159, 286, and 157 mm of rainfall, respectively. Year 2005 broke all records for th e number of hurricanes and named tropical storms in U.S., but none of them directly hit the area where the experiment was carried out. Nonetheless, during the data collection period of 2005, 40% of the days had rainfall events, with a considerable amount of precipitation, 732 mm, a nd an average of 135 mm/month (Figure 2-18). In the course of 2004, (Figure 2-17) even when it rained less frequently, 31% of the days, the cumulative ra infall for the experimental period was even larger, with 944 mm, and 190+ mm/month on aver age. However, most of this rainfall (530 mm, or 56%) occurred during the tropi cal storm and the two hurricanes. If these events were not considered, an average of 84 mm/month fell during 2004. Figure 2-17 shows that in 2004 most of th e rain fell during August and September (793 mm), and the least rain fell in Octobe r and November (116 mm). A relatively “dry” period occurred from 21 October to 24 Novemb er (35 days), with only two small rainfall events of 1.5 and 2.5 mm. A similar situa tion happened in 2005 (Figure 2-18) during April and May, when 223 mm fell over 17 rainfa ll events. On the ot her hand, June, July, and August were rainier months with 478 mm on 44 rainfall events. Irrigation Events Figures 2-19 and 2-20 show the evolution of the cumulative number of irrigation events allowed per treatment in 2004 and 2005, respectively. There were a greater number of irrigation events as the irrigati on frequency increased (Parts B vs. C vs. D, from Figures 2-19 and 2-20). The treatm ent without-rain-sensor (2-WORS) was programmed to run 2 d/w, independently of th e weather and/or soils moisture conditions and, as expected, irrigation cycles were not bypassed. However, to avoid possible damage

PAGE 53

37 to the equipment, power was turned off and no data were collected from 26 September through 30 September 2004, due to hurricane Jea nne. This is reflected on Figure 2-19, Part A, when through these dates the slope of the curves looks hor izontal (no irrigation events). Nevertheless, this period was rainy and very short in proportion, so final results should not be significantly affected. Treatments 2-WRS and 2-DWRS were contro lled by the same rain sensor and set to run the same days so, not surprisingly, they overrode the same amount of irrigation cycles, and only one line can be seen for both treatments (Figure 2-19, Part A). In spite of this result, it is important to know what proportion of the scheduled irrigation cycles (SIC) were finally overridd en by the different treatments (Table 2-3). The statistical analysis (P<0.0001) showed that as expected, 2-WORS was different from the rest of the treatments (overriding 0% of the SIC), and that 2-WRS and 2-DWRS were not statistically different. These last two treatments bypassed 30% and 37% of the possible irrigation events in 2004 and 2005, respectively, and more than a third as an average for both seasons. Regarding the SMS-based treatments, th ere was not a clear difference between them, except for the IMs. Sensors from br ands AC, RB, and WW, overrode significantly more SIC than 2-WRS and 2-DWRS, ranging from 70% to 92% in total, and considering all frequencies tested. The proportion of SIC overridden by IMs, however, ranged from 32% to 50% the first year, from 24% to 70% in 2005, and from 28% to 56% for both years together. In the first year, 2-IM ove rrode almost the same proporti on of irrigation cycles than the control 2-WRS (32% vs. 30%, respectiv ely). In 2005, however, 2-IM was the only

PAGE 54

38 treatment that resulted in more irrigation cy cles than the control treatment, overriding only 24% of the possible ones compared to 37% by 2-WRS. As a result, 2-IM was not statistically different from 2-WRS. Th e other two IM treatments did not differ statistically, but overrode more irrigation cycles compared to 2-WRS and less than the rest of the SMS-based treatments. These results suggest that IMs, ex cept for the said 2IM in 2005, would be able to respond to rainfall events at least in a similar proportion than a rain sensor device set at 6 mm. In order to corroborate the effectiveness of the SMSs, it was important to detect when actual irrigation cycles occurred a nd how were they related to rainfall and evapotranspiration conditions. Figures 2-21 and 2-22 show the maximum weekly irrigation water requirement–or weekly rainfall – ETo difference (RED)–fo r years 2004 and 2005, respectively. In 2004, August and September had only one week each with a negative RED. However, after 2 October, eight of eleven weeks showed a ne gative RED. On the other hand, in 2005, it can be seen that every month had at least one deficitary week. However, the longest negative REDs happened in April, May, and July of 2005, with three, four, and two consecutive weeks, respectively. Table 2-4 shows the percent of irriga tion cycles allowed by the SMS-based treatments through the experimental mont hs of 2004 and 2005. In 2004, on average, a lesser amount of irrigation events were allo wed in August and September (21% and 18%, respectively) compared to October and N ovember (44% and 46%, respectively). In 2005, a greater proportion of irrigati on cycles were allowed in April, May, and July (31% and 46%, and 24%, respectively), and a fewer pr oportion in June and August (13% and 17%,

PAGE 55

39 respectively). These tendencies were concorda nt with the dryer/rain ier periods and, when correlating the monthly RED values and the percent of irrigation cycles allowed per month, r values of –0.93 and –0.76 were f ound for 2004 and 2005, respectively (Table 24). Looking in detail, for 2004, ACs followed this tendency allowing fewer irrigation cycles during the rainy mont hs of August and September, and more cycles during the dryer period between October and November In the case of 1-AC, the only irrigation cycles allowed occurred in November. 2-AC allowed less irrigation cycles on August and September (22% and 11%) and more cycles during October and November (22% and 33%, respectively); the same tendency as 7AC (10%, 0%, 3%, and 17%, for the same months, respectively). RBs also responded to this tendency on th e 1 d/w and 7 d/w frequencies, allowing a greater proportion of SIC to occur during th e end of the year; but 2-RB did not follow this, and actually run more cycles at the begi nning and at the end of the year (33% vs. 22%), with no irrigation cycles during September and October. A similar situation occurred with the IM tr eatments. The first and second half of the season were clearly different for 1-IM (betw een 20% 25% vs. 50% 100% of the SIC, respectively), and for 2-IM (33% vs. 100%, re spectively). However, 7-IM showed a more even conduct for the first three months (84% 73%, 87%, respectivel y), and then dropped down in November (23%). Finally, for 2004, WW treatments were also more active at the end of the season, allowing between 33% and 75% of the SIC (with the exception of 2-WW, on November, with 11%), compared to 0% to 29% for the first two months.

PAGE 56

40 During 2005, ACs also followed the dryer/ wetter period’s tendency in every frequency tested. In the case of 1-AC, it ra n 50% and 75% of the possible times on April and May (the driest months), and did not al low irrigations on June, July, and August (the wetter months). On these same last three months, 2-AC and 7-AC treatments allowed no more than 13% and 10% of the poten tial irrigation cycles respectively. A similar situation happened with WW sens ors. No irrigation cycles were allowed during the rainy months of June, July, and A ugust by the 1 and 2 d/w frequencies. In the case of 7-WW, it followed this tendency in June and August (with 10% and 6%, respectively), but in July–month that had two consecutive weeks of negative RED–it showed a higher number of irrigation events (32%). In 2005, RB sensors allowed few irrigation cy cles to start during all the different months. The 1 d/w frequency permitted be tween 20 and 25% of the SIC, the 2 d/w frequency resulted in 11%, and 7 d/w freque ncy between 6% and 19%. Nevertheless, all the frequencies had a month when they did not allow any irrigation cycles (April, June, and August for 1, 2, and 7 d/w, respectively). IM, in general, permitted more irrigation cycles than the other sensors. The 1-IM treatment showed similar behavior to the other brands during April, May, and June. Nevertheless, during July and August, 60% a nd 75% of irrigation events were allowed, respectively. Even in the rainy months of June, July, and August, 2-IM allowed 75% to 89% of the potential irrigation cycles to o ccur. A smaller variation in the number of irrigation events allowed between the differe nt months was shown by 7-IM (from 16% to 45%), and exhibited a closer pattern to the ot her brands, being more active in the months

PAGE 57

41 of April and May. However, it also permitted a greater amount of irrigation cycles per month than the other brands at this frequency. These relationships are clearer when looki ng at the VMC of the plots for year 2004, when evident differences in weather conditions through time were found. Figure 2-23 shows the VMC of treatment 0-NI where all th e increments in VMC were due to rainfall events. The differences between the dry and th e wet periods were reflected in the soil moisture content. Figures 2-24 to 2-35 show the VMC in plots that contained the SMSs controlling the irrigation treatments. These fi gures, show the results of the scheduled irrigation cycles (SIC), where the blue dots represent bypassed SIC, the red dots represent allowed SIC, and the red lines represent th e range of VMC when the SIC were allowed. When an increment in the VMC does not have a red dot, it means that a rainfall event occurred. These results show that, in general, the SMS-based treatments were able to follow the dryer and wetter periods, c ontrolling the amount of water to be delivered to the different treatments, and suggesting that this technology could be a usef ul tool to achieve automation of landscape-turfgrass irrigation, even when a RS is not present or nonfunctional. However, these are not precisi on instruments, which was evident because sometimes they bypassed irrigation cycles, and sometimes they did not, even reading the same or lower VMC. Moreover, according to the range of VMC over which the different SMS brands allowed irrigation, AC and RB ha d the narrowest average range (2.8% and 3.6%, respectively) suggesting that they were more accurate and consistent to measure the VMC than WW and IM (that had and av erage range of 7.0 and 8.9%, respectively).

PAGE 58

42 The IM controllers were set at positio n #1, which corresponds to –10 kPa (e.g. field capacity) according to the manufacturer. Howe ver, according to the results obtained in Chapter 4, at –10 kPa the GMSs were actually sensing a dryer soil condition, between -17 and -23 kPa. This explains why the IMs allowed irrigation cycles when not necessary. Therefore, setting the IM controllers at posi tion #2 or #3 would have resulted in increased irrigation savings. Irrigation Application Comparisons Tables 2-5, 2-6, and 2-7 show (for years 2004, 2005, and 2004+2005, respectively) the irrigation depth applied to treatments, statistical comparisons between them, and percent of water savings achieved by the tr eatments compared to 2-WRS, 2-DWRS, and 2-WORS. All the statistical analyses in thes e Tables showed a high level of confidence (P<0.0001), and are discussed below. Time-based treatments vs. SMS-based treatments Comparing the average of the time-based treatments with the average of the SMSbased treatments, Table 2-7 (Column A) show s that there was a significant difference between them; with 1044 and 420 mm of cumula tive irrigation depth, respectively. The same statistical difference was found in 2004 and 2005 (Tables 2-5 and 2-6). This means that the SMS-based treatments, on average, were more efficient as a means to save water than the time-based treatments, even when a rain sensor was an important component on two of the three time-based treatments. Time-based treatments The three time-based treatments (2 -WORS, 2-WRS, and 2-DWRS) were significantly different from each other during the whole period of study (Tables 2-5, 2-6 and 2-7; Column B).

PAGE 59

43 Treatment 2-WRS (two days/week, with a ra in sensor) was established to mimic a homeowner who complies with the irrigation laws and regul ations and sets the timer according to recommended practices; therefore, it was employed as the control treatment for water use volume. During the first seas on, this treatment accounted for 481 mm of water, or an equivalent of 98 mm/month, and 514 mm, or 96 mm/month, for 2005. A recent study, carried out by Hale y et al. (2006) in Central Florida, within the St. Johns River Water Management District (SJRWMD) found that homeowners with automatic irrigation systems applied 149 mm/month on aver age. Therefore, the comparisons made here may be considered conservative and diffe rences in the results for actual homeowners could be larger4. The well-managed or water conservative hom eowner profile, imitated by treatment 2-DWRS (two days/week, with a rain sensor and 60% of 2-WRS), applied 64% and 61% of the water used by 2-WRS in years 2004 and 2005, respectively, close to the 60% desired5. The yearly depths were 310 and 313 mm (or an equivalent of 63 and 59 mm/month), in 2004 and 2005, respectively. Haley et al. (2006) found within this homeowner profile (also programmed to replace 60% of historical ET) an irrigation water use of 105 mm/month6. The treatment simulating an irrigation system with an absent or non-functional rain sensor (2-WORS) accounted for 696 and 818 mm in the first and second season, or 141 4 Differences could be due to a better irrigation scheduling in this experiment, which was adjusted monthly. In the Haley et al. (2006) experiment, these homeowners set their own controller run times, which generally were not adjusted seasonally, and tended to over-irrigate in the late fall. 5 Equivalent to 36% and 39% in water savings compared to 2-WRS (Tables 2-4 and 2-5). 6 This difference could be due to less rainfall (an average of 122 mm/month during Haley’s research vs. 163 mm/month in this one), different soil conditions, or because some of the homeowners in this profile probably did not have a rain sensor, or maybe they had one, but it may have been non-functional.

PAGE 60

44 and 153 mm/month, respectively. It means that this treatment applied 45% and 59% more water than the treatment with a functional RS (2-WRS), and 52% more water during both years, on average. These results demonstrat e the importance not onl y for the presence, but also for the need of a functional and well-maintained rain shut-off device on all automated irrigation systems in Florida; wher e rainy weather, particularly in the warm months, is common (NOAA, 2003). Moreover, as the study prepared by Whitc omb (2005) recently found, just 25% of the surveyed homeowners in Fl orida with automatic irrigati on systems reported having a RS, and the author suggests that they ar e often incorrectly in stalled. Therefore, appropriately installed and properly work ing rain sensors could signify not only substantial water savings to homeowners, but could also lead to sound environmental and economic benefits to the state. In addition, their payback period could be less than a year; depending on the weather conditions, the area to be irrigated, the cost of water, and the cost of installed rain sensors (see Chapter 3). Comparisons between SMS-irrigation frequencies When the averages of the three different SMS irrigation-frequencies were analyzed (Tables 2-5 to 2-7, Column C), the 2 d/w fre quency used a significantly higher volume of water, followed by the 1 d/w frequency during both seasons and as a total, with 478 and 420 mm of total cumulative water depth, re spectively. The 7 d/w frequency was as high as the 2 d/w frequency in 2004, but resulted in the least water applied in 2005 (mostly due to the decrease in wate r application by 7-IM between 2004 and 2005). Therefore, 7 d/w was significantly the lowest of all three frequencies, w ith an average of 362 mm, in total cumulative water applied in this experi ment (Table 2-7). This was probably because

PAGE 61

45 more frequently scheduled irrigation events can be bypassed as a result of frequent rainfall. These results suggest that, in a long r un, to schedule low volume-high frequency irrigation cycles (7 d/ w) in Closed Control Loop irrigation systems7, appears to be a better strategy regarding water conservation in turfgrass irrigation, than to schedule them for some specific days during the week (1 or 2 d/w) and with a higher volume during each irrigation cycle. Soil moisture sensor-brands comparison The different brands were compared in terms of irrigation wa ter applied. Figure 236 shows the cumulative water applied by brand on years 2004 and 2005. As an overall comparison, IM sensors resulted in signifi cantly more irrigation during 2004 (P<0.0001), with 420 mm on average, followed by WWs, 18 8 mm, and then by sensors from brands AC and RB–which were not statistically diffe rent–and showed the lowest water use rate, with 116 and 100 mm, respectively. The same relationship was found in 2005, except that in this year, AC used a significantly higher ir rigation depth than RB, with all four brands statistically different, with 451, 164, 135 a nd 105 mm on average for IMs, WWs, ACs and RBs, respectively. However, these averages could not be directly compared to find out which brand was better in every case, because an interaction between brand and frequency was evidenced by the statistical analysis (P<0.0001) This implies that some SMS brands performed better at a certain fr equency than other ones, but not as good at another frequency. Hence, differences between SMSs brands were evaluated separately within each irrigation frequency. 7 In Closed Control Loop irrigation systems the decision to whether initiate or not an irrigation cycle is regulated by a SMS.

PAGE 62

46 Brand comparisons within irrigation frequencies From Tables 2-5 to 2-7 (Column D), it can be seen that there were statistical differences between brands within each frequency, for 2004, 2005, and as a total cumulative depth for both seasons. In 2004, th e highest water use rate in the 1 d/w frequency was displayed by IM, and then by WW, RB, and AC, respectively; all statistically different from each other. A similar trend was ex hibited at the 7 d/w, except that AC and RB were not signi ficantly different at this fre quency. At the 2 d/w frequency, however, the least amount of water was applied by brands RB and WW (not statistically different between them), then by AC, and then by IM. During 2005, results from 1 and 2 d/w frequenc ies were similar. At 1 d/w all four brands were statistically diffe rent from each other. However, comparatively, WW applied less water than in 2004 and, conversely, AC applied more. So, for 2005, IM applied the most, followed by AC, RB, and WW, respectivel y. Finally, at the 7 d/w frequency, the same statistical differences between br ands from 2004 were found in 2005, showing a high consistency through time. These similarities and differences, comp aring both years, could have happened because the number of total SIC was rela tively greater in the 7 d/w frequency, and relatively smaller in the 2 and 1 d/w fre quencies (302, 87, and 43 times, respectively). This means that timing of the rain events could have had a higher impact at the 1 d/w and 2 d/w level, but lower at th e 7 d/w frequency. Basically, th ese differences took place in the dry months of April and May 2005, when 1-AC, 1-WW, and 2-WW allowed some of the SIC to run (Table 2-4). The total cumulative irrigation depth applied in both seasons (Table 2-7, Comparison Column D) showed that, in the 1 d/w frequency, only IM used significantly

PAGE 63

47 more water than the other brands. For the 2 d/w frequency, all brands were significantly different, where IM applied more water, and then, in decreasing order, AC, WW, and RB. As 2004 and 2005, IM was the brand that signifi cantly applied more water at the 7 d/w frequency, followed by WW, and then by AC and RB (which were not statistically different). Summarizing, IMs always appl ied significantly more water than the other brands in every frequency tested, and in both seasons. This could be because of their reported limitations to timely sense the differences in so il water content, their hysteretic behavior, the high variability of its read ings, and their drawbacks for its use in sandy soils, where low tension values are necessary to prevent plant stress (Irmak and Haman 2001; Taber et al., 2002: Intrigliolo and Ca stel, 2004; McCann et al., 1992) Therefore, IMs do not appear to be the best choice fo r these climatic-soil conditions. Regarding the other brands, the best choi ce depends on the local restrictions and regulations concerning the freque ncy of landscape irrigation. Wh en irrigation is limited to 1 d/w, no differences appeared to be eviden t between the brands. For 2 d/w restrictions, according to these results, the best choice w ould be, in decreasing order, RB, WW, and AC. If no frequency limitations were presen t, AC or RB looks like the best choice, followed by WW. Overall comparison Tables 2-5, 2-6 and 2-7, (Column E), show the statistical differences between every irrigation treatment, independently if they we re time-based or SMS-based. As expected, 2-WORS always applied significantly more water than the other treatments. During 2004, following 2-WORS, were 2-WRS, 2-IM, a nd 7-IM, with no significant difference between them, and these treatments were followed by 2-DWRS and 1-IM, which were

PAGE 64

48 similar between them. However, during 2005 and 2004 + 2005, all these six treatments were significantly different between them, and showed significantly higher levels of water use than the rest of the treatments. On the other extreme, 7-AC and 7-RB always resulted in significantly less cumulative ir rigation depth throughout this experiment. Table 2-7 shows that 2-IM was the only SMS-based treatment that applied significantly more water (11%) than the control treatment 2WRS. Conversely, the other two IM treatments saved a significant amount of water (20% and 28%, by 1-IM and 7IM, respectively) compared to 2-WRS. Howeve r, these last proportions were far from the water savings achieved by the other SMS-based treatments, when compared to 2-WRS: AC sensors recorded irrigation water savings ranging from 65% to 88%, RBs from 72% to 85%, and WWs from 54% to 73%, depending on the irrigation freque ncy tested. It is important to remark that these water savings were on top of those already achieved by 2WRS. Therefore, these results show that, in ge neral, SMSs can also act as “rain sensors”, but with a superior performan ce in terms of water savings. When the irrigation treatments were compared to almost 75% of the surveyed homeowners in Florida (Whitcomb, 2005), this is with a non-functiona l or absent rain sensor (2-WORS), the significant difference in water savings increased, ranging from 77% to 92% for ACs, 81% to 90% for RBs, and 69% to 82%, for WWs. Even 2-IM (which applied 11% more water than 2-WR S) showed significant water savings, 27% with respect to 2-WORS, indicating that th is sensor was working but did not bypass as much SIC as other SMS-based treatments. Moreover, when compared to the water conservative 2-DWRS, treatments from brands AC, RB and WW also show ed significant water savings that ranged from 44% to

PAGE 65

49 80%, 55% to 76%, and 26 to 57%, respectivel y. On the other hand, all IM-frequencies applied significantly more irri gation than 2-WORS, with values that ranged from 15% to 77% more water. These results clearly demons trate that the use of SMSs, along with traditional timers in residential irrigation systems, could lead to important water savings. However, the correct choice of the SMS, and its technology to meas ure or “sense” the soil water status, is of great consequence. Automation of Irrigation Systems A complete automation of a residential ir rigation system, based on SMSs, could be achieved programming the timer to run every day as a scheduling strategy. Then, the SMSs will allow the system to initiate the irrigation cycles only when it is actually needed by the turfgrass, and override them wh en the sensed water content is over a preset threshold. In this experiment, this was confirmed when most of the SMS-based treatments were able to follow the dry and rainy peri ods–controlling the amount of water to be delivered to the different treatments (Table 2-4)–when the 7 d/w irrigation frequency used significantly less water than the other frequencies (Table 2-7, Column C), and when two of the SMS-based treatments programmed to run 7 d/w consistently used the smallest amount of water (Tables 2-5, 2-6, and 2-7). In effect, treatments 7-AC and 7-RB recorded total water savings of 85% or more, when co mpared to 2-WRS, and 90% or more when compared to 2-WORS (Tables 2-7). It is interesting to note th at this concept (with a pote ntial irrigation frequency of seven days a week) is contradictory to the wa ter use regulations and restrictions imposed by the Water Management Districts and/or muni cipalities in Florida (where irrigation is

PAGE 66

50 allowed only one or two days per week). Howe ver, these results suggest that choosing the right type of sensor, and programming the auto matic irrigation system to run everyday for a short period of time (allowing the SMS to deci de whether or not to irrigate), could save large amounts of potable water used for irri gation purposes, and l ooks like a better BMP than the current ones. Moreover, this concept is not in oppos ition to the general recommendation for deeper and less frequent irri gation for turfgrass, because these treatments (7-AC and 7RB) finally overrode more than 85% of the SIC (Table 2-4), resulting in a low actual irrigation frequency, which was supplem ented by large rainfall events. Turfgrass Quality No differences in turfgrass quality, in cluding non-irrigated plots, were found among treatments. This could be explained in part by the generally wet climatic conditions that happened through almost all the time of the experiment, which favored the growth and development of the bermudagr ass. Another factor contributing to this, even during the “dry” periods, could be found in the species itself. Common bermudagrass is known as a more drought-toler ant grass compared to the pervasive St. Augustinegrass [Stenotaphrum secundatum (Walt.) Kuntze] found in North-Central Florida landscapes (Harivandi et al., 2001; Baldwin et al., 2005; Turgeon, 2005). As a result, the treatment effects were buffere d with respect of th e turfgrass quality parameters, and it could be concluded that no irrigation was necessary to maintain an acceptable turf quality during the experiment ti me-period. Figure 2-37 shows, in Part A, an homogeneous good quality between the diffe rent plots, and, in Part B, when bermudagrass went dormant on all plots at the same time. Having analogous weather conditions, Jordan et al. ( 2003) obtained similar results working with bentgrass.

PAGE 67

51 Summary and Conclusions High frequency rainfall events and a la rge amount of cumulative precipitation prevailed during the time frame of this rese arch, except for some weeks at the end of 2004 and at the early spring of 2005, when ETo exceeded rainfall. When the monthly deficit rainfall values (relative to ETo) we re correlated with the percent of irrigation cycles allowed per month, r values of –0.93 and –0.76 were found for 2004 and 2005, respectively. It was inferred that most of the SMS-based treatments automatically canceled the majority of the irrigation cycl es during the rainy periods, and responded to dry periods by allowing irrigations to occur. The three time-based treatments (2 -WORS, 2-WRS, and 2-DWRS) were significantly different from each other dur ing the whole study period. The treatment without a rain sensor (2-WORS) used, on av erage, 52% more water than the treatment with a functional one (2-WRS), showing the importance of a well-maintained rain shutoff device in all automated irrigation systems in Florida. On the other hand, treatment 2DWRS, applied close to 60% of the water applied by 2-WRS. These time-based treatments were establis hed to mimic the operation of irrigation systems carried out by different homeowner profiles. However, according to the results of this research, these treatments were fairly well managed compared to homeowners’ actual operation practices in the Centra l Florida Ridge. Therefore, results in water use from this experiment can be considered conservative and differences for actual homeowners could be even larger. When the time-based treatments were compared to the SMS-based treatments, results showed that, on average, the SMSbased treatments were significantly more efficient as a means to save water than the traditional time-based treatments. However,

PAGE 68

52 not all SMS-treatments tested performed the same. The 2-IM treatment was the only SMS-based treatment that applied significantl y more water than the control 2-WRS (11% more). The other two IM treatments, 1-IM a nd 7-IM, used significan tly less water than 2WRS (20% and 28%, respectively) but always applied significantly more water than the other brands/treatments in every frequency test ed. Therefore, IMs do not appear to be the best choice for the weather and soil conditions of this study. All the other brands (AC, RB, and WW ) recorded significant irrigation water savings compared to the control 2-WRS, which ranged from 54% to 88%, depending on the irrigation frequency. These results showed that most SMSs except for 2-IM, can also act as rain sensors, with supe rior performance in terms of wa ter savings. When these last brands were compared to 2-WORS, the di fferences in water savings increased, and ranged from 69% to 92% over the 308-days study period. Irrigation frequencies were also compar ed at the end of the study. All three frequencies tested (1, 2, and 7 d/w) were significantly di fferent. The 2 d/w frequency used the highest volume of water, followe d by the 1 d/w frequency, and 7 d/w was the one that used the least amount of water. More over, and being part of this last frequency, treatments 7-AC and 7-RB significantly and consistently used the smallest amount of water regarding all treatments, during both seas ons. They recorded total water savings of 85% or more, when compared to 2-WRS, and 90% or more when compared to 2-WORS. These results suggest that scheduling low volume-high fr equency irrigation cycles (7 d/w) in Closed Control Loop irrigation systems, appears to be a better strategy regarding water conservation for turfgrass ir rigation in Florida’s sandy soils during rainy periods, than scheduling irriga tion cycles one or two days per week. Moreover, it was

PAGE 69

53 concluded that no irrigation wa s necessary to maintain an acceptable turf quality during the experimental period, which was evidenced by acceptable quality in non-irrigated plots. Therefore, SMS-based technology could lead not only to a complete automation of residential irrigation systems, but to save substantial irrigation water if implemented.

PAGE 70

54 Table 2-1. Irrigation treatmen t codes and descriptions. Treatment Codes Irrigation Frequency (days/week) Soil Moisture Sensor Brand or Treatment Description Soil Moisture Sensor-Based 1-AC 1 Acclima 1-RB 1 Rainbird 1-IM 1 Irrometer 1-WW 1 Water Watcher 2-AC 2 Acclima 2-RB 2 Rainbird 2-IM 2 Irrometer 2-WW 2 Water Watcher 7-AC 7 Acclima 7-RB 7 Rainbird 7-IM 7 Irrometer 7-WW 7 Water Watcher Time-Based 2-WRS 2 With rain sensor 2-WORS 2 Without rain sensor 2-DWRS 2 Deficit with rain sensor, 60% of 2-WRS 0-NI 0 No irrigation Table 2-2. Monthly irrigation depth to repl ace historical evapotranspiration, assuming system efficiency of 60%, and c onsidering effective rainfall. Month Irrigation depth (mm) January 0 February 0 March 112 April 112 May 183 June 142 July 137 August 178 September 137 October 122 November 91 December 91 Source: Based on Dukes and Haman (2002a)

PAGE 71

55 Table 2-3. Total number and percent of ove rridden scheduled irri gation cycles; 2004 and 2005. 2004 2005 2004 + 2005 Treatment Scheduled OverriddenSchedul ed OverriddenScheduled Overridden (#) (%) (#) (%) (#) (%) 2-WORS 40 0 46 0 86 0 f 2-WRS 40 30 46 37 86 34 e 2-DWRS 40 30 46 37 86 34 e 2-IM 41 32 46 24 87 28 e 1-IM 20 50 23 48 43 49 d 7-IM 142 41 160 70 302 56 d 1-RB 20 55 23 83 43 70 c 1-WW 20 63 23 87 43 76 c 7-WW 142 65 160 75 302 71 c 2-AC 41 73 46 83 87 78 bc 1-AC 20 85 23 78 43 81 abc 2-WW 41 85 46 78 87 82 abc 2-RB 41 85 46 91 87 89 ab 7-AC 142 92 160 92 302 92 a 7-RB 142 89 160 93 302 91 a P<0.0001

PAGE 72

56 Table 2-4. Percent of irrigation cycles allo wed by the SMS-based treatments through the experimental months of 2004 and 2005. Year 2004 (%) Year 2005 (%) Treatment Aug Sep Oct Nov Apr May Jun Jul Aug 1-AC 0 0 0 75 50 75 0 0 0 2-AC 22 11 22 33 22 33 13 11 11 7-AC 10 0 3 17 17 10 0 6 10 1-RB 0 40 100 75 0 25 20 25 20 2-RB 33 0 0 22 11 11 0 11 11 7-RB 6 3 13 17 7 19 7 6 0 1-IM 25 20 50 100 25 100 20 75 60 2-IM 33 33 100 100 67 100 75 89 67 7-IM 84 73 87 23 40 45 17 35 16 1-WW 0 20 75 50 25 50 0 0 0 2-WW 11 0 33 11 67 44 0 0 0 7-WW 29 20 42 33 37 42 10 32 6 Average 21 18 44 46 31 46 13 24 17 RED (mm) 184 397 -16 -17 -14 -23 76 -25 30 CC (r) -0.93 -0.76 RED = Rainfall ETo difference CC = Correlation coefficient

PAGE 73

57 Table 2-5. Cumulative irrigation depth applie d to treatments, statistical comparisons between them, and percent of water sa vings compared to 2-WRS, 2-DWRS, and 2-WORS; year 2004. Comparisons+ Water savings (%) vs. Treatment Cumulative depth 2004 (mm) A B C D E 2-WRS2-WORS 2-DWRS SMS-Based 1-AC 95 d g 80 86 69 1-RB 128 c f 73 82 59 1-IM 318 a c 34 54 -3 1-WW 209 b e 57 70 33 1-Avg188 b 2-AC 196 b e 59 72 37 2-RB 87 c gh 82 88 72 2-IM 470 a b 2 32 -52 2-WW 94 c g 81 87 70 2-Avg212 a 7-AC 57 c h 88 92 82 7-RB 85 c gh 82 88 73 7-IM 471 a b 2 32 -52 7-WW 261 b d 46 63 16 7-Avg218 a SMS-Avg 206 b Time-Based 2-WORS 696 a a -45 0 -125 2-WRS 481 b b 0 31 -55 2-DWRS 310 c c 36 55 0 Time-Avg 496 a SMS=Soil moisture sensor P<0.0001 Avg=Average +A=Time-based treatments vs. SMS-based treatments B=Between time-based treatments C=Between irrigation frequency averages D=Within irrigation frequency E=Overall comparison

PAGE 74

58 Table 2-6. Cumulative irrigation depth applie d to treatments, statistical comparisons between them, and percent of water sa vings compared to 2-WRS, 2-DWRS, and 2-WORS; year 2005. Comparisons+ Water savings (%) vs. Treatment Cumulative depth 2005 (mm) A B C D E 2-WRS2-WORS 2-DWRS SMS-Based 1-AC 188 b gh 63 77 40 1-RB 153 c i 70 81 51 1-IM 475 a d 7 42 -52 1-WW 114 d j 78 86 64 1-Avg232 b 2-AC 152 b i 70 81 51 2-RB 101 c j 80 88 68 2-IM 635 a b -24 22 -103 2-WW 177 b h 66 78 44 2-Avg266 a 7-AC 65 c k 87 92 79 7-RB 62 c k 88 92 80 7-IM 244 a f 52 70 22 7-WW 202 b g 61 75 36 7-Avg143 c SMS-Avg 214 b Time-Based 2-WORS 818 a a -59 0 -161 2-WRS 514 b c 0 37 -64 2-DWRS 313 c e 39 62 0 Time-Avg 548 a SMS=Soil moisture sensor P<0.0001 Avg=Average +A=Time-based treatments vs. SMS-based treatments B=Between time-based treatments C=Between irrigation frequency averages D=Within irrigation frequency E=Overall comparison

PAGE 75

59 Table 2-7. Total cumulative irrigation depth a pplied to treatments, st atistical comparisons between them, and percent of water sa vings compared to 2-WRS, 2-DWRS, and 2-WORS; years 2004 + 2005. Comparisons+ Water savings (%) vs. Treatment Total cumulative depth 2004+2005 (mm) A B C D E 2-WRS2-WORS 2-DWRS SMS-Based 1-AC 283 b i 72 81 55 1-RB 281 b i 72 81 55 1-IM 793 a d 20 48 -27 1-WW 323 b h 68 79 48 1-Avg420 b 2-AC 348 b h 65 77 44 2-RB 188 d j 81 88 70 2-IM 1105 a b -11 27 -77 2-WW 270 c i 73 82 57 2-Avg478 a 7-AC 122 c k 88 92 80 7-RB 147 c k 85 90 76 7-IM 715 a e 28 53 -15 7-WW 463 b g 54 69 26 7-Avg362 c SMS-Avg 420 b Time-Based 2-WORS 1514 a a -52 0 -143 2-WRS 995 b c 0 34 -60 2-DWRS 623 c f 37 59 0 Time-Avg 1044 a SMS=Soil moisture sensor P<0.0001 Avg=Average +A=Time-based treatments vs. SMS-based treatments B=Between time-based treatments C=Between irrigation frequency averages D=Within irrigation frequency E=Overall comparison

PAGE 76

60 0 5 10 15 20 25 30 35 05101520253035 Tension (kPa)Calibrated VMC (%) Figure 2-1. Soil water retent ion curve from tensiometers and calibrated ECH2O probe readings.

PAGE 77

61 Figure 2-2. Soil moisture sensor brands tested in this study.

PAGE 78

62 Figure 2-3. Irrigation controls as installed for this study: soil moisture sensors-controllers brands: A) Rain Bird, B) Water Watcher, C) Acclima, and D) Irrometer, and irrigation timer E) Rain Bird.

PAGE 79

63 Figure 2-4. Rain sensor installed for this study.

PAGE 80

64 Figure 2-5. Catch-can display for uni formity tests on turfgrass plots.

PAGE 81

65 Figure 2-6. General view of the irri gation controls used in this study. Figure 2-7. Pipes, flowmeters, va lves, and wirings for this study.

PAGE 82

66 Figure 2-8. Control board showing timers, so il moisture sensor-controllers, solenoid valves wiring, and flowmeters-datalogge r (details are shown in the next Figures).

PAGE 83

67 Figure 2-9. Control board detail show ing the solenoid valves control box. Figure 2-10. Control board detail, flowme ter-datalogger boxes showing A) multiplexers, B) CR 10X datalogger used for this study.

PAGE 84

68 Figure 2-11. Automated weather stati on near turf plots for this study.

PAGE 85

69 Figure 2-12. ECH2O probe, capacitance soil moisture probe shown with a HOBO data logger as installed for this study.

PAGE 86

70 ABCDEF1215525471786012 1156664863407611 1041516153647910 94621505843669 82746492660648 74951534459677 63549465658266 55131523452765 46264476455684 34561403656553 25869375133392 16155642852631ABCDEF Figure 2-13. Plot plan showing the low-quarter distribution un iformity testing results on each plot.

PAGE 87

71 ABCDEF1215.26.47.48.85.64.412 1114.05.45.411.46.84.011 106.46.26.810.26.07.010 96.06.07.410.08.07.69 84.65.69.67.89.48.48 74.45.66.47.89.29.07 65.44.66.69.48.05.86 53.67.45.89.27.65.25 44.45.05.67.88.46.44 35.86.06.07.47.25.63 25.85.66.26.86.65.82 14.47.26.06.68.28.41ABCDEF Figure 2-14. Plot plan showi ng average volumetric water cont ent (%) on each plot during a relatively “dry” period. Pl ots in red were discarded, and plots in green were used for placement of SMSs.

PAGE 88

72 ABCDEF1220.811.011.612.211.012.412 1127.811.612.212.810.89.411 1012.48.810.410.410.89.810 911.411.211.011.410.410.89 811.410.612.612.211.810.68 710.29.612.011.011.011.27 610.611.610.210.89.69.26 59.212.210.49.610.49.85 49.810.29.89.48.89.24 39.89.812.010.410.88.63 212.011.210.211.214.010.62 112.211.611.212.014.815.81ABCDEF Figure 2-15. Plot plan showi ng average volumetric water content (%) on each plot during a relatively “wet” condition. Plots in re d were discarded, and plots in green were used for placement of SMSs.

PAGE 89

73 A BCDE F 122-WORS.22-IM.30-NI.31-RB.212 11 7-IM.2 1-IM.47-RB.4 2-AC.2 11 101-AC.4 7-IM.3 1-RB.42-WRS.21-AC.22-DWRS.310 97-AC.3 2-WW.1 7-AC.22-DWRS.2 2-AC.3 7-RB.39 8 2-AC.17-IM.4 1-WW.4 7-WW.2 2-WORS.48 72-DWRS.1 2-WW.4 2-RB.22-RB.42-DWRS.47 67-AC.42-IM.22-WRS.31-WW.2 2-WW.2 1-RB.36 51-IM.31-AC.10-NI.22-WORS.32-RB.15 41-AC.30-NI.1 7-WW.3 7-RB.21-RB.14 3 7-WW.4 7-RB.12-WORS.12-RB.32-IM.4 7-WW.1 3 27-AC.12-WRS.1 7-IM.1 1-WW.11-IM.12-IM.12 12-WRS.4 2-WW.3 1-IM.21-WW.3 2-AC.4 0-NI.41ABCDEF Figure 2-16. Plot plan with the modified completely randomized design (same color depicts treatment repetitions).

PAGE 90

74 157 269 104 35 944 0 5 10 15 20 25 3020-Jul 27-Jul 3-Aug 10-Aug 17-Aug 24-Aug 31-Aug 7-Sep 14-Sep 21-Sep 28-Sep 5-Oct 12-Oct 19-Oct 26-Oct 2-Nov 9-Nov 16-Nov 23-Nov 30-Nov 7-Dec 14-DecDate (2004)Daily rainfall (mm)0 100 200 300 400 500 600 700 800 900 1000Cumulative rainfal (mm) Figure 2-17. Daily and cumulative rainfall in 2004. Note: rainfall for 5 Sep. (188 mm) and 6 Sep. (81 mm) is shown as a cumulative total (269 mm). 39 42 33 42 51 732 0 5 10 15 20 25 3025-Mar 1-Apr 8-Apr 15-Apr 22-Apr 29-Apr 6-May 13-May 20-May 27-May 3-Jun 10-Jun 17-Jun 24-Jun 1-Jul 8-Jul 15-Jul 22-Jul 29-Jul 5-Aug 12-Aug 19-Aug 26-AugDate (2005)Daily rainfall (mm)0 100 200 300 400 500 600 700 800Cumulative rainfall (mm) Figure 2-18. Daily and cu mulative rainfall in 2005.

PAGE 91

75 40 28 -5 5 15 25 35 45 55 65 75 85 20-Jul19-Aug18-Sep18-Oct17-Nov17-Dec Date (2004)Cummulative Number of Irrigation Events 2-WRS 2-WORS 2-DWRS 3 8-10 -5 5 15 25 35 45 55 65 75 85 20-Jul19-Aug18-Sep18-Oct17-Nov17-Dec Date (2004)Cummulative Number of Irrigation Events 1-AC 1-RB 1-IM 1-WW 11 28 6 -5 5 15 25 35 45 55 65 75 85 20-Jul19-Aug18-Sep18-Oct17-Nov17-Dec Date (2004)Cummulative Number of Irrigation Events 2-AC 2-RB 2-IM 2-WW 11 16 84 49 -5 5 15 25 35 45 55 65 75 85 20-Jul19-Aug18-Sep18-Oct17-Nov17-Dec Date (2004)Cummulative Number of Irrigation Events 7-AC 7-RB 7-IM 7-WW Figure 2-19. Cumulative number of irrigati on events per treatment in 2004; A) timebased treatments, and soil moisture sensor-based treatments at irrigation frequencies of B) 1 d/w, C) 2 d/w, and D) 7 d/w. Note: In part A), treatments 2-WRS and 2-DWRS were controlled by th e same rain sensor and were set to run at the same days, so only one li ne can be seen for both treatments. A B C D

PAGE 92

76 5 4 12 3-5 0 5 10 15 20 25 30 35 40 45 50 25-Mar24-Apr24-May23-Jun23-Jul22-Aug Date (2005)Cummulative Number of Irrigation Events 1-AC 1-RB 1-IM 1-WW 8 4 35 10-5 0 5 10 15 20 25 30 35 40 45 50 25-Mar24-Apr24-May23-Jun23-Jul22-Aug Date (2005)Cummulative Number of Irrigation Events 2-AC 2-RB 2-IM 2-WW 13 12 48 40 -5 0 5 10 15 20 25 30 35 40 45 50 25-Mar24-Apr24-May23-Jun23-Jul22-Aug Date (2005)Cummulative Number of Irrigation Events 7-AC 7-RB 7-IM 7-WW 45 29 -5 0 5 10 15 20 25 30 35 40 45 50 25-Mar24-Apr24-May23-Jun23-Jul22-Aug Date (2005)Cummulative Number of Irrigation Events 2-WRS 2-WORS 2-DWRS Figure 2-20 Cumulative number of irrigation events per treatment in 2005; A) time-based treatments, and soil moisture sensor-based treatments at irrigation frequencies of B) 1 d/w, C) 2 d/w, and D) 7 d/w. Note: In part A), treatments 2-WRS and 2-DWRS were controlled by the same rain sensor and were set to run at the same days, so only one line can be seen for both treatments. D C A B

PAGE 93

77 153 276 101 -40 -20 0 20 40 60 80Aug-7 Aug-14 Aug-21 Aug-28 Sep-4 Sep-11 Sep-18 Sep-25 Oct-2 Oct-9 Oct-16 Oct-23 Oct-30 Nov-6 Nov-13 Nov-20 Nov-27 Dec-4 Dec-11 Dec-14Week (2004)Weekly rainfall ETo difference (mm) Figure 2-21. Maximum weekly irrigation wate r requirement (rainfall ETo difference); year 2004. -40 -20 0 20 40 60 80Mar-31 Apr-7 Apr-14 Apr-21 Apr-28 May-5 May-12 May-19 May-26 Jun-2 Jun-9 Jun-16 Jun-23 Jun-30 Jul-7 Jul-14 Jul-21 Jul-28 Aug-4 Aug-11 Aug-18 Aug-25 Aug-31Week (2005)Weekly rainfall ETo difference (mm) Figure 2-22. Maximum weekly irrigation wate r requirement (rainfall ETo difference); year 2005.

PAGE 94

78 Figure 2-23. Volumetric moisture content through time, on treatment 0-NI, year 2004.

PAGE 95

79 Figure 2-24. Volumetric moisture content (V MC) through time, showing resu lts of the scheduled irriga tion cycles (SIC), where th e blue dots represent bypassed SIC, the red dots represent allowed SIC, and the red lines re present the range of VMC when the SIC were allowed; treatment 1-AC, year 2004.

PAGE 96

80 Figure 2-25. Volumetric moisture content (V MC) through time, showing resu lts of the scheduled irriga tion cycles (SIC), where th e blue dots represent bypassed SIC, the red dots represent allowed SIC, and the red lines re present the range of VMC when the SIC were allowed; treatment 1-IM, year 2004.

PAGE 97

81 Figure 2-26. Volumetric moisture content (V MC) through time, showing resu lts of the scheduled irriga tion cycles (SIC), where th e blue dots represent bypassed SIC, the red dots represent allowed SIC, and the red lines re present the range of VMC when the SIC were allowed; treatment 1-RB, year 2004.

PAGE 98

82 Figure 2-27. Volumetric moisture content (V MC) through time, showing resu lts of the scheduled irriga tion cycles (SIC), where th e blue dots represent bypassed SIC, the red dots represent allowed SIC, and the red lines re present the range of VMC when the SIC were allowed; treatment 1-WW, year 2004.

PAGE 99

83 Figure 2-28. Volumetric moisture content (V MC) through time, showing resu lts of the scheduled irriga tion cycles (SIC), where th e blue dots represent bypassed SIC, the red dots represent allowed SIC, and the red lines re present the range of VMC when the SIC were allowed; treatment 2-AC, year 2004.

PAGE 100

84 Figure 2-29. Volumetric moisture content (V MC) through time, showing resu lts of the scheduled irriga tion cycles (SIC), where th e blue dots represent bypassed SIC, the red dots represent allowed SIC, and the red lines re present the range of VMC when the SIC were allowed; treatment 2-IM, year 2004.

PAGE 101

85 Figure 2-30. Volumetric moisture content (V MC) through time, showing resu lts of the scheduled irriga tion cycles (SIC), where th e blue dots represent bypassed SIC, the red dots represent allowed SIC, and the red lines re present the range of VMC when the SIC were allowed; treatment 2-RB, year 2004.

PAGE 102

86 Figure 2-31. Volumetric moisture content (V MC) through time, showing resu lts of the scheduled irriga tion cycles (SIC), where th e blue dots represent bypassed SIC, the red dots represent allowed SIC, and the red lines re present the range of VMC when the SIC were allowed; treatment 2-WW, year 2004.

PAGE 103

87 Figure 2-32. Volumetric moisture content (V MC) through time, showing resu lts of the scheduled irriga tion cycles (SIC), where th e red dots represent allowed SIC (8% of the SIC), and the red line s represent the range of VMC when the SIC were allowed; treatment 7-AC, year 2004.

PAGE 104

88 Figure 2-33. Volumetric moisture content (V MC) through time, showing resu lts of the scheduled irriga tion cycles (SIC), where th e red dots represent allowed SIC (59% of the SI C), and the red lines represent the range of VMC when the SIC were allowed; treatment 7-IM, year 2004.

PAGE 105

89 Figure 2-34. Volumetric moisture content (V MC) through time, showing resu lts of the scheduled irriga tion cycles (SIC), where th e red dots represent allowed SIC (11% of the SI C), and the red lines represent the range of VMC when the SIC were allowed; treatment 7-RB, year 2004.

PAGE 106

90 Figure 2-35. Volumetric moisture content (V MC) through time, showing resu lts of the scheduled irriga tion cycles (SIC), where th e red dots represent allowed SIC (35% of the SI C), and the red lines represent the range of VMC when the SIC were allowed; treatment 7-WW, year 2004.

PAGE 107

91 2004 100 c 116 c 188 b 420 a 0 100 200 300 400 500 IMWWACRB BrandAverage irrigation depth (mm) 2005 105 d 135 c 164 b 451 a 0 100 200 300 400 500 IMWWACRBBrandAverage irrigation depth (mm) Figure 2-36. Average irrigation depth a pplied by brand; years 2004 and 2005 (P<0.0001).

PAGE 108

92 Figure 2-37. View of differen t plots where no evident turfgr ass quality differences could be detected; A) good quality, B) dormant.

PAGE 109

93 CHAPTER 3 EXPANDING DISK RAIN SENSOR PERFORMANCE AND POTENTIAL IRRIGATION WATER SAVINGS Rain Sensors A rain sensor (RS), also called rain shut -off device or rain switch (Figures 3-1 and 3-2), is a device designed to interrupt a schedu led cycle of an automatic irrigation system controller (i.e. timer), when a specific amount of rain fall has occurred (Dukes and Haman, 2002b; Hunter Industries, Inc., 2006). Rain shut-off devices are the most comm on type of irrigation-sensor, due to an increasing number of munici palities throughout the country that have mandates and/or cost-saving programs for their use, on ne w and existing residential and commercial irrigation systems. In addition, and except for th e most arid environments, they appear to be a useful tool for water conservation, at a relatively low cost (Dewey, 2003). Currently, there are mandates for the use of RSs in various municipalities in New Jersey, North and South Carolina, Georgia, Texas, Minnesota and Connecticut (Dewey, 2003). However, Florida is the only state in th e nation with an overall RS statute. Florida law requires an automatic rain sensor shut -off device that is properly installed and functioning on all automatic irrigation systems installe d after 1 May 1991 (Florida Statutes, Chapter 373.62, n.d.; Florida Statutes 2001). Moreover, some local laws also require older systems to be retrof itted with rain shut-off switches1 (SJRWMD, 2006). 1 Soil Moisture Sensors, could also be considered as “rain shut-off switches” because they can bypass irrigation cycles after sufficient rainfall (see Chapter 2, Soil moisture sensor-based treatments). However, in this chapter, rain sensors are referred as their most common usage; i.e., those devices that directly sense rainfall.

PAGE 110

94 Rain sensors can be easily hooked up to any automatic irrigation system controller and mounted in an open area where they are exposed to rainfall. Some new irrigation controllers have a special connect ion, which allows a RS to be attached directly. If this connection is not available, it can always be “hard-wired” into th e controller, connecting the RS in series with the common wire. When a specific amount of rainfall has occurred, the RS will interrupt the system common wire which disables the solenoid valves until the sensor dries out (Dukes and Haman, 2002b). Advantages According to Dukes and Haman (2002b) the use of RSs has several advantages: by eliminating unnecessary irrigation events th ey conserve water, reduce wear on the irrigation system, reduce disease and weed pr essure, and reduce the runoff and/or deep percolation that carries pollutants–such as fe rtilizers and pesticides–into storm drains and groundwater. RSs also save money, because they reduce utility bills and turf maintenance costs. It could be added that these benef its are supplemented by a relatively low cost, easy installation, low maintenance, and long dur ability (more than ten years according to manufacturers, and a 5-year warranty). Types and Methods Several types and models of RSs, which differ in their operation method, have been developed. Some of them have a receptacle to weigh the amount of water and, after a preset weight of water is co llected, the connection to the automatic irrigation valve is interrupted until a portion of water in the receptacle evaporates reducing the weight below a critical level. Other models also us e a receptacle but, inst ead of weight, they detect the water level with a set of electro des. The distance between the bottom of the receptacle and the electrodes can be adjusted so the irrigation system is not switched off

PAGE 111

95 by small rain events. The primar y disadvantage of this type of device is that any other external volume/weight (debris, small animal s, etc.) can turn off the irrigation system (Dukes and Haman, 2002b). The third and most widely used method empl oys an expanding material to sense the amount of rainfall (Figure 3-1). Hygroscopic disks (Figure 3-3) absorb water and expand proportionally to rainfall amount. As the mo isture-laden disks e xpand, they activate a switch that interrupts the programmed irriga tion cycle. The switch remains open as long as the disks are swollen. When the rain has passed the disks begin to dry out, they contract, and the switch closes ag ain (Hunter Industries, Inc., 2006). Different RS models have, in general, so me kind of regulation to activate them after a specific amount of rainfall. The expanding hygroscopic disk s-type Mini Click model (Hunter Industries, Inc. ), very common in Florida, has five different settings (Figure 3-1) that can bypass an irrigation cycle after rain fall quantities of 3, 6, 13, 19, or 25 mm To adjust to the desired shut-off quantity it is necessary to rotate the cap on the switch housing, so that the pin is locate d in the proper slot (Figure 3-1). The time that it takes the Mini Click to rese t for normal sprinkler operation after the rain has stopped is determined by weathe r conditions (temperature, wind, sunlight, humidity, etc.), which will determine how fast the hygroscopic discs dry out. To compensate for the drying rate of the site 's soil or for an “overly sunny” installation location, these sensors have an adjustment capability, the “vent ri ng” (Figure 3-1). By closing it, the hygroscopic discs will dry more slowly. Howeve r, drying time has not been quantitatively evaluated.

PAGE 112

96 Finally, a new version of these devices (a lso with hygroscopic discs inside) is the radio-controlled or wireless RS (Figure 3-2). The components of this system are a sensor unit installed in an area subj ect to rainfall and a receive r unit hooked up to the timer. Some advantages of these sensors include a qui cker and easier insta llation, and additional mounting locations to choose from (up to 90 m away from the receiv er), especially for sites that present difficulty in routing wire as well as for retrofit applications (Hunter Industries, Inc., 2006). A new feature promoted by industry with rega rd to wireless RSs is their quick shut down of the irrigation system af ter it starts to rain (they do not include specific preset adjustments for a certain precipitation amount), and their ability to bypass irrigation for a short period of time once it stops raining. Similar to the Mini Clicks, the wireless RSs can be adjusted to keep the irrigation system off after the rain stops, depending on the climatic conditions and by setting its adjust able ventilation windows (Figure 3-2), which control the dry-out time (Hunt er Industries, Inc., 2006). Installation In residential and light commercial irri gation applications, a RS is typically installed near the roofline on the side of a building. Nevertheless, manufacturers recommend mounting it on any surface where it will be exposed to unobstructed rainfall. The RS location should receive about the same amount of sun and sh ade as the turf, but should not be in the path of sprinkler spray (Hunter Industries, Inc., 2006). Objectives The objectives of this experiment were as follows: a) evaluate the reliability of two commercially available expanding disk RS-type s with respect to number of irrigation cycles bypassed, accuracy of set point with rainfall depth, and duration in irrigation

PAGE 113

97 bypass mode, b) quantify the amount of water that RSs could save compared to timebased irrigation schedules without RS, and c) estimate the payback period of RSs at different set points. Materials and Methods Twelve Mini-Click (MC) and four Wirele ss Rain-Click (WL) rain sensor models (Figures 3-1 and 3-2), for a total of sixteen devices (Hunter Industries, Inc., San Marcos, CA) were placed in a completely randomized design (Figure 3-4) at the Agricultural and Biological Engineering Department turfgrass re search facility, University of Florida, Gainesville, Florida. The experiment t ook place from 25 March through 31 December 2005. Data Each time a rain sensor changed status (from allowing irrigation, to bypass mode, or vice versa), the date and time was automa tically recorded, at a one-minute sampling interval, by means of two AM16/32 multiplexers (Campbell Scientific, Logan, UT), which were hooked up to a CR 10X model da talogger (Campbell Scientific, Logan, UT) (Figure 3-4). Climatic conditions were recorded by an automated weather station containing a CR 10X model data logger (Campbell Scientif ic, Logan, UT), located within 15 m of the experimental site. Rainfall was measured by means of a tipping bucket rain gauge, which was calibrated against a manual rain gauge. Re sults show that both methods were highly similar (R2= 0.998) (Figure 3-5). Total rainfall data, as well as the other weather measurements (see Chapter 2), were collected every 15 minutes. However, because this sampling interval was too long to quantify the precise amount of precipitation that fell before the sensors switched off,

PAGE 114

98 rainfall data were recorded at intervals of 0.25 mm after 29 June (day of year, hour, and minute were logged). Therefore, total rainfa ll before each RS switched to bypass mode was calculated, in order to evaluate the accuracy of the rainfall thresholds. The total time that each RS remained in the irri gation bypass mode was also computed. Treatments Four treatments with four replications each were established (Table 3-1). For the MCs, three different settings were establis hed: 3, 13, and 25 mm thresholds (treatment codes 3-MC, 13-MC, and 25-MC, respectively). The vent rings of the MCs were kept completely open. In the case of the WLs, th e dry-out ventilation windows were set half open. None of these treatments were connected to irrigation timers. So, in order to estimate how many cycles these settings w ould have overridden and how much water could have been potentially saved, they we re compared to the treatment With-RainSensor from the SMS-experiment (code 2WRS, Chapter 2). In that experiment, treatment 2-WRS used a Mini Click rain sensor set at 6 mm, so, in this chapter, it will be referred as 6-MC. Treatment 6-MC was hooked up to a timer, which was scheduled to run two days per week (Sunday and Thursday), beginning at 0100 h; to simulate watering restrictions imposed in Fl orida (FDEP, 2002; Florida Statutes, Chapter 373.62, n.d.). Thus, if the rain sensors were in bypass mode as a result of rainfall during a scheduled irrigation event, this bypass was considered “potential” i rrigation savings. The weekly irrigation depth (see Table 2-2) was set to replace the hist orical ET, as recommended by Dukes and Haman (2002a).

PAGE 115

99 Statistical Analysis Data were analyzed using the General Linear Model (GLM) procedure of SAS (SAS, 2000). If significant F values (P < 0.05) were detected, Duncan’s Multiple Range Test was used to separate means. Results and Discussion Climatic Conditions Figure 3-6 shows the daily and cumulative rainfall during the experiment. During the 282-day experiment, 174 days exhibited ra infall (62%), including 11 days with more than 25 mm. The cumulative precipitation was 1112 mm, an amount that is not uncommon in this region. However, there wa s one dry period in the late fall, from 25 October through 20 November, when ju st one event of 0.5 mm occurred. Number of Times in Bypass Mode The cumulative number of times that sensors switched to bypass mode, averaged by treatment, is shown in Figure 3-7. It can be seen that the cumulative number of times in bypass mode were statistically different where WL > 3-MC > 13-MC > 25-MC, with 81, 43, 30, and 8 events, respectively, in the 282 -day experiment. However, as seen in Figures 3-8 to 3-11, the number of times in bypass mode within treatments was variable. The most variable treatments were 3-MC and 13-MC (between 30 and 54 times, and between 22 and 39 times, respectively). The four replications of th e WL treatment (Figure 3-8) were extremely consistent, with a similar number of events in bypass m ode (between 78 and 83). However, this was not the case of 3-MC (Figure 3-9). All four 3-MC sensors behaved similarly for the first thirteen rainfall events. After 3 June, two un its (A and B) continued to have the same behavior, while C and D units had similar performance to each other but did not bypass

PAGE 116

100 as many events as A and B, with 30-36 vs. 53-54 times, respectively. Similar to 3-MC, treatment 13-MC also showed an irregular performance betw een replicates (Figure 3-10). With the exception of the first two rain events which were not sensed by replicate D, all replications switched to bypass mode on the sa me dates until 3 June (similar to 3-MC). After that date, replic ate A bypassed more times than the ot her replicates (39 times versus 32, 28, and 22 times, for replications A, B, C, and D, respectively). Replicates from treatment 25-MC performed similarly (Figure 3-11), shutting off between 7 to 8 times. All sensors worked identically for the first four rainfall events and then replicates A and C operated similarl y while the performan ce of the other two replicates was slightly different. The diffe rence in sensor perf ormance for the 25-MC treatment was not as pronounced as the other MC-treatments, in part due to the fewer number of rain events greater than 25 mm. It should be noted that although these RS units would have bypassed irrigation 7 to 8 times, th ere were 11 rainfall ev ents greater than 25 mm. Depth of Rainfall Before Shut Off According to Figliola and B easley (2000), the accuracy of an instrument refers to its ability to indicate a true value exactl y. Accuracy is related to absolute error, which is defined as the difference between the true value of a measurement and the indicated value of the instrument: = true value – indicated value [3-1] from which the percent accuracy, A, is found by: 100 1 value true A [3-2]

PAGE 117

101 The average depth of rainfall before the rain sensors switched to bypass mode is shown in Table 3-2. Treatment WL shut off on average after 1.4 mm of rain but, because this model does not have a specific set point accuracy was not calculated. Treatments 3MC, 13-MC and 25-MC shut off after 3.4, 10.0, and 24.5 mm, resulting in accuracies of 88%, 77%, and 98%, respectively. These average accuracies suggest that, in general, the MCs responded close to their settings, with 25-MC and 3-MC operating closest to their set point. However, some rainfall events, large e nough to meet the RS settings and to theoretically shut off the irrigation system, we re not detected by some units. For example, on treatment 3-MC (Table 3-3) replications C and D did not detect rainfall events between 11.4 and 122.0 mm on ten occasions. On treatment 13-MC (Table 3-4), one or more of three units did not bypass some ra in events between 19.1 and 122.0 mm on seven different occasions. In the case of 25-MC, five rain events larger than 33 mm were not sensed by one or more units (Table 3-5). No relationship between this behavior and rain intensity or other clim atic condition was found. In addition, on seven occasions some units fr om 3-MC (Table 3-6) shut off several hours after the rain had stopped (even more than 24 h later). The same situation happened with some units from 13-MC in twelve diffe rent occasions (Table 3-7). Moreover, on 7 April, replication D from 13-MC switched to bypass mode after 11.7 mm of rain, then switched to ON when it was still raining, and did not switch to bypass mode again; even when it rained an additional 28 mm. These observations clearly show that some MC-units tested had different sensitivities to specific se ttings and, additionally, some times they responded properly

PAGE 118

102 according to their settings, and sometimes they did not. Moreover, units that had inconsistent behavior showed problems when they were fairly new, and these units had the most of both types of behaviors (i.e. not bypassing events or bypassing events after rainfall stopped). On the other hand, WL treatments switched to bypass mode in absence of rainfall (Table 3-8). The number of times that this happened ranged betw een 11 and 22, with an average of 16 times; remaining in that mode for a minimum of one minute, a maximum exceeding 10 hours, and an average of more th an 3 hours. These situations were triggered when high relative humidities occurred (95% on average) or, on five occasions, minutes before a rainfall event began. Th is shows that sometimes these sensors are too sensitive, with the drawback that they could bypass a scheduled irrigation cycle, even when no rainfall occurred, a situation that ha ppened twice during this experiment. Duration in Irrigation Bypass Mode (Dry-Out Period) Figures 3-12 to 3-14 show histograms and frequency distribution for intervals of time of 6 h in bypass mode for treatments WL, 3-MC and 13-MC. In the case of 25-MC, because of the small number of occurrences (7-8 times) no interval of time had a number of occurrences 5, hence no histogram and frequenc y distribution could be plotted (Figliola and Beasley, 2000). Results showed that half the time WL-sen sors remained in bypass mode between 0 and 12 h (Figure 3-12), 80% of time they remain ed in that status for less than 24 h, and only 5% of the events lasted between 54 and 78 h. This is concordant with manufacturers’ advertisements, in order that WL will remain in that status shortly after the rain stops (Hunter Industries, Inc., 2006). Treatment 3-MC (Figure 3-13) remained in bypass mode less than 24 h most of the time (with a peak between 18 and 24 h), and more

PAGE 119

103 than 80% of the time remained in that stat us for less than 48 h. Fo r 13-MC, most of the time in bypass mode was for less than 24 h, similar to 3-MC, and more than 80% of time they did not stay in that status for more than 36 h (Figure 3-14). Although it was not possible to generate a histogram for 25-M C, the maximum length in bypass mode was just over 30 h. Hence, the lower set points tended to stay in bypass mode for a longer period of time. This is explained by the larger number of successive small rainfall events that occurred, keeping the sensors with lowe r settings in bypass mode for a longer period of time. Potential Water Savings Treatment 6-MC bypassed the irrigation system on 16 occasions (37% of all the scheduled irrigation cycles), ac counting for 304 mm in water savings2. The total potential water savings for the other treatments, by replic ation and as an average per treatment, are shown in Table 3-9. Results revealed that, ev en when treatment averages showed logical and statistical differences between them (363, 245, 142, and 25 mm for WL, 3-MC, 13MC, and 25-MC, respectively), replications fr om MC treatments were highly variable, with CVs of 28%, 61%, and 34%, for 3-MC, 13-MC, and 25-MC, respectively. Table 3-9 also shows that, on treatment 3-MC, the final amou nt of the potential water savings was 304 mm for two replicati ons, and 195 and 178 mm for the other two, resulting in a treatment average of 245 mm, compared to 304 mm for 6-MC, contrary to what was expected. Again, this is a consequen ce of the erratic behavi or of units C and D from 3-MC. 2 This corresponds to the difference between treatments 2-WORS and 2-WRS (2 days-per-week Without Rain Sensor vs. 2 days-per-week With Rain Sensor, respectively) detailed in Chapter 2.

PAGE 120

104 Treatment 13-MC showed potential wate r savings ranging from 78 to 266 mm. As expected, all replications s howed lesser total potential water savings than 6-MC. Conversely, replication A from 13-MC showed larger water savings (266 mm) than two of the 3-MC replications (C and D, with 195 and 178 mm, respectively), again demonstrating the variability betwee n different MC units (Table 3-9). The 25-MC replications showed small poten tial water savings (between 12 and 29 mm) compared to the other treatments (Table 39), despite the fact that there were 11 rain events of 25 mm or more. However, these ev ents would have had to occur a maximum of 30 h or less before the irrigation window so that the RS would bypass the scheduled irrigation event. Conversely to the erratic behavior of the MC treatments, WL showed a high consistency among its replications and account ed for the highest potential water savings among the treatments, between 342 and 380 mm (Table 3-9), and between 38 and 76 more mm of irrigation wa ter savings than 6-MC. Payback Period In spite of the variability found, it is interesting to quantify how much money the potential water savings could represent, and also to calculate a payback period for the rain sensors. According to Augustin (2000), th e historical net irrigation requirements for this period of study, for the Gainesville ar ea, are around 65% of the total requirements per year; therefore, the water savings per year would have been 558, 377, 468, 218, and 38 mm for WL, 3-MC, 6-MC, 13-MC, and 25-MC respectively. If a system irrigates 1000 m2 of turf, each mm of water a pplied is equivalent to 1 m3 applied to this surface. Assuming a cost of $75 for the WL and $25 fo r the MC units, plus $50 for installation, Table 3-10 shows the potential payback period pe r treatment at different water costs. If

PAGE 121

105 the water cost was $0.264/m3 ($1.00/TG), the payback period would have been less than a year for WL, 3-MC, and 6-MC; and between 1.3 and 7.4 years for treatments 13-MC and 25-MC, respectively. According to this analysis, except for 25MC, the installation and maintenance of a RS appears to be strongly justified. Howeve r, this contrasts profoundly with reality. As the study by Whitcomb (2005) recently found, ju st 25% of the surveyed homeowners in Florida with automatic irrigation systems repor ted having a RS, and the author speculates that they are often incorrectly installed. Finally, it is important to remember that in the soil moisture sensor-experiment, due to favorable weather conditions, no irrigati on was necessary to maintain an acceptable turf quality during the experi ment’s time-period; hence, every bypassed irrigation cycle would have led to valuable water savi ngs. In this example, 363, 245, 142, and 25 m3 of fresh water that could have b een lost to deep percolation or runoff, would have been saved by WL, 3-MC, 13-MC, and 25-MC, respectively. Summary and Conclusions The experiment was carried out during a rainy period, where 62% of the days had rainfall. The cumulative number of times th at sensors switched to bypass mode, when averaged by treatment, were inversel y proportional to their set points. Accuracy test results suggest that, on average, the MCs responded close to their set points. However, some replications show ed erratic behavior, sometimes responding properly according to their sett ings, and sometimes not detecting rainfall events five or more times their set points, or even shutti ng off several hours after the rain had stopped. This explains the range of variation in th e number of times that individual RSs switched

PAGE 122

106 to bypass mode. On the other hand, high relati ve humidities some times caused WL units to switch to bypass mode, in absence of rainfall. In general, the lower set poi nts on the MC treatments tende d to stay in bypass mode for a longer period of time, because of the larger number of successive small rainfall events that occurred. Treatment WL tended to stay in bypass mode for a shorter period of time than MC treatments. The potential water savings of the va rious RS set point s were inversely proportional to their set point. Depending on the area to be ir rigated and on the cost of water, the payback period would have been close to a year for WL, 3-MC, and 13-MC. However, setting the MC at 25 mm is not recommended in Central Florida, because it showed small potential water savi ngs, even in a rainy year. Finally, rain sensors, depending on their se t points, showed that they can be a useful and highly recommended tool when used by homeowners as a means to save water in Florida, but not when accuracy is required.

PAGE 123

107 Table 3-1. Treatments description. Treatment Model Set point Vent window 3-MC Mini-click II 3 mm completely open 13-MC Mini-click II 13 mm completely open 25-MC Mini-click II 25 mm completely open WL Wireless ---[z] half open [z] WL does not have an adjustable set point. Table 3-2. Average depth of rainfall before rain sensors switched to bypass mode. Set point Rainfall depth Accuracy Treatment (mm) (mm) (%) 3-MC 3 3.4 88 13-MC 13 10.0 77 25-MC 25 24.5 98 WL --1.4 ---[z] [z] Because these instruments do not declare a specific set point, accuracy was not calculated. Table 3-3. Large rainfall events not bypassed by treatment 3-MC. Rainfall Replicate Date (mm) A B C D 7-Jun 17.0 X X 8-Jun 11.4 X X 12-Jun 20.3 X X 2-Jul 24.6 X X 3-Aug 16.3 X X 7-Aug 16.5 X X 8-Aug 12.1 X[z] X[z] 10-Aug 17.5 X X 6-Oct 79.2 X X 17-Dec 122.0 X X [z] The day before it rained 16.5 mm extra.

PAGE 124

108 Table 3-4. Large rainfall events not bypassed by treatment 13-MC. Rainfall Replicate Date (mm) A B C D 26-Mar 28.5 X X 1-Apr 19.1 X X 5-May 41.7 X[z] X[z] 12-Jun 20.3 X X 2-Jul 24.6 X X X 6-Oct 79.2 X 17-Dec 122.0 X[y] X [z] Shut off after 45 mm of rainfall. [y] The day before it rained 12.5 mm extra. Table 3-5. Large rainfall events not bypassed by treatment 25-MC. Rainfall Replicate Date (mm) A B C D 27-Jun 41.7 X 29-Jun 38.6 X 20-Aug 33.3 X X X 6-Oct 79.2 X X X X 17-Dec 122.0 X X X Table 3-6. Hours after rain stopped and sens ors switched to bypass mode; treatment 3MC. Replicate (h) Date A B C D 3-Jul 6 1-Aug 6 21-Sep 6 4 30-Nov 18 10-Dec X 16-Dec 18 20-Dec X X X = more than 24 h.

PAGE 125

109 Table 3-7. Hours after rain stopped and sens ors switched to bypass mode; treatment 13MC. Replicate (h) Date A B C D 7-Apr X[z] 6-May X 4-Jul X 10 X 6-Aug 19 7-Aug 14 11-Aug 18 14-Aug 7 31-Aug 10 2-Sep X 5-Oct 5 21-Nov 3 10-Dec X X = more than 24 h. X[z] Switched to ON when it was raini ng. After that, it rained 28 mm extra. Table 3-8. WL replications that switched to bypass mode in absence of rainfall, elapsed time that they remained in bypass mode, and relative humidity at the time when this occurred. Elapsed time in bypass mode (hh: mm) Relative Humidity (%) ReplicationsReplications ABCD Avg ABCD Avg Minimum 01:2500:0101:5100:08 00:51 91868986 88 Maximum 07:1010:3406:2707:34 07:56 98989798 98 Average 03:4102:3403:1602:54 03:06 96959693 95 Times (#)17221211 16 17221211 16 Table 3-9. Total potential water savings per treatment. Replications (mm) CV Treatment A B C D Average (%) WL 380 365 364 342 363 a 4 3-MC 304 304 195 178 245 b 28 13-MC 266 133 89 78 142 c 61 25-MC 29 29 29 12 25 d 34 P< 0.05

PAGE 126

110 Table 3-10. Potential payback period per treatment. Water cost Payback period per treatment (years) ($/TG) ($/m3) WL 3-MC 6-MC 13-MC 25-MC 0.50 0.13 1.7 1.5 1.2 2.6 14.8 1.00 0.26 0.8 0.8 0.6 1.3 7.4 1.50 0.40 0.6 0.5 0.4 0.9 4.9 2.00 0.53 0.4 0.4 0.3 0.7 3.7

PAGE 127

111 Figure 3-1. Mini-Click (Hunter I ndustries, Inc.) rain sensor. A) Rain threshold set slots, B) vent ring. Figure 3-2. Wireless Rain-Click (Hunter Indu stries, Inc.) rain sensor. A) Ventilation window adjustment knob, B) ventilation windows, C) antenna.

PAGE 128

112 Figure 3-3. The expanding material of a rain shut-off switch. Figure 3-4. Rain sensor experi ment layout: A) Wireless Rain-C lick rain sensors, B) MiniClick rain sensors, C) Wireless Rain -Click receivers, D) multiplexers, E) CR 10X datalogger.

PAGE 129

113 y = 0.9962x + 0.1109 R2 = 0.9980 10 20 30 40 50 60 0102030405060 Tipping bucket rain gauge (mm)Manual rain gauge (mm) Figure 3-5. Manual rain gauge measurements compared to tipping bucket rain gauge measurements. 39 42122 79 33 42 51 1112 0 5 10 15 20 25 302 5 M a r 9 A p r 2 4 A p r 9 M a y 2 4 M a y 8 J u n 2 3 J u n 8 J u l 2 3 J u l 7 A u g 2 2 A u g 6 S e p 2 1 S e p 6 O c t 2 1 O c t 5 N o v 2 0 N o v 5 D e c 2 0D e cDate (2005)Daily rainfall (mm)0 200 400 600 800 1000 1200Cumulative rainfall (mm) Figure 3-6. Daily and cumulative rainfall.

PAGE 130

114 81a 43b 30c 8d 0 10 20 30 40 50 60 70 80 90 24-Mar24-Apr25-May25-Jun26-Jul26-Aug26-Sep27-Oct27-Nov28-Dec DateCumulative events in bypass mode WL 3-MC 13-MC 25-MC Figure 3-7. Cumulative number of times rain sensors switched to bypass mode; average per treatment. Different letters indicate a significant difference by Duncan’s Multiple Range Test (P<0.05)

PAGE 131

115 81 82 83 78 0 10 20 30 40 50 60 70 80 90 24-Mar24-Apr25-May25-Jun26-Jul26-Aug26-Sep27-Oct27-Nov28-Dec DateCumulative events in bypass mode A B C D Figure 3-8. Cumulative number of times ra in sensors switched to bypass mode; WL treatment, with replicates indicated by A-D. 54 53 36 30 0 10 20 30 40 50 60 70 80 90 24-Mar24-Apr25-May25-Jun26-Jul26-Aug26-Sep27-Oct27-Nov28-Dec DateCumulative events in bypass mode A B C D Figure 3-9. Cumulative number of times rain sensors switched to bypass mode; 3-MC treatment, with replicates indicated by A-D.

PAGE 132

116 39 32 28 22 0 10 20 30 40 50 60 70 80 90 24-Mar24-Apr25-May25-Jun26-Jul26-Aug26-Sep27-Oct27-Nov28-Dec DateCumulative events in bypass mode A B C D Figure 3-10. Cumulative number of times rain sensors switched to bypass mode; 13-MC treatment, with replicates indicated by A-D. 8 7 0 10 20 30 40 50 60 70 80 90 24-Mar24-Apr25-May25-Jun26-Jul26-Aug26-Sep27-Oct27-Nov28-Dec DateCumulative events in bypass mode A B C D Figure 3-11. Cumulative number of times rain sensors switched to bypass mode; 25-MC treatment, with replicates indicated by A-D.

PAGE 133

117 Figure 3-12. Histogram and frequency distri bution for 6-hour intervals in bypass mode; WL. Figure 3-13. Histogram and frequency distri bution for 6-hour intervals in bypass mode; 3-MC.

PAGE 134

118 Figure 3-14. Histogram and frequency distri bution for 6-hour intervals in bypass mode; 13-MC.

PAGE 135

119 CHAPTER 4 GRANULAR MATRIX SENSOR PERFORMANCE COMPARED TO TENSIOMETER IN A SANDY SOIL Numerous methods exist to measure so il water content, including gravimetric measurements, neutron scatteri ng, resistance blocks, tensiome ters, and granular matrix sensors (GMS). These methods are fairly co mmon, some of them have been used for several decades, and continue to be used extensively in irrigati on scheduling (Gardner, 1986; Seyfried, 1993; Leib et al., 200 2; Leib et al., 2003; Or, 2001). In this research, two of these methods we re compared, tensiometers and granular matrix sensors (GMS), in a sandy soil. Tensiometers Tensiometers have been used for many year s to measure soil water tension in the field. Of all the methods available for m onitoring water potential for irrigation, the tensiometers are perhaps the most wi dely used (Campbell and Mulla, 1990). A tensiometer (Figure 4-1) consists of a sealed, water-filled tube with a permeable porous cup on one end, an airtight seal on th e other end and some means of measuring tension (a gauge, manometer, or electronic pre ssure transducer). The device is installed in the soil with the ceramic tip in close contact with the soil at the desired depth. Water is pulled out through the ceramic tip into the soil creating a tension in the closed tube. As the soil is re-wetted (e.g., fr om rain or irrigation), the te nsion gradient reduces, causing water to flow into the ceramic tip (Ley et al., 2000). At equilibrium, the water pressure (tension) in the tensiometer is equal in magn itude to the soil matric potential (Cassell and

PAGE 136

120 Klute, 1986). When tensiometers installed at the root zone reach a certain reading, they can be used as indicators of the need for irrigation, based on soil texture and crop type (Ley et al., 2000). Most commercially available tensiometers use a vacuum gauge to read the tension created and have a scale from 0 to 100 kilo Pascals (kPa) (100 kP a = 100 centibars = 1 bar = 14.7 psi). The practical operating range is fr om 0 to 75 kPa. If the water column is intact, a zero reading indi cates saturated soil conditions Readings of around 10 kPa correspond to field capacity for coarse-textu red soils, while readings of around 30 kPa can approximate field capacity for some fi ner-textured soils. The upper limit of 75 kPa corresponds to as much as 90% depletion of total available water fo r the coarse-textured soils, but is only about 30% depletion for silt loam, clay loams, and other fine-textured soils. This limits the practical use of tensiometers to coarse-textured soils or to high frequency irrigation where soil water cont ent is maintained high (Ley et al., 2000). Tensiometers have been designed for use in situations where tensions above 30 kPa are rarely expected (e.g. sandy soils), when finer resolution near saturation is needed, and/or in conditions where rapi dly changing moisture tensions need to be observed. This is the case of the MLT-RSU (Miniature Low Tension – Remote Sensing Unit)tensiometer developed by Irrometer (Irromete r Company, Inc., Riverside, CA), shown on Figure 4-1. Careful installation of tensiometers is required for reliable results. The ceramic tip must be in intimate and complete contact with the soil, and installation sites should represent the field in terms of water application patterns, soil types, slopes, and exposure; and should be installed out of the way of tr affic and cultivation. In freezing climates,

PAGE 137

121 tensiometers must be insulated or removed during winter months, because it takes only a small frost to knock the vacuum gauges out of calibration (Ley et al., 2000). Granular Matrix Sensors The GMS (Figure 4-2) is a device that meas ures soil electrical resistance, that can be converted to soil water tension (SWT), e ither using a calibration formula provided in the literature for sandy soils (Irmak and Haman, 2001) and silt loam soils (Eldredge et al., 1993), or calibrating them for a specific soil type (Hanson et al ., 2000b; Intrigliolo and Castel, 2004). Since the development of the GMS, many researchers (Eldredge et al., 1993; Mitchell and Shock, 1996; Bausch and Bernard, 1996) have used it in agricultural water management, including irrigation scheduling. Ho wever, some limitations in its use have been found. For example, in coarse–texture d soils (e.g. sand) a lack of soil–sensor interface might be observed, and consequently may lead to incorrect estimation of SWT (Irmak and Haman, 2001). Also, GMS has a high uncertainty in the wet range and do not respond to changes at a SWT lower than 10 kPa and, therefore, may not be a suitable tool in those cases where irrigation practices maintain a low SWT (Irmak and Haman, 2001; Taber et al., 2002: Intrigliolo and Castel, 2004). In additi on, McCann et al. (1992) found that GMS has a hysteretic behavior when induc ed to rapid drying or partial rewetting of the soil, which could affect the performance of the GMS in estimating the actual soil water status. Moreover, there is also evidence that the most important drawback on its use for irrigation scheduling is the high variability of its readings (CV of 35–50%), increasing at the lower SWT range (Taber et al., 2002; Intrigliolo and Ca stel, 2004). Finally, calibration appears to be excl usive for each individual sensor (Egbert et al., 1992; Hanson et al., 2000b; Leib et al., 2003; Intrigliolo and Castel, 2004).

PAGE 138

122 In spite of all these limitations, GMS may be useful when a relative indication of soil wetness is needed, as i ndicated by reports of their successful use for irrigation scheduling in onion (Shock et al., 1998a), potato (Shock et al ., 1998b), tomato, and walnut trees (Hanson et al., 2000a). GMS – Tensiometer Comparison Previous research found that GMSs functi oned consistently ov er a range of SWT from 10 kPa to 200 kPa (Hanson et al., 2000a; Le ib et al., 2003; Intrigliolo and Castel, 2004). So, GMS can operate in a drier range than tensiometers, but with a lower resolution at the wet end of SWT (Egbert et al., 1992; Irmak and Haman, 2001). This is an important limitation for the use of GMS in the predominantly coarse-textured soils of Florida. Bausch and Bernard (1996) evaluated the va lidity of the granular matrix sensor SWT values calculated using Thomson and Armstrong (1987) and Shock et al. (1996) calibration equations with the tensiomete r-measured SWT. However, both equations underestimated SWT for sandy clay loam soil. Thomson et al. (1996) compared Thomson and Armstrong (1987) and McCann et al. ( 1992) equations and indicated that the equations deviated signifi cantly for estimating SWT. Objectives The objectives of this research were to: a) compare GMS to tensiometer readings on a sandy soil, b) establish a relationship between them, if any, and c) evaluate the performance of the GMSs in a sandy soil. Materials and Methods The experiment was conducted in 2004 in the Water Resources Lab at the University of Florida, Gainesville, Flor ida. Three PVC cylinders [19.14 L (0.30 m high

PAGE 139

123 0.285 m diameter)], were hand-packed with an Arredondo fine sand (loamy, siliceous, semiactive, hyperthermic Grossarenic Paleudul ts) (USDA, 2003), taken from the field where the soil moisture sensor experiment was carried out. Experimental Set-Up In each PVC cylinder, three 200SS mode l Watermark GMSs (Irrometer Company, Inc., Riverside, CA) were installed vertical ly, in a uniformly spaced pattern. (Prior to installation, GMSs were immersed in wate r for two days, according to manufacturer instructions.) A temperature sensor (Irrome ter Company, Inc., Riverside, Calif.) was installed in the center of th e PVC cylinders to correct th e GMS readings. In addition, three MLT-RSU model Tensiometers (Irro meter Company, Inc., Riverside, CA), hereafter-called tensiometers, were also pl aced near the GMSs. Finally, a capacitance ECH2O probe (Decagon Devices, Inc., Pullman, WA) was also placed in each PVC cylinder. Figures 4-1 through 4-4 show pictur es of the devices. GMSs were placed with their center at 9 cm soil depth. All other devices were installed with their sensing section at the depth of GMS centers. The experi mental layout is shown in Figure 4-5. ECH2O Probes Calibration Before the experiment was carried out, the ECH2O probes were calibrated through the thermogravimetric soil sampling method (Gardner, 1986). Undisturbed soil samples were collected (using a core sampler of 137.4 cm3) from each PVC cylinder at the same depth where ECH2O probes were placed. Samples were taken from a saturated through a dry condition (28.4% to 4.8% of VMC, re spectively, based on gravimetric data). The samples were weighed and then ove n-dried at 104C for 24 h. Then, the dry samples were re-weighed. Percent soil water co ntent on a dry mass or gravimetric basis, Pw, was determined by the following formula:

PAGE 140

124 100x weight dry sample weight dry sample weight sample wet Pw [4-1] To convert from a gravimetric basis to water content on a volumetric basis, Pv, the gravimetric soil water content was multiplied by the soil bulk density (BD): Pv = Pw x BD [4-2] and soil bulk density was determined by: dry soil of volume dry sample of weight BD [4-3] The volumetric soil water content of each sample was then compared to the ECH2O probe readings at the same date and time wh en the samples were taken and, then, these data were compared to calibrate the ECH2O probe readings. Treatments All three PVC cylinders had a ceramic porous plate placed at the bottom to withdraw water from the cylinders to accel erate drying. Suction was applied, using a vacuum pump connected to the porous plate. A different initial suction time was applied to each cylinder to create different rates of soil drying, defining the treatments (Table 41): 0, 5, 15, and 50 min for treatments T0, T5, T15, and T50, respectively. Data The tensiometers, GMSs, and temperature sensor were connected to a Watermark Monitor (Figure 4-6). This is a data l ogger developed by Irrometer (Irrometer Company, Inc., Riverside, Calif.), which includes an automatic temperature compensation of the soil moisture readings, and gives GMS and te nsiometer readings in kPa. The ECH2O probes were connected to a HOBO Micro Station (F igure 4-7) data logger (Onset Computer Corporation, Bourne, MA). All sensor data were recorded every 15 minutes.

PAGE 141

125 Results and Discussion Calibration of the ECH2O Probe Data from the three ECH2O probes compar ed to thermogravimetric measurements is given in Figure 4-8. It can be seen that ECH2O measurements and gravimetric measurements were highly correlated (R2 = 0.95). However, the ECH20 probes under predicted the gravimetric water conten t by about 20% as seen in Figure 4-8. GMSs versus Tensiometers. In previous research, when comparing GMS with other SMC-methods, authors took electrical resistance measurements from GMS, and then converted these data to tension, through calibration equations (Irmak and Haman, 2001; Eldredge et al., 1993; Hanson et al., 2000b). In this research, the Watermark Monitor-datalogger gave the GMS readings expressed immediatel y in tension, so no conversion with a calibration curve was theoretically necessary. The results from the different treatments are displayed in Figures 4-9 through 4-16. It can be seen that the curves of GMSs a nd tensiometers followed a similar trend, but tension readings were different Their soil water tension curves crossed at average values in the range of 5.8 through 6.7 kPa, depending on the treatment, with an overall average of 6.2 kPa (Table 4-2). At tensions less than this level, GMSs gave lower tensions than the tensiometers, and the opposite occurred at tensions higher than these values, where GMSs showed a consistently drier estimate of the soil moisture content compared to tensiometers. Relating the average of GMS and tensiomete r readings (Figure 4-17), a coefficient of determination of R2=0.9582 was found for a linear relationship between them, and a coefficient of determination of R2=0.9916 for a logarithmic relationship between them.

PAGE 142

126 Previous studies reported that GMSs do not respond accurately at tensions less than 10 kPa (Egbert et al., 1992; Irmak and Haman, 2001) Moreover, Eldredge et al. (1993) did not include these data when performing thei r analysis. In the present experiment, GMS values at tensions less than 7 kPa were not stable, showing fluctuations between 1kPa (Figures 4-13 and to 4-14). When repl otting the average of GMS and tensiometer readings without including tensions below 10 kPa (Figure 4-18) the coefficient of determination increased to R2=0.9809 for a linear relations hip, and increased to R2=0.9956 for a logarithmic relationship. In the sandy soils of North-Central Florida, the trigger point to start an irrigation cycle is often reported as 10-30 kPa. If that is the case, when tensiometers were reading 10 kPa, GMSs were reading between 17 and 23 kPa, and an average of 20 kPa. This could have a major consequence if an auto matic irrigation system is controlled by a GMS, because is highly probable that they would allow irrigation cycles when actually they are not necessary, which is exactly what happened with the GMS-controlled irrigation in Chapter 2. According to these resu lts, it is evident that calibration of GMS units in this type of soil is necessary. Conclusions Tension readings from tensiometers and GM Ss were the same, at levels close to 6.2 kPa. Below this tension, GMSs readings were lower than those from tensiometers, and the opposite occurred at tensions higher than this value. At the same time as tensions continued to increase, the difference betw een both methods increased. In addition, GMS readings below 7 kPa were not stab le, fluctuating between 1kPa. The set point (kPa) to initiate irrigation could be of great consequence regarding water use efficiency, when GMSs are set to control an automatic irrigation system.

PAGE 143

127 According to these results, it was evident that calibration of GMS units in this type of soil is necessary to obtain readings closer to reality, and achieve adequate irrigation management.

PAGE 144

128 Table 4-1. Treatments. Treatment Suction Time (minutes) A 0 B 5 C 15 D 15 E 50 Table 4-2. GMS-Tensiometer crossing points. Treatment (-kPa) T0 T5 T15 T50 GMS # Tensiometer # Each Avg EachAvg EachAvg EachAvg Avg 1 5.5 5.3 6.2 1 2 6.1 6.15.6 5.86.4 6.4 6.1 3 6.6 6.4 6.5 1 5.6 5.5 6.8 2 2 6.2 6.15.8 6.06.8 6.8 6.3 3 6.7 6.6 6.8 1 6.2 5.3 5.3 6.2 3 2 6.4 6.46.0 5.95.6 5.86.4 6.4 6.1 3 6.7 6.5 6.4 6.5 Average 6.4 6.0 5.8 6.5 6.2

PAGE 145

129 Figure 4-1. MLT-RSU Tensiometer. Figure 4-2. Watermark GMS.

PAGE 146

130 Figure 4-3. Temperature sensor. Figure 4-4. ECH2O probe

PAGE 147

131 Figure 4-5. Experimental layout (top view). A) Tensiometers, B) Granular matrix sensors, C) ECH2O probe, and D) Thermometer. C A A A B D B B

PAGE 148

132 Figure 4-6. Watermark monitor. Figure 4-7. ECH2O probe hooked up to a HOBO Micro Station datalogger.

PAGE 149

133 y = 0.8052x 0.0273 R2 = 0.9486 0 5 10 15 20 25 30 051015202530 VMC-Gravimetric (%)VMC-Probe (%) Figure 4-8. Volumetric moisture co ntent (VMC) from all three ECH2O probes compared to gravimetric measurements.

PAGE 150

134 0 10 20 30 40 50 09619228838448057667276886496010561152 Hours after suctionTension (kPa) Tens-1 Tens-2 Tens-3 GMS-1 GMS-2 GMS-3 Figure 4-9. Soil water tension through time; treatment T0. 0 10 20 30 40 50 04896144192240288336384432480528576624672720 Hours after suctionTension (kPa) Tens-1 Tens-2 Tens-3 GMS-1 GMS-2 GMS-3 Figure 4-10. Soil water tension through time; treatment T5.

PAGE 151

135 0 10 20 30 40 50 04896144192240288336384432480528576624672720768816 Hours after suctionTension (kPa) Tens-1 Tens-2 Tens-3 GMS-1 GMS-2 GMS-3 Figure 4-11. Soil water tension through time; treatment T15. 0 10 20 30 40 50 04896144192240 Hours after suctionTension (kPa) Tens-1 Tens-2 Tens-3 GMS-1 GMS-2 GMS-3 Figure 4-12. Soil water tension through time; treatment T50.

PAGE 152

136 0 5 10 15 20 25 0961922883844805766727688649601056 Hours after suctionTension (kPa) Tens-1 Tens-2 Tens-3 GMS-1 GMS-2 GMS-3 Figure 4-13. Soil water tension through time; detail showing when curves from GMS and tensiometers cross; treatment T0. 0 5 10 15 20 25 04896144192240288336384432480528 Hours after suctionTension (kPa) Tens-1 Tens-2 Tens-3 GMS-1 GMS-2 GMS-3 Figure 4-14. Soil water tension through time; detail showing when curves from GMS and tensiometers cross; treatment T5.

PAGE 153

137 0 5 10 15 20 25 04896144192240288336384 Hours after suctionTension (kPa) Tens-1 Tens-2 Tens-3 GMS-1 GMS-2 GMS-3 Figure 4-15. Soil water tension through time; detail showing when curves from GMS and tensiometers cross; treatment T15. 0 5 10 15 20 25 049141924283338 Hours after suctionTension (kPa) Tens-1 Tens-2 Tens-3 GMS-1 GMS-2 GMS-3 Figure 4-16. Soil water tension through time; detail showing when curves from GMS and tensiometers cross; treatment T50.

PAGE 154

138 Figure 4-17. Relation between the average soil matric potential (SMP) from tensiometers and GMS.

PAGE 155

139 Figure 4-18. Relation between the average soil matric potential (SMP) from tensiometers and GMS; excluding GMS data < 10 kPa.

PAGE 156

140 CHAPTER 5 CONCLUSIONS AND FUTURE WORK Conclusions The goals of this research were to find out if different SMS-systems (sensor with a proprietary controlle r) could reduce irrigation water application–while maintaining acceptable turf quality–compared to current pr actices and, on the other hand, to collect evidence related to RS performance and re liability. The main objectives of this experiment were to quantify differences in irrigation water use and turf quality between: 1) a soil moisture sensor-based irrigation sy stem compared to a tim e-based scheduling, 2) different commercial irrigati on soil moisture sensor (S MSs), and 3) a time-based scheduling system with or without a rain se nsor (RS). The secondary objectives were to: a) evaluate the reliability of two commercially availabl e expanding disk RS-types, b) quantify the amount of water that RSs coul d save compared to time-based irrigation schedules without RS, and c) estimate the pa yback period of RSs at different set points. Results showed that no significant di fferences in turfgrass quality among treatments were detected, which was evid enced by good quality in non-irrigated plots. This was a consequence of the high freque ncy rainfall events and large amount of cumulative precipitation that prevailed during the time frame of this research and, on the other hand, because of the documented char acteristics of bermudagrass as a droughttolerant plant. Regarding the time-based treatments, 2-WORS (without-rain-sensor) used significantly (52%) more wate r than 2-WRS (with-rain-senso r), showing the importance

PAGE 157

141 not only for the presence but also for the need of a well-maintained rain shut-off device in all automated irrigation systems. However, treatment 2-WRS was fairly well managed and conservative, compared to homeowners’ actual operation practi ces, so “real” water savings on residential landscap es could be even larger. It was inferred that, in general, SMS-base d treatments were able to follow and detect fairly well when sufficient rain occu rred, overriding pre-set irrigation cycles, and allowing the rest of them to run when necessary. SMS-based treatments were, on average, sign ificantly more efficient as a means to save water than the time-based treatments. However, not all SMSs tested performed the same. Sensors from brand Irrometer always applied significantly more water than the other brands/treatments in every frequency. All the other brands (AC, RB, and WW) recorded significant irrigation water savings compared to 2-WRS, which ranged from 54% to 88%, depending on the irrigation fre quency. When compared to 2-WORS, the differences increased, and ranged from 69% to 92% in water savings. Therefore SMS-systems represent a prom ising technology, because of the water savings that they can accomplish, while maintaining an acceptable turfgrass quality. The correct choice of a SMS should take into consideration featur es like its technology, response-time, irrigation scheduling st rategy, and cost, among other aspects. Regarding the RS treatments, on averag e, treatments WL, 3-MC, 13-MC, and 25MC responded close to their rain fall set points (1.4, 3.4, 10.0, and 24.5 mm, respectively). However, some replications showed erratic behavior through time. The number of times that these sensor s shut off irrigation was inversely proportional to the magnitude of their set point (81, 43, 30, and 8 times, respectively)

PAGE 158

142 with potential water savings following a similar trend (363, 245, 142, and 25 mm, respectively). Under the relatively wet testing conditions typical to Florida, the payback period of the RSs tested could be less than a y ear, except for 25-MC (around 7 years). Consequently, RSs are strongly recommended fo r use by homeowners as a means to save water, but not when accuracy is required. Moreover, as the study prepared by Whitc omb (2005) recently found, just 25% of the surveyed homeowners in Fl orida with automatic irrigati on systems reported having a RS, and the author suggests that they ar e often incorrectly in stalled. Therefore, appropriately installed and properly work ing rain sensors could signify not only substantial water savings to homeowners, but could also lead to sound environmental and economic benefits to the state. Future Work The SMS-system technology should be tested under real homeowner conditions in order to validate these results. Also, future information obtained from the same research field will give an evaluation of this SMS-based system performance over a longer period of time, regarding its consistency and durabi lity, and its response under different weather conditions. In addition, these SMSs were buried at 7-10 cm, meaning that they were placed where the soil is more susceptible to changes in its moisture content. Theoretically (and this should be tested), burying them a little bit deeper and/or setting the SMS-controllers to a dryer condition could pr omote the turfgrass to produce a longer root system and, consequently, could result in even less act ual irrigation frequency and water use.

PAGE 159

143 Finally, these SMS-based results, reinfo rced by other experiments using this technology, open the possibility of redefini ng the BMPs for residential turfgrass irrigation, and for review and further discussi on of the state’s wate ring restrictions as well.

PAGE 160

144 APPENDIX A LIST OF ABBREVIATIONS AC : Acclima BMP : Best Management Practices CCL : Closed Control Loop DWRS : 60% deficit with rain sensor d/w : days per week ET : Evapotranspiration ETo : Potential or Reference Evapotranspiration FDR : Frequency Domain Reflectrometery GMS : Granular Matrix Sensor IM : Irrometer MC : Mini-Click rain sensor NI : No irrigation RB : Rain Bird RS : Rain Sensor SIC : Scheduled Irrigation Cycles SJRWMD : St. Johns River Water Management District SMS : Soil Moisture Sensor SWB : Simplified Water Balance SWT : Soil Water Tension TDR : Time Domain Reflectometry U.S. : United States of America VMC : Volumetric Moisture Content WL : Wireless Rain-Click rain sensor WMD : Water Management District WORS : Without rain sensor WRS : With rain sensor WW : Water Watcher

PAGE 161

145 APPENDIX B STATISTICAL ANALYSES The following are the SAS codes and the output text files for the statistical analyses performed. DRY-WET CODES options nodate nonumber center formdlim= "*" linesize= 85 ; data sms; input plot$ rep$ dry wet; cards ; /* Data is inputted here */ ; data sms; set sms; proc glm data =sms; title 'Dry' ; class plot; model dry = plot / ss3 ; means plot/ duncan ; run ; data sms; set sms; proc glm data =sms; title 'Wet' ; class plot; model wet = plot / ss3 ; means plot/ duncan ; run ;

PAGE 162

146 DRY-WET ANALYSIS OUTPUT ************************************************************************************* Dry The GLM Procedure Class Level Information Class Levels Values plot 72 A01 A02 A03 A04 A05 A06 A07 A08 A09 A10 A11 A12 B01 B02 B03 B04 B05 B06 B07 B08 B09 B10 B11 B12 C01 C02 C03 C04 C05 C06 C07 C08 C09 C10 C11 C12 D01 D02 D03 D04 D05 D06 D07 D08 D09 D10 D11 D12 E01 E02 E03 E04 E05 E06 E07 E08 E09 E10 E11 E12 F01 F02 F03 F04 F05 F06 F07 F08 F09 F10 F11 F12 Number of Observations Read 360 Number of Observations Used 360 ************************************************************************************* At once Dry The GLM Procedure Dependent Variable: dry Sum of Source DF Squares Mean Square F Value Pr > F Model 71 1524.922222 21.477778 7.19 <.0001 Error 288 860.800000 2.988889 Corrected Total 359 2385.722222 R-Square Coeff Var Root MSE dry Mean 0.639187 24.79612 1.728840 6.972222 Source DF Type III SS Mean Square F Value Pr > F plot 71 1524.922222 21.477778 7.19 <.0001 *************************************************************************************

PAGE 163

147 Dry The GLM Procedure Duncan's Multiple Range Test for dry NOTE: This test controls the Type I comparisonwise error rate, not the experimentwise error rate. Alpha 0.05 Error Degrees of Freedom 288 Error Mean Square 2.988889 Number of Means 2 3 4 5 6 7 8 9 10 11 12 Critical Range 2.152 2.266 2.341 2.397 2.441 2.476 2.506 2.531 2.553 2.572 2.589 Number of Means 13 14 15 16 17 18 19 20 21 22 23 Critical Range 2.605 2.618 2.631 2.642 2.653 2.663 2.671 2.680 2.687 2.695 2.701 Number of Means 24 25 26 27 28 29 30 31 32 33 34 Critical Range 2.708 2.714 2.719 2.725 2.730 2.734 2.739 2.743 2.747 2.751 2.755 Number of Means 35 36 37 38 39 40 41 42 43 44 45 Critical Range 2.758 2.761 2.765 2.768 2.771 2.773 2.776 2.779 2.781 2.784 2.786 Number of Means 46 47 48 49 50 51 52 53 54 55 56 Critical Range 2.788 2.790 2.792 2.794 2.796 2.798 2.800 2.801 2.803 2.804 2.806 Number of Means 57 58 59 60 61 62 63 64 65 66 67 Critical Range 2.807 2.809 2.810 2.811 2.813 2.814 2.815 2.816 2.817 2.818 2.819 Number of Means 68 69 70 71 72 Critical Range 2.820 2.821 2.822 2.823 2.824 Means with the same letter are not significantly different. Duncan Grouping Mean N plot A 15.200 5 A12 A A 14.000 5 A11 B 11.400 5 D11 B C B 10.200 5 D10 C B C B D 10.000 5 D09 C B D C E B D 9.600 5 C08 C E B D F C E B D 9.400 5 D06 F C E B D F C E B D 9.400 5 E08

PAGE 164

148 F C E B D F C E B D G 9.200 5 D05 F C E B D G F C E B D G 9.200 5 E07 F C E B D G F C E B H D G 9.000 5 F07 F C E H D G F C E I H D G 8.800 5 D12 F C E I H D G F C E J I H D G 8.400 5 F01 F C E J I H D G F C E J I H D G 8.400 5 F08 F C E J I H D G F C E J I H D G 8.400 5 E04 F C E J I H D G F C K E J I H D G 8.200 5 E01 F C K E J I H D G F L C K E J I H D G 8.000 5 E09 F L C K E J I H D G F L C K E J I H D G 8.000 5 E06 F L C K E J I H D G M F L C K E J I H D G 7.800 5 D07 M F L C K E J I H D G M F L C K E J I H D G 7.800 5 D08 M F L C K E J I H D G M F L C K E J I H D G 7.800 5 D04 M F L C K E J I H D G M F L C K E J I H D G N 7.600 5 F09 M F L C K E J I H D G N M F L C K E J I H D G N 7.600 5 E05 M F L K E J I H D G N M F L K E J I H D G N 7.400 5 C12 M F L K E J I H D G N M F L K E J I H D G N 7.400 5 D03 M F L K E J I H D G N M F L K E J I H D G N 7.400 5 C09 M F L K E J I H D G N M F L K E J I H D G N 7.400 5 B05 M F L K E J I H G N M F L O K E J I H G N 7.200 5 E03 M F L O K E J I H G N M F L O K E J I H G N 7.200 5 B01 M F L O K E J I H G N M F L O K E J I H P G N 7.000 5 F10 M F L O K J I H P G N M F L O K J I H P G N 6.800 5 C10 M F L O K J I H P G N M F L O K J I H P G N 6.800 5 D02 M F L O K J I H P G N M F L O K J I H P G N 6.800 5 E11 M L O K J I H P G N M L O K Q J I H P G N 6.600 5 E02 M L O K Q J I H P G N M L O K Q J I H P G N 6.600 5 C06 M L O K Q J I H P G N M L O K Q J I H P G N 6.600 5 D01 M L O K Q J I H P N

PAGE 165

149 M L O K Q J I H P N 6.400 5 A10 M L O K Q J I H P N M L O K Q J I H P N 6.400 5 B12 M L O K Q J I H P N M L O K Q J I H P N 6.400 5 C07 M L O K Q J I H P N M L O K Q J I H P N 6.400 5 F04 M L O K Q J I P N M R L O K Q J I P N 6.200 5 C02 M R L O K Q J I P N M R L O K Q J I P N 6.200 5 B10 M R L O K Q J P N M R L O K Q J P N 6.000 5 B03 M R L O K Q J P N M R L O K Q J P N 6.000 5 C03 M R L O K Q J P N M R L O K Q J P N 6.000 5 A09 M R L O K Q J P N M R L O K Q J P N 6.000 5 E10 M R L O K Q J P N M R L O K Q J P N 6.000 5 C01 M R L O K Q J P N M R L O K Q J P N 6.000 5 B09 M R L O K Q J P N M R L O K Q J P N 5.800 5 A02 M R L O K Q J P N M R L O K Q J P N 5.800 5 F02 M R L O K Q J P N M R L O K Q J P N 5.800 5 A03 M R L O K Q J P N M R L O K Q J P N 5.800 5 F06 M R L O K Q J P N M R L O K Q J P N 5.800 5 C05 M R L O K Q P N M R L O K Q P N 5.600 5 C04 M R L O K Q P N M R L O K Q P N 5.600 5 F03 M R L O K Q P N M R L O K Q P N 5.600 5 B08 M R L O K Q P N M R L O K Q P N 5.600 5 B07 M R L O K Q P N M R L O K Q P N 5.600 5 E12 M R L O K Q P N M R L O K Q P N 5.600 5 B02 M R L O Q P N M R L O Q P N 5.400 5 B11 M R L O Q P N M R L O Q P N 5.400 5 C11 M R L O Q P N M R L O Q P N 5.400 5 A06 M R O Q P N M R O Q P N 5.200 5 F05 R O Q P N R O Q P N 5.000 5 B04 R O Q P R O Q P 4.600 5 A08

PAGE 166

150 R O Q P R O Q P 4.600 5 B06 R Q P R Q P 4.400 5 A01 R Q P R Q P 4.400 5 A07 R Q P R Q P 4.400 5 A04 R Q P R Q P 4.400 5 F12 R Q R Q 4.000 5 F11 R R 3.600 5 A05 ************************************************************************************* Wet The GLM Procedure Class Level Information Class Levels Values plot 72 A01 A02 A03 A04 A05 A06 A07 A08 A09 A10 A11 A12 B01 B02 B03 B04 B05 B06 B07 B08 B09 B10 B11 B12 C01 C02 C03 C04 C05 C06 C07 C08 C09 C10 C11 C12 D01 D02 D03 D04 D05 D06 D07 D08 D09 D10 D11 D12 E01 E02 E03 E04 E05 E06 E07 E08 E09 E10 E11 E12 F01 F02 F03 F04 F05 F06 F07 F08 F09 F10 F11 F12 Number of Observations Read 360 Number of Observations Used 360 *************************************************************************************

PAGE 167

151 Wet The GLM Procedure Dependent Variable: wet Sum of Source DF Squares Mean Square F Value Pr > F Model 71 2421.600000 34.107042 18.19 <.0001 Error 288 540.000000 1.875000 Corrected Total 359 2961.600000 R-Square Coeff Var Root MSE wet Mean 0.817666 12.11776 1.369306 11.30000 Source DF Type III SS Mean Square F Value Pr > F plot 71 2421.600000 34.107042 18.19 <.0001 ************************************************************************************* Wet The GLM Procedure Duncan's Multiple Range Test for wet NOTE: This test controls the Type I comparisonwise error rate, not the experimentwise error rate. Alpha 0.05 Error Degrees of Freedom 288 Error Mean Square 1.875 Number of Means 2 3 4 5 6 7 8 9 10 11 12 Critical Range 1.705 1.794 1.854 1.899 1.933 1.961 1.985 2.005 2.022 2.037 2.051 Number of Means 13 14 15 16 17 18 19 20 21 22 23 Critical Range 2.063 2.074 2.084 2.093 2.101 2.109 2.116 2.122 2.129 2.134 2.140 Number of Means 24 25 26 27 28 29 30 31 32 33 34 Critical Range 2.145 2.149 2.154 2.158 2.162 2.166 2.169 2.173 2.176 2.179 2.182 Number of Means 35 36 37 38 39 40 41 42 43 44 45 Critical Range 2.185 2.187 2.190 2.192 2.194 2.197 2.199 2.201 2.203 2.205 2.206 Number of Means 46 47 48 49 50 51 52 53 54 55 56 Critical Range 2.208 2.210 2.211 2.213 2.215 2.216 2.217 2.219 2.220 2.221 2.222

PAGE 168

152 Number of Means 57 58 59 60 61 62 63 64 65 66 67 Critical Range 2.224 2.225 2.226 2.227 2.228 2.229 2.230 2.231 2.231 2.232 2.233 Number of Means 68 69 70 71 72 Critical Range 2.234 2.235 2.235 2.236 2.237 Means with the same letter are not significantly different. Duncan Grouping Mean N plot A 27.8000 5 A11 B 20.8000 5 A12 C 15.8000 5 F01 C C 14.8000 5 E01 D 12.8000 5 D11 D E D 12.6000 5 C08 E D E D F 12.4000 5 A10 E D F E D F 12.4000 5 F12 E D F E D F 12.2000 5 A01 E D F E D F 12.2000 5 D08 E D F E D F 12.2000 5 C11 E D F E D F 12.2000 5 D12 E D F E D F 12.2000 5 B05 E D F E G D F 12.0000 5 A02 E G D F E G D F 12.0000 5 C07 E G D F E G D F 12.0000 5 E02 E G D F E G D F 12.0000 5 D01 E G D F E G D F 12.0000 5 C03 E G D F H E G D F 11.8000 5 E08 H E G D F H E G D F I 11.6000 5 B06 H E G D F I H E G D F I 11.6000 5 B01 H E G D F I H E G D F I 11.6000 5 C12 H E G D F I H E G D F I 11.6000 5 B11

PAGE 169

153 H E G D F I H E G D J F I 11.4000 5 A08 H E G D J F I H E G D J F I 11.4000 5 A09 H E G D J F I H E G D J F I 11.4000 5 D09 H E G D J F I H E G D J F I 11.2000 5 D02 H E G D J F I H E G D J F I 11.2000 5 B02 H E G D J F I H E G D J F I 11.2000 5 F07 H E G D J F I H E G D J F I 11.2000 5 C01 H E G D J F I H E G D J F I 11.2000 5 B09 H E G D J F I H E G D J F I 11.0000 5 E12 H E G D J F I H E G D J F I 11.0000 5 B12 H E G D J F I H E G D J F I 11.0000 5 E07 H E G D J F I H E G D J F I 11.0000 5 D07 H E G D J F I H E G D J F I 11.0000 5 C09 H E G D J F I H E G K D J F I 10.8000 5 E10 H E G K D J F I H E G K D J F I 10.8000 5 D06 H E G K D J F I H E G K D J F I 10.8000 5 E03 H E G K D J F I H E G K D J F I 10.8000 5 E11 H E G K D J F I H E G K D J F I 10.8000 5 F09 H E G K J F I H E G K L J F I 10.6000 5 A06 H E G K L J F I H E G K L J F I 10.6000 5 F02 H E G K L J F I H E G K L J F I 10.6000 5 B08 H E G K L J F I H E G K L J F I 10.6000 5 F08 H E G K L J F I H E G K L J F I 10.4000 5 D10 H E G K L J F I H E G K L J F I 10.4000 5 D03 H E G K L J F I H E G K L J F I 10.4000 5 C10 H E G K L J F I H E G K L J F I 10.4000 5 E09 H E G K L J F I H E G K L J F I 10.4000 5 E05 H E G K L J F I H E G K L J F I 10.4000 5 C05 H G K L J F I

PAGE 170

154 H G K L J F I 10.2000 5 C06 H G K L J F I H G K L J F I 10.2000 5 B04 H G K L J F I H G K L J F I 10.2000 5 C02 H G K L J F I H G K L J F I 10.2000 5 A07 H G K L J I H G K L J I 9.8000 5 A04 H G K L J I H G K L J I 9.8000 5 B03 H G K L J I H G K L J I 9.8000 5 F10 H G K L J I H G K L J I 9.8000 5 A03 H G K L J I H G K L J I 9.8000 5 C04 H G K L J I H G K L J I 9.8000 5 F05 H K L J I H K L J I 9.6000 5 E06 H K L J I H K L J I 9.6000 5 B07 H K L J I H K L J I 9.6000 5 D05 K L J I K L J I 9.4000 5 D04 K L J I K L J I 9.4000 5 F11 K L J K L J 9.2000 5 F06 K L J K L J 9.2000 5 F04 K L J K L J 9.2000 5 A05 K L K L 8.8000 5 B10 K L K L 8.8000 5 E04 L L 8.6000 5 F03

PAGE 171

155 PLOTS WITH SMS CODES options nodate nonumber center formdlim= "*" linesize= 85 ; data sms; input plot$ rep$ dry wet; cards ; /* Data is inputted here */ ; data sms; set sms; proc glm data =sms; title 'Plots with SMS-Dry' ; class plot; model dry = plot / ss3 ; means plot/ duncan ; run ; data sms; set sms; proc glm data =sms; title 'Plots with SMS-Wet ; class plot; model wet = plot / ss3 ; means plot/ duncan ; run ;

PAGE 172

156 PLOTS WITH SMS ANALYSIS OUTPUT ************************************************************************************* Plots with SMS-Dry The GLM Procedure Class Level Information Class Levels Values plot 12 A02 B03 B08 B09 C02 C05 D02 E02 F02 F03 F04 F05 Number of Observations Read 60 Number of Observations Used 60 ************************************************************************************* Plots with SMS-Dry The GLM Procedure Dependent Variable: dry Sum of Source DF Squares Mean Square F Value Pr > F Model 11 11.3833333 1.0348485 0.43 0.9350 Error 48 115.6000000 2.4083333 Corrected Total 59 126.9833333 R-Square Coeff Var Root MSE dry Mean 0.089644 25.93672 1.551881 5.983333 Source DF Type III SS Mean Square F Value Pr > F plot 11 11.38333333 1.03484848 0.43 0.9350 ************************************************************************************* Plots with SMS-Dry The GLM Procedure Duncan's Multiple Range Test for dry NOTE: This test controls the Type I comparisonwise error rate, not the experimentwise error rate.

PAGE 173

157 Alpha 0.05 Error Degrees of Freedom 48 Error Mean Square 2.408333 Number of Means 2 3 4 5 6 7 8 9 10 11 12 Critical Range 1.973 2.075 2.143 2.191 2.228 2.258 2.282 2.302 2.320 2.334 2.347 Means with the same letter are not significantly different. Duncan Grouping Mean N plot A 6.8000 5 D02 A A 6.6000 5 E02 A A 6.4000 5 F04 A A 6.2000 5 C02 A A 6.0000 5 B03 A A 6.0000 5 B09 A A 5.8000 5 A02 A A 5.8000 5 C05 A A 5.8000 5 F02 A A 5.6000 5 F03 A A 5.6000 5 B08 A A 5.2000 5 F05 ************************************************************************************* Plots with SMS-Wet The GLM Procedure Class Level Information Class Levels Values plot 12 A02 B03 B08 B09 C02 C05 D02 E02 F02 F03 F04 F05 Number of Observations Read 60 Number of Observations Used 60 *************************************************************************************

PAGE 174

158 Plots with SMS-Wet The GLM Procedure Dependent Variable: wet Sum of Source DF Squares Mean Square F Value Pr > F Model 11 59.3333333 5.3939394 5.22 <.0001 Error 48 49.6000000 1.0333333 Corrected Total 59 108.9333333 R-Square Coeff Var Root MSE wet Mean 0.544676 9.712070 1.016530 10.46667 Source DF Type III SS Mean Square F Value Pr > F plot 11 59.33333333 5.39393939 5.22 <.0001 ************************************************************************************* Plots with SMS-Wet The GLM Procedure Duncan's Multiple Range Test for wet NOTE: This test controls the Type I comparisonwise error rate, not the experimentwise error rate. Alpha 0.05 Error Degrees of Freedom 48 Error Mean Square 1.033333 Number of Means 2 3 4 5 6 7 8 9 10 11 12 Critical Range 1.293 1.360 1.403 1.435 1.460 1.479 1.495 1.508 1.519 1.529 1.537 Means with the same letter are not significantly different. Duncan Grouping Mean N plot A 12.0000 5 A02 A A 12.0000 5 E02 A

PAGE 175

159 B A 11.2000 5 D02 B A B A 11.2000 5 B09 B A B A C 10.6000 5 F02 B A C B A C 10.6000 5 B08 B C B C 10.4000 5 C05 B C B C 10.2000 5 C02 B C B D C 9.8000 5 B03 B D C B D C 9.8000 5 F05 D C D C 9.2000 5 F04 D D 8.6000 5 F03 *************************************************************************************

PAGE 176

160 CUMULATIVE IRRIGATION DEPTH, YEAR 2004 options nodate nonumber center formdlim= "*" linesize= 85 ; data sms; input tmt$ day$ brand$ type$ based$ timeb$ mm; cards ; /* Data is inputted here */ ; data sms2; set sms(where=(type = 'sms' )); proc glm data =sms2; title 'TOTAL Cumulative mm (21 July 2004 – 14 Dec 2004)' ; class brand day; model mm = day brand day(brand) / ss3 ; test h =brand e =day(brand); means day/ duncan ; means brand/ duncan ; means brand/ duncan e =day(brand); run ; proc glm data =sms2; title 'Comparison of Interaction' ; class brand day; model mm = brand*day / ss3 ; means brand*day/ duncan ; run ; data sms3; set sms(where=(day = '1' )); proc glm data =sms3; title 'Cumulative mm-Once per Week' ; class brand ; model mm = brand/ ss3 ; means brand/ duncan ; run ; data sms3; set sms(where=(day = '2' )); proc glm data =sms3; title 'Cumulative mm-Twice per Week' ; class brand ; model mm = brand/ ss3 ; means brand/ duncan ; run ; data sms3; set sms(where=(day = '7' )); proc glm data =sms3; title 'Cumulative mm-Everyday' ; class brand ; model mm = brand/ ss3 ; means brand/ duncan ; run ; data sms; set sms; proc glm data =sms; title 'Comparison of Time-based treatments' ; class timeb; model mm = timeb / ss3 ;

PAGE 177

161 means timeb/ duncan ; run ; proc glm data =sms; title 'Comparison of Sensor Type -By Individual Treatment (SMS, WRS, WORS, DWRS)' ; class tmt; model mm = tmt / ss3 ; means tmt/ duncan ; run ; ************************************************************************************* TOTAL Cumulative mm (21 July 2004 14 Dec 2004) The GLM Procedure Class Level Information Class Levels Values brand 4 ac im rb ww day 3 1 2 7 Number of Observations Read 48 Number of Observations Used 48 ************************************************************************************* TOTAL Cumulative mm (21 July 2004 14 Dec 2004) The GLM Procedure Dependent Variable: mm Sum of Source DF Squares Mean Square F Value Pr > F Model 11 950553.2500 86413.9318 201.04 <.0001 Error 36 15474.0000 429.8333 Corrected Total 47 966027.2500 R-Square Coeff Var Root MSE mm Mean 0.983982 10.07039 20.73242 205.8750

PAGE 178

162 Source DF Type III SS Mean Square F Value Pr > F day 2 8419.6250 4209.8125 9.79 0.0004 brand 3 784008.7500 261336.2500 607.99 <.0001 day(brand) 6 158124.8750 26354.1458 61.31 <.0001 Tests of Hypotheses Using the Type III MS for day(brand) as an Error Term Source DF Type III SS Mean Square F Value Pr > F brand 3 784008.7500 261336.2500 9.92 0.0097 ************************************************************************************* TOTAL Cumulative mm (21 July 2004 14 Dec 2004) The GLM Procedure Duncan's Multiple Range Test for mm NOTE: This test controls the Type I comparisonwise error rate, not the experimentwise error rate. Alpha 0.05 Error Degrees of Freedom 36 Error Mean Square 429.8333 Number of Means 2 3 Critical Range 14.87 15.63 Means with the same letter are not significantly different. Duncan Grouping Mean N day A 218.438 16 7 A A 211.625 16 2 B 187.563 16 1 ************************************************************************************* TOTAL Cumulative mm (21 July 2004 14 Dec 2004) The GLM Procedure Duncan's Multiple Range Test for mm NOTE: This test controls the Type I comparisonwise error rate, not the experimentwise error rate.

PAGE 179

163 Alpha 0.05 Error Degrees of Freedom 36 Error Mean Square 429.8333 Number of Means 2 3 4 Critical Range 17.17 18.05 18.62 Means with the same letter are not significantly different. Duncan Grouping Mean N brand A 419.667 12 im B 187.917 12 ww C 116.000 12 ac C C 99.917 12 rb ************************************************************************************* TOTAL Cumulative mm (21 July 2004 14 Dec 2004) The GLM Procedure Duncan's Multiple Range Test for mm NOTE: This test controls the Type I comparisonwise error rate, not the experimentwise error rate. Alpha 0.05 Error Degrees of Freedom 6 Error Mean Square 26354.15 Number of Means 2 3 4 Critical Range 162.2 168.1 171.0 Means with the same letter are not significantly different. Duncan Grouping Mean N brand A 419.67 12 im B 187.92 12 ww B B 116.00 12 ac B B 99.92 12 rb

PAGE 180

164 ************************************************************************************* Comparison of Interaction The GLM Procedure Class Level Information Class Levels Values brand 4 ac im rb ww day 3 1 2 7 Number of Observations Read 48 Number of Observations Used 48 ************************************************************************************* Comparison of Interaction The GLM Procedure Dependent Variable: mm Sum of Source DF Squares Mean Square F Value Pr > F Model 11 950553.2500 86413.9318 201.04 <.0001 Error 36 15474.0000 429.8333 Corrected Total 47 966027.2500 R-Square Coeff Var Root MSE mm Mean 0.983982 10.07039 20.73242 205.8750 Source DF Type III SS Mean Square F Value Pr > F brand*day 11 950553.2500 86413.9318 201.04 <.0001 ************************************************************************************* Comparison of Interaction The GLM Procedure Level of Level of --------------mm------------brand day N Mean Std Dev ac 1 4 95.250000 2.0615528 ac 2 4 196.000000 8.7939373 ac 7 4 56.750000 2.6299556

PAGE 181

165 im 1 4 318.000000 23.1084400 im 2 4 470.000000 12.8322510 im 7 4 471.000000 62.4980000 rb 1 4 127.750000 12.3659479 rb 2 4 86.750000 6.0759087 rb 7 4 85.250000 6.6520673 ww 1 4 209.250000 8.9953692 ww 2 4 93.750000 7.8049130 ww 7 4 260.750000 9.4295634 ************************************************************************************* Cumulative mm-Once per Week The GLM Procedure Class Level Information Class Levels Values brand 4 ac im rb ww Number of Observations Read 16 Number of Observations Used 16 ************************************************************************************* Cumulative mm-Once per Week The GLM Procedure Dependent Variable: mm Sum of Source DF Squares Mean Square F Value Pr > F Model 3 118333.6875 39444.5625 204.35 <.0001 Error 12 2316.2500 193.0208 Corrected Total 15 120649.9375 R-Square Coeff Var Root MSE mm Mean 0.980802 7.407234 13.89319 187.5625 Source DF Type III SS Mean Square F Value Pr > F brand 3 118333.6875 39444.5625 204.35 <.0001

PAGE 182

166 Cumulative mm-Once per Week The GLM Procedure Duncan's Multiple Range Test for mm NOTE: This test controls the Type I comparisonwise error rate, not the experimentwise error rate. Alpha 0.05 Error Degrees of Freedom 12 Error Mean Square 193.0208 Number of Means 2 3 4 Critical Range 21.40 22.40 23.01 Means with the same letter are not significantly different. Duncan Grouping Mean N brand A 318.000 4 im B 209.250 4 ww C 127.750 4 rb D 95.250 4 ac ************************************************************************************* Cumulative mm-Twice per Week The GLM Procedure Class Level Information Class Levels Values brand 4 ac im rb ww Number of Observations Read 16 Number of Observations Used 16 ************************************************************************************* Cumulative mm-Twice per Week The GLM Procedure Dependent Variable: mm

PAGE 183

167 Sum of Source DF Squares Mean Square F Value Pr > F Model 3 385960.2500 128653.4167 1514.31 <.0001 Error 12 1019.5000 84.9583 Corrected Total 15 386979.7500 R-Square Coeff Var Root MSE mm Mean 0.997365 4.355480 9.217284 211.6250 Source DF Type III SS Mean Square F Value Pr > F brand 3 385960.2500 128653.4167 1514.31 <.0001 ************************************************************************************* Cumulative mm-Twice per Week The GLM Procedure Duncan's Multiple Range Test for mm NOTE: This test controls the Type I comparisonwise error rate, not the experimentwise error rate. Alpha 0.05 Error Degrees of Freedom 12 Error Mean Square 84.95833 Number of Means 2 3 4 Critical Range 14.20 14.86 15.27 Means with the same letter are not significantly different. Duncan Grouping Mean N brand A 470.000 4 im B 196.000 4 ac C 93.750 4 ww C C 86.750 4 rb

PAGE 184

168 Cumulative mm-Everyday The GLM Procedure Class Level Information Class Levels Values brand 4 ac im rb ww Number of Observations Read 16 Number of Observations Used 16 ************************************************************************************* Cumulative mm-Everyday The GLM Procedure Dependent Variable: mm Sum of Source DF Squares Mean Square F Value Pr > F Model 3 437839.6875 145946.5625 144.28 <.0001 Error 12 12138.2500 1011.5208 Corrected Total 15 449977.9375 R-Square Coeff Var Root MSE mm Mean 0.973025 14.55996 31.80442 218.4375 Source DF Type III SS Mean Square F Value Pr > F brand 3 437839.6875 145946.5625 144.28 <.0001 ************************************************************************************* Cumulative mm-Everyday The GLM Procedure Duncan's Multiple Range Test for mm NOTE: This test controls the Type I comparisonwise error rate, not the experimentwise error rate. Alpha 0.05 Error Degrees of Freedom 12 Error Mean Square 1011.521

PAGE 185

169 Number of Means 2 3 4 Critical Range 49.00 51.29 52.68 Means with the same letter are not significantly different. Duncan Grouping Mean N brand A 471.00 4 im B 260.75 4 ww C 85.25 4 rb C C 56.75 4 ac ************************************************************************************* Comparison of Time-based treatments The GLM Procedure Class Level Information Class Levels Values timeb 3 dwrs wors wrs Number of Observations Read 60 Number of Observations Used 12 ************************************************************************************* Comparison of Time-based treatments The GLM Procedure Dependent Variable: mm Sum of Source DF Squares Mean Square F Value Pr > F Model 2 299253.5000 149626.7500 251.96 <.0001 Error 9 5344.7500 593.8611 Corrected Total 11 304598.2500 R-Square Coeff Var Root MSE mm Mean 0.982453 4.915636 24.36927 495.7500

PAGE 186

170 Source DF Type III SS Mean Square F Value Pr > F timeb 2 299253.5000 149626.7500 251.96 <.0001 ************************************************************************************* Comparison of Time-based treatments The GLM Procedure Duncan's Multiple Range Test for mm NOTE: This test controls the Type I comparisonwise error rate, not the experimentwise error rate. Alpha 0.05 Error Degrees of Freedom 9 Error Mean Square 593.8611 Number of Means 2 3 Critical Range 38.98 40.69 Means with the same letter are not significantly different. Duncan Grouping Mean N timeb A 696.00 4 wors B 481.25 4 wrs C 310.00 4 dwrs ************************************************************************************* Comparison of Sensor Type -By Individual Treatment (SMS, WRS, WORS, DWRS) The GLM Procedure Class Level Information Class Levels Values tmt 15 1ac 1im 1rb 1ww 2ac 2dwrs 2im 2rb 2wors 2wrs 2ww 7ac 7im 7rb 7ww Number of Observations Read 60 Number of Observations Used 60 ************************************************************************************* Comparison of Sensor Type -By Individual Treatment (SMS, WRS, WORS, DWRS)

PAGE 187

171 The GLM Procedure Dependent Variable: mm Sum of Source DF Squares Mean Square F Value Pr > F Model 14 2056470.900 146890.779 317.51 <.0001 Error 45 20818.750 462.639 Corrected Total 59 2077289.650 R-Square Coeff Var Root MSE mm Mean 0.989978 8.151996 21.50904 263.8500 Source DF Type III SS Mean Square F Value Pr > F tmt 14 2056470.900 146890.779 317.51 <.0001 ************************************************************************************* Comparison of Sensor Type -By Individual Treatment (SMS, WRS, WORS, DWRS) The GLM Procedure Duncan's Multiple Range Test for mm NOTE: This test controls the Type I comparisonwise error rate, not the experimentwise error rate. Alpha 0.05 Error Degrees of Freedom 45 Error Mean Square 462.6389 Number of Means 2 3 4 5 6 7 8 Critical Range 30.63 32.21 33.25 34.00 34.58 35.03 35.41 Number of Means 9 10 11 12 13 14 15 Critical Range 35.72 35.98 36.20 36.40 36.56 36.71 36.84 Means with the same letter are not significantly different. Duncan Grouping Mean N tmt A 696.00 4 2wors B 481.25 4 2wrs B B 471.00 4 7im

PAGE 188

172 B B 470.00 4 2im C 318.00 4 1im C C 310.00 4 2dwrs D 260.75 4 7ww E 209.25 4 1ww E E 196.00 4 2ac F 127.75 4 1rb G 95.25 4 1ac G G 93.75 4 2ww G H G 86.75 4 2rb H G H G 85.25 4 7rb H H 56.75 4 7ac

PAGE 189

173 CUMULATIVE IRRIGATION DEPTH, YEAR 2005 options nodate nonumber center formdlim= "*" linesize= 85 ; data sms; input tmt$ day$ brand$ type$ based$ timeb$ mm; cards ; /* Data is inputted here */ ; data sms2; set sms(where=(type = 'sms' )); proc glm data =sms2; title 'Cumulative mm (March 25 Aug 31/2005)' ; class brand day; model mm = day brand day(brand) / ss3 ; test h =brand e =day(brand); means day/ duncan ; means brand/ duncan ; means brand/ duncan e =day(brand); run ; proc glm data =sms2; title 'Comparison of Interaction' ; class brand day; model mm = brand*day / ss3 ; means brand*day/ duncan ; run ; data sms3; set sms(where=(day = '1' )); proc glm data =sms3; title 'Cumulative mm-Once per Week' ; class brand ; model mm = brand/ ss3 ; means brand/ duncan ; run ; data sms3; set sms(where=(day = '2' )); proc glm data =sms3; title 'Cumulative mm-Twice per Week' ; class brand ; model mm = brand/ ss3 ; means brand/ duncan ; run ; data sms3; set sms(where=(day = '7' )); proc glm data =sms3; title 'Cumulative mm-Everyday' ; class brand ; model mm = brand/ ss3 ; means brand/ duncan ; run ; data sms; set sms; proc glm data =sms; title 'Comparison of Time-based treatments' ; class timeb; model mm = timeb / ss3 ; means timeb/ duncan ;

PAGE 190

174 run ; proc glm data =sms; title 'Comparison of Sensor Type -By Individual Treatment (SMS, WRS, WORS, DWRS)' ; class tmt; model mm = tmt / ss3 ; means tmt/ duncan ; run ; ************************************************************************************* Cumulative mm (March 25 Aug 31/2005) The GLM Procedure Class Level Information Class Levels Values brand 4 ac im rb ww day 3 1 2 7 Number of Observations Read 48 Number of Observations Used 48 ************************************************************************************* Cumulative mm (March 25 Aug 31/2005) The GLM Procedure Dependent Variable: mm Sum of Source DF Squares Mean Square F Value Pr > F Model 11 1296396.167 117854.197 620.33 <.0001 Error 36 6839.500 189.986 Corrected Total 47 1303235.667 R-Square Coeff Var Root MSE mm Mean 0.994752 6.443418 13.78354 213.9167

PAGE 191

175 Source DF Type III SS Mean Square F Value Pr > F day 2 128785.0417 64392.5208 338.93 <.0001 brand 3 923152.6667 307717.5556 1619.68 <.0001 day(brand) 6 244458.4583 40743.0764 214.45 <.0001 Tests of Hypotheses Using the Type III MS for day(brand) as an Error Term Source DF Type III SS Mean Square F Value Pr > F brand 3 923152.6667 307717.5556 7.55 0.0184 ************************************************************************************* Cumulative mm (March 25 Aug 31/2005) The GLM Procedure Duncan's Multiple Range Test for mm NOTE: This test controls the Type I comparisonwise error rate, not the experimentwise error rate. Alpha 0.05 Error Degrees of Freedom 36 Error Mean Square 189.9861 Number of Means 2 3 Critical Range 9.88 10.39 Means with the same letter are not significantly different. Duncan Grouping Mean N day A 266.125 16 2 B 232.313 16 1 C 143.313 16 7 ************************************************************************************* Cumulative mm (March 25 Aug 31/2005) The GLM Procedure Duncan's Multiple Range Test for mm NOTE: This test controls the Type I comparisonwise error rate, not the experimentwise error rate.

PAGE 192

176 Alpha 0.05 Error Degrees of Freedom 36 Error Mean Square 189.9861 Number of Means 2 3 4 Critical Range 11.41 12.00 12.38 Means with the same letter are not significantly different. Duncan Grouping Mean N brand A 451.417 12 im B 163.917 12 ww C 135.083 12 ac D 105.250 12 rb ************************************************************************************* Cumulative mm (March 25 Aug 31/2005) The GLM Procedure Duncan's Multiple Range Test for mm NOTE: This test controls the Type I comparisonwise error rate, not the experimentwise error rate. Alpha 0.05 Error Degrees of Freedom 6 Error Mean Square 40743.08 Number of Means 2 3 4 Critical Range 201.6 209.0 212.6 Means with the same letter are not significantly different. Duncan Grouping Mean N brand A 451.42 12 im B 163.92 12 ww B B 135.08 12 ac B B 105.25 12 rb *************************************************************************************

PAGE 193

177 Comparison of Interaction The GLM Procedure Class Level Information Class Levels Values brand 4 ac im rb ww day 3 1 2 7 Number of Observations Read 48 Number of Observations Used 48 ************************************************************************************* Comparison of Interaction The GLM Procedure Dependent Variable: mm Sum of Source DF Squares Mean Square F Value Pr > F Model 11 1296396.167 117854.197 620.33 <.0001 Error 36 6839.500 189.986 Corrected Total 47 1303235.667 R-Square Coeff Var Root MSE mm Mean 0.994752 6.443418 13.78354 213.9167 Source DF Type III SS Mean Square F Value Pr > F brand*day 11 1296396.167 117854.197 620.33 <.0001 ************************************************************************************* Comparison of Interaction The GLM Procedure Level of Level of --------------mm------------brand day N Mean Std Dev ac 1 4 188.000000 13.5892114 ac 2 4 152.000000 14.1185457 ac 7 4 65.250000 2.9860788 im 1 4 475.250000 25.5000000

PAGE 194

178 im 2 4 634.750000 29.8817112 im 7 4 244.250000 5.0579970 rb 1 4 152.500000 10.5356538 rb 2 4 101.250000 1.2583057 rb 7 4 62.000000 3.5590261 ww 1 4 113.500000 9.3273791 ww 2 4 176.500000 7.1414284 ww 7 4 201.750000 7.4105780 ************************************************************************************* Cumulative mm-Once per Week The GLM Procedure Class Level Information Class Levels Values brand 4 ac im rb ww Number of Observations Read 16 Number of Observations Used 16 ************************************************************************************* Cumulative mm-Once per Week The GLM Procedure Dependent Variable: mm Sum of Source DF Squares Mean Square F Value Pr > F Model 3 325874.6875 108624.8958 420.65 <.0001 Error 12 3098.7500 258.2292 Corrected Total 15 328973.4375 R-Square Coeff Var Root MSE mm Mean 0.990581 6.917196 16.06951 232.3125 Source DF Type III SS Mean Square F Value Pr > F brand 3 325874.6875 108624.8958 420.65 <.0001

PAGE 195

179 Cumulative mm-Once per Week The GLM Procedure Duncan's Multiple Range Test for mm NOTE: This test controls the Type I comparisonwise error rate, not the experimentwise error rate. Alpha 0.05 Error Degrees of Freedom 12 Error Mean Square 258.2292 Number of Means 2 3 4 Critical Range 24.76 25.91 26.61 Means with the same letter are not significantly different. Duncan Grouping Mean N brand A 475.25 4 im B 188.00 4 ac C 152.50 4 rb D 113.50 4 ww ************************************************************************************* Cumulative mm-Twice per Week The GLM Procedure Class Level Information Class Levels Values brand 4 ac im rb ww Number of Observations Read 16 Number of Observations Used 16 Cumulative mm-Twice per Week The GLM Procedure Dependent Variable: mm Sum of

PAGE 196

180 Source DF Squares Mean Square F Value Pr > F Model 3 736501.2500 245500.4167 857.77 <.0001 Error 12 3434.5000 286.2083 Corrected Total 15 739935.7500 R-Square Coeff Var Root MSE mm Mean 0.995358 6.357048 16.91769 266.1250 Source DF Type III SS Mean Square F Value Pr > F brand 3 736501.2500 245500.4167 857.77 <.0001 ************************************************************************************* Cumulative mm-Twice per Week The GLM Procedure Duncan's Multiple Range Test for mm NOTE: This test controls the Type I comparisonwise error rate, not the experimentwise error rate. Alpha 0.05 Error Degrees of Freedom 12 Error Mean Square 286.2083 Number of Means 2 3 4 Critical Range 26.06 27.28 28.02 Means with the same letter are not significantly different. Duncan Grouping Mean N brand A 634.75 4 im B 176.50 4 ww B B 152.00 4 ac C 101.25 4 rb

PAGE 197

181 Cumulative mm-Everyday The GLM Procedure Class Level Information Class Levels Values brand 4 ac im rb ww Number of Observations Read 16 Number of Observations Used 16 ************************************************************************************* Cumulative mm-Everyday The GLM Procedure Dependent Variable: mm Sum of Source DF Squares Mean Square F Value Pr > F Model 3 105235.1875 35078.3958 1374.50 <.0001 Error 12 306.2500 25.5208 Corrected Total 15 105541.4375 R-Square Coeff Var Root MSE mm Mean 0.997098 3.525034 5.051815 143.3125 Source DF Type III SS Mean Square F Value Pr > F brand 3 105235.1875 35078.3958 1374.50 <.0001 ************************************************************************************* Cumulative mm-Everyday The GLM Procedure Duncan's Multiple Range Test for mm NOTE: This test controls the Type I comparisonwise error rate, not the experimentwise error rate. Alpha 0.05 Error Degrees of Freedom 12 Error Mean Square 25.52083

PAGE 198

182 Number of Means 2 3 4 Critical Range 7.783 8.147 8.367 Means with the same letter are not significantly different. Duncan Grouping Mean N brand A 244.250 4 im B 201.750 4 ww C 65.250 4 ac C C 62.000 4 rb ************************************************************************************* Comparison of Time-based treatments The GLM Procedure Class Level Information Class Levels Values timeb 3 dwrs wors wrs Number of Observations Read 60 Number of Observations Used 12 ************************************************************************************* Comparison of Time-based treatments The GLM Procedure Dependent Variable: mm Sum of Source DF Squares Mean Square F Value Pr > F Model 2 517260.6667 258630.3333 1654.35 <.0001 Error 9 1407.0000 156.3333 Corrected Total 11 518667.6667 R-Square Coeff Var Root MSE mm Mean 0.997287 2.280936 12.50333 548.1667

PAGE 199

183 Source DF Type III SS Mean Square F Value Pr > F timeb 2 517260.6667 258630.3333 1654.35 <.0001 ************************************************************************************* Comparison of Time-based treatments The GLM Procedure Duncan's Multiple Range Test for mm NOTE: This test controls the Type I comparisonwise error rate, not the experimentwise error rate. Alpha 0.05 Error Degrees of Freedom 9 Error Mean Square 156.3333 Number of Means 2 3 Critical Range 20.00 20.88 Means with the same letter are not significantly different. Duncan Grouping Mean N timeb A 818.000 4 wors B 513.500 4 wrs C 313.000 4 dwrs ************************************************************************************* Comparison of Sensor Type -By Individual Treatment (SMS, WRS, WORS, DWRS) The GLM Procedure Class Level Information Class Levels Values tmt 15 1ac 1im 1rb 1ww 2ac 2dwrs 2im 2rb 2wors 2wrs 2ww 7ac 7im 7rb 7ww Number of Observations Read 60 Number of Observations Used 60 ************************************************************************************* Comparison of Sensor Type -By Individual Treatment (SMS, WRS, WORS, DWRS) The GLM Procedure

PAGE 200

184 Dependent Variable: mm Sum of Source DF Squares Mean Square F Value Pr > F Model 14 2886198.233 206157.017 1124.97 <.0001 Error 45 8246.500 183.256 Corrected Total 59 2894444.733 R-Square Coeff Var Root MSE mm Mean 0.997151 4.821510 13.53719 280.7667 Source DF Type III SS Mean Square F Value Pr > F tmt 14 2886198.233 206157.017 1124.97 <.0001 ************************************************************************************* Comparison of Sensor Type -By Individual Treatment (SMS, WRS, WORS, DWRS) The GLM Procedure Duncan's Multiple Range Test for mm NOTE: This test controls the Type I comparisonwise error rate, not the experimentwise error rate. Alpha 0.05 Error Degrees of Freedom 45 Error Mean Square 183.2556 Number of Means 2 3 4 5 6 7 8 Critical Range 19.28 20.27 20.93 21.40 21.76 22.05 22.28 Number of Means 9 10 11 12 13 14 15 Critical Range 22.48 22.64 22.79 22.91 23.01 23.11 23.19 Means with the same letter are not significantly different. Duncan Grouping Mean N tmt A 818.000 4 2wors B 634.750 4 2im C 513.500 4 2wrs

PAGE 201

185 D 475.250 4 1im E 313.000 4 2dwrs F 244.250 4 7im G 201.750 4 7ww G H G 188.000 4 1ac H H 176.500 4 2ww I 152.500 4 1rb I I 152.000 4 2ac J 113.500 4 1ww J J 101.250 4 2rb K 65.250 4 7ac K K 62.000 4 7rb

PAGE 202

186 CUMULATIVE IRRIGATION DEPTH, YEAR 2004 + 2005 options nodate nonumber center formdlim= "*" linesize= 85 ; data sms; input tmt$ day$ brand$ type$ based$ timeb$ mm; cards ; /* Data is inputted here */ ; data sms2; set sms(where=(type = 'sms' )); proc glm data =sms2; title 'TOTAL Cumulative mm (21 July 2004 Aug 31/2005)' ; class brand day; model mm = day brand day(brand) / ss3 ; test h =brand e =day(brand); means day/ duncan ; means brand/ duncan ; means brand/ duncan e =day(brand); run ; proc glm data =sms2; title 'Comparison of Interaction' ; class brand day; model mm = brand*day / ss3 ; means brand*day/ duncan ; run ; data sms3; set sms(where=(day = '1' )); proc glm data =sms3; title 'Cumulative mm-Once per Week' ; class brand ; model mm = brand/ ss3 ; means brand/ duncan ; run ; data sms3; set sms(where=(day = '2' )); proc glm data =sms3; title 'Cumulative mm-Twice per Week' ; class brand ; model mm = brand/ ss3 ; means brand/ duncan ; run ; data sms3; set sms(where=(day = '7' )); proc glm data =sms3; title 'Cumulative mm-Everyday' ; class brand ; model mm = brand/ ss3 ; means brand/ duncan ; run ; data sms; set sms; proc glm data =sms; title 'Comparison of Time-based treatments' ; class timeb;

PAGE 203

187 model mm = timeb / ss3 ; means timeb/ duncan ; run ; proc glm data =sms; title 'Comparison of Sensor Type -By Individual Treatment (SMS, WRS, WORS, DWRS)' ; class tmt; model mm = tmt / ss3 ; means tmt/ duncan ; run ; ************************************************************************************* TOTAL Cumulative mm (21 July 2004 Aug 31/2005) The GLM Procedure Class Level Information Class Levels Values brand 4 ac im rb ww day 3 1 2 7 Number of Observations Read 48 Number of Observations Used 48 ************************************************************************************* TOTAL Cumulative mm (21 July 2004 Aug 31/2005) The GLM Procedure Dependent Variable: mm Sum of Source DF Squares Mean Square F Value Pr > F Model 11 3957976.917 359816.083 574.61 <.0001 Error 36 22543.000 626.194 Corrected Total 47 3980519.917 R-Square Coeff Var Root MSE mm Mean 0.994337 5.961023 25.02388 419.7917 Source DF Type III SS Mean Square F Value Pr > F day 2 107648.167 53824.083 85.95 <.0001

PAGE 204

188 brand 3 3393706.750 1131235.583 1806.52 <.0001 day(brand) 6 456622.000 76103.667 121.53 <.0001 Tests of Hypotheses Using the Type III MS for day(brand) as an Error Term Source DF Type III SS Mean Square F Value Pr > F brand 3 3393706.750 1131235.583 14.86 0.0035 ************************************************************************************* TOTAL Cumulative mm (21 July 2004 Aug 31/2005) The GLM Procedure Duncan's Multiple Range Test for mm NOTE: This test controls the Type I comparisonwise error rate, not the experimentwise error rate. Alpha 0.05 Error Degrees of Freedom 36 Error Mean Square 626.1944 Number of Means 2 3 Critical Range 17.94 18.86 Means with the same letter are not significantly different. Duncan Grouping Mean N day A 477.750 16 2 B 419.875 16 1 C 361.750 16 7 ************************************************************************************* TOTAL Cumulative mm (21 July 2004 Aug 31/2005) The GLM Procedure Duncan's Multiple Range Test for mm NOTE: This test controls the Type I comparisonwise error rate, not the experimentwise error rate. Alpha 0.05 Error Degrees of Freedom 36 Error Mean Square 626.1944

PAGE 205

189 Number of Means 2 3 4 Critical Range 20.72 21.78 22.47 Means with the same letter are not significantly different. Duncan Grouping Mean N brand A 871.08 12 im B 351.83 12 ww C 251.08 12 ac D 205.17 12 rb ************************************************************************************* TOTAL Cumulative mm (21 July 2004 Aug 31/2005) The GLM Procedure Duncan's Multiple Range Test for mm NOTE: This test controls the Type I comparisonwise error rate, not the experimentwise error rate. Alpha 0.05 Error Degrees of Freedom 6 Error Mean Square 76103.67 Number of Means 2 3 4 Critical Range 275.6 285.6 290.6 Means with the same letter are not significantly different. Duncan Grouping Mean N brand A 871.1 12 im B 351.8 12 ww B B 251.1 12 ac B B 205.2 12 rb ************************************************************************************* Comparison of Interaction

PAGE 206

190 The GLM Procedure Class Level Information Class Levels Values brand 4 ac im rb ww day 3 1 2 7 Number of Observations Read 48 Number of Observations Used 48 ************************************************************************************* Comparison of Interaction The GLM Procedure Dependent Variable: mm Sum of Source DF Squares Mean Square F Value Pr > F Model 11 3957976.917 359816.083 574.61 <.0001 Error 36 22543.000 626.194 Corrected Total 47 3980519.917 R-Square Coeff Var Root MSE mm Mean 0.994337 5.961023 25.02388 419.7917 Source DF Type III SS Mean Square F Value Pr > F brand*day 11 3957976.917 359816.083 574.61 <.0001 ************************************************************************************* Comparison of Interaction The GLM Procedure Level of Level of --------------mm------------brand day N Mean Std Dev ac 1 4 283.25000 14.6600364 ac 2 4 348.00000 17.4547033 ac 7 4 122.00000 3.8297084 im 1 4 793.25000 47.0345618 im 2 4 1104.75000 28.2297125 im 7 4 715.25000 57.4536045 rb 1 4 280.25000 16.5201897

PAGE 207

191 rb 2 4 188.00000 4.8304589 rb 7 4 147.25000 8.1802608 ww 1 4 322.75000 12.4465524 ww 2 4 270.25000 9.6046864 ww 7 4 462.50000 7.7244202 ************************************************************************************* Cumulative mm-Once per Week The GLM Procedure Class Level Information Class Levels Values brand 4 ac im rb ww Number of Observations Read 16 Number of Observations Used 16 ************************************************************************************* Cumulative mm-Once per Week The GLM Procedure Dependent Variable: mm Sum of Source DF Squares Mean Square F Value Pr > F Model 3 748014.7500 249338.2500 349.34 <.0001 Error 12 8565.0000 713.7500 Corrected Total 15 756579.7500 R-Square Coeff Var Root MSE mm Mean 0.988679 6.362870 26.71610 419.8750 Source DF Type III SS Mean Square F Value Pr > F brand 3 748014.7500 249338.2500 349.34 <.0001 ************************************************************************************* Cumulative mm-Once per Week The GLM Procedure Duncan's Multiple Range Test for mm

PAGE 208

192 NOTE: This test controls the Type I comparisonwise error rate, not the experimentwise error rate. Alpha 0.05 Error Degrees of Freedom 12 Error Mean Square 713.75 Number of Means 2 3 4 Critical Range 41.16 43.08 44.25 Means with the same letter are not significantly different. Duncan Grouping Mean N brand A 793.25 4 im B 322.75 4 ww B B 283.25 4 ac B B 280.25 4 rb ************************************************************************************* Cumulative mm-Twice per Week The GLM Procedure Class Level Information Class Levels Values brand 4 ac im rb ww Number of Observations Read 16 Number of Observations Used 16 ************************************************************************************* Cumulative mm-Twice per Week The GLM Procedure Dependent Variable: mm Sum of Source DF Squares Mean Square F Value Pr > F Model 3 2147901.500 715967.167 2352.90 <.0001 Error 12 3651.500 304.292

PAGE 209

193 Corrected Total 15 2151553.000 R-Square Coeff Var Root MSE mm Mean 0.998303 3.651273 17.44396 477.7500 Source DF Type III SS Mean Square F Value Pr > F brand 3 2147901.500 715967.167 2352.90 <.0001 ************************************************************************************* Cumulative mm-Twice per Week The GLM Procedure Duncan's Multiple Range Test for mm NOTE: This test controls the Type I comparisonwise error rate, not the experimentwise error rate. Alpha 0.05 Error Degrees of Freedom 12 Error Mean Square 304.2917 Number of Means 2 3 4 Critical Range 26.88 28.13 28.89 Means with the same letter are not significantly different. Duncan Grouping Mean N brand A 1104.75 4 im B 348.00 4 ac C 270.25 4 ww D 188.00 4 rb

PAGE 210

194 Cumulative mm-Everyday The GLM Procedure Class Level Information Class Levels Values brand 4 ac im rb ww Number of Observations Read 16 Number of Observations Used 16 ************************************************************************************* Cumulative mm-Everyday The GLM Procedure Dependent Variable: mm Sum of Source DF Squares Mean Square F Value Pr > F Model 3 954412.5000 318137.5000 369.69 <.0001 Error 12 10326.5000 860.5417 Corrected Total 15 964739.0000 R-Square Coeff Var Root MSE mm Mean 0.989296 8.109189 29.33499 361.7500 Source DF Type III SS Mean Square F Value Pr > F brand 3 954412.5000 318137.5000 369.69 <.0001 ************************************************************************************* Cumulative mm-Everyday The GLM Procedure Duncan's Multiple Range Test for mm NOTE: This test controls the Type I comparisonwise error rate, not the experimentwise error rate. Alpha 0.05 Error Degrees of Freedom 12 Error Mean Square 860.5417

PAGE 211

195 Number of Means 2 3 4 Critical Range 45.20 47.31 48.59 Means with the same letter are not significantly different. Duncan Grouping Mean N brand A 715.25 4 im B 462.50 4 ww C 147.25 4 rb C C 122.00 4 ac ************************************************************************************* Comparison of Time-based treatments The GLM Procedure Class Level Information Class Levels Values timeb 3 dwrs wors wrs Number of Observations Read 60 Number of Observations Used 12 ************************************************************************************* Comparison of Time-based treatments The GLM Procedure Dependent Variable: mm Sum of Source DF Squares Mean Square F Value Pr > F Model 2 1602266.167 801133.083 1127.52 <.0001 Error 9 6394.750 710.528 Corrected Total 11 1608660.917 R-Square Coeff Var Root MSE mm Mean 0.996025 2.553434 26.65573 1043.917

PAGE 212

196 Source DF Type III SS Mean Square F Value Pr > F timeb 2 1602266.167 801133.083 1127.52 <.0001 ************************************************************************************* Comparison of Time-based treatments The GLM Procedure Duncan's Multiple Range Test for mm NOTE: This test controls the Type I comparisonwise error rate, not the experimentwise error rate. Alpha 0.05 Error Degrees of Freedom 9 Error Mean Square 710.5278 Number of Means 2 3 Critical Range 42.64 44.50 Means with the same letter are not significantly different. Duncan Grouping Mean N timeb A 1514.00 4 wors B 994.75 4 wrs C 623.00 4 dwrs ************************************************************************************* Comparison of Sensor Type -By Individual Treatment (SMS, WRS, WORS, DWRS) The GLM Procedure Class Level Information Class Levels Values tmt 15 1ac 1im 1rb 1ww 2ac 2dwrs 2im 2rb 2wors 2wrs 2ww 7ac 7im 7rb 7ww Number of Observations Read 60 Number of Observations Used 60

PAGE 213

197 Comparison of Sensor Type -By Individual Treatment (SMS, WRS, WORS, DWRS) The GLM Procedure Dependent Variable: mm Sum of Source DF Squares Mean Square F Value Pr > F Model 14 9299750.433 664267.888 1032.98 <.0001 Error 45 28937.750 643.061 Corrected Total 59 9328688.183 R-Square Coeff Var Root MSE mm Mean 0.996898 4.656238 25.35865 544.6167 Source DF Type III SS Mean Square F Value Pr > F tmt 14 9299750.433 664267.888 1032.98 <.0001 ************************************************************************************* Comparison of Sensor Type -By Individual Treatment (SMS, WRS, WORS, DWRS) The GLM Procedure Duncan's Multiple Range Test for mm NOTE: This test controls the Type I comparisonwise error rate, not the experimentwise error rate. Alpha 0.05 Error Degrees of Freedom 45 Error Mean Square 643.0611 Number of Means 2 3 4 5 6 7 8 Critical Range 36.12 37.98 39.20 40.09 40.77 41.30 41.74 Number of Means 9 10 11 12 13 14 15 Critical Range 42.11 42.42 42.68 42.91 43.11 43.28 43.44 Means with the same letter are not significantly different. Duncan Grouping Mean N tmt A 1514.00 4 2wors

PAGE 214

198 B 1104.75 4 2im C 994.75 4 2wrs D 793.25 4 1im E 715.25 4 7im F 623.00 4 2dwrs G 462.50 4 7ww H 348.00 4 2ac H H 322.75 4 1ww I 283.25 4 1ac I I 280.25 4 1rb I I 270.25 4 2ww J 188.00 4 2rb K 147.25 4 7rb K K 122.00 4 7ac

PAGE 215

199 LIST OF REFERENCES Allen R.G., L.S. Pereira, D. Raes and M. Smith. 1998. Crop evapotranspiration: Guidelines for computing crop requiremen ts. Irrigation and Drainage Paper No. 56, FAO, Rome, Italy. ASAE (American Society of Agricultura l Engineers). 2000. Te st procedure for determining the uniformity of water distri bution of center pivo t and lateral move irrigation machines equipped with spray or sprinkler nozzles. Am erican Society of Agricultural Engineers Standards, ANSI/ASAE S436.1, 48th ed. St. Joseph, MI. Augustin, B.J. 2000. Water requirements of Fl orida turfgrasses. BUL200, Institute of Food and Agricultural Sciences, University of Florida, Gainesville, FL. Available at: http://edis.ifas.ufl.edu/EP024 Accessed 7 January 2006. Augustin, B.J. and G.H. Snyder. 1984. Mois ture sensor-controlled irrigation for maintaining bermudagrass turf. Agron. J. 76: 848-850. AWWA (American Water Works Associatio n). 2004. Florida County restricts lawns. WaterWiser, September 2004. Available at: http://www.awwa.org/waterwiser/watch/archive.cfm Accessed 7 January 2006 Baldwin, C.M., H. Liu, L.B. McCarty and W. L. Bauerle. 2005. Drought Tolerance of Six Bermudagrass Cultivars. TurfGrass Trends, Apr 1, 2005. Available at: http://www.turfgrasstrends.com/turfgra sstrends/article /articleDetail.jsp?id=156340 Accessed 7 January 2006. Baum, M.C., M.D. Dukes, and G.L. Miller. 2003. Residential irrigation uniformity and efficiency in Florida. ASAE Mee ting Paper FL03-100. St. Joseph, MI: ASAE. Bausch, W.C., and T.M. Bernard. 1996. Validit y of the Watermark sensor as a soil moisture measuring device. In: Proc. of the Intl. Conf. Evapotranspiration and Irrigation Scheduling. C.R. Camp, E.J. Sa dler, and R.E. Yoder (eds.). St. Joseph, MI: ASAE. pp. 933-938. Biran, I., B. Bravdo, I. Bushkin-Harav, a nd E. Rawitz. 1981. Water consumption and growth rate of 11 turfgrasses as affect ed by mowing height, irrigation frequency and soil moisture. Agron. J. 73: 85–90.

PAGE 216

200 Boman, B., S. Smith, and B. Tullos. 2002. Control and automation in citrus microirrigation systems. CH194, Institute of Food and Agricultural Sciences, University of Florida, Gainesville, FL. Available at: http://edis.ifas.ufl.edu/CH194 Accessed 7 January 2006. Bonos, S.A. and J.A. Murphy. 1999. Growth responses and performance of Kentucky bluegrass under summer stress. Crop Sci. 39:770-774. Burney, L., T. Swihart, and J. Llewe llyn. 1998. Water Supply Planning in Florida. Florida Water Resource Journal, October 1998: 27-28. Burt, C.M.; A.J. Clemmens, and K.H. Strelkoff. 1997. Irrigation performance measurements: efficiency and uniformity. Journal of Irrigation and Drainage Engineering 123(6): 423-442. Campbell, G.S., and D.J. Mulla. 1990. Measur ement of soil water content and potential. In: Irrigation of agricultural crops Agronomy Monogra ph No. 30. ASA-CSSASSSA, Madison, WI. pp. 127-142. Carriker, R.R. 2000. Florida's water: supply, use, and public policy. Department of Food and Resource Economics, Report FE 207, University of Florida, Gainesville, FL. Cassell, D.K., and A. Klute. 1986. Water pot ential: tensiometry. In : A. Klute (ed.) Methods of soil analysis. Part 1. 2nd ed. Agronomy 9. American Society of Agronomy, Madison, WI. pp. 563-596. Charlesworth, P. 2000. Soil water monitoring. Irrigation Insights No.1. CSIRO Land and Water, Griffith, NSW. 96 p. Christiansen, J.E. 1942. Irrigation by sprinkli ng. California Agric. Exp. Stn. Bull. 670. University of California, Berkley, CA. Clark, G.A., C.D. Stanley, and D.N. Maynar d. 1994. Tensiometer control vs. tomato crop coefficients for irrigation schedulin g. ASAE Meeting Paper 94-2118, St. Joseph, MI: ASAE. Colbaugh, P.F., and C.L. Elmore. 1985. In fluence of water on pest activity. In V.A. Gibeault and S.T. Cockerham (ed.) Turf grass water conservation. Cooperative Extension, University of California. pp. 113-129. Davis, D.B., and P.H. Dernoeden, 1991. Summ er patch and Kentucky bluegrass quality as influenced by cultural practices. Agron. J. 83:6702677. Dewey, C. 2003. Sensors at work. Irri gation and green industry, September 2003. Available at: http://www.igin.com/Irr igation/0903sensors.html Accessed 7 January 2006.

PAGE 217

201 Dukes, M. D., E.H. Simonne, W.E. Davis, D.W. Studstill, and R. Hochmuth. 2003. Effect of sensor-based high frequency irrigation on bell pepper yield and water use. In Proceedings 2nd International Conferen ce on Irrigation and Drainage, Phoenix, AZ: 665-674. Dukes, M.D. and D. Z. Haman. 2002a. Operat ion of residential ir rigation controllers. CIR1421, Institute of Food and Agricultura l Sciences, University of Florida, Gainesville, FL. Dukes, M.D. and D.Z. Haman. 2002b. Reside ntial irrigation system rainfall shutoff devices. ABE325, Institute of Food and Ag ricultural Sciences, University of Florida, Gainesville, FL. Available at: http://edis.ifas.ufl.edu/AE221 Accessed 7 January 2006. Dukes, M.D. and J.M. Scholberg. 2005. Soil moisture controlled subsurface drip irrigation on sandy soils. Applied Engineering in Agriculture 21(1): 89-101. Dukes, M.D. and R. Muoz-Carpena. 2005. Soil water sensor-based automatic irrigation of vegetable crops. In: Encyclopedia of Wate r Science. S.W. Trimble, B.A. Stewart and T.A. Howell (eds). Marcel-Dekker, Inc., NY. pp. 1-5. Edis, R. and B. George, 2000. Time-domain refl ectometry: an introdu ction. In: Soil water monitoring. P. Charlesworth (ed.). Irrigation Insights No.1. CSIRO Land and Water, Griffith, NSW. pp. 74-81. Egbert, J.A., S. Baker, J.M. Baker. 1992. Calibration of Watermark soil moisture sensors for soil matrix potential and temperature. Plant Soil 143:213–217. Eldredge, E.P., C.C. Schock, and T. Stieber. 1993. Calibration of granular matrix sensors for irrigation management. Agron. J. 85:1228–1232. FDEP (Florida Department of Environmenta l Protection). 2002. Offi cial notice of rule development: water shortage, NO 6240.412. Publication Date: July 19, 2002. Available at: http://tlhora6.dep.state.fl .us/onw/publications/2-6240RuleNoticewithrulete xtRevisedJuly2DT.pdf. Accessed 26 January 2005. Figliola, R.S. and D.E. Beasley. 2000. Theory and design for mechanical measurements. 3rd ed. John Wiley and Sons, Inc. New York. Florida Statutes, Part VI, Chapter 373.62. n.d. Water Conservation; automatic sprinkler systems. Available at: http://www.floridadep.org/water /stormwater/npdes/docs/ch373.pdf Accessed 16 February 2006. Florida Statutes. 2001. Changes to Chapter 373. Available at: http://www.dep.state.fl.us/c mp/federal/files/373ana01.pdf Accessed 16 February 2006.

PAGE 218

202 Gardner, W.H. 1986. Water content. In: Methods of soil analysis. Part 1. Physical and Mineralogical Methods, 2nd ed. A. Klute (ed.). Agronomy Monogr. No. 9. ASA and SSSA, Madison, WI. pp. 635–662. Haley, M.B., M.D. Dukes, and G.L. Miller 2006. Residential irrigation water use in Central Florida. Journal of Irrigation and Drainage Engineering (in press). Hanson R.B., S. Orloff, and D. Peters. 2000a Monitoring soil moisture helps refine irrigation management. Calif. Agric. 54:38–42 Hanson R.B., S. Orloff, and D. Peters. 2000b. Eff ectiveness of tensiometers and electrical resistance sensors varies with soil condition. Calif. Agric. 54:47–50 Harivandi, M.A., V.A. Gibeault, M.J. Henr y, L. Wu, P.M. Geisel, and C.L. Unruh. 2001. Turfgrass selection for the home landscap e. Publication 8035, University of California, Division of Agriculture and Natural Resources. Available at http://anrcatalog.ucdavis.edu/pdf/8035.pdf. Accessed 16 February 2006. Hodges, A.W., J.J. Haydu, P.J. van Blokla nd, and A.P. Bell. 1994. Contribution of the turfgrass industry to Florida’s ec onomy, 1991-1992: A value-added approach. University of Florida, Institute of Food and Agricultural Sciences, Food and Resource economics Department. Economic Report ER. Hunter Industries, Inc. 2006. Products information. Available at: http://www.hunterindustries.com Accessed 16 February 2006. Irmak, S. and D.Z. Haman. 2001. Performance of the Watermark granular matrix sensor in sandy soils. Appl. Eng. Agric. 17: 787–795. Irrigation Association. 2003. Landscape irrigation scheduling and water management. Irrigation Association Wa ter Management Committee. Falls Church, VA. Intrigliolo, D.S. and J.R. Castel. 2004. Conti nuous measurement of plant and soil water status for irrigation scheduling in plum. Irrig. Sci. 23: 93–102. Jordan, J.E., R.H. White, D.M. Vietor, T.C. Hale, J.C. Thomas, and M.C. Engelke. 2003. Effect of irrigation frequency on turf quali ty, shoot density, and root length density of five bentgrass cultivars. Crop Science 43: 282–287. http://crop.scijournals.o rg/cgi/reprint/43/1/282 Kackley, K.E., A.P. Grybauskas, P.H. Dernoeden, and R.L. Hill. 1990. Role of drought stress in the development of summer patc h in field inoculated Kentucky bluegrass. Phytopathology 80: 655-658. Leib, B.G., M. Hattendorf, T. Elliott, and G. Matthews. 2002. Adoption and adaptation of scientific irrigation scheduling: tr end from Washington, USA as of 1998. Agric. Water Manage. 55:105–120.

PAGE 219

203 Leib, B.G., J.D. Jabro, and G.R. Matthew s. 2003. Field evaluation and performance comparison of soil moisture sensors. Soil Science: 168(6) 396-408. Ley, T.W., R.G. Stevens, R.R. Topielec and W.H. Neibling. 2000. Soil water monitoring and measurement. Pacific Nort hwest, WA. 33 p. Available at: http://cru.cahe.wsu.edu/CEPublications/pnw0475/pnw0475.html Accessed 13 November 2003. Marella, R.L. 1992. Water withdrawals, us e, and trends in Florida, 1990. Water Resources Investigations Report 92-4140. United States Department of the Interior, U.S. Geological Survey, Tallahassee, FL. Mayer, P.W., W.B. DeOreo, E.M. Opitz, J.C. Kiefer, W.Y. Davis, B. Dziegielewski, and J.O. Nelson. 1999. Residential end uses of water. American Water Works Association Research Foundation. Denver, CO. McCann, I.R., D.C. Kincaid, and D. Wang. 1992. Operational characteristics of the watermark model 200 soil water potential sensor for irrigation management. Appl. Eng. Agric. 8:603–609 McCarty, L.B. 2005. Best Golf C ourse Management Practices. 2nd ed. Upper Saddle River, NJ. Prentice-Hall Inc. Merriam, J.L., and J. Keller. 1978. Farm i rrigation system evaluation: a guide for management. Department of Agricultural and Irrigation Engineering, Utah State University, Logan, UT. Mitchell, A.R., and C.C. Shock. 1996. A Watermark datalogging system for ET measurement. In: Proc. of the Intl. Conf. Evapotranspira tion and Irrigation Scheduling. C. R. Camp, E. J. Sadler, and R. E. Yoder (eds.). St. Joseph, MI: ASAE. pp. 468-473. Muoz-Carpena, R. 2004. Field devices for monitoring soil water content. BUL343, Institute of Food and Agricultural Sciences, University of Florida, Gainesville, FL. Available at: http://edis.ifas.ufl.edu/AE266 Accessed 7 January 2006. Muoz-Carpena, R. and M.D. Dukes. 2005. Automatic irrigation based on soil moisture for vegetable crops. ABE356, Institute of Food and Agricultural Sciences, University of Florida, Gainesville, FL. Available at: http://edis.ifas.ufl.edu/AE354 Accessed 7 January 2006. Muoz-Carpena, R., H. Bryan, W. Klassen, and M.D. Dukes. 2003. Automatic soil moisture-based drip irrigation for impr oving tomato production. Proceedings of the Florida State Horticultu ral Society 116: 80-85. Muoz-Carpena, R., M.D. Dukes, Y. Li, and W. Klassen. 2005. Field comparison of tensiometer and granular matrix sensor automatic drip irrigation on tomato. HortTechnology 15(3): 584-590.

PAGE 220

204 National Research Council. 1996. A new era in irrigation. National Academies Press. Washington D.C. NOAA (National Oceanic and Atmospheric Administration). 2003. Monthly station normals of temperature, precipitation, a nd heating and cooling degree days 1971 – 2000; 08 Florida. Climatography of The Un ited States No. 81. 28p. Available at: http://hurricane.ncdc.noaa.gov/c limatenormals/clim81/FLnorm.pdf Accessed 29 April 2006. Nogueira, L.C., M.D. Dukes, D.Z. Haman, J.M. Scholberg, and C. Cornejo. 2002. Data acquisition and irrigation controller based on CR10X da talogger and TDR sensor. Proceedings Soil and Crop Science Society of Florida 2002, 62: 38-46. ODR (Office of Economic and Demographic Research.). 2006. Demographic Estimating Conference; Florida Demogra phic Forecast. Available at: http://edr.state.fl.us/conferen ces/population/demographic.htm Accessed 29 April 2006 Or, D. 2001. Who invented the tensiometer? Soil Sci. Soc. Am. J. 65: 1–3. Pathan, S.M., L. Barton and T.D. Colmer. 2003. Evaluation of a soil moisture sensor to reduce water and nutrient leaching in turf. Horticulture Australia project number TU 02006. 21p. Qualls, R.J., J.M. Scott, and W.B. De Oreo. 2001. Soil moisture sensors for urban landscape irrigation: effec tiveness and reliability. Journal of the American Water Resources Association 37(3): 547-559. Sarasota County. 2006. Water efficient land scaping ordinance; Ordinance #2001-081. Available at: http://sarasota.extension.uf l.edu/WEL/ord/docs/ord.htm Accessed 24 April 2006. SAS (Statistical Analysis System). 2000. SAS/STAT User’s Guide, Ver. 9. Cary, N.C.: SAS Institute, Inc. Schmugge, T.J., T.J. Jackson, and H.L. McKim. 1980. Survey of methods for soil moisture determination. Water Resour. Res. 16: 961-979. Seyfried, M.S. 1993. Field ca libration and monitoring of soil water content with fiberglass electrical resistance sensors. Soil Sc. Soc. Am. J. 57: 1432–1436. Shock, C.C., E. Fibert, and M. Saunders 1996. Malheur Experiment Station Annual Report. Special Report 964, Oregon State University, Ontario, OR. Shock, C.C., E.B.G. Feibert, and L.D. Saunde rs. 1998a. Onion yield and quality affected by soil water potential as irrigation threshold. HortScience 33: 1188–1191

PAGE 221

205 Shock, C.C., E.B.G. Feibert, and L.D. Saunde rs. 1998b. Potato yield and quality response to deficit irrigation. HortScience 33: 655–659 Shock, C.C., E.B.G. Feibert, L.D. Saunde rs, and E.P. Eldredge. 2002. Automation of subsurface drip irrigation for crop resear ch. In: Proceedings of the World Congress of Computers in Agriculture and Natural Re sources. F.S. Zazueta and J. Xin eds. Iguacu Falls, Brazil. pp. 809-816 SJRWMD (Saint John’s River Water Manage ment District). 2006. Waterwise Florida landscapes. Available at: http://sjrwmd.com/programs/outreach/c onservation/landscape/principle5.html Accessed 29 January 2006. Skogley, C.R., and C.D. Sawyer. 1992. Fiel d research. In: Turfgrass. Agron. Monogr. 32. D.V. Waddington, R.N. Carrow, and R.C. Shearman (eds.). ASA, CSSA, and SSSA, Madison, WI. pp. 589–614 Smajstrla, A.G. and R.C. Koo. 1986. Use of tensiometers for scheduling of citrus irrigation. Proceedings of the Florida St ate Horticultural Society 99: 51-56. Smajstrla, A.G. and S.J. Locascio. 1994. Irr igation cutback effects on drip-irrigated tomato yields. Proceedings of the Flor ida State Horticultural Society 107: 113-118. Solley, W.B., R.R. Pierce, and H.A. Perlman. 1998. Estimated use of water in the United States in 1995. United States Geol ogical Survey Circular 1200. 78 p. SWUCA (Southern Water Use Caution Ar ea Recovery Strategy). 2006. Southwest Florida Water Management District. Re vised Draft, March 2006. 145 p. Available at: http://www.swfwmd.state.fl.us /waterman/swuca/SWUCA.htm Accessed 29 January 2006. Taber H.G., V. Lawson, B. Smith, and D. Shogren. 2002. Scheduling microirrigation with tensiometers or Watermarks. Int. Water Irrig. 22: 22–26 Tampa Bay Water. 2005. Evaluating implementa tion of multiple irrigation and landscape ordinances in the Tampa Bay region. April 2005. Available at: http://www.tampabaywater.org/ conservation/reportsdocs.aspx Accessed 19 January 2006. Thomas, B.D., E. Cummings, and W.H. Whittstruck. 1985. Soil survey of Alachua County, Florida. USDA/NRCS in cooperati on with the University of Florida, Institute of Food and Agricultural Scienc es, Soil Science Department and the Florida Department of Agriculture and Consumer Services. Thomson, S.J., and C.F. Armstrong. 1987. Calibration of the Watermark model 200 soil moisture sensor. Applied Engineering in Agriculture 3(2): 186–189.

PAGE 222

206 Thomson, S.J., T. Younos, and K. Wood. 1996. Evaluation of calibra tion equations and application methods for the Watermark gr anular matrix soil moisture sensor. Applied Engineering in Agriculture 12(1): 99–103. Topp, G.C. 2003. State of the art of measuring soil water content. Hydrological Processes 17: 2993–2996. Turgeon, A.J. 2005. Turfgrass Management. 7th ed. Upper Saddle River, NJ. PrenticeHall Inc. 415p. USCB (United States Census Bureau). 2004a. Population estimat es. Washington, DC. Available at: http://www.census.gov/popest/estimates.php Accessed 19 January 2006. USCB (United States Census Bureau). 2004b. Housing unit estimates. Washington, DC. Available at: http://www.census.gov/popest/housing/ Accessed 19 January 2006. USDA (United States Department of Ag riculture). 1981. Land resource regions and major land resource areas of the united states. Soil Conservation Service Handbook 256. USDA, Washington, DC. USDA (United States Department of Agricult ure). 2003. Official soil series descriptions. Natural Resource and Conservation Serv ice, Washington, DC. Available at: http://soils.usda .gov/technical/classification/osd/index.html Accessed 13 April 2005. USDA (United States Department of Ag riculture). 2005. Florida State agriculture overview – 2005. National Agricultural St atistics Service. Available at: http://www.nass.usda.gov/Statistics_ by_State/Ag_Overview/AgOverview_FL.pdf Accessed 13 April 2006. Walker, J.P., G.R. Willgoose and J.D. Ka lma. 2004. In situ measurement of soil moisture: a comparison of techniques. Journal of Hydrology 293: 85-99. Whitcomb, J.B. 2005. Florida water rates evalua tion of single-family homes. Southwest Florida Water Management District. Available at: http://www.swfwmd.state.fl.us/docum ents/reports/water_rate_report.pdf Accessed 29 January 2006. Yoder, R.E., D.L. Johnson, J.B. Wilkerson, and D.C. Yoder. 1998. Soil water sensor performance. Appl. Eng. Agric. 14: 121-133. Youngner, V.B., A.W. Marsh, R.A. Strohma n, V.A. Gibeault, and S. Spaulding. 1981. Water use and turfgrasses. Calif. Turfgrass Cult. 31: 1-4.

PAGE 223

207 Zazueta, F.S., A.G. Smajstrla and G.A. Clark. 2002. Irrigation system controllers. SSAGE-22, Institute of Food and Agricultural Sciences, University of Florida, Gainesville, FL. Available at: http://edis.ifas.ufl.edu/AE077 Accessed 10 April 2005.

PAGE 224

208 BIOGRAPHICAL SKETCH Bernard Cardenas-Lailhacar was born in Sant iago, Chile. He received a Bachelor of Science degree in agricultural engineering from the Universidad Austral de Chile. Later, he accepted a graduate assistantship positi on in the Agricultural and Biological Engineering Department at the University of Florida, and began studying toward a Master of Science degree. On August 2006, Cardenas-Lailhacar received his M.Sc. degree after defending his dissertation untitle d “Sensor-based Automation of Irrigation of Bermudagrass.” While at the University of Florida, he obt ained the highest grade of his class (GPA: 4.0), received five awards, and was invited to be part of two Honor Societies, on the basis of his class performance. Moreover, he obt ained the second place at the 2006 Graduate Student Research Award of the American Society of Agricultural and Biological Engineers (ASABE). He is proficient in three languages (Englis h, Spanish, and French), and has traveled to 12 different countries in Europe and the Americas.


Permanent Link: http://ufdc.ufl.edu/UFE0015863/00001

Material Information

Title: Sensor-Based Automation of Irrigation in Bermudagrass
Physical Description: Mixed Material
Copyright Date: 2008

Record Information

Source Institution: University of Florida
Holding Location: University of Florida
Rights Management: All rights reserved by the source institution and holding location.
System ID: UFE0015863:00001

Permanent Link: http://ufdc.ufl.edu/UFE0015863/00001

Material Information

Title: Sensor-Based Automation of Irrigation in Bermudagrass
Physical Description: Mixed Material
Copyright Date: 2008

Record Information

Source Institution: University of Florida
Holding Location: University of Florida
Rights Management: All rights reserved by the source institution and holding location.
System ID: UFE0015863:00001


This item has the following downloads:


Full Text












SENSOR-BASED AUTOMATION OF IRRIGATION OF BERMUDAGRASS


By

BERNARD CARDENAS-LAILHACAR













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

UNIVERSITY OF FLORIDA


2006





























Copyright 2006

by

Bernard Cardenas-Lailhacar



































To my parents and sons















ACKNOWLEDGMENTS

In the first place, I wish to thank my parents for their always enormous and

unconditional love, support, and guidance; my sons for being the most adorable human

beings I have ever met; my ex-wife for taking care of them while I was completing my

studies, her understanding, patience, and sacrifice; and my amorcita for her enormous

support, understanding, patience and, most of all, her immense love. Next, I would like to

thank all my thesis committee members for being not just professors, but great human

beings: Dr. Dorota Z. Haman and Dr. Grady L. Miller, also for their guidance and

patience, and a huge thank you goes to Dr. Michael D. Dukes, for giving me the

opportunity to work with him, which was always a lot of work, but also a pleasure. Also,

I wish to give a special thank you to Melissa B. Haley for being always ready to help me.

Lastly, I would also like to thank Engineer Larry Miller, Senior Engineering Technician

Danny Burch, and students Mary Shedd, Stephen Hanks, Clay Coarsey, Brent Addison,

Jason Frank, and Clay Breazeale for their assistance on this research. This research was

supported by the Pinellas-Anclotte Basin Board of the Southwest Water Management

District, the Florida Nursery and Landscape Growers Association, and the Florida

Agricultural Experiment Station.
















TABLE OF CONTENTS



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

LIST OF TABLES .......... ... .......... ...... ........... .. .... ....... ............ .. viii

LIST OF FIGURES .............. ............................................ ..x

A B S T R A C T ......... .................................. ................................................x v

CHAPTER

1 IN TR O D U C T IO N ............................................................. .. ......... ...... .....

W ate r ............................................................. ..... ........................................... .
W after D em and ................................... ........................... ................ 2
W after U se .............................................................................. 2
W after U se R restrictions .................................................................. ....................... 3
L landscapes in F lorida ................................................................... ......... .... 5
Irrigation ............................................................. . 6
Irrig atio n T im ers .................................................................................. 6
Soil Moisture Content Measurement .......................................... ...............8
G ranular m atrix sensor ............................................. ................................. 8
M odem soil m oisture sensors...................................... ......................... 9
Controllers ................. ...... ...... .... .............. ...... .................... 11
A utom atic C control of Irrigation..................................... .................................... 12
Rain Sensors ................................................................. ............... 13
Irrigation and Turfgrass Quality .................................................... ........ ...........15

2 SENSOR-BASED AUTOMATION OF IRRIGATION OF BERMUDAGRASS .... 20

In tro du ctio n .......................................... .........20..........
M materials an d M eth od s ...................................................................... ....................2 5
T re a tm e n ts ...................................................................................................... 2 7
Uniformity Test ............... ...... ...............................................28
D ry-W et A analysis ............ ..... .. .......................................... ................. .. 29
Plot Irrigation Management and Data Collection.................... ..... ...........30
D ata A n a ly sis .................................................................................................. 3 3
R results and D discussion ............................. .................... .. ........ .. .............34
U n form ity T ests............ ... ....................................................................... ......... 34


v









Dry-W et Analysis .................. .......................... .. ....... ................. 34
R a in fa ll ................................................................3 5
Irrigation E vents ....................................... ......................... ........ 36
Irrigation Application Comparisons............................................................42
Time-based treatments vs. SMS-based treatments.................. ............42
T im e-based treatm ents ................................................ ............... ... 42
Comparisons between SMS-irrigation frequencies....................... ...44
Soil moisture sensor-brands comparison..............................................45
Brand comparisons within irrigation frequencies .....................................46
O v erall com p arison ........................................................... .....................4 7
Autom ation of Irrigation System s .............................. ............. ................49
T urfgrass Q quality ......................................... ...... ....... .............. 50
Sum m ary and C conclusions ............................................................... ............... 51

3 EXPANDING DISK RAIN SENSOR PERFORMANCE AND POTENTIAL
IRRIGATION WATER SAVINGS ........................................ ....................... 93

R a in S e n so rs ............................................................................................................... 9 3
A dv an tag e s ................................................................9 4
T ypes and M ethods............. ................................................ .... ...... .... ..... 94
In sta lla tio n .................................................................. .................................9 6
O b j e c tiv e s .................................................................. ..................................9 6
M materials and M methods ....................................................................... ..................97
D a ta ......................................................................... 9 7
T re a tm e n ts ..................................................................................................... 9 8
Statistical A analysis .......................... ............ ...........................99
R results and D iscu ssion .............................. ........................ .. ...... .... ...... ...... 99
C lim atic C conditions ................................................... .................. ............ 99
Number of Times in Bypass Mode ............................................. ...............99
Depth of Rainfall Before Shut Off ...............................................................100
Duration in Irrigation Bypass Mode (Dry-Out Period) ...................................102
Potential W ater Savings ............................................................................. 103
Payback Period .................. ................................................... 104
Sum m ary and Conclusions ......................................................... .............. 105

4 GRANULAR MATRIX SENSOR PERFORMANCE COMPARED TO
TENSIOMETER IN A SANDY SOIL ............... .............................................119

Tensiometers ................ ......... ....................... ....... ....... 119
G ranular M atrix Sensors........................................ ..... ................... ............... 121
GM S Tensiom eter Com prison ................................... ............................. ....... 122
O objectives ..............................................122.............................
M materials and M methods ........................................... ....................................... 122
E x p erim ental Set-U p ........................................... ........................................ 12 3
E C H 20 Probes C alibration ......... ............................................ ...... ..... ... 123
T re atm en ts ................................................... ............. ................ 12 4
D ata ................................................ .......... ............... 12 4









R esu lts an d D iscu ssion ................................................................................... 12 5
Calibration of the ECH20 Probe ............. ................................ ...............125
GM Ss versus Tensiom eters. ........................................... ......................... 125
C o n c lu sio n s.......................................................................................................... 12 6

5 CONCLUSIONS AND FUTURE WORK.......................................................140

C o n c lu sio n s .......................................................................................................... 14 0
Future W ork .............. ..................................................................... 142

APPENDIX

A LIST OF ABBREVIATIONS...................... ....... ............................. 144

B STATISTICAL ANALYSES ...........................................................................145

L IST O F R E FE R E N C E S ......... ................. ...................................... ..........................199

B IO G R A PH IC A L SK E T C H ........................................... ...........................................208
















LIST OF TABLES


Table page

2-1. Irrigation treatment codes and descriptions.......................................................54

2-2. Monthly irrigation depth to replace historical evapotranspiration, assuming
system efficiency of 60%, and considering effective rainfall................................54

2-3. Total number and percent of overridden scheduled irrigation cycles; 2004 and
2005................. .............. .............................55

2-4. Percent of irrigation cycles allowed by the SMS-based treatments through the
experimental months of 2004 and 2005. ...................... ...... ........... .............. 56

2-5. Cumulative irrigation depth applied to treatments, statistical comparisons
between them, and percent of water savings compared to 2-WRS, 2-DWRS, and
2-W ORS; year 2004. ............................................... ...............57

2-6. Cumulative irrigation depth applied to treatments, statistical comparisons
between them, and percent of water savings compared to 2-WRS, 2-DWRS, and
2-W OR S; year 2005. ............................................... ...............58

2-7. Total cumulative irrigation depth applied to treatments, statistical comparisons
between them, and percent of water savings compared to 2-WRS, 2-DWRS, and
2-W O R S; years 2004 + 2005. ....................................................... .....................59

3-1. Treatm ents description. ................................................ ............................... 107

3-2. Average depth of rainfall before rain sensors switched to bypass mode ............... 107

3-3. Large rainfall events not bypassed by treatment 3-MC .....................................107

3-4. Large rainfall events not bypassed by treatment 13-M C ............... ....................108

3-5. Large rainfall events not bypassed by treatment 25-MC ...................................108

3-6. Hours after rain stopped and sensors switched to bypass mode; treatment 3-MC. .108

3-7. Hours after rain stopped and sensors switched to bypass mode; treatment 13-MC.109









3-8. WL replications that switched to bypass mode in absence of rainfall, elapsed time
that they remained in bypass mode, and relative humidity at the time when this
o ccu rred ................................ .......... ... .......... ............................. 10 9

3-9. Total potential water savings per treatment ............................................... 109

3-10. Potential payback period per treatment. ....................... ...................110

4-1. T reatm ents. ........................................................................ 12 8

4-2. GM S-Tensiometer crossing points. ........................................................................128
















LIST OF FIGURES


Figure page

1-1. Components of an automated irrigation system: A) timer, B) power supply, C)
soil moisture sensor-controller circuitry, D) soil moisture sensor, and E)
solenoid valve............................................................................................. 16

1-2. G ranular m atrix sensors (GM S) ........................................... ......................... 17

1-3. Components of an automated irrigation system. 1) Timer, and 2) soil moisture
sensor-controllers from different brands. ...................................... ...............18

1-4 R ain shut-off sw itch ........................................................................... ...... 19

1-5. The expanding material of a rain shut-off switch............... ............ ............... 19

2-1. Soil water retention curve from tensiometers and calibrated ECH20 probe
read in g s. .......................................................... ................ 6 0

2-2. Soil moisture sensor brands tested in this study...................................................... 61

2-3. Irrigation controls as installed for this study: soil moisture sensors-controllers
brands: A) Rain Bird, B) Water Watcher, C) Acclima, and D) Irrometer, and
irrigation tim er E) Rain Bird. ...... ...................................................................... 62

2-4. R ain sensor installed for this study.......................................................................... 63

2-5. Catch-can display for uniformity tests on turfgrass plots ....................................64

2-6. General view of the irrigation controls used in this study............... ... ...............65

2-7. Pipes, flowmeters, valves, and wirings for this study. ............................................65

2-8. Control board showing timers, soil moisture sensor-controllers, solenoid valves
wiring, and flowmeters-datalogger (details are shown in the next s).....................66

2-9. Control board detail showing the solenoid valves control box. .............................67

2-10. Control board detail, flowmeter-datalogger boxes showing A) multiplexers, B)
CR 10X datalogger used for this study. ...................................... ............... 67

2-11. Automated weather station near turf plots for this study.......................................68









2-12. ECH20 probe, capacitance soil moisture probe shown with a HOBO data logger
as installed for this study ................................ .............. .............. ......... 69

2-13. Plot plan showing the low-quarter distribution uniformity testing results on each
p lot .................................. ................... ...... ........ ............................ 7 0

2-14. Plot plan showing average volumetric water content (%) on each plot during a
relatively "dry" period. Plots in red were discarded, and plots in green were used
for place ent of SM Ss. ......................... .................... ... .... .... ............... 71

2-15. Plot plan showing average volumetric water content (%) on each plot during a
relatively "wet" condition. Plots in red were discarded, and plots in green were
used for place ent of SM Ss........................................................ ............... 72

2-16. Plot plan with the modified completely randomized design (same color depicts
treatment repetitions)................... ..... .. .. .............. ...........73

2-17. Daily and cumulative rainfall in 2004. Note: rainfall for 5 Sep. (188 mm) and 6
Sep. (81 mm) is shown as a cumulative total (269 mm). .................................74

2-18. D aily and cum ulative rainfall in 2005. .......................................... ............... 74

2-19. Cumulative number of irrigation events per treatment in 2004; A) time-based
treatments, and soil moisture sensor-based treatments at irrigation frequencies of
B) 1 d/w, C) 2 d/w, and D) 7 d/w ........................................ ......................... 75

2-20 Cumulative number of irrigation events per treatment in 2005; A) time-based
treatments, and soil moisture sensor-based treatments at irrigation frequencies of
B) 1 d/w C) 2 d/w and D) 7 d/w ........................................ ........................ 76

2-21. Maximum weekly irrigation water requirement (rainfall ETo difference); year
2 0 04 .................................................................................7 7

2-22. Maximum weekly irrigation water requirement (rainfall ETo difference); year
2005................. .............. .............................77

2-23. Volumetric moisture content through time, on treatment 0-NI, year 2004.............78

2-24. Volumetric moisture content (VMC) through time, showing results of the
scheduled irrigation cycles (SIC); treatment 1-AC, year 2004. ............................79

2-25. Volumetric moisture content (VMC) through time, showing results of the
scheduled irrigation cycles (SIC); treatment 1-IM, year 2004.............................80

2-26. Volumetric moisture content (VMC) through time, showing results of the
scheduled irrigation cycles (SIC); treatment 1-RB, year 2004. ............................81









2-27. Volumetric moisture content (VMC) through time, showing results of the
scheduled irrigation cycles (SIC); treatment 1-WW, year 2004. ...........................82

2-28. Volumetric moisture content (VMC) through time, showing results of the
scheduled irrigation cycles (SIC); treatment 2-AC, year 2004 ............................83

2-29. Volumetric moisture content (VMC) through time, showing results of the
scheduled irrigation cycles (SIC); treatment 2-IM, year 2004.............................84

2-30. Volumetric moisture content (VMC) through time, showing results of the
scheduled irrigation cycles (SIC); treatment 2-RB, year 2004 ............................85

2-31. Volumetric moisture content (VMC) through time, showing results of the
scheduled irrigation cycles (SIC); treatment 2-WW, year 2004. ...........................86

2-32. Volumetric moisture content (VMC) through time, showing results of the
scheduled irrigation cycles (SIC); treatment 7-AC, year 2004 ............................87

2-33. Volumetric moisture content (VMC) through time, showing results of the
scheduled irrigation cycles (SIC); treatment 7-IM, year 2004 ................................88

2-34. Volumetric moisture content (VMC) through time, showing results of the
scheduled irrigation cycles (SIC); treatment 7-RB, year 2004 ............................89

2-35. Volumetric moisture content (VMC) through time, showing results of the
scheduled irrigation cycles (SIC); treatment 7-WW, year 2004. ...........................90

2-36. Average irrigation depth applied by brand; years 2004 and 2005 (P<0.0001)........91

2-37. View of different plots where no evident turfgrass quality differences could be
detected; A) good quality, B) dormant .... ........... ......................................... 92

3-1. Mini-Click (Hunter Industries, Inc.) rain sensor. A) Rain threshold set slots, B)
vent ring ............................................................................................... 111

3-2. Wireless Rain-Click (Hunter Industries, Inc.) rain sensor. A) Ventilation window
adjustment knob, B) ventilation windows, C) antenna. ............... ...............111

3-3. The expanding material of a rain shut-off switch ................................................... 112

3-4. Rain sensor experiment layout: A) Wireless Rain-Click rain sensors, B) Mini-
Click rain sensors, C) Wireless Rain-Click receivers, D) multiplexers, E) CR
10X datalogger ................................................................ ... ........ 112

3-5. Manual rain gauge measurements compared to tipping bucket rain gauge
m easurem ents ................................................................ .... .........113

3-6. D aily and cum ulative rainfall. ........................................................ ............... 113









3-7. Cumulative number of times rain sensors switched to bypass mode; average per
treatment. Different letters indicate a significant difference by Duncan's
M multiple R ange Test (P<0.05) ........................................... ............................... 114

3-8. Cumulative number of times rain sensors switched to bypass mode; WL
treatm ent, with replicates indicated by A-D ...................................... ................ 115

3-9. Cumulative number of times rain sensors switched to bypass mode; 3-MC
treatment, with replicates indicated by A-D .................................................. 115

3-10. Cumulative number of times rain sensors switched to bypass mode; 13-MC
treatment, with replicates indicated by A-D.....................................................116

3-11. Cumulative number of times rain sensors switched to bypass mode; 25-MC
treatment, with replicates indicated by A-D.............. ................... ................... 116

3-12. Histogram and frequency distribution for 6-hour intervals in bypass mode; WL. 117

3-13. Histogram and frequency distribution for 6-hour intervals in bypass mode; 3-
M C ................................................... ..................... ................ 1 1 7

3-14. Histogram and frequency distribution for 6-hour intervals in bypass mode; 13-
M C ...................................... .................................................... 1 1 8

4-1. M LT-R SU Tensiom eter ......... ................. ..................................... ............... 129

4-2. Watermark GMS...................... .................... ............ 129

4-3. Tem perature sensor............... ....................... ........ ........ .. .. ............ 130

4-4. E CH 20 probe ......... ..................... ............ .. .......... ............ ............. 130

4-5. Experimental layout (top view). A) Tensiometers, B) Granular matrix sensors, C)
ECH20 probe, and D) Thermometer............................................................. 131

4-6. W term ark m onitor............................................................................... ..... ..... 132

4-7. ECH20 probe hooked up to a HOBO Micro Station datalogger.............................132

4-8. Volumetric moisture content (VMC) from all three ECH20 probes compared to
gravim etric m easurem ents.......................................................... ............... 133

4-9. Soil water tension through time; treatment TO. ............................... ............... 134

4-10. Soil water tension through time; treatment T5. .............................................. 134

4-11. Soil water tension through time; treatment T15 ............ ..............................135

4-12. Soil water tension through time; treatment T50. ................................................135









4-13. Soil water tension through time; detail showing when curves from GMS and
tensiom eters cross; treatm ent TO ............................................ .......................... 136

4-14. Soil water tension through time; detail showing when curves from GMS and
tensiom eters cross; treatm ent T5 ...................................................................... 136

4-15. Soil water tension through time; detail showing when curves from GMS and
tensiometers cross; treatment T15.................................. ............... 137

4-16. Soil water tension through time; detail showing when curves from GMS and
tensiometers cross; treatment T50.............................................137

4-17. Relation between the average soil matric potential (SMP) from tensiometers and
G M S ........................................................................... 13 8

4-18. Relation between the average soil matric potential (SMP) from tensiometers and
GM S; excluding GMS data < 10 kPa........................ ......... .............. 139















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 Science

SENSOR-BASED AUTOMATION OF IRRIGATION OF BERMUDAGRASS

By

Bernard Cardenas-Lailhacar

August 2006

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

Turfgrass in landscapes contributes to substantial cropped area in Florida. New

irrigation technologies could improve irrigation efficiency, promoting water conservation

and reducing the environmental impacts. The objectives of this research were to quantify

differences in irrigation water use and turf quality among 1) a soil moisture sensor-based

irrigation system compared to a time-based scheduling, 2) different commercial irrigation

soil moisture sensor (SMSs), 3) a time-based scheduling system with or without a rain

sensor (RS), and 4) the reliability of two commercially available expanding disk RS-

types. The experimental area consisted of common bermudagrass (Cynodon dactylon L.)

plots (3.66 x 3.66 m) in a completely randomized design, located in Gainesville, Florida.

The monitoring period for the irrigation treatments took place from 20 July through 14

December of 2004 and from 25 March through 31 August of 2005. Treatments consisted

of irrigating one, two, or seven days a week, each with four different commercial SMSs

brands. A non-irrigated control and time-based treatments were also implemented. In









addition, twelve Mini-Click (MC) and four Wireless Rain-Click (WL) rain sensor models

not connected to irrigation were monitored from 25 March through 31 December 2005.

For the MCs, three different thresholds were established: 3, 13, and 25 mm (codes 3-MC,

13-MC, and 25-MC, respectively). No significant differences in turfgrass quality among

irrigation treatments were detected. On average, SMS-based treatments reduced irrigation

water application compared to time-based treatments. The treatment without-rain-sensor

(2-WORS) used significantly (52%) more water than the with-rain-sensor treatment (2-

WRS). Most brands recorded significant irrigation water savings compared to 2-WRS,

which ranged from 54% to 88%, for the best performing sensors, and depending on the

irrigation frequency. Therefore SMS-systems represent a promising technology, because

of the water savings that they can accomplish, while maintaining an acceptable turfgrass

quality during rainy periods (944 and 732 mm of rainfall, for seasons 2004 and 2005,

respectively). On average, RS treatments WL, 3-MC, 13-MC, and 25-MC responded

close to their rainfall set points (1.4, 3.4, 10.0, and 24.5 mm, respectively). However,

some replications showed erratic behavior through time. The number of times that these

sensors shut off irrigation was inversely proportional to the magnitude of their set point

(81, 43, 30, and 8 times, respectively) with potential water savings following a similar

trend (363, 245, 142, and 25 mm, respectively). Under the relatively wet testing

conditions typical to Florida, the payback period could be less than a year, except for 25-

MC (around 7 years). Consequently, RSs are strongly recommended for use by

homeowners as a means to save water, but not when accuracy is required.














CHAPTER 1
INTRODUCTION

Turfgrass is the main cultivated crop in Florida with nearly four times the acreage

as the next largest crop, citrus (Hodges et al., 1994; United States Department of

Agriculture [USDA], 2005). Irrigation of residential, industrial, commercial, and

recreational turf areas is necessary to ensure acceptable turf quality. As a consequence of

problems related to drought, coupled with a steadily increasing demand for water

resources, the state of Florida has imposed restrictions on irrigation water use. Water

used for turfgrass irrigation, however, remains to be publicly discussed. The development

of Best Management Practices (BMPs) for irrigation water use in landscapes has become

an undeniable strategic, economic, and environmental issue for the state. New irrigation

technologies could improve irrigation efficiency, promoting water conservation and

reducing the environmental impacts of turfgrass culture, which is a major component of

landscapes in Florida.

Water

Florida receives an average of around 1400 mm of rainfall a year (National Oceanic

and Atmospheric Administration [NOAA], 2003). Unlike many areas dependent on

irrigation, annual rainfall in Florida typically exceeds evapotranspiration. Nevertheless,

irrigation is required because total annual rainfall for Florida typically varies both

geographically and temporally (USDA, 1981; Carriker, 2000). Such rainfall variation has

a direct impact on surface water and groundwater supplies. Lack of rainfall for even a

few days causes depletion of moisture in sandy soils commonly found in Florida; along









with reduction of stream flow and groundwater recharge (Carriker, 2000; National

Research Council, 1996).

Water Demand

Florida has the second largest withdrawal of groundwater for public supply in the

United States (Solley et al., 1998). Groundwater was the source of more than 88% of the

water withdrawn for public supply in 1990 (Carriker, 2000). In 1995, nearly 93% of

population in Florida used groundwater as a drinking water source (Solley et al., 1998).

Water withdrawals for public supply in Florida have increased rapidly, from 600,000

m3/day in 1950 to 7.3 million m3/day in 1990 (Carriker, 2000). The population served by

public-supply systems increased from 5.42 million in 1970 to 11.23 million in 1990

(Marella, 1992).

Florida has a fast-growing population with a net inflow of more than 1100 people a

day, and ranks as the second largest net gain in the nation. The population of 17 million

in 2004 is projected to exceed 21 million people by 2015, becoming the third most

populous state in the nation (United States Census Bureau [USCB], 2004a). The U.S.

Census Bureau estimated 156.8 thousand single-family housing starts and 56.7 thousand

multi-family housing starts in Florida in 2003, accounting for approximately 11% of all

new homes constructed in the United States, the largest amount in any single state in the

U.S. (USCB, 2004b). As urban populations swell, pressures on limited supplies of clean

water are increasing, and it may become a scarce resource.

Water Use

Indoor water use per person in the U.S. is relatively constant across all geographic

and social lines. Depending on climate, residential outdoor water use can account for

22% to 67% of total annual water use (Mayer et al., 1999). The primary use of residential









outdoor water is irrigation. Historically, Florida exhibit dry and warm spring and fall

weather, as well as sporadic large rain events in the summer. These climatic conditions,

coupled with low water holding capacity of the soil, make irrigation indispensable for the

high quality landscapes desired by homeowners (Haley et al., 2006; National Research

Council, 1996).

Recent studies in the U.S. indicate that, on average, 58% of potable water is used

for landscape irrigation (Mayer et al., 1999). In the Central Florida Ridge, this average

has been show to be as high as 74% (Haley et al., 2006). Consequently, proper irrigation

water use clearly represents a substantial opportunity for residential water savings.

Furthermore, residential water use research, carried out by Mayer et al. (1999),

found that homeowners with a standard landscape used 77 mm per month, on average, for

irrigation purposes in U.S. However, in Central Florida, Haley et al. (2006) found that

typical homeowners with a standard landscape for the region, which consisted of

approximately three-quarters turfgrass across the irrigated area, used an average of 149

mm per month. Therefore, opportunities that result in better irrigation scheduling by

homeowners may lead to substantial savings in irrigation water use.

Water Use Restrictions

The Florida Water Resources Act of 1972 established a form of administrative

water law that brought all waters of the state under regulatory control. Five Water

Management Districts (WMDs) were formed, encompassing the entire state (National

Research Council, 1996; Burney et al., 1998). These agencies have the legal authority

and financial capacity to manage water comprehensively, and can impose conservation

and water shortage management (National Research Council, 1996).









The Florida Department of Environmental Protection (FDEP, 2002) specifies some

water use classifications to be employed when implementing water use restrictions,

describing landscape irrigation as "the outdoor irrigation of grass, trees and other plants

in places such as residences, businesses, golf courses, parks, recreational areas,

cemeteries, and public buildings."

When a WMD declares a water shortage, it will impose water use restrictions in

different phases depending upon the severity of the shortage. The phase names and their

specific goals in water use reduction are: I Moderate: 15%, II Severe: 30%, III Extreme:

45%, and IV Critical: 60%. Moreover, any local government has the right to impose even

stronger water restrictions (FDEP, 2002).

Where there is a year-round watering rule, it applies to everyone who uses water

outdoors-homes, businesses, parks, golf courses, etc.-regardless of the water source,

whether private well, public utility or surface water. However, there are some exceptions

to the water restrictions, such as when reclaimed or reuse water is being used (St. John's

River Water Management District [SJRWMD], 2006).

Much of Florida is under Phase II water restrictions. Basically, this means that lawn

watering is limited to two days a week (Wednesdays and Saturdays for odd-number

addresses, Thursdays and Saturdays for evens), and restricted to certain hours to reduce

evaporative and wind losses (before 1000 h and after 1600 h in the Orlando area, for

instance, or before 0800 h and after 1600 h in parts of South Florida). As of the end of

March 2001, the densely populated southernmost part of the state-including Palm Beach,

Broward, Dade and Monroe counties-was under even tougher regulations. Lawn

watering was allowed for only three hours, one day a week (SJRWMD, 2006). Since









1991, there have been water restrictions enforced by the St. Johns River Water

Management District (SJRWMD), district where this study was carried out. Residential

irrigation is limited to two days per week and prohibited between 1000 h and 1600 h,

regardless of the water source (SJRWMD, 2006).

Violating Florida's water restrictions is punishable with penalties of up to $500,

with additional fees as applicable. South Florida is enforcing a tough zero-tolerance

policy (SJRWMD, 2006).

Landscapes in Florida

Florida homeowners now maintain more than 1.5 million hectares of lawn with

20,000 hectares of new grass planted every year (American Water Works Association

[AWWA], 2005).

In an effort to meet Florida's water conservation goals, Volusia County has passed

an ordinance requiring new homes to have less grass. The ordinance mandates that new

yards at homes and businesses have landscapes requiring little or no irrigation.

Homeowners can have up to 75% of the yard with grass if the rest of the landscape

retains the original, natural vegetation without irrigation. Under the ordinance, 50 percent

of a new landscape can be irrigated up to 25 mm of water per week (AWWA, 2005).

Likewise, in Sarasota County, according to its Ordinance #2001-081, from year 2001,

new single and multi-family residences will have no more than 50% of the total irrigated

landscape dedicated to high irrigation water use zones including turf, annuals and

vegetable gardens (Sarasota County, 2006). Similar restrictions have been in effect in the

Tampa Bay area (Tampa Bay Water, 2005).

These types of ordinances that limit plant type assume that turfgrass water needs

are responsible for excessive water application. However, recent research in Central









Florida indicates that excessive water application is due to homeowner mis-management

of irrigation (Haley et al., 2006). Similar conclusions were found in the Tampa Bay

region, where approximately 30 percent of irrigation water use is wasted due to

inefficient irrigation system design, installation, operation, or maintenance (Tampa Bay

Water, 2005).

Irrigation

An efficient irrigation schedule is the application of water in the correct amount

and only when needed. Under-irrigation and over-irrigation can negatively affect

turfgrass quality. Over-irrigation tends to have environmentally costly effects because of

wasted water and energy, leaching of nutrients and/or agricultural chemicals into

groundwater supplies, degradation of surface water supplies by sediment-laden irrigation

water runoff, and erosion (Ley et al., 2000), and increased evapotranspiration (Biran et

al., 1981). Increasing irrigation efficiency, using just the appropriate amount of water to

irrigate lawns, can be achieved by a number of different methods.

Irrigation Timers

Irrigation time clock controllers, or timers, are an integral part of an automatic

irrigation system. They are an essential tool to apply water in the necessary quantity and

at the right time; however, through incorrect programming, timers can result in over-

irrigation. Time clock controllers have been available for many years in the form of

mechanical and electromechanical irrigation timers. These devices have evolved into

electronic systems that rely on solid state and integrated circuits, so they tend to be very

flexible and provide a large number of features at a relatively low cost, allowing accurate

control of water, while responding to environmental changes and plant demands (Zazueta

et al., 2002; Boman et al., 2002).









Two general types of timers are used in automatic irrigation systems: Open Control

Loop systems and Closed Control Loop systems. Open Control Loop systems apply a

preset action, as is done with simple mechanical irrigation timers. In a Closed Control

Loop (CCL) the system receives feedback from one or more sensors, make decisions, and

apply the results of these decisions to the irrigation system (Zazueta et al., 2002). First, it

is necessary to set up a general strategy in the timer. Then, the control system takes over

and makes decisions of whether or not to apply water based on data from the sensorss.

For example, soil moisture sensors can avoid irrigation when adequate soil moisture is

already present, rain sensors can prevent irrigation during or after significant rain, wind

sensors can stop the system when a speed-threshold is surpassed, sensors can be used to

detect pressure and shut the system down if the pump is not primed or to initiate flush

cycles in filters, etc. (Zazueta et al., 2002; Boman et al., 2002).

The simplest form of a CCL system is to set up a high-frequency irrigation in the

timer, which could be interrupted by a soil moisture sensor. The sensor is wired into the

line that supplies power from the timer to the electric solenoid valve (Figure 1-1). The

sensor operates as a switch that responds to soil moisture content. When sufficient soil-

moisture is available, the sensor maintains an open circuit between the timer and the

solenoid valve. When soil-moisture drops below a certain threshold, the sensing device

closes the circuit. Thus, the irrigation control system can bypass a pre-programmed

schedule, or maintain the soil water content within a specified range. These two

approaches are known as bypass and on-demand, respectively (Dukes and Mufioz-

Carpena, 2005). Bypass configurations skip an entire timed irrigation event based on the









soil water status at the beginning of that event or by checking the soil water status at

intervals within a time-based event (Mufioz-Carpena and Dukes, 2005).

Soil Moisture Content Measurement

The standard method of measuring soil moisture content is the thermogravimetric

method, which requires oven drying of a known volume of soil at 105 OC and

determining the weight loss. This method is time consuming and destructive to the

sampled soil, meaning that it cannot be used for repetitive measurements at the same

location. However, it is indispensable as a standard method for calibration and evaluation

purposes (Walker et al., 2004).

Among the widely used on-site soil moisture measurement techniques are neutron

scattering, gamma ray attenuation, soil electrical conductivity (including electrical

conductivity probes, electrical resistance blocks and electromagnetic induction),

tensiometry, hygrometry (including electrical resistance, capacitance, piezoelectric

sorption, infra-red absorption and transmission, dimensionally varying element, dew

point, and psychometric), and soil dielectric constant (including capacitance and time

domain reflectometry). Reviews on the advantages, disadvantages, and basis of these

measurement techniques may be found in Schmugge et al., 1980; Campbell and Mulla,

1990; Charlesworth, 2000; Ley et al., 2000; Topp, 2003; Mufioz-Carpena, 2004; and

Walker et al., 2004.

Granular matrix sensor

The granular matrix sensor (GMS) is a device that measures soil electrical

resistance, that can be converted to soil water tension (SWT), either using a calibration

formula provided in the literature for sandy soils (Irmak and Haman 2001) and silt loam









soils (Eldredge et al., 1993), or calibrating them for a specific soil type (Hanson et al.,

2000b; Intrigliolo and Castel, 2004).

The GMS (Figure 1-2) is made of a porous ceramic external shell with an internal

granular matrix material, which approximates compressed fine sand, containing two

electrodes. A synthetic porous membrane for protection against deterioration surrounds

the matrix material. The GMS includes an internal gypsum cylindrical tablet, which

provides buffering against salinity effects that may cause erroneous readings. A stainless

steel casing, with holes drilled in it, surrounds the synthetic porous membrane.

The GMS operates on the electrical resistance principle: water conditions in the

unit change with corresponding variations in water conditions in the soil, and changes

within the block are reflected by differences in resistance between the electrodes.

The transmission matrix material was designed to respond faster than gypsum

blocks to SWT in the 0 to 100 kPa range. Some commercial GMSs exhibit good

sensitivity to SWT over a range from 0 to 200 kPa. This makes them more adaptable to a

wider range of soil textures and irrigation regimes than traditional gypsum blocks and

tensiometers (Thomson et al., 1996; Charlesworth, 2000). Also, the GMSs are much

more stable and have a longer life than gypsum blocks and, compared to tensiometers,

require little maintenance and can be left in the soil under freezing conditions (Ley et al.,

2000).

Modern soil moisture sensors

The concept of connecting to timers one or more soil moisture sensors (SMSs) to

determine irrigation needs, and to automate irrigation systems, has moved forward in

recent years. Over the last decade, the SMS industry has advanced dramatically. Two

basic reasons can explain this advancement. The first has been the major development of









computer technology (with more powerful, smaller and economical integrated circuits).

The other phenomenon has been the significant advances in the application of

electromagnetic methods to the measurement of soil water content. These methods make

use of the high relative permittivity dielectricc constant) of the water in soil for estimating

the water content. The relative permittivity of water is about 80, whereas the other

components in soil, including air, have relative permittivities in the range of one to seven.

Hence, methods that measure the relative permittivity are effective for the measurement

of soil water content (Topp, 2003).

Combining the computer technology and the soil dielectric concept has allowed

manufacturers to produce a number of different types of inexpensive SMSs for irrigation

scheduling. An increasing adoption of the dielectric methods has been observed, because

they are non-destructive, provide almost instantaneous measurements, do not require

maintenance, and can provide continuous readings through automation. However, they

have important differences in terms of calibration requirements, accuracy, cost,

installation and maintenance requirements, etc. (Mufioz-Carpena and Dukes, 2005).

The main techniques used by these sensors can be classified as Time Domain

Reflectometry (TDR) and Frequency Domain Reflectrometery (FDR) (Leib et al., 2003).

Time domain reflectometry. The speed of an electromagnetic signal passing

through a material varies with the dielectric of the material. Most TDR instruments

operate by sending a step pulse signal down steel rods (called wave-guides) buried in the

soil. The signal reaches the end of the probes and is reflected back to the TDR control

unit where it is detected and analyzed. The time taken for the pulse to return varies with









the soil dielectric, which is related to the water content of the soil surrounding the probe

(Topp, 2003).

According to Charlesworth (2000) and Edis and George (2000), TDR instruments

give the most robust soil water content data, with little need for recalibration between

different soil types. An important advantage of TDRs in turfgrass irrigation management,

is that accurate measurements may be made near the surface compared to techniques such

as the neutron probe (Ley et al., 2000).

Frequency Domain Reflectometry. Frequency domain reflectometry (FDR)

measures the soil dielectric by placing the soil (in effect) between two electrical plates to

form a capacitor. Hence 'capacitance' is the term commonly used to describe what these

instruments measure. When a voltage is applied to the electric plates a frequency can be

measured. This frequency varies with the soil dielectric (Charlesworth, 2000).

In spite of the advances and advantages of these modern SMSs, when comparing

the performance of different brand/types, significant differences were found in respect to

set-up requirements, accuracy, data interpretation, maintenance, and initial cost (Ley et

al., 2000) and the ability to repeat measurements accurately over time and under various

moisture regimes after initial calibration (Yoder et al., 1998).

Controllers

Modern commercially available SMS-systems include a controller. This piece of

equipment is the one that sends the signal to the buried SMS and reads the soil moisture

content. The controller has an adjustable threshold (Figure 1-3), which can be set

between relatively dry to wet soil moisture conditions; depending on the plant material,

soil type, depth-installation of the SMS, etc. In general, manufacturers recommend

setting the thresholds 24 hours after a significant rainfall event or after an irrigation that









filled the soil profile with water to field capacity. The controller is connected in series

with the residential irrigation timer and acts as a switch depending on the pre-set soil

moisture threshold.

Automatic Control of Irrigation

An automatic SMS-based irrigation system seeks to maintain a desired soil

moisture range in the root zone that is optimal or adequate for plant growth and/or

quality. This type of system adapts the amount of water applied according to plant

requirements without managers having to undertake daily monitoring or make

adjustments according to actual weather conditions (Mufioz-Carpena and Dukes, 2005;

Pathan et al., 2003).

The continuous monitoring of the soil moisture status becomes particularly

important in sandy soils. A wide range of applications to automatically control irrigation

events has been investigated in coarse textured soils. In Florida, switching tensiometers

have been studied for agricultural production (Smaj strla and Koo, 1986, Clark et al.,

1994; Smaj strla and Locascio, 1994; Mufioz-Carpena et al., 2003, Mufioz-Carpena et al.,

2005), and for maintaining bermudagrass turf (Augustin and Snyder, 1984). Although

they found water savings, these investigations suggest that tensiometers require

calibration and frequent maintenance, up to twice per week. Consequently, the adoption

of this technology will not lead to automatically controlled irrigation since it will not

eliminate human interaction in irrigation management.

Other types of sensors have been adapted to automate irrigation based on soil

moisture status. Nogueira et al. (2002) used TDR sensors to maintain soil moisture within

two preset limits (upper and lower soil moisture thresholds). Dukes and Scholberg (2005)

and Dukes et al. (2003) found 11% and 50% in water savings, without diminishing yields









on sweet corn and green bell pepper, using TDR probes and a commercially available

dielectric sensor, respectively. Granular matrix sensors (GMSs) have also been used to

automatically irrigate agricultural products (Mufioz-Carpena et al., 2003; Shock et al.,

2002) and, as with other solid-state sensors, do not require as much maintenance as

tensiometers. Although TDR and GMS, as well as similar types of sensors, have been

successfully used in agriculture, they have found limited use in residential landscape

irrigation (Qualls et al., 2001).

Rain Sensors

A rain sensor (RS), also called rain shut-off device (Figure 1-4), is a piece of

equipment designed to interrupt a scheduled cycle of an automatic irrigation system

controller when a specific amount of rainfall has occurred and, depending on the weather

conditions, after the said rainfall (Dukes and Haman, 2002b; Hunter Industries Inc.,

2006).

Florida law requires a RS device on all automatic lawn sprinkler systems (Florida

Statutes, Chapter 373.62, n.d.). The original text said: "Any person who purchases and

installs an automatic lawn sprinkler system after May 1, 1991, shall install a rain sensor

device or switch which will override the irrigation cycle of the sprinkler system when

adequate rainfall has occurred." In 2001, this Chapter was amended to require the owner

not only to install, but also to maintain and operate a RS device or switch (Florida

Statutes, 2001). Moreover, some local laws also require older systems to be retrofitted

with rain shut-off switches (SJRWMD, 2006).

Florida is the only state in the nation with an overall RS statute. However, recently,

Georgia Gov. Sonny Perdue has signed into law H1277 requiring RSs on newly installed









irrigation systems in the Atlanta metro region. The new law affects systems installed after

January 1, 2005 (AWWA, 2004).

As with soil moisture sensors, rain sensors can be connected to any automatic

irrigation system controller and mounted in an open area where they are exposed to

rainfall. The new irrigation timers have a special connection, which allows a RS to be

attached directly. If it is not available, or the sensor does not work with a given timer, the

sensor can always be "hard-wired" into the controller, wiring the RS in series with the

common wire. When a specific amount of rainfall has occurred, the RS will interrupt the

irrigation system common wire, which disables the solenoid valves until the sensor dries

(Dukes and Haman, 2002b).

Figure 1-4 shows a simple and low cost RS. Rain causes the hygroscopic porous

disks in the device to swell and open a micro-switch (Figure 1-5). The switch remains

open as long as the disks are swollen. When the rain has passed and the disks dry out, the

switch will close again.

According to Dukes and Haman (2002b), the use of rain sensors has several

advantages: they conserve water, preventing irrigation after recent rain events; reduce

wear on the irrigation system, because the system runs only when necessary; reduce

disease and weeds development, by eliminating unnecessary irrigation events; help

protect surface and groundwater, by reducing the runoff and deep percolation that carries

pollutants, such as fertilizers and pesticides; and, finally, RSs save money, because they

reduce utility bills and maintenance costs.

Rain sensors should be mounted on any surface where they will be exposed to

unobstructed rainfall, but should not be in the path of sprinkler spray. These sensors are









typically installed near the roofline on the side of a building, but manufacturers

recommend mounting it in a location that receives about the same amount of sun and

shade as the turf (Hunter Industries Inc., 2006).

Irrigation and Turfgrass Quality

Under-irrigation and over-irrigation can negatively affect turfgrass quality. It has

been reported that, deeper and reduced irrigation frequency improves turfgrasses quality.

Augustin and Snyder (1984) concluded that this practice tends to reduce N leaching in

sandy soils, increasing N utilization, resulting in a better color rating (better quality).

Bonos and Murphy (1999) reported an increase in a Kentucky bluegrass (Poapratensis

L.) cultivar root growth as drought stress was imposed. Recently, Jordan et al. (2003)

found that bentgrass irrigated every 4 days produced a significantly denser and deeper

root system, a higher shoot density, and greater overall plant health, resulting in better

turf quality, than grass watered every 1 or 2 days (even under putting green management

conditions). McCarty (2005) summarizes that drier conditions slow shoot growth and

increase root growth and leaf water content.

Moreover, limitations to establishment and survival of some turfgrass weeds

(Colbaugh and Elmore, 1985; Youngner et al., 1981), and reduction of some pathogens

severity (Davis and Dernoeden, 1991; Kackley et al., 1990) has been associated with

deep, infrequent irrigation.
















~~4"4


*


Figure 1-1. Components of an automated irrigation system: A) timer, B) power supply,
C) soil moisture sensor-controller circuitry, D) soil moisture sensor, and E)
solenoid valve.


^r~.... :
tf't^
k u
.~LiJ'


























































Figure 1-2. Granular matrix sensors (


r





18









twoI









Figure 1-3. Components of an automated irrigation system. 1) Timer, and 2) soil moisture
sensor-controllers from different brands.
m --l-- I '



.-.....


































Figure 1-4. Rain shut-off switch.


Figure 1-5. The expanding material of a rain shut-off switch.














CHAPTER 2
SENSOR-BASED AUTOMATION OF IRRIGATION OF BERMUDAGRASS

Introduction

Turfgrass in landscape applications is the most extensively cultivated crop in

Florida (Hodges et al., 1994; USDA, 2005). Irrigation of residential, industrial,

commercial, and recreational turf areas is commonly employed to ensure acceptable turf

quality. As a consequence of problems related to drought, coupled with a steadily

increasing demand for water, the state of Florida has imposed restrictions on irrigation

water use. The development of Best Management Practices (BMPs) for irrigation water

use in turf has become an undeniable strategic, economic, and environmental issue for the

state. New irrigation technologies could improve irrigation efficiency promoting water

conservation and reducing the environmental impacts of the landscapes, which are often

composed of turfgrass as a major portion of the irrigated area.

Florida receives an average of around 1400 mm of rainfall a year, which typically

exceeds evapotranspiration. Nevertheless, irrigation is required because total annual

rainfall for Florida typically varies both geographically and temporally (USDA, 1981;

Carriker, 2000; NOAA, 2003), and lack of rainfall for even a few days causes depletion

of moisture in Florida's predominately sandy soils (Carriker, 2000; National Research

Council, 1996).

Florida has the second largest withdrawal of groundwater for public supply in the

United States. In 1995, nearly 93% of population in Florida used groundwater as a

drinking water source (Solley et al., 1998). Florida has a fast-growing population with a









net inflow of more than 1100 people a day. By 2025, it is projected to be the third most

populous state in the nation (Office of Economic and Demographic Research [ODR],

2006; USCB 2004a). The U.S. Census Bureau estimated that Florida accounted for

approximately 11% of all new homes constructed in the U.S. in 2003, the largest amount

in any single state in the U.S. (USCB, 2004b), the majority of them with in-ground

irrigation systems1 (Tampa Bay Water, 2005). As urban populations swell, pressures on

limited supplies of clean water are increasing. Even saltwater intrusion in groundwater

from the Floridan aquifer have been found in coastal Hillsborough, Manatee and Sarasota

counties (Southern Water Use Caution Area Recovery Strategy [SWUCA], 2006)

The primary use of residential outdoor water is irrigation. Recent studies in the

U.S. indicate that, on average, 58% of potable water is used for landscape irrigation, that

households that use automatic timers to control their irrigation systems used 47% more

water outdoors than those without timers, and that homes with in-ground sprinkler

systems use 35% more water outdoors than those without in-ground systems (Mayer et

al., 1999). In the Central Florida Ridge, the potable water used for landscape irrigation is

as high as 74%, with an average of 64% (Haley et al., 2006), and even when irrigation is

restricted to two days a week and from 1000 h to 1600 h (SJRWMD, 2006), typically

homeowners tended to over-irrigate (Haley et al., 2006).

Over-irrigation or under-irrigation can negatively affect turfgrass quality. It has

been reported that deeper and reduced irrigation frequency improves turfgrass quality.

Augustin and Snyder (1984) concluded that this practice tended to reduce N leaching in



1 57% and 85% of new homes built in Pasco and Hillsborough counties, respectively, have in-ground
irrigation systems. Actual percentages may be higher since many homeowners install irrigation systems
after moving into the home. In the Tampa region, 70% of homes are estimated to have in-ground irrigation.









sandy soils, increasing N utilization, resulting in a better color rating (better quality).

Bonos and Murphy (1999) reported an increase in a Kentucky bluegrass (Poapratensis

L.) cultivar root growth as drought stress was imposed. Recently, Jordan et al. (2003)

found that bentgrass irrigated every 4 days produced a significantly larger and deeper root

system, a higher shoot density, and an overall plant health-resulting in greater turf

quality-than that watered every 1 or 2 days (even under golf putting green management

conditions). McCarty (2005) summarizes that drier conditions slow shoot growth, and

increase root growth and leaf water content. Moreover, limitations to the establishment

and survival of some turfgrass weeds (Colbaugh and Elmore, 1985; Youngner et al.,

1981), and reduction of some pathogens severity (Davis and Dernoeden, 1991; Kackley

et al., 1990) have been associated with deep, infrequent irrigation. Hence, better irrigation

scheduling by homeowners may lead to improved turfgrass quality coupled with potential

savings in irrigation water use.

Over the last decade, the soil moisture sensor (SMS) industry has advanced

dramatically. Two basic reasons can explain this advancement. The first has been the

major development of computer technology (with more powerful, smaller and more

economical integrated circuits), and the other phenomenon has been the significant

advances in the application of electromagnetic methods to the measurement of soil water

content. These methods make use of the high relative permittivity dielectricc constant) of

the water in soil for estimating the water content. The relative permittivity of water is

about 80, whereas the other components in soil, including air, have relative permittivities

in the range of one to seven. Hence, methods that measure the relative permittivity are

effective for the measurement of the soil water content (Topp, 2003).









Combining the computer technology and the soil dielectric concept has allowed

manufacturers to design and produce a number of different types of inexpensive SMSs

for irrigation scheduling. However, when comparing the performance of different

brand/types of sensors for measurement of soil moisture, differences were found. For

example, Ley et al. (2000) found significant differences between sensors with respect to

set-up requirements, accuracy, data interpretation, maintenance, and initial cost. Yoder et

al. (1998) obtained differences related with error, accuracy, reliability, durability,

installation factors, and the ability to repeat measurements accurately over time and under

various moisture regimes after initial calibration.

Automation of irrigation systems, based on SMSs, has the potential to provide

maximum water use efficiency, by maintaining soil moisture between a desired range that

is optimal or adequate for plant growth and/or quality; allowing irrigation only when

necessary (Mufioz-Carpena and Dukes, 2005).

A wide range of applications to automatically control irrigation events have been

investigated in coarse textured soils. In Florida, switching tensiometers have been studied

for agricultural production (Smajstrla and Koo, 1986, Clark et al., 1994; Smajstrla and

Locascio, 1994; Mufioz-Carpena et al., 2003; Mufioz-Carpena et al., 2005), and for

maintaining bermudagrass turf (Augustin and Snyder, 1984). Although they found water

savings, these investigations suggest that tensiometers require calibration and frequent

maintenance, up to twice per week. Consequently, the adoption of this technology will

not lead to an automatically controlled irrigation system, since it will not eliminate

human interaction in irrigation management.









Other types of sensors have been adapted to automate irrigation based on soil

moisture status in Florida. Nogueira et al. (2002) used TDR sensors to maintain soil

moisture within two preset limits (upper and lower soil moisture thresholds). Dukes and

Scholberg (2005) and Dukes et al. (2003) found 11% and 50% in water savings-without

diminishing yields-using TDR probes on sweet corn, and a commercially available

dielectric sensor on green bell pepper, respectively. Granular matrix sensors (GMSs)

have also been used to automatically irrigate agricultural products (Mufioz-Carpena et al.,

2003; Shock et al., 2002) and, as with other solid-state sensors, do not require as much

maintenance as tensiometers.

Although SMSs have been successfully used in agriculture, they have found limited

use in residential landscape irrigation and further investigation is required to provide

evidence of their potential use in this area. A study using GMSs to control urban

landscape irrigation in Colorado, used 533 mm of water for irrigation when compared to

the theoretical requirement of 726 mm, a reduction of 27% (Qualls et al., 2001).

Since 1991, Florida law requires a rain sensor device or switch hooked up to all

automatic lawn sprinkler systems (Florida Statutes, Chapter 373.62, n.d.). A rain sensor

(RS) is a piece of equipment designed to interrupt a scheduled cycle of an automatic

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

Haman, 2002b; Hunter Industries Inc., 2005). Benefits and advantages of its use are

similar to those of SMSs, and have been summarized by Dukes and Haman (2002b).

Even when this law has been in effect for a long time, and RSs have been commercially

available for many years, little evidence related to their usefulness and/or to quantify their

water savings exists









The goals of this research were to find out if different SMS-systems (sensor with a

proprietary controller) could reduce irrigation water application-while maintaining

acceptable turf quality-compared to current practices. The objectives of this experiment

were to quantify differences in irrigation water use and turf quality between: 1) a SMS-

based irrigation system compared to a time-based scheduling, 2) different commercial

irrigation SMSs, and 3) a time-based scheduling system with or without a RS.

Materials and Methods

The experimental area was located at the Agricultural and Biological Engineering

Department facilities, University of Florida, Gainesville, Florida; on an Arredondo fine

sand (loamy, siliceous, semiactive, hyperthermic Grossarenic Paleudults) (Thomas et.al,

1985; USDA, 2003). This soil has a field capacity of 7% (Figure 2-1), as determined

from repacked soil columns (see Chapter 4 for methodology details).

Seventy-two 3.66 m x 3.66 m plots were established on a field covered with

common bermudagrass (Cynodon dactylon L.). Each plot was sprinkler irrigated by four

quarter-circle pop-up spray heads, with an application rate of 38 mm/hr and regulated at

172 kPa (Hunter 12A, Hunter Industries, Inc., San Marcos, CA). Much of the irrigation

hardware was in place from a previous research project; however, extensive renovations

were performed to make the equipment serviceable.

Plots were mowed twice weekly at a height of 5.5 cm. Chemicals were applied as

needed to control weeds and pests. Nutrient applications were made using ammonium

sulfate (21-0-0), at a N rate of 50 kg hal-, on April and May of 2004, before the beginning









of the experiment. Then, a granulated N controlled-release fertilizer (Polyon, PTI,

Sylacauga, AL)2 was applied at a rate of 180 kg hal-, on July 2004 and April 2005.

Four commercially available SMSs were selected for evaluation (Figure 2-2):

Acclima Digital TDT RS-500 (Acclima Inc., Meridian, ID), Watermark 200SS-5

(Irrometer Company, Inc., Riverside, CA), Rain Bird MS-100 (Rain Bird International,

Inc., Glendora, CA), and Water Watcher DPS-100 (Water Watcher, Inc., Logan, UT),

codified as AC, IM, RB, and WW, respectively. In order to find similar outcomes to

those that homeowners would encounter, sensors were not calibrated, and were used

directly "out of the box."

Each one of these SMSs systems includes a SMS and a controller (Figure 2-3). The

controller's thresholds can be adjusted between "dry" and "wet" on the RB (on a 1 to 8

scale), and between "moist" and "dry" on the WW (on a -3 to 3 scale). The IM can be set

at a specific soil water tension (kPa) and the AC can be set directly to a specific soil

volumetric moisture content (VMC), expressed in percent.

As recommended by manufacturers, all controller thresholds, except for the AC,

were set 24 hours after a significant rainfall event (on 20 July 2004, after four days of

rain with a total of 107 mm) that filled the soil profile with water. On RB controllers, the

thresholds were set by adjusting the dial until the LED turned off and on. On the WW,

initially the unit could not be calibrated since the soil moisture was outside the range of

the controller. After discussion with the manufacturer, a resistor was added between the

solenoid valve wire and the valve common wire. The calibration procedure consisted of

activating the reset button, which allowed its auto-calibration. The IM controller was set

2 The mention of trade and company names is for the benefit of the reader and does not imply an
endorsement of the product.









at number 1 (equivalent to 10 kPa, and approximately to field capacity, according to the

manufacturer), whereas the AC controller was set on their display at a VMC of 7%, based

on the measured soil water release curve of the soil. All these controllers were connected

in series with typical residential irrigation timers (see description of timers under

Treatments sub-heading).

Treatments

Two basic types of treatments were defined: SMS-based treatments, and time-based

treatments (Table 2-1). In the SMS-based treatments, all four brands were tested with

three irrigation frequencies: one, two, and seven days per week (1 d/w, 2 d/w and 7 d/w,

respectively). The 1 d/w and 2 d/w watering frequencies represent typical watering

restrictions imposed in Florida (FDEP, 2002; SJRWMD, 2006).

Within the time-based treatments, a frequency of 2 d/w was defined (the most

common in Florida, and current watering restriction in the area of study). Two treatments

were connected to a rain sensor (2-WRS and 2-DWRS), to simulate requirements

imposed on homeowners by Florida Statutes (Chapter 373.62, n.d.). The rain sensor

(Figure 2-4) (Mini-click II, Hunter Industries, Inc., San Marcos, CA) was set at 6 mm

rainfall threshold. A without-rain-sensor treatment (2-WORS) was also included, in order

to simulate homeowner irrigation systems with an absent or non-functional rain sensor.

Finally, a non-irrigated treatment (0-NI) was also implemented as a control for turfgrass

quality. All experimental treatments were repeated four times, for a total of 64 plots, in a

modified completely randomized design3


3 See Dry-Wet Analysis subheading for details









The weekly irrigation depth was set to replace the historical ET-based irrigation

schedule recommended by Dukes and Haman (2002a) for the area where this experiment

was carried out (Table 2-2). All treatments were programmed to have the equal

opportunity to apply the same amount of irrigation per week, except for treatments 2-

DWRS (deficit-with-rain-sensor, 60% of this amount), and 0-NI (non-irrigated). The

irrigation depths were adjusted monthly.

The irrigation cycles were programmed on two ESP-6, and three ESP-4Si model

timers (Rain Bird International, Inc., Glendora, CA) (Figure 2-3). They were

programmed to start between 0100 and 0500 h, with the purpose of diminishing wind

drift and decreasing evaporation.

Uniformity Test

An irrigation uniformity test measures the relative distribution application of water

depth over a given area. This concept results in a numeric value to quantify the variability

in depth of sprinkler irrigation over a target area. Two methods have been developed to

quantify uniformity: distribution uniformity (DU) and Christiansen's coefficient of

uniformity (CU).

According to Merriam and Keller (1978), the low-quarter irrigation distribution

uniformity (DUiq) can be calculated with the following equation:


DUz, =_- [2-1]
tot

where D1q is the lower quarter of the average of a group of catch-can measurements, and

Dtot is the total average of a group of catch-can measurements. This method emphasizes

the areas that receive the least irrigation by focusing on the lowest quarter. Although a









system may have even distribution, over-irrigation can occur because of mismanagement

(Burt et al., 1997).

On the other hand, the CU treats over-irrigation and under-irrigation equally as

compared to the mean, and can be calculated by the Christiansen (1942) formula:



CU = 1 [2-2]

i=1

where V equals the volume in a given catch-can, and V refers to the mean volume.

To carry out the uniformity test on the field, 16 catch-cans on a 0.9 m x 0.9 m

square grid pattern were placed on each plot. To minimize edge effects, this grid was

positioned 0.4 m inside the plot boundaries (Figure 2-5). The cans had an opening

diameter of 15.9 cm and a depth of 20.3 cm. Pressure at the two farthest plots was

verified with a pressure gauge. The system was set to run for 35 min, to ensure that the

average water application depth was at least 13 mm. Wind velocity during the test period

was measured with a hand held anemometer. The American Society of Agricultural

Engineers (ASAE) standards (ASAE, 2000) allow uniformity testing with wind speeds up

to 5 m/s. However, if wind was over 2.5 m/s or the distribution was affected by wind

gusts, the test was discontinued.

Dry-Wet Analysis

In accordance with manufacturer recommendations for the products tested, the

SMS should be buried in the driest zone of a multiple-zone system. Thus, that particular

zone would receive sufficient irrigation, whereas the other zones would be slightly over-

irrigated. Accordingly, to identify the driest and wettest plots in the experimental area, a









survey was carried out on each plot, before the beginning of the experiment. In addition,

because a total of 64 plots were required, this analysis was used to discard 8 plots from a

pool of 72 plots available.

On 12 March 2004, after 14 days without rainfall, a relatively "dry" soil moisture

condition was evident. The VMC was measured in each plot by means of a hand held

TDR device, which measured the moisture in the top 20 cm (Field Scout 300, Spectrum

Technologies, Inc., Plainfield, IL). Measurements were taken at five locations in the

center 1 m X 1 m of each plot and averaged. On 17 March 2004, 24 hr after a 23 mm

rainfall filled the soil profile, the VMC on a "wet" condition was measured as well. After

selecting the driest plots, the SMS that controlled a particular treatment was buried in the

center of one of these plots, thereby controlling all four replications. In all cases, SMSs

were installed in the top 7-10 cm of the soil, where most of the roots were present.

Plot Irrigation Management and Data Collection

Figures 2-6 to 2-10 show the set up of the different control features of the

experiment. Each plot was managed individually, and data were collected independently

from each plot as well. Pulse-type positive displacement flowmeters (PSMT 20mm x

190mm, Amco Water Metering Systems, Inc., Ocala, FL) were connected to nine

AM16/32 multiplexers (Campbell Scientific, Logan, UT), which were hooked up to a CR

10X model datalogger (Campbell Scientific, Logan, UT), to continually measure

irrigation volume and frequency applied to each plot (Figure 2-10). In addition, meters

were read manually each week.

Weather data were collected by an automated weather station (Campbell Scientific,

Logan, UT), located within 1 m of the experimental site (Figure 2-11). Measurements,

made every fifteen minutes, included air temperature, relative humidity, wind speed,









wind direction, solar radiation, barometric pressure, and soil heat flux. Rainfall was

recorded continuously by a manual rain gauge during 2004 and by a tipping bucket rain

gauge in 2005.

Reference evapotranspiration (ETo) was calculated from the Penman-Monteith

equation described in FAO-56 (Allen et al., 1998) as follows:


900
0.408A(R G)+/ 90 u2(e -e)
ETo= T+273 [2-5]
A + /(1 + 0.34u2)


40980.6108exp 17.27T.
A = [2-6]
(T+237.3)2

Rn = Rns Rnl [2-7]


RnI= -,K +TmmK 0.34-0.14,J 1.35R -0.35 [2-8]
2 R o

Rns = (1- a)Rs [2-9]

Rso = (0.75 + z(2 x 10-5))Ra [2-10]

Ra 24(60) Gd, [, sin(o) sin(3) + sin(, ) cos(() cos(3)] [2-11]


dr = 1 + 0.033 cos [2-12]
(365


6 = 0.409 sin 2---J-1.39 [2-13]
365

cos = arcos [-tan(ip)tan(6)] [2-14]

Se(Tmax) + e(Tm,,n)
es = [2-15]
2









RH RH
eo (Tmn ) Rmax + eo (Tm) Rmm)
m 100 max 100
ea= 100 -100 [2-16]
2


e(T) = 0.6108exp 1727T [2-17]
_T+237.3

where:
ETo = Potential evapotranspiration, mm/day
A = slope of the vapor pressure curve, kPa C-1
Rn= net radiation of the turf surface, MJ m-2 day1
Rn1= net outgoing longwave radiation, MJ m-2 day1
Rn = net solar or shortwave radiation, MJ m-2 day1
Rso = clear sky solar radiation, MJ m-2 day1
Rs = measured solar radiation W/m2 x 0.0864, MJ m-2 day1
Ra = extraterrestrial radiation, MJ m-2 day1
G = measured soil heat flux density, MJ m-2 day1
Gs, = solar constant, 0.0820 MJ m-2 min'
T = measured air temperature at a 1.5 m height, C
U2 = measured wind speed at a 2 m height, m s-1
es = saturation vapor pressure, kPa
ea = actual vapor pressure, kPa
e(T) = saturation vapour pressure at air temperature, kPa
RH = relative humidity at 1.5 m height, %
dr = inverse relative distance Earth-Sun
Cos = sunset hour angle, rad
6 = solar declination, rad
y = psychrometric constant, 0.067 kPa C1
y = Stefan-Boltzmann constant, 4.903 x 10-9 MJ K-4 m-2
J = Julian day



The soil moisture content was monitored with a capacitance soil water probe (20

cm ECH20, Decagon Devices, Inc., Pullman, WA) installed in each plot (Figure 2-12).

These probes were connected to HOBO micro-loggers (Onset Computer Corp., Bourne,

MA). Each HOBO datalogger had four probes connected to it, and readings were


recorded every 15 minutes during 2004 and every hour in 2005.









Before the beginning of the experiment, calibration of the ECH20 probes was

performed at the research site using the thermogravimetric soil sampling method

(Gardner, 1986). Four probes connected to a HOBO datalogger were installed in the field.

Undisturbed soil samples were collected from the field at less than 20 cm from the

probes, and at the same depth where the probes were placed. Samples were taken from a

saturated through a dry condition. When each sample was removed, date and time was

recorded. The volumetric soil water content of each sample was then compared to the

ECH20 probe readings at the same date and time when the samples were taken (see

Chapter 4 for methodology details) and a site-specific calibration curve was developed.

Turfgrass quality was visually assessed and rated using a scale of 1 to 9, where 1

represents brown, dormant or dead turf, and 9 represents the best quality (Skogley and

Sawyer, 1992). A rating of 5 was considered the minimum acceptable turf quality for a

homeowner. Ratings were carried out in July, October and December of 2004, and in

April, May and July of 2005.

The data were obtained from 20 July through 14 December of 2004-when the

turfgrass went dormant due to cool temperatures, and irrigation was discontinued-and

from 25 March through 31 August of 2005.

Data Analysis

Statistical data analyses were performed using the general linear model (GLM)

procedure of the Statistical Analysis System software (SAS, 2000). Analysis of Variance

was used to determine treatment differences and Duncan's Multiple Range Test was used

to identify mean differences. The combined data from both years were analyzed. When

interactions between years and other parameters were detected, each year's data were

analyzed separately and, when needed, a monthly data analysis was made as well.









Results and Discussion

Uniformity Tests

The uniformity tests resulted in a wide range of DUlq values across the plots (15%

to 78%), with an average of 52% that, according to the Irrigation Association (2003)

overall system quality ratings, is considered "fair." Obvious problems such as leaks and

broken heads were repaired prior to testing, but in some cases problems were discovered

as a result of testing (Figure 2-13, plots A8, A12, B9, and Dl) and action was taken to

correct the problem. Baum et al. (2003) performed uniformity tests on irrigation systems

of homes in Central Florida having spray heads. That research found an average DUiq of

41%, with a range of 12% to 67%.

The average CU for all the plots was 71%, with a range of 50% to 89%. Baum et al.

(2003), found a CU average of 59%, with a range between 50% and 72%. Therefore,

these experimental plots had a better distribution application of water depth, expressed as

DUlq and CU, than actual spray irrigation zones on homes sampled in Central Florida.

It is interesting to mention that, considering all the catch-cans of the experiment

responsible for the lowest readings, 99% of them were placed on the edges of the plots,

indicating that substantial edge effects occurred in the testing. This is common for

sprinkler irrigation systems and did not negatively impact the results, because soil

moisture and turf quality ratings of each plot were always taken inside this perimeter. In

addition, this situation tended to minimize the effect of irrigation overlapping between

plots.

Dry-Wet Analysis

The data collected during the dry-wet analysis are shown in Figures 2-14 and 2-15.

The statistical analysis (Appendix B, Dry) revealed that at the dry condition, plots Al









and A12 were similar, but had significantly higher VMC than the other plots. Under the

wet condition, again plots Al 1 and A12 were significantly different from the rest, also

having a higher VMC.

Therefore, plots Al 1 and A12 were discarded for this experiment because they

were too "wet" (Figures 2-14 and 2-15). On the other hand, plots A4, A5, A7, A8, F11,

and F12 were also discarded, because they appeared to have the lowest VMC values of

all plots, coupled with a comparatively lower turfgrass quality before the beginning of the

experiment. These discarded plots appear in red in the plot plans shown in Figures 2-14

and 2-15. Consequently, a total of 64 plots were left for this research. When a statistical

analysis was performed (Appendix B, Wet) with the 64 plots included, the only

significant difference was that Fl and El showed higher values at the wet condition than

the rest, so they were discarded as locations for SMS placement.

The green plots shown in Figures 2-14 and 2-15 were selected to bury the SMS for

the control plots, because they appeared to be similarly dry and on the dryer end, and

because of the practical convenience of burying the cables of the different SMS on the

same trench, at the same time, and closer to the control board (Figure 2-8). A statistical

analysis on these plots (Appendix B, Plots with SMS) indicated that they were not

statistically different (P>0.94). Figure 2-16 shows the plot plan containing all the

treatments and repetitions, in a modified completely randomized design, where the same

color depicts treatment repetitions (plots showing an X were the discarded ones).

Rainfall

Both 2004 and 2005 were rainy years (Figures 2-17 and 2-18), with high frequency

rainfall events and a large amount of cumulative precipitation, which is not uncommon in

this region. During 2004, a tropical storm and two hurricanes-Frances and Jeanne-passed









over the research area during the experiment, resulting in 159, 286, and 157 mm of

rainfall, respectively. Year 2005 broke all records for the number of hurricanes and

named tropical storms in U.S., but none of them directly hit the area where the

experiment was carried out.

Nonetheless, during the data collection period of 2005, 40% of the days had rainfall

events, with a considerable amount of precipitation, 732 mm, and an average of 135

mm/month (Figure 2-18). In the course of 2004, (Figure 2-17) even when it rained less

frequently, 31% of the days, the cumulative rainfall for the experimental period was even

larger, with 944 mm, and 190+ mm/month on average. However, most of this rainfall

(530 mm, or 56%) occurred during the tropical storm and the two hurricanes. If these

events were not considered, an average of 84 mm/month fell during 2004.

Figure 2-17 shows that in 2004 most of the rain fell during August and September

(793 mm), and the least rain fell in October and November (116 mm). A relatively "dry"

period occurred from 21 October to 24 November (35 days), with only two small rainfall

events of 1.5 and 2.5 mm. A similar situation happened in 2005 (Figure 2-18) during

April and May, when 223 mm fell over 17 rainfall events. On the other hand, June, July,

and August were rainier months with 478 mm on 44 rainfall events.

Irrigation Events

Figures 2-19 and 2-20 show the evolution of the cumulative number of irrigation

events allowed per treatment in 2004 and 2005, respectively. There were a greater

number of irrigation events as the irrigation frequency increased (Parts B vs. C vs. D,

from Figures 2-19 and 2-20). The treatment without-rain-sensor (2-WORS) was

programmed to run 2 d/w, independently of the weather and/or soils moisture conditions

and, as expected, irrigation cycles were not bypassed. However, to avoid possible damage









to the equipment, power was turned off and no data were collected from 26 September

through 30 September 2004, due to hurricane Jeanne. This is reflected on Figure 2-19,

Part A, when through these dates the slope of the curves looks horizontal (no irrigation

events). Nevertheless, this period was rainy and very short in proportion, so final results

should not be significantly affected.

Treatments 2-WRS and 2-DWRS were controlled by the same rain sensor and set

to run the same days so, not surprisingly, they overrode the same amount of irrigation

cycles, and only one line can be seen for both treatments (Figure 2-19, Part A).

In spite of this result, it is important to know what proportion of the scheduled

irrigation cycles (SIC) were finally overridden by the different treatments (Table 2-3).

The statistical analysis (P<0.0001) showed that, as expected, 2-WORS was different from

the rest of the treatments (overriding 0% of the SIC), and that 2-WRS and 2-DWRS were

not statistically different. These last two treatments bypassed 30% and 37% of the

possible irrigation events in 2004 and 2005, respectively, and more than a third as an

average for both seasons.

Regarding the SMS-based treatments, there was not a clear difference between

them, except for the IMs. Sensors from brands AC, RB, and WW, overrode significantly

more SIC than 2-WRS and 2-DWRS, ranging from 70% to 92% in total, and considering

all frequencies tested. The proportion of SIC overridden by IMs, however, ranged from

32% to 50% the first year, from 24% to 70% in 2005, and from 28% to 56% for both

years together.

In the first year, 2-IM overrode almost the same proportion of irrigation cycles than

the control 2-WRS (32% vs. 30%, respectively). In 2005, however, 2-IM was the only









treatment that resulted in more irrigation cycles than the control treatment, overriding

only 24% of the possible ones compared to 37% by 2-WRS. As a result, 2-IM was not

statistically different from 2-WRS. The other two IM treatments did not differ

statistically, but overrode more irrigation cycles compared to 2-WRS and less than the

rest of the SMS-based treatments. These results suggest that IMs, except for the said 2-

IM in 2005, would be able to respond to rainfall events at least in a similar proportion

than a rain sensor device set at 6 mm.

In order to corroborate the effectiveness of the SMSs, it was important to detect

when actual irrigation cycles occurred and how were they related to rainfall and

evapotranspiration conditions.

Figures 2-21 and 2-22 show the maximum weekly irrigation water requirement-or

weekly rainfall ETo difference (RED)-for years 2004 and 2005, respectively. In 2004,

August and September had only one week each with a negative RED. However, after 2

October, eight of eleven weeks showed a negative RED. On the other hand, in 2005, it

can be seen that every month had at least one deficitary week. However, the longest

negative REDs happened in April, May, and July of 2005, with three, four, and two

consecutive weeks, respectively.

Table 2-4 shows the percent of irrigation cycles allowed by the SMS-based

treatments through the experimental months of 2004 and 2005. In 2004, on average, a

lesser amount of irrigation events were allowed in August and September (21% and 18%,

respectively) compared to October and November (44% and 46%, respectively). In 2005,

a greater proportion of irrigation cycles were allowed in April, May, and July (31% and

46%, and 24%, respectively), and a fewer proportion in June and August (13% and 17%,









respectively). These tendencies were concordant with the dryer/rainier periods and, when

correlating the monthly RED values and the percent of irrigation cycles allowed per

month, r values of -0.93 and -0.76 were found for 2004 and 2005, respectively (Table 2-

4).

Looking in detail, for 2004, ACs followed this tendency allowing fewer irrigation

cycles during the rainy months of August and September, and more cycles during the

dryer period between October and November. In the case of 1-AC, the only irrigation

cycles allowed occurred in November. 2-AC allowed less irrigation cycles on August and

September (22% and 11%) and more cycles during October and November (22% and

33%, respectively); the same tendency as 7-AC (10%, 0%, 3%, and 17%, for the same

months, respectively).

RBs also responded to this tendency on the 1 d/w and 7 d/w frequencies, allowing a

greater proportion of SIC to occur during the end of the year; but 2-RB did not follow

this, and actually run more cycles at the beginning and at the end of the year (33% vs.

22%), with no irrigation cycles during September and October.

A similar situation occurred with the IM treatments. The first and second half of the

season were clearly different for 1-IM (between 20% 25% vs. 50% 100% of the SIC,

respectively), and for 2-IM (33% vs. 100%, respectively). However, 7-IM showed a more

even conduct for the first three months (84%, 73%, 87%, respectively), and then dropped

down in November (23%).

Finally, for 2004, WW treatments were also more active at the end of the season,

allowing between 33% and 75% of the SIC (with the exception of 2-WW, on November,

with 11%), compared to 0% to 29% for the first two months.









During 2005, ACs also followed the dryer/wetter period's tendency in every

frequency tested. In the case of 1-AC, it ran 50% and 75% of the possible times on April

and May (the driest months), and did not allow irrigations on June, July, and August (the

wetter months). On these same last three months, 2-AC and 7-AC treatments allowed no

more than 13% and 10% of the potential irrigation cycles respectively.

A similar situation happened with WW sensors. No irrigation cycles were allowed

during the rainy months of June, July, and August by the 1 and 2 d/w frequencies. In the

case of 7-WW, it followed this tendency in June and August (with 10% and 6%,

respectively), but in July-month that had two consecutive weeks of negative RED-it

showed a higher number of irrigation events (32%).

In 2005, RB sensors allowed few irrigation cycles to start during all the different

months. The 1 d/w frequency permitted between 20 and 25% of the SIC, the 2 d/w

frequency resulted in 11%, and 7 d/w frequency between 6% and 19%. Nevertheless, all

the frequencies had a month when they did not allow any irrigation cycles (April, June,

and August for 1, 2, and 7 d/w, respectively).

IM, in general, permitted more irrigation cycles than the other sensors. The 1-IM

treatment showed similar behavior to the other brands during April, May, and June.

Nevertheless, during July and August, 60% and 75% of irrigation events were allowed,

respectively. Even in the rainy months of June, July, and August, 2-IM allowed 75% to

89% of the potential irrigation cycles to occur. A smaller variation in the number of

irrigation events allowed between the different months was shown by 7-IM (from 16% to

45%), and exhibited a closer pattern to the other brands, being more active in the months









of April and May. However, it also permitted a greater amount of irrigation cycles per

month than the other brands at this frequency.

These relationships are clearer when looking at the VMC of the plots for year 2004,

when evident differences in weather conditions through time were found. Figure 2-23

shows the VMC of treatment 0-NI where all the increments in VMC were due to rainfall

events. The differences between the dry and the wet periods were reflected in the soil

moisture content. Figures 2-24 to 2-35 show the VMC in plots that contained the SMSs

controlling the irrigation treatments. These figures, show the results of the scheduled

irrigation cycles (SIC), where the blue dots represent bypassed SIC, the red dots represent

allowed SIC, and the red lines represent the range of VMC when the SIC were allowed.

When an increment in the VMC does not have a red dot, it means that a rainfall event

occurred.

These results show that, in general, the SMS-based treatments were able to follow

the dryer and wetter periods, controlling the amount of water to be delivered to the

different treatments, and suggesting that this technology could be a useful tool to achieve

automation of landscape-turfgrass irrigation, even when a RS is not present or non-

functional. However, these are not precision instruments, which was evident because

sometimes they bypassed irrigation cycles, and sometimes they did not, even reading the

same or lower VMC. Moreover, according to the range of VMC over which the different

SMS brands allowed irrigation, AC and RB had the narrowest average range (2.8% and

3.6%, respectively) suggesting that they were more accurate and consistent to measure

the VMC than WW and IM (that had and average range of 7.0 and 8.9%, respectively).









The IM controllers were set at position #1, which corresponds to -10 kPa (e.g. field

capacity) according to the manufacturer. However, according to the results obtained in

Chapter 4, at -10 kPa the GMSs were actually sensing a dryer soil condition, between -17

and -23 kPa. This explains why the IMs allowed irrigation cycles when not necessary.

Therefore, setting the IM controllers at position #2 or #3 would have resulted in increased

irrigation savings.

Irrigation Application Comparisons

Tables 2-5, 2-6, and 2-7 show (for years 2004, 2005, and 2004+2005, respectively)

the irrigation depth applied to treatments, statistical comparisons between them, and

percent of water savings achieved by the treatments compared to 2-WRS, 2-DWRS, and

2-WORS. All the statistical analyses in these Tables showed a high level of confidence

(P<0.0001), and are discussed below.

Time-based treatments vs. SMS-based treatments

Comparing the average of the time-based treatments with the average of the SMS-

based treatments, Table 2-7 (Column A) shows that there was a significant difference

between them; with 1044 and 420 mm of cumulative irrigation depth, respectively. The

same statistical difference was found in 2004 and 2005 (Tables 2-5 and 2-6). This means

that the SMS-based treatments, on average, were more efficient as a means to save water

than the time-based treatments, even when a rain sensor was an important component on

two of the three time-based treatments.

Time-based treatments

The three time-based treatments (2-WORS, 2-WRS, and 2-DWRS) were

significantly different from each other during the whole period of study (Tables 2-5, 2-6

and 2-7; Column B).









Treatment 2-WRS (two days/week, with a rain sensor) was established to mimic a

homeowner who complies with the irrigation laws and regulations and sets the timer

according to recommended practices; therefore, it was employed as the control treatment

for water use volume. During the first season, this treatment accounted for 481 mm of

water, or an equivalent of 98 mm/month, and 514 mm, or 96 mm/month, for 2005. A

recent study, carried out by Haley et al. (2006) in Central Florida, within the St. Johns

River Water Management District (SJRWMD), found that homeowners with automatic

irrigation systems applied 149 mm/month on average. Therefore, the comparisons made

here may be considered conservative and differences in the results for actual homeowners

could be larger4.

The well-managed or water conservative homeowner profile, imitated by treatment

2-DWRS (two days/week, with a rain sensor, and 60% of 2-WRS), applied 64% and 61%

of the water used by 2-WRS in years 2004 and 2005, respectively, close to the 60%

desired5. The yearly depths were 310 and 313 mm (or an equivalent of 63 and 59

mm/month), in 2004 and 2005, respectively. Haley et al. (2006) found within this

homeowner profile (also programmed to replace 60% of historical ET) an irrigation water

use of 105 mm/month6.

The treatment simulating an irrigation system with an absent or non-functional rain

sensor (2-WORS) accounted for 696 and 818 mm in the first and second season, or 141

4 Differences could be due to a better irrigation scheduling in this experiment, which was adjusted monthly.
In the Haley et al. (2006) experiment, these homeowners set their own controller run times, which generally
were not adjusted seasonally, and tended to over-irrigate in the late fall.

5 Equivalent to 36% and 39% in water savings compared to 2-WRS (Tables 2-4 and 2-5).

6 This difference could be due to less rainfall (an average of 122 mm/month during Haley's research vs.
163 mm/month in this one), different soil conditions, or because some of the homeowners in this profile
probably did not have a rain sensor, or maybe they had one, but it may have been non-functional.









and 153 mm/month, respectively. It means that this treatment applied 45% and 59% more

water than the treatment with a functional RS (2-WRS), and 52% more water during both

years, on average. These results demonstrate the importance not only for the presence,

but also for the need of a functional and well-maintained rain shut-off device on all

automated irrigation systems in Florida; where rainy weather, particularly in the warm

months, is common (NOAA, 2003).

Moreover, as the study prepared by Whitcomb (2005) recently found, just 25% of

the surveyed homeowners in Florida with automatic irrigation systems reported having a

RS, and the author suggests that they are often incorrectly installed. Therefore,

appropriately installed and properly working rain sensors could signify not only

substantial water savings to homeowners, but could also lead to sound environmental and

economic benefits to the state. In addition, their payback period could be less than a year;

depending on the weather conditions, the area to be irrigated, the cost of water, and the

cost of installed rain sensors (see Chapter 3).

Comparisons between SMS-irrigation frequencies

When the averages of the three different SMS irrigation-frequencies were analyzed

(Tables 2-5 to 2-7, Column C), the 2 d/w frequency used a significantly higher volume of

water, followed by the 1 d/w frequency during both seasons and as a total, with 478 and

420 mm of total cumulative water depth, respectively. The 7 d/w frequency was as high

as the 2 d/w frequency in 2004, but resulted in the least water applied in 2005 (mostly

due to the decrease in water application by 7-IM between 2004 and 2005). Therefore, 7

d/w was significantly the lowest of all three frequencies, with an average of 362 mm, in

total cumulative water applied in this experiment (Table 2-7). This was probably because









more frequently scheduled irrigation events can be bypassed as a result of frequent

rainfall.

These results suggest that, in a long run, to schedule low volume-high frequency

irrigation cycles (7 d/w) in Closed Control Loop irrigation systems7, appears to be a

better strategy regarding water conservation in turfgrass irrigation, than to schedule them

for some specific days during the week (1 or 2 d/w) and with a higher volume during

each irrigation cycle.

Soil moisture sensor-brands comparison

The different brands were compared in terms of irrigation water applied. Figure 2-

36 shows the cumulative water applied by brand on years 2004 and 2005. As an overall

comparison, IM sensors resulted in significantly more irrigation during 2004 (P<0.0001),

with 420 mm on average, followed by WWs, 188 mm, and then by sensors from brands

AC and RB-which were not statistically different-and showed the lowest water use rate,

with 116 and 100 mm, respectively. The same relationship was found in 2005, except that

in this year, AC used a significantly higher irrigation depth than RB, with all four brands

statistically different, with 451, 164, 135 and 105 mm on average for IMs, WWs, ACs

and RBs, respectively. However, these averages could not be directly compared to find

out which brand was better in every case, because an interaction between brand and

frequency was evidenced by the statistical analysis (P<0.0001). This implies that some

SMS brands performed better at a certain frequency than other ones, but not as good at

another frequency. Hence, differences between SMSs brands were evaluated separately

within each irrigation frequency.

7 In Closed Control Loop irrigation systems the decision to whether initiate or not an irrigation cycle is
regulated by a SMS.









Brand comparisons within irrigation frequencies

From Tables 2-5 to 2-7 (Column D), it can be seen that there were statistical

differences between brands within each frequency, for 2004, 2005, and as a total

cumulative depth for both seasons. In 2004, the highest water use rate in the 1 d/w

frequency was displayed by IM, and then by WW, RB, and AC, respectively; all

statistically different from each other. A similar trend was exhibited at the 7 d/w, except

that AC and RB were not significantly different at this frequency. At the 2 d/w frequency,

however, the least amount of water was applied by brands RB and WW (not statistically

different between them), then by AC, and then by IM.

During 2005, results from 1 and 2 d/w frequencies were similar. At 1 d/w all four

brands were statistically different from each other. However, comparatively, WW applied

less water than in 2004 and, conversely, AC applied more. So, for 2005, IM applied the

most, followed by AC, RB, and WW, respectively. Finally, at the 7 d/w frequency, the

same statistical differences between brands from 2004 were found in 2005, showing a

high consistency through time.

These similarities and differences, comparing both years, could have happened

because the number of total SIC was relatively greater in the 7 d/w frequency, and

relatively smaller in the 2 and 1 d/w frequencies (302, 87, and 43 times, respectively).

This means that timing of the rain events could have had a higher impact at the 1 d/w and

2 d/w level, but lower at the 7 d/w frequency. Basically, these differences took place in

the dry months of April and May 2005, when 1-AC, 1-WW, and 2-WW allowed some of

the SIC to run (Table 2-4).

The total cumulative irrigation depth applied in both seasons (Table 2-7,

Comparison Column D) showed that, in the 1 d/w frequency, only IM used significantly









more water than the other brands. For the 2 d/w frequency, all brands were significantly

different, where IM applied more water, and then, in decreasing order, AC, WW, and RB.

As 2004 and 2005, IM was the brand that significantly applied more water at the 7 d/w

frequency, followed by WW, and then by AC and RB (which were not statistically

different).

Summarizing, IMs always applied significantly more water than the other brands in

every frequency tested, and in both seasons. This could be because of their reported

limitations to timely sense the differences in soil water content, their hysteretic behavior,

the high variability of its readings, and their drawbacks for its use in sandy soils, where

low tension values are necessary to prevent plant stress (Irmak and Haman 2001; Taber et

al., 2002: Intrigliolo and Castel, 2004; McCann et al., 1992). Therefore, IMs do not

appear to be the best choice for these climatic-soil conditions.

Regarding the other brands, the best choice depends on the local restrictions and

regulations concerning the frequency of landscape irrigation. When irrigation is limited to

1 d/w, no differences appeared to be evident between the brands. For 2 d/w restrictions,

according to these results, the best choice would be, in decreasing order, RB, WW, and

AC. If no frequency limitations were present, AC or RB looks like the best choice,

followed by WW.

Overall comparison

Tables 2-5, 2-6 and 2-7, (Column E), show the statistical differences between every

irrigation treatment, independently if they were time-based or SMS-based. As expected,

2-WORS always applied significantly more water than the other treatments. During 2004,

following 2-WORS, were 2-WRS, 2-IM, and 7-IM, with no significant difference

between them, and these treatments were followed by 2-DWRS and 1-IM, which were









similar between them. However, during 2005 and 2004 + 2005, all these six treatments

were significantly different between them, and showed significantly higher levels of

water use than the rest of the treatments. On the other extreme, 7-AC and 7-RB always

resulted in significantly less cumulative irrigation depth throughout this experiment.

Table 2-7 shows that 2-IM was the only SMS-based treatment that applied

significantly more water (11%) than the control treatment 2-WRS. Conversely, the other

two IM treatments saved a significant amount of water (20% and 28%, by 1-IM and 7-

IM, respectively) compared to 2-WRS. However, these last proportions were far from the

water savings achieved by the other SMS-based treatments, when compared to 2-WRS:

AC sensors recorded irrigation water savings ranging from 65% to 88%, RBs from 72%

to 85%, and WWs from 54% to 73%, depending on the irrigation frequency tested. It is

important to remark that these water savings were on top of those already achieved by 2-

WRS. Therefore, these results show that, in general, SMSs can also act as "rain sensors",

but with a superior performance in terms of water savings.

When the irrigation treatments were compared to almost 75% of the surveyed

homeowners in Florida (Whitcomb, 2005), this is with a non-functional or absent rain

sensor (2-WORS), the significant difference in water savings increased, ranging from

77% to 92% for ACs, 81% to 90% for RBs, and 69% to 82%, for WWs. Even 2-IM

(which applied 11% more water than 2-WRS) showed significant water savings, 27%

with respect to 2-WORS, indicating that this sensor was working but did not bypass as

much SIC as other SMS-based treatments.

Moreover, when compared to the water conservative 2-DWRS, treatments from

brands AC, RB and WW also showed significant water savings, that ranged from 44% to









80%, 55% to 76%, and 26 to 57%, respectively. On the other hand, all IM-frequencies

applied significantly more irrigation than 2-WORS, with values that ranged from 15% to

77% more water.

These results clearly demonstrate that the use of SMSs, along with traditional

timers in residential irrigation systems, could lead to important water savings. However,

the correct choice of the SMS, and its technology to measure or "sense" the soil water

status, is of great consequence.

Automation of Irrigation Systems

A complete automation of a residential irrigation system, based on SMSs, could be

achieved programming the timer to run every day as a scheduling strategy. Then, the

SMSs will allow the system to initiate the irrigation cycles only when it is actually

needed by the turfgrass, and override them when the sensed water content is over a pre-

set threshold.

In this experiment, this was confirmed when most of the SMS-based treatments

were able to follow the dry and rainy periods-controlling the amount of water to be

delivered to the different treatments (Table 2-4)-when the 7 d/w irrigation frequency

used significantly less water than the other frequencies (Table 2-7, Column C), and when

two of the SMS-based treatments programmed to run 7 d/w consistently used the smallest

amount of water (Tables 2-5, 2-6, and 2-7). In effect, treatments 7-AC and 7-RB recorded

total water savings of 85% or more, when compared to 2-WRS, and 90% or more when

compared to 2-WORS (Tables 2-7).

It is interesting to note that this concept (with a potential irrigation frequency of

seven days a week) is contradictory to the water use regulations and restrictions imposed

by the Water Management Districts and/or municipalities in Florida (where irrigation is









allowed only one or two days per week). However, these results suggest that choosing the

right type of sensor, and programming the automatic irrigation system to run everyday for

a short period of time (allowing the SMS to decide whether or not to irrigate), could save

large amounts of potable water used for irrigation purposes, and looks like a better BMP

than the current ones.

Moreover, this concept is not in opposition to the general recommendation for

deeper and less frequent irrigation for turfgrass, because these treatments (7-AC and 7-

RB) finally overrode more than 85% of the SIC (Table 2-4), resulting in a low actual

irrigation frequency, which was supplemented by large rainfall events.

Turfgrass Quality

No differences in turfgrass quality, including non-irrigated plots, were found

among treatments. This could be explained in part by the generally wet climatic

conditions that happened through almost all the time of the experiment, which favored

the growth and development of the bermudagrass. Another factor contributing to this,

even during the "dry" periods, could be found in the species itself. Common

bermudagrass is known as a more drought-tolerant grass compared to the pervasive St.

Augustinegrass [Stenotaphrum secundatum (Walt.) Kuntze] found in North-Central

Florida landscapes (Harivandi et al., 2001; Baldwin et al., 2005; Turgeon, 2005). As a

result, the treatment effects were buffered with respect of the turfgrass quality

parameters, and it could be concluded that no irrigation was necessary to maintain an

acceptable turf quality during the experiment time-period. Figure 2-37 shows, in Part A,

an homogeneous good quality between the different plots, and, in Part B, when

bermudagrass went dormant on all plots at the same time. Having analogous weather

conditions, Jordan et al. (2003) obtained similar results working with bentgrass.









Summary and Conclusions

High frequency rainfall events and a large amount of cumulative precipitation

prevailed during the time frame of this research, except for some weeks at the end of

2004 and at the early spring of 2005, when ETo exceeded rainfall. When the monthly

deficit rainfall values (relative to ETo) were correlated with the percent of irrigation

cycles allowed per month, r values of -0.93 and -0.76 were found for 2004 and 2005,

respectively. It was inferred that most of the SMS-based treatments automatically

canceled the majority of the irrigation cycles during the rainy periods, and responded to

dry periods by allowing irrigations to occur.

The three time-based treatments (2-WORS, 2-WRS, and 2-DWRS) were

significantly different from each other during the whole study period. The treatment

without a rain sensor (2-WORS) used, on average, 52% more water than the treatment

with a functional one (2-WRS), showing the importance of a well-maintained rain shut-

off device in all automated irrigation systems in Florida. On the other hand, treatment 2-

DWRS, applied close to 60% of the water applied by 2-WRS.

These time-based treatments were established to mimic the operation of irrigation

systems carried out by different homeowner profiles. However, according to the results of

this research, these treatments were fairly well managed compared to homeowners' actual

operation practices in the Central Florida Ridge. Therefore, results in water use from this

experiment can be considered conservative and differences for actual homeowners could

be even larger.

When the time-based treatments were compared to the SMS-based treatments,

results showed that, on average, the SMS-based treatments were significantly more

efficient as a means to save water than the traditional time-based treatments. However,









not all SMS-treatments tested performed the same. The 2-IM treatment was the only

SMS-based treatment that applied significantly more water than the control 2-WRS (11%

more). The other two IM treatments, 1-IM and 7-IM, used significantly less water than 2-

WRS (20% and 28%, respectively), but always applied significantly more water than the

other brands/treatments in every frequency tested. Therefore, IMs do not appear to be the

best choice for the weather and soil conditions of this study.

All the other brands (AC, RB, and WW) recorded significant irrigation water

savings compared to the control 2-WRS, which ranged from 54% to 88%, depending on

the irrigation frequency. These results showed that most SMSs, except for 2-IM, can also

act as rain sensors, with superior performance in terms of water savings. When these last

brands were compared to 2-WORS, the differences in water savings increased, and

ranged from 69% to 92% over the 308-days study period.

Irrigation frequencies were also compared at the end of the study. All three

frequencies tested (1, 2, and 7 d/w) were significantly different. The 2 d/w frequency

used the highest volume of water, followed by the 1 d/w frequency, and 7 d/w was the

one that used the least amount of water. Moreover, and being part of this last frequency,

treatments 7-AC and 7-RB significantly and consistently used the smallest amount of

water regarding all treatments, during both seasons. They recorded total water savings of

85% or more, when compared to 2-WRS, and 90% or more when compared to 2-WORS.

These results suggest that scheduling low volume-high frequency irrigation cycles

(7 d/w) in Closed Control Loop irrigation systems, appears to be a better strategy

regarding water conservation for turfgrass irrigation in Florida's sandy soils during rainy

periods, than scheduling irrigation cycles one or two days per week. Moreover, it was






53


concluded that no irrigation was necessary to maintain an acceptable turf quality during

the experimental period, which was evidenced by acceptable quality in non-irrigated

plots. Therefore, SMS-based technology could lead not only to a complete automation of

residential irrigation systems, but to save substantial irrigation water if implemented.









Table 2-1. Irrigation treatment codes and descriptions.
Treatment Irrigation Frequency Soil Moisture Sensor Brand
Codes (days/week) or Treatment Description

Soil Moisture Sensor-Based
1-AC 1 Acclima
1-RB 1 Rainbird
1-IM 1 Irrometer
1-WW 1 Water Watcher
2-AC 2 Acclima
2-RB 2 Rainbird
2-IM 2 Irrometer
2-WW 2 Water Watcher
7-AC 7 Acclima
7-RB 7 Rainbird
7-IM 7 Irrometer
7-WW 7 Water Watcher

Time-Based
2-WRS 2 With rain sensor
2-WORS 2 Without rain sensor
2-DWRS 2 Deficit with rain sensor, 60% of 2-WRS
0-NI 0 No irrigation

Table 2-2. Monthly irrigation depth to replace historical evapotranspiration, assuming
system efficiency of 60%, and considering effective rainfall.


Month
January
February
March
April
May
June
July
August
September
October
November


Irrigation depth (mm)
0
0
112
112
183
142
137
178
137
122
91


December 91
Source: Based on Dukes and Haman (2002a)










Table 2-3. Total number and percent of overridden scheduled irrigation cycles; 2004 and
2005.


2004
Treatment Scheduled Overridden
(#) (%)
2-WORS 40 0
2-WRS 40 30
2-DWRS 40 30
2-IM 41 32
1-IM 20 50
7-IM 142 41
1-RB 20 55
1-WW 20 63
7-WW 142 65
2-AC 41 73
1-AC 20 85
2-WW 41 85
2-RB 41 85
7-AC 142 92
7-RB 142 89
P<0.0001


2005
Scheduled Overridden
(#) (%)
46 0
46 37
46 37
46 24
23 48
160 70
23 83
23 87
160 75
46 83
23 78
46 78
46 91
160 92
160 93


2004 + 2005
Scheduled Overridden
(#) (%)
86 0 f
86 34 e
86 34 e
87 28 e
43 49 d
302 56 d
43 70 c
43 76 c
302 71 c
87 78 bc
43 81 abc
87 82 abc
87 89 ab
302 92 a
302 91 a










Table 2-4. Percent of irrigation cycles allowed by the SMS-based treatments through the
experimental months of 2004 and 2005.
Treatment Year 2004 (%) Year 2005 (%)
Treatment
Aug Sep Oct Nov Apr May Jun Jul Aug
1-AC 0 0 0 75 50 75 0 0 0
2-AC 22 11 22 33 22 33 13 11 11
7-AC 10 0 3 17 17 10 0 6 10
1-RB 0 40 100 75 0 25 20 25 20
2-RB 33 0 0 22 11 11 0 11 11
7-RB 6 3 13 17 7 19 7 6 0
1-IM 25 20 50 100 25 100 20 75 60
2-IM 33 33 100 100 67 100 75 89 67
7-IM 84 73 87 23 40 45 17 35 16
1-WW 0 20 75 50 25 50 0 0 0
2-WW 11 0 33 11 67 44 0 0 0
7-WW 29 20 42 33 37 42 10 32 6
Average 21 18 44 46 31 46 13 24 17
RED (mm) 184 397 -16 -17 -14 -23 76 -25 30
CC (r) -0.93 -0.76
RED = Rainfall ETo difference
CC = Correlation coefficient









Table 2-5. Cumulative irrigation depth applied to treatments, statistical comparisons
between them, and percent of water savings compared to 2-WRS, 2-DWRS,
and 2-WORS; year 2004.

treatment Cumulative depth Comparisons+ Water savings (%) vs.
2004 (mm) A B C D E 2-WRS 2-WORS 2-DWRS

SMS-Based
1-AC 95 d g 80 86 69
1-RB 128 c f 73 82 59
1-IM 318 a c 34 54 -3
1-WW 209 b e 57 70 33
1-Avg 188 b
2-AC 196 b e 59 72 37
2-RB 87 c gh 82 88 72
2-IM 470 a b 2 32 -52
2-WW 94 c g 81 87 70
2-Avg 212 a
7-AC 57 c h 88 92 82
7-RB 85 c gh 82 88 73
7-IM 471 a b 2 32 -52
7-WW 261 b d 46 63 16
7-Avg 218 a
SMS-Avg 206 b

Time-Based
2-WORS 696 a a -45 0 -125
2-WRS 481 b b 0 31 -55
2-DWRS 310 c c 36 55 0
Time-Avg 496 a


SMS=Soil moisture sensor
Avg=Average
AA=Time-based treatments vs.


P<0.0001


SMS-based treatments


B=Between time-based treatments
C=Between irrigation frequency averages
D=Within irrigation frequency
E=Overall comparison









Table 2-6. Cumulative irrigation depth applied to treatments, statistical comparisons
between them, and percent of water savings compared to 2-WRS, 2-DWRS,
and 2-WORS; year 2005.

treatment Cumulative depth Comparisons+ Water savings (%) vs.
2005 (mm) A B C D E 2-WRS 2-WORS 2-DWRS

SMS-Based
1-AC 188 b gh 63 77 40
1-RB 153 c i 70 81 51
1-IM 475 a d 7 42 -52
1-WW 114 d j 78 86 64
1-Avg 232 b
2-AC 152 b i 70 81 51
2-RB 101 c j 80 88 68
2-IM 635 a b -24 22 -103
2-WW 177 b h 66 78 44
2-Avg 266 a
7-AC 65 c k 87 92 79
7-RB 62 c k 88 92 80
7-IM 244 a f 52 70 22
7-WW 202 b g 61 75 36
7-Avg 143 c
SMS-Avg 214 b

Time-Based
2-WORS 818 a a -59 0 -161
2-WRS 514 b c 0 37 -64
2-DWRS 313 c e 39 62 0
Time-Avg 548 a


SMS=Soil moisture sensor
Avg=Average
A=Time-based treatments vs.


P<0.0001


SMS-based treatments


B=Between time-based treatments
C=Between irrigation frequency averages
D=Within irrigation frequency
E=Overall comparison









Table 2-7. Total cumulative irrigation depth applied to treatments, statistical comparisons
between them, and percent of water savings compared to 2-WRS, 2-DWRS,
and 2-WORS; years 2004 + 2005.


Total cumulative Comparisons+ Water savings (%) vs.
Treatment depth 2004+2005
(mm) A B C D E 2-WRS 2-WORS 2-DWRS

SMS-Based
1-AC 283 b i 72 81 55
1-RB 281 b i 72 81 55
1-IM 793 a d 20 48 -27
1-WW 323 b h 68 79 48
1-Avg 420 b
2-AC 348 b h 65 77 44
2-RB 188 d j 81 88 70
2-IM 1105 a b -11 27 -77
2-WW 270 c i 73 82 57
2-Avg 478 a
7-AC 122 c k 88 92 80
7-RB 147 c k 85 90 76
7-IM 715 a e 28 53 -15
7-WW 463 b g 54 69 26
7-Avg 362 c
SMS-Avg 420 b

Time-Based
2-WORS 1514 a a -52 0 -143
2-WRS 995 b c 0 34 -60
2-DWRS 623 c f 37 59 0
Time-Avg 1044 a


SMS=Soil moisture sensor
Avg=Average
A=Time-based treatments vs.


P<0.0001


SMS-based treatments


B=Between time-based treatments
C=Between irrigation frequency averages
D=Within irrigation frequency
E=Overall comparison











35


30


25


S20


T 15


10


5


0
0 5 10 15 20 25 30
Tension (kPa)

Figure 2-1. Soil water retention curve from tensiometers and calibrated ECH20 probe
readings.











^y VI -------
Rainbird
Irrometer


Water Watcher


Acclima


Figure 2-2. Soil moisture sensor brands tested in this study.


'I






62






: =-- .








S-
--- I
.- l-~-- -- ^^^^^^^^^^B


--
nn
.1


I3


Acclma""'.
Acclimat2 ii- -


Figure 2-3. Irrigation controls as installed for this study: soil moisture sensors-controllers
brands: A) Rain Bird, B) Water Watcher, C) Acclima, and D) Irrometer, and
irrigation timer E) Rain Bird.









































figure 2-4. Kain sensor installea tor tnis study.


































2V'


igure 2-5. Catch-can display for uni


rmity tests on turfgrass plots.


































Figure 2-6. General view of the irrigation controls used in this study.


Figure 2-7. Pipes, flowmeters, valves, and wirings for this study.






66
















.......... ... .':"I..j .. ..







Figure 2-8. Control board showing timers, soil moisture sensor-controllers, solenoid
valves wiring, and flowmeters-datalogger (details are shown in the next
Figures).

































Figure 2-9. Control board detail showing the solenoid valves control box.











A A A A













Figure 2-10. Control board detail, flowmeter-datalogger boxes showing A) multiplexers,
B) CR 10X datalogger used for this study.























































Figure z- 11. Automated weather station near turt plots tor tnis stay.










i.WS


figure 2-12. EC-H2U probe, capacitance soil moisture probe shown with a HU-BU data
logger as installed for this study.








A B C D E F

12 15 52 54 71 78 60 12


11 56 66 48 63 40 76 11


o1 41 51 61 53 64 79 io


9 46 21 50 58 43 66 9


8 27 46 49 26 60 64 8


7 49 51 53 44 59 67 7


6 35 49 46 56 58 26 6


5 51 31 52 34 52 76 5


4 62 64 47 64 55 68 4


3 45 61 40 36 56 55 3


2 58 69 37 51 33 39 2


1 61 55 64 28 52 63 1

A B C D E F
Figure 2-13. Plot plan showing the low-quarter distribution uniformity testing results on
each plot.








A B C D E F

6.4 7.4 8.8 5.6


5.4 5.4 11.4 6.8


6.2


6.8


10.2


6.0


7.0


7.4 10.0 8.0 7.6


9.6 7.8 9.4 8.4


4.6


7.4


5.0


6.4


6.6


7.8


9.4


9.2


8.0


9.0


5.8


5.6


7.2


6.0 6.6 8.2 8.4


A B C D E F
Figure 2-14. Plot plan showing average volumetric water content (%) on each plot during
a relatively "dry" period. Plots in red were discarded, and plots in green were
used for placement of SMSs.


o1 6.4


9 6.0


6


5.4


3 5.8


1


4.4








A B C D E F

11.0 11.6 12.2 11.0

11.6 12.2 12.8 10.8


8.8


10.4


10.4


10.8


11.0 11.4 10.4 10.8

12.6 12.2 11.8 10.6


11.6

12.2

10.2


12.0

10.2


11.0

10.8


11.0


9.6


11.2

9.2


11.2


1 12.2 11.6


11.2 12.0


14.8 15.8


A B C D E F
Figure 2-15. Plot plan showing average volumetric water content (%) on each plot during
a relatively "wet" condition. Plots in red were discarded, and plots in green
were used for placement of SMSs.


9.8


io 12.4

9 11.4


6 10.6


3 9.8


















































11-VWV.3 0-NI.4

A B C D E F
Figure 2-16. Plot plan with the modified completely randomized design (same color
depicts treatment repetitions).














30


25


20
E


E 15


0 10


5


0


104 35 269


1000
344
900

800

700 E
E
600

500

400 3
E
300

200

100


S0 ) Q K> > > > > U U
U0 0000000 0W Q
0 4 i f o M 0 1- V -. co W o0 gc6 Ol C D C. IO i. V

Date (2004)

Figure 2-17. Daily and cumulative rainfall in 2004. Note: rainfall for 5 Sep. (188 mm)
and 6 Sep. (81 mm) is shown as a cumulative total (269 mm).


30 -.- "


25


E 20
E

E 15


m 10


5


0
O IMI. M C CI

0Date 2005)
Date (2005)


600 -
E
500 =

400 -

300 .
E
200

100


Figure 2-18. Daily and cumulative rainfall in 2005.













85
75
65
' 55
Z L 45
> 035
35
E 25
15
5
-5
20-Jul


40

28





19-Aug 18-Sep 18-Oct 17-Nov 17-Dec
Date (2004)

2-WRS 2-WORS 2-DWRS


85
75
65
E 55
Z 45
S0 35
S25 28
E
S 15 1
5 6
-5
20-Jul 19-Aug 18-Sep 18-Oct 17-Nov 17-Dec

C Date (2004)
-2-AC -2-RB -2-IM -2-WVV


85
75
65
E 55
z L 45
-> 35
| 25
E 5
15
0


8-10
3


20-Jul 19-Aug 18-Sep 18-Oct 17-Nov 17-Dec
Date (2004)

S -1-AC -1-RB -1-IM -1-WW


85
75
S 65
55

Z 45
S35
E E 25
E
15

-5
Jul
D


49




16



19-Aug 18-Sep 18-Oct 17-Nov 17-Dec
Date (2004)
I-7-AC -7-RB -7-IM -7-VVWW


Figure 2-19. Cumulative number of irrigation events per treatment in 2004; A) time-

based treatments, and soil moisture sensor-based treatments at irrigation

frequencies of B) 1 d/w, C) 2 d/w, and D) 7 d/w. Note: In part A), treatments

2-WRS and 2-DWRS were controlled by the same rain sensor and were set to

run at the same days, so only one line can be seen for both treatments.












50
45
40
8 35
E 30
Z 25
20
I M15
E 10
0 5

5
25-Mar 24-Apr 24-May 23-Jun 23-Jul 22-Aug

A Date (2005)
-2-WRS -2-WORS -2-DWRS


50
45 45
40
S35
29 E 30
zL 25
S20
-)15
E 10
S10

0


12
5
4
3


25-Mar 24-Apr 24-May 23-Jun 23-Jul 22-Aug

B Date (2005)
--1-AC -1-RB -1-IM -1-VVW


50 50 48
45 45 -
40 5 40 40
35 35 35
S30 30
z 25 Zw 25
o020 >.- 20
E 15 15 13
10 10 E-10 12
a 8 =5
5 4 0
0 0
-5 ---- -5
25-Mar 24-Apr 24-May 23-Jun 23-Jul 22-Aug 25-Mar 24-Apr 24-May 23-Jun 23-Jul 22-Aug
C Date (2005) D Date (2005)
-2-AC -2 -2M -2-WW A 7] --7-AC --7-RB --7-IM -7-WW

Figure 2-20 Cumulative number of irrigation events per treatment in 2005; A) time-based
treatments, and soil moisture sensor-based treatments at irrigation frequencies
of B) 1 d/w, C) 2 d/w, and D) 7 d/w. Note: In part A), treatments 2-WRS and
2-DWRS were controlled by the same rain sensor and were set to run at the
same days, so only one line can be seen for both treatments.












80 m ",m

E
S60
U
C
0 40


20





-g -20


-40
r co 0 co O Cm 0) eD C o 0 C o o 0 ,- V
3 0 6. 6. C. C) U) 0 > > >
4) a0) a) 0)00 C C ) 0 O) <0)
44 CO Week (2004)

Figure 2-21. Maximum weekly irrigation water requirement (rainfall ETo difference);
year 2004.






80


E 60
U

S40


S -20

I--I I
.4-
0


yr 20200


-40 -
V W CO N C' 0 CD C' 0 CD C0 0 r- CO N r C0 M0 -W

( C 4 L (U L > >. >. 3 3 C C C -, 0) 0 0) 0)

Week (2005)

Figure 2-22. Maximum weekly irrigation water requirement (rainfall ETo difference);
year 2005.


























25





S20

U










-OO
15
0
E


E 10











0


C Co D CD D CD CD CD 0 C C C Co C Co 0 CC C Co C Co C C 0 Co C Co 0 CC C
O C O O Co O O C O CD) O CD CD Co C O C CO O CD O C O O C) O O CO O O O O CD 0D

<- O C CD ID CD OO OC 0D CD-a C C t- O CO --O CO CO -l -

CD CD C I= CD Ca a' C C CD a' CD CD CO O O OO O O C O OC D C O O C C O O O 3 O O O 2 N O O O C O O D
r--- r-- Ca CO 0 00 00 CO 00 O a' ) a' a' 0-) a' a' 0 00 0C 0C 0C 0 '



CD CD C3 CD 3 C3 CD CD CD CD CD CD CD CD C D CD =
Date


Figure 2-23. Volumetric moisture content through time, on treatment 0-NI, year 2004.













































7.1




4.1





C C C C C C C C C C CD CD C C' C C' CM C CM C C C C C C C C C C CM C C C=

C C C C O C O C -O C C C Cl C C I CT Cl C' 3 C` 3 CV C C C C C







Figure 2-24. Volumetric moisture content (VMC through time, showing results of the scheduled irrigation cycles (SIC), where the

















F I


25 -




1 20 -
D

0


t 15



1 12.3


> 8.0.7
S__8.0 8.7 8.9 8.8


5 -




0

CD CD CD C O CD CD CD C) O C) C) CO I= O C) C CD CD CD CD CD CD CD CD O C> C ) 9 O C3 C)
0 0 0 CA0 0 0 0 0 0 0 0 0 0 0 0 0 00 0 0 0 0 0 000 0 0 0 0 0 0
o 00 o 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
o o 0 0 0 0 0 0 0 0 0 0 0 0 0Q 0 0 0 0 0 00 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0


0 0 00 0 0 00 00 00) 0 0 0 0 0 000 000 00
Cl Cl Cl CD C D C C c 0 0 C1 C( -- 0 0 C>l i- Cl 03 0 l ( 0 0 0

Date

Figure 2-25. Volumetric moisture content (VMC) through time, showing results of the scheduled irrigation cycles (SIC), where the
blue dots represent bypassed SIC, the red dots represent allowed SIC, and the red lines represent the range of VMC when
the SIC were allowed; treatment 1-IM, year 2004.

















30




25


Fe
S20 -

0
o


i 15




2 10





5.2 5.2 5.9



0
C? C? C? C? C? C? C? C? C? C? C? C? C? C? C? Ci Ci Ci Ci C C ? C ? C ? C? C ? C ? ci 3 C C C C C:' C' C:' C:' C' CC C
F c2 cV c t ci cm ci ci c5 content (VMC) through tci cs h ing reslt of th schedued: irrigation cyclCC C: C (IC), C e C C h
o oD CD CD CD CD CD CD CD CD CD CD CD CD CD CD CD CD CD CD CD CD CD CD CD CD CD CD CD CD CD CD CD CD CD CD CD CD CD CD CD CD CD
CD N c DN c CN c MN C MN C M C M(N C (N C (N C M C ( CD ( Ci N, D C C C 4-- C -- C -- C -- Cl C- --
O i- C C O i- C 2- C 2- C 2- C - - C CN CN CN CN CN CN C

SO G CD CD CD CD CD CD M0 00 00 000 00000000 C C C C C

Date

Figure 2-26. Volumetric moisture content (VMC) through time, showing results of the scheduled irrigation cycles (SIC), where the
blue dots represent bypassed SIC, the red dots represent allowed SIC, and the red lines represent the range of VMC when
the SIC were allowed; treatment 1-RB, year 2004.









































6.2 8.9

5 6.2 6.2






S C C CD C CD C C CD CD CD CD CD CC CZ C3 CZ C3 CZ CZ CZ CZ CZ CZ CZ CD CD CD CD CD C C C CD C C C C C
=1 M2 o2 to D mtri mo tM o tc o tnt t b oC th o in ct is co ing sltC of h in cO in C ign C in c in c l b i) b t
o d sre p set b y- Css S t h t r e p5 P al l o wd) and c Or lie r Op esn P rn V P we
o t cN mC l l cD ; tcrm co m cN I c o N e Cr c20 C c c r- z l m L L
M "l M "l fl Ml M M C" CN M CN M M- C- CN O C= = = -

rCi ri rN m o o M' r' r' moo--- CN C N CCC CN CN (N CCC CN CN CN C N -
fr-fr-fr-fr-m mm mm mm m5m mm mm mm m 85 85 n 8 8 CN CN C c c
O O M M M M M M 0 0 0 0 0 0 0 0 C C CZ CZ) - - - - - -
Date

Figure 2-27. Volumetric moisture content (VMC) through time, showing results of the scheduled irrigation cycles (SIC), where the
blue dots represent bypassed SIC, the red dots represent allowed SIC, and the red lines represent the range of VMC when
the SIC were allowed; treatment 1-WW, year 2004.





























20

o
U
L-
15

E


E 10 I I I I





5
Zu 11.2 V ,f .h ,











0
9. 0 0 0 807 9.0










C I C3 CD 0 1 0 C0 1 0 C) 0 D C CD C0D 0 D 0 D 0 IDO C0 D C0 CDO C3 CD ID 0 CD 0










blue dots represent bypassed SIC, the red dots represent allowed SIC, and the red lines represent the range of VMC when
the SIC were allowed; treatment t 02-AC, year 2004. t (- D CD (N co 0 r- DD CD
Co C C0 0 C 0 0 -D D CD C0 = CD CD I=> CD 0- C CD C0 C D CD CD 0 D 0 0= CD I=,

D U0) CN ID 0) Q C O ID 0 CD CO UJ C C ID CD CD U-) 'D CN D- L CID 0 CD CD
C D C- D- CD CD CD CD CD CD C CO C C D C CD CI DD CI CDD C D CD CI I> DD CI CD 0 0 0 0D CD 0 0 CD CD ID
0 0 0 0 00 00 00 0 0 0 0 0 0 D D C. C: C:)
Date

Figure 2-28. Volumetric moisture content (VMC) through time, showing results of the scheduled irrigation cycles (SIC), where the
blue dots represent bypassed SIC, the red dots represent allowed SIC, and the red lines represent the range of VMC when
the SIC were allowed; treatment 2-AC, year 2004.



















JU





25





20 -


0017.8


W 15 I I 1 I 15.8 15.9 15.7

E 1 X 3 3 14- 3 .7" 14-1 14.5
. 0113.8 13.6 1 74 13'
.1 2, 3\ .a 13.2 14 713 "7 1 1 4 13.7
12.3






5-
0-t^ ^ ^ ^ ^^ ^ ^ ^ ^ ^ 15^ ^ ^ ^ ^^ ^ ^ ^ ^ ^ ^^ ^ ^ ^ ^ ^


0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
o 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
C D CD CD D C DO CD D CD Ct 0D O In In CD 0 CD CD CI CD C D In CD CD In CD CD
IO In In O In In CD O I 0O SO CO CI C I n CI CD CO ICD LL CO IO CD C


0n 0- 4 r- 0 0- 0 l 0 0 I- L 0 "- I l0n
C- CO CO CO CO CO CO CO In CI In In I4 In In 0 0 0 0 0 0 0
o o 0 0)00 0 0 0)0 0D C CD 0
Date


o o 0 0
0 0 0 0
D t 0 D CD
o1 C- CO I
o o 6


o St CO CN
O C0 01
In: In n


Figure 2-29. Volumetric moisture content (VMC) through time, showing results of the scheduled irrigation cycles (SIC), where the

blue dots represent bypassed SIC, the red dots represent allowed SIC, and the red lines represent the range of VMC when

the SIC were allowed; treatment 2-IM, year 2004.


31.136.8 37.4


30.2 31.2 30.2