Agricultural Water Management 96 (2009) 1828–1836

Contents lists available at ScienceDirect

Agricultural Water Management

journal homepage: www.elsevier.com/locate/agwat

Landscape irrigation by evapotranspiration-based irrigation controllers under dry

conditions in Southwest Florida
S.L. Davis a, M.D. Dukes a,*, G.L. Miller b
a
Agricultural and Biological Engineering Department, University of Florida, P.O. Box 110570, Gainesville, FL 32611-0570, USA
b
Crop Science Department, North Carolina State University, P.O. Box 7620, Raleigh, NC, 27695-7620, USA

A R T I C L E I N F O A B S T R A C T

Article history: Due to high demand for aesthetically pleasing urban landscapes from continually increasing population
Received 16 December 2008 in Florida, new methods must be explored for outdoor water conservation. Three brands of
Received in revised form 6 August 2009 evapotranspiration (ET) controllers were selected based on positive water savings results in arid
Accepted 7 August 2009
climates. ET controllers were evaluated on irrigation application compared to a time clock schedule
Available online 5 September 2009
intended to mimic homeowner irrigation schedules. Three ET controllers were tested: Toro Intelli-sense;
ETwater Smart Controller 100; Weathermatic SL1600. Other time-based treatments were TIME, based on
Keywords:
the historical net irrigation requirement and RTIME that was 60% of TIME. Each treatment was replicated
Drought
four times for a total of twenty St. Augustinegrass plots which were irrigated through individual
Turfgrass
Water conservation irrigation systems. Treatments were compared to each other and to a time-based schedule without rain
sensor (TIME WORS) derived from TIME. The study period, August 2006 through November 2007, was
dry compared to 30-year historical average rainfall. The ET controllers averaged 43% water savings
compared to a time-based treatment without a rain sensor and were about twice as effective and
reducing irrigation compared to a rain sensor alone. There were no differences in turfgrass quality across
all treatments over the 15-month study. The controllers adjusted their irrigation schedules to the
climatic demand effectively, with maximum savings of 60% during the winter 2006–2007 period and
minimum savings of 9% during spring 2007 due to persistent dry conditions. RTIME had similar savings
to the ET controllers compared to TIME WORS indicating that proper adjustment of time clocks could
result in substantial irrigation savings. However, the ET controllers would offer consistent savings once
programmed properly.
ß 2009 Elsevier B.V. All rights reserved.

1. Introduction Evapotranspiration (ET) is defined as the evaporation from the

Water is a limited resource as evidenced by water shortages seen
in areas all over the world despite differences in climate. Water
shortages in Florida have become more prevalent in the last few
decades. Florida has the second largest withdrawal of groundwater
used for public supply in the United States (Solley et al., 1998) and
the largest net gain in population with an inflow of approximately
1100 people per day (United States Census Bureau, 2005). New home
construction has increased to accommodate the large influx of
people and most new homes include in-ground automated irrigation
systems. However, homes with these systems have been shown to
increase outdoor water use by 47% (Mayer et al., 1999). The need for
landscape irrigation will continually grow with increased popula-
tion and home construction if the demand for the current type of
urban landscapes does not change.
* Corresponding author. Tel.: +1 352 392 1864x205; fax: +1 352 392 4092.
E-mail address: mddukes@ufl.edu (M.D. Dukes).
soil surface and the transpiration through plant canopies (Allen
et al., 1998). ET is a part of a balanced energy budget that
exchanges energy for outgoing water at the surface of the plant.
The components of ET are solar radiation, air temperature, relative
humidity, and wind speed (ASCE-EWRI, 2005). Reference ET (ETo)
is the evapotranspiration from a hypothetical reference crop
assumed to be similar to an actively growing, well-watered, dense
green grass of uniform height (ASCE-EWRI, 2005).
Evapotranspiration-based controllers, also known as ET con-
trollers, are irrigation controllers that use an estimation of ET to
schedule irrigation. Each controller works differently depending on
manufacturer, but is typically programmed with landscape-
specific conditions intended to make them more efficient (Riley,
2005). ET controllers receive ETo information in three general
ways, consequently dividing ET controllers into three main types:
(1) standalone controllers, (2) signal-based controllers, and (3)
historical-based controllers. Standalone controllers receive cli-
matic data from on-site sensors and use calculations to determine
ETo whereas signal-based controllers receive ETo calculated off-site

0378-3774/$ – see front matter ß 2009 Elsevier B.V. All rights reserved.
doi:10.1016/j.agwat.2009.08.005
 S.L. Davis et al. / Agricultural Water Management 96 (2009) 1828–1836
from local weather stations. Historical-based controllers rely on
historical ETo information to adjust irrigation based on general
climate patterns, but are not as efficient as other methods because
actual changes in weather are not taken into account.
ET controllers have been used frequently over the last five
years for studies performed by irrigation districts and other
agencies in the western United States. Savings are usually
reported in terms of actual or potential. Potential savings is
defined by Hunt et al. (2001) as the ‘‘difference between actual
outdoor water applied and what should have been applied taking
weather into account.’’ Actual savings is determined by compar-
ing current use to some reference use which is usually based on
water use history.
A study conducted in 2002 in west San Fernando Valley,
California by the Los Angeles Department of Water and Power
showed 17% actual savings by a WeatherTRAK enabled controller
relative to a normalized weather year found through statistical
modeling from the pre-retrofit time period and 78% of potential
savings (Bamezai, 2004). A residential runoff reduction study was
conducted using a modified Sterling irrigation controller to accept
a broadcast signal from the WeatherTRAK ET Everywhere service in
Irvine California; the ET controller group potentially reduced dry
weather runoff 49% and saved 71% compared to the control groups
(Diamond, 2003). Aquacraft Inc. performed an ET controller study
in Colorado to determine savings compared to ETo for the area and
six sites were already irrigating below historical ETo. The first year
resulted in 94% of ETo replacement by irrigation with 20% error
between sites and achievement of 88% of the potential savings while
the second year resulted in 71% of ETo replacement and achievement
of 92% of the potential savings (Aquacraft Inc., 2002, 2003). Devitt
et al. (2008) found that using signal-based ET controllers in Las Vegas
homeowner landscapes reduced water applied by 20% on average
compared to sites without an ET-based controller. Results showed
that 13 out of 16 ET controller sites reduced water applied compared
to 4 of 10 sites without ET controllers. To date, results from ET
controller studies generally have not been published in peer-
reviewed journals. Additionally, these controllers have not been
evaluated in a subtropical climate such as Florida.
The objective of this study was to evaluate the ability of three
brands of ET-based controllers to schedule irrigation by comparing
irrigation application to a time clock schedule intended to mimic
homeowner irrigation schedules, while maintaining acceptable
turfgrass quality.
2. Materials and methods
This study was conducted at the University of Florida Gulf Coast
Research and Education Center (GCREC) in Wimauma, Florida and
at the University of Florida Agricultural and Biological Engineering
Department in Gainesville, Florida. There were a total of twenty
plots at the GCREC that measured 7.62 m  12.2 m, with 3.05 m
buffer zones between adjacent plots. Each plot consisted of 65% St.
Augustinegrass (Stenotaphrum secundatum cv. ‘Floratam’) and 35%
mixed ornamentals to represent a typical residential landscape in
Florida. This research focuses only on the turfgrass. Landscapes
were maintained through mowing, pruning, edging, mulching,
fertilization, and pest and weed control according to current
University of Florida Institute of Food and Agricultural Sciences
(UF-IFAS) recommendations (Sartain, 1991; Black and Ruppert,
1998). The controllers set up in Gainesville were connected only to
a CR10X data logger (Campbell Scientific, Logan, UT) to record run
times to study the variability in water application between ET
controllers of the same brand.
Weather data available on site included rainfall, solar radiation,
wind speed, air temperature and relative humidity at 15 min
intervals from a Florida Automated Weather Network (FAWN)
1829
station. The FAWN station was located within 100 m of the test
site.
Five treatments were established at the GCREC that were
replicated four times for a total of twenty plots in a completely
randomized block design. The irrigation treatments were as
follows: Weathermatic SL1600 controller with SLW15 weather
monitor (Dallas, TX); Toro Intelli-sense (Riverside, CA) utilizing the
WeatherTRAK ET Everywhere service (Hydropoint Datasystems
Inc., Petaluma, CA); ETwater Smart Controller 100 (Corte Madera,
CA); TIME, a time-based treatment determined by UF-IFAS
recommendations (Dukes and Haman, 2002); and RTIME, a
time-based treatment that was 60% of TIME. All treatments
utilized rain sensors to bypass irrigation after 6 mm of rainfall.
Individual valve and flow meter combinations were used to
supply and monitor irrigation to each zone (separate irrigation
zones for turfgrass and ornamentals) of each plot. The flow meters
(15.9 mm V100 w/Pulse Output, AMCO Water Metering Systems,
Ocala, FL) used to monitor irrigation water application were
connected to five Campbell Scientific SDM-SW8A switch closure
input modules that in turn were connected to a CR10X data logger.
The CR10X data logger monitored switch closures every 18.9 l from
the water meters. The meters were also read manually each week.
Irrigation sprinklers specified for the turfgrass portions of the
plots consisted of Rain Bird (Glendora, CA) 1806 15 cm pop up
spray bodies and Rain Bird R13-18 black rotary nozzles. In each
plot, there were four sprinklers with a 1808 arc (R13-18H) and a
center sprinkler with a 3608 arc (R13-18F). The application rate of
the sprinklers was specified by the manufacturer as 15.5 mm/h.
Thirty-year historical rainfall averages were calculated from
monthly rainfall data collected by the National Oceanic and
Atmospheric Administration (NOAA, 2005) from 1975 through
2005. The closest NOAA weather station from the project site with
available rainfall data was located approximately 28 km away, in
Parrish, FL.
There were five periods of data collection: 13 August 2006
through 30 November 2006 as fall 2006; 1 December 2006 through
26 February 2007 as winter 2006–2007; 27 February 2007 through
31 May 2007 as spring 2007; 1 June 2007 through 31 August 2007
as summer 2007; and 1 September 2007 through 30 November
2007 as fall 2007. All five treatments were set up with two days per
week watering restrictions during fall 2006 and winter 2006–2007,
Wednesday and Saturday, and no watering between 10 am and 4
pm. Also, the ET controller treatments were established based on
the site location without accounting for system efficiency (Table 1).
The Weathermatic controller was set to apply 100% of the
calculated water requirement while the Toro and ETwater
controllers were set to the maximum controller efficiency of
95%. The monthly irrigation depth for TIME was 60% of the net
irrigation requirement derived from historical ET and effective
rainfall specific to south Florida (Dukes and Haman, 2002) and
RTIME was a reduced treatment, applying 60% of the irrigation
depth calculated from TIME equaling 36% of the net irrigation
requirement (Table 2). Spring, summer, and fall 2007 differed from
the previous two periods in that the ET controller treatments could
irrigate any day of the week and up to everyday instead of two days
per week and were updated with a system efficiency of 80%
determined from irrigation uniformity testing instead of 100% or
95% as described above (Table 1). TIME was increased to apply
irrigation to replace 100% of the net irrigation requirement instead
of 60% used during the first three periods (Table 2). Once again,
RTIME applied 60% of TIME resulting in the reduced treatment
applying 60% of the net irrigation requirement. The first two
testing periods were meant to simulate a worst-case scenario of
minimal irrigation; whereas, the last three testing periods were
intended to simulate typical ET controller settings and a reasonable
homeowner time clock schedule.
1830 S.L. Davis et al. / Agricultural Water Management 96 (2009) 1828–1836

Table 1
Program settings for each brand of ET controller for a warm season turfgrassa on a sandy soilb and a slope of 08.

Controller Fall 2006 and winter 2006–2007 Spring, summer, and fall 2007

Weathermatic Toro ETwater Weathermatic Toro ETwater

Sprinkler typec 15.2 mm/h 15.5 mm/h 15.5 mm/h 15.2 mm/h 15.5 mm/h 15.5 mm/h
Root depth NAd 152 mm 152 mm NA 152 mm 152 mm
Efficiencye 100% 95% 95% 80% 80% 80%
Zip codef 33598 NA NA 33598 NA NA
Microclimate NA Full sun Full Sun NA Full sun Full sun
Days allowedg Wed, Sat Wed, Sat Wed, Sat Any day Any day Any day
a
The plant type setting is used to choose crop coefficients to calculate plant evapotranspiration.
b
The soil type setting is used to determine the depth of available water for the root zone.
c
Sprinkler type is a term commonly used by ET controllers to specify the application rate of an irrigation zone.
d
NA applies to controller settings that were not applicable to a particular controller.
e
Scheduling efficiency is used to calculate gross irrigation once net irrigation is determined.
f
Zip code is used to find the latitude to determine the monthly solar radiation for ET calculations.
g
Days allowed refers to the days irrigation was allowed to occur per week.

Table 2
Runtimes and application amounts per irrigation eventa for the time-based treatment (TIME) operating on a twice-weekly schedule.

Month Fall 2006 and winter 2006–2007b Spring, summer, and fall 2007

Time (min) Application amount (mm) Time (min) Application amount (mm)

January 23 6 39 10
February 24 6 41 11
March 35 9 58 15
April 37 10 62 16
May 34 9 56 15
June 31 8 51 13
July 48 12 80 21
August 53 14 88 23
September 31 8 52 14
October 32 8 53 14
November 33 8 55 14
December 29 7 48 12

Total potential annual irrigation 840 1424

a
Application rate used to calculate runtimes was 15.5 mm/h.
b
Assumed 60% system efficiency and estimated effective rainfall for south Florida with 60% net irrigation replacement.

Results were quantified by comparing all treatments to a 3. Results

time-based treatment without a rain sensor (TIME WORS). The
time-based treatment without a rain sensor was calculated from
TIME by including water application from irrigation events that
were bypassed due to rain and was not an actual treatment.
Irrigation runtimes for this treatment were adjusted monthly
according to Table 2. It is important to note that the TIME WORS
comparison was scheduled to apply 1424 mm/yr (Table 2);
however, studies have shown that homeowners in Central
Florida apply as much as 1788 mm/yr (Haley et al., 2007). Thus,
this comparison schedule may be conservative relative to water
use in some areas.
Turfgrass quality was measured monthly using the National
Turfgrass Evaluation Program (NTEP) standards (Shearman and
Morris, 1998). The turfgrass was rated on a scale from 1 to 9 where
1 represented dead turfgrass or bare ground, 9 represented an ideal
turfgrass, and 5 was considered minimally acceptable quality for a
residential setting. Each rating was determined monthly by the
same person examining aspects of color, density, uniformity,
texture, and disease or environmental stress.
Water application was summed into weekly totals for statistical
comparisons between treatments using weeks as repeated
measures. The SAS statistical software (SAS Institute Inc., Cary,
NC) was used for all statistical analyses, utilizing the General Linear
Model (GLM) procedure and the mixed procedure with a 95%
confidence level. Means separation was conducted using Duncan’s
multiple range test and least squares means separation was
conducted using the Tukey–Kramer test for pairwise comparisons.
All months received less rain than the historical average except
for April 2007, 69% higher than average, and October 2007, 104%
higher than average (Fig. 1). Overall, both years were drier than the
historical average with a total of 1326 mm of rainfall for the
approximately 16-month study period occurring from August
2006 through November 2007. This rainfall was 33% less than the
historical total from the local NOAA weather station. There were
145 rain events over 472 days; 69% of the study period contained
dry days.
Irrigation water application data were collected from the three
replications of each brand of ET controller at the Gainesville
turfgrass plots (Table 3) to assess variability between different
units of controllers within a brand. It was determined through an
ANOVA that there were no differences between the Weathermatic
replications (P = 0.926), the Toro replications (P = 0.999), or the ET
Water replications (P = 0.999).
3.1. Fall 2006
All plots in fall 2006 suffered from an infestation of chinch bugs
(Blissus insularis ‘Barber’) and a fungal disease known as Curvularia.
Chinch bugs are small pests that inhabit areas of thatch in St.
Augustinegrass and live off plant fluids causing the turfgrass to die
(Buss, 1993). Curvularia is caused by a pathogen that typically
attacks stressed plant material (Wong et al., 2005). Damaged
turfgrass was replaced with new sod during the week following 26
 S.L. Davis et al. / Agricultural Water Management 96 (2009) 1828–1836 1831

Fig. 1. Comparison of rainfall for the 2006–2007 study period and average historical rainfall on a monthly and cumulative basis for southwest Florida.

Table 3 controller was not functional during this period due to circuitry
Average daily irrigation water applicationa (mm) for the three replications of ET
problems. All treatments irrigated less than the TIME WORS
controllers located in Gainesville from 22 May 2007 through 30 November 2007.
treatment (totaling 317 mm).
Replication ET controller brand The ET controller treatments applied less irrigation than the
Weathermatic Toro ETwater TIME WORS treatment except for the month of October as can be
1 1.1 a 1.5 a 1.2 a
seen in the steeper slopes of the lines (Fig. 2). October 2006 had less
2 1.2 a 1.5 a 1.2 a time-based irrigation because the schedule derived from Dukes
3 1.1 a 1.5 a 1.2 a and Haman (2002) contained an error for October in south Florida.
Averageb 1.1 B 1.5 A 1.2 B
Irrigation application for the time-based treatments should have
a
resembled September since October had less rainfall and no more
Numbers with different lower case letters indicate differences across replicates
than a 4% difference in ET, totaling 119 mm in September and
of the same brand (columns) at the 95% confidence level using Duncan’s Multiple
Range Test. 115 mm in October.
b
Statistical analysis was performed on controller brands (row) and where Rainfall that occurred within 24 h of a scheduled irrigation
different upper case letters indicate differences across controllers. event caused many of the scheduled events to be bypassed by all
treatments (Table 4). The Weathermatic controller irrigated 19
times and bypassed more events than any other treatment,
September 2006; no more than 25% of any plot was re-sodded and averaging 1.3 events per week. The increased bypassing of events
most of the damage was located along the edges of the plots. was due to the mandatory 48-h bypass period initiated for each
TIME irrigated the most by applying 230 mm whereas RTIME rainfall event greater than 6 mm in the early part of the period.
irrigated the least, applying 144 mm (Fig. 2). Cumulatively, the Since the controller was only allowed to irrigate two days per week
Weathermatic and Toro controllers applied similar depths over the to follow watering restrictions, there were limited opportunities
period totaling 197 mm and 193 mm, respectively. The ETwater for this controller to allow irrigation to occur. However, the
Weathermatic controller scheduled larger irrigation depths per
event when allowed to irrigate in the latter part of the period
resulting in similar cumulative irrigation as the Toro controller.
The Toro controller irrigated the same average number of events

Table 4
Weekly water application and the average number of irrigation events that
occurred per week for each study period.

Treatment Fall 2006 Winter Spring Summer Fall

2006 2007 2007 2007

Weekly water applicationa (mm)

Weathermatic 12 bc 7c 32 ab NFb 20 c
Toro 12 bc 6c 30 b 26 bc 15 d
ETwater NF NF NF 24 c 18 cd
TIME 15 b 11 b 29 b 30 b 31 b
RTIME 9c 7c 17 d 16 d 18 cd
TIME WORSc 20 a 14 a 35 a 44 a 36 a

Average number of irrigation events per week

Weathermatic 1.3 1.3 6.2 NF 5.2
Toro 1.5 0.9 4.5 4.2 1.7
ETwater NF NF NF 6.5 4.6
TIME 1.5 1.6 1.7 1.4 1.7
RTIME 1.5 1.6 1.7 1.3 1.7
TIME WORS 2.0 2.0 2.0 2.0 2.0
a
Numbers with different letters in columns indicate differences at the 95%
confidence level using Duncan’s multiple range test.
b
Fig. 2. Fall 2006 cumulative water application and daily rainfall (13 August–30 NF indicates nonfunctional treatments that were over the specified time period.
c
November). TIME WORS is a time-based treatment without bypassed rain events in TIME.
1832 S.L. Davis et al. / Agricultural Water Management 96 (2009) 1828–1836

Table 5 All treatments resulted in savings compared to the TIME WORS

Savings compared to TIME WORSa and turfgrass qualityb for each study period.
treatment (Table 5). RTIME showed the most savings at 55% due to
Treatment Fall 2006 Winter Spring Summer Fall the extremely low water application in October. TIME had 28%
2006 2007 2007 2007 savings due in part to the low watering schedule in October.
Savingsc compared to TIME WORS (%) Savings from the ET controllers, Weathermatic and Toro, fell
Weathermatic 38 50 9 NFd 43 between the other treatments by saving 38% and 39%, respectively.
Toro 39 60 15 41 59
ETwater NF NF NF 45 50
3.2. Winter 2006–2007
TIME 28 20 18 31 15
RTIME 55 49 50 63 50
Winter water application was less than that of any other period
Turfgrass qualitye
Weathermatic 4.8 a 5.7 a 6.2 a NF 6.4 a
due to the reduced climatic demand. Irrigation (Fig. 3) ranged from
Toro 4.9 a 5.9 a 6.4 a 6.1 a 7.1 a 84 mm for the Toro controller to 169 mm for TIME. The ETwater
ETwater NF NF NF 6.1 a 7.0 a controller did not function during this period due to the fact that
TIME 4.7 a 6.0 a 6.2 a 6.1 a 6.6 a continued circuitry problems and results were not reported. It is
RTIME 4.8 a 5.7 a 6.1 a 5.8 a 6.5 a
important to note that the problems with the ETwater controller
a
TIME WORS is a time-based treatment without bypassed rain events in TIME. were isolated to this individual unit. The same model controller on
b
Turfgrass quality ratings used a 1–9 scale where 1 was of lowest quality, 9 was other sites functioned reliably. Rainfall totaled 167 mm over the
of highest quality, and 5 was minimally acceptable.
c
Savings were calculated using cumulative period totals.
88-day period. Irrigation events were less frequent for the ET
d
NF indicates nonfunctional treatments that were over the specified time period. controllers; the Toro controller irrigated 12 times and the
e
Numbers with different letters in columns indicate differences at the 95% Weathermatic controller irrigated 16 times out of a possible 25
confidence level using Duncan’s multiple range test. irrigation days compared to 20 events for the time-based
treatments, TIME and RTIME (Table 4). Water savings occurred
per week as TIME and RTIME, each averaging 1.5 events. Weekly for all treatments compared to the TIME WORS treatment, ranging
water application by the Toro controller compared to the TIME and from 20% to 60% (Table 5).
RTIME treatments was not statistically different. There were differences among treatments (P < 0.0001), but not
There were differences among treatments (P < 0.0001), but not among replications (P = 0.948) for weekly water application
among replications (P = 0.807) for fall 2006 weekly water (Table 4). The Weathermatic controller, Toro controller, and
application (Table 4). The TIME (15 mm/wk) schedule applied RTIME had similar weekly application rates of 6–7 mm/wk.
6 mm/wk more irrigation (P = 0.0002) than RTIME (9 mm/wk). The However, these three treatments irrigated less than TIME
ET controllers, Weathermatic and Toro, were not different from (P < 0.0001), at 11 mm/wk. Finally, TIME WORS (P  0.0004)
each other, both averaging 12 mm/wk. There were also no irrigated more than any of the other irrigation treatments; water
differences between TIME and both ET controllers. The ET application averaged 14 mm/wk for this period. The time-based
controllers were not different (P = 0.063 and P = 0.106, respec- treatments, TIME and RTIME, irrigated the most events per week,
tively) compared to RTIME. All treatments applied less irrigation on average as 1.6 events. The ET controllers irrigated similar depths
than the TIME WORS treatment with an average of 20 mm/wk of irrigation on a weekly basis, but applied water less often. The
(P  0.0002). Weathermatic controller averaged 1.3 events per week and the
Average turfgrass quality ratings were below the minimally Toro controller averaged 0.9 events per week. Turfgrass quality
acceptable value of 5.0 for all treatments due to pest and disease ratings ranged from 5.7 to 6.0 and were not different across
problems as described above (Table 5). Water application amount treatments (Table 5). There was no correlation between water
was not correlated with turfgrass quality (P = 0.4503). application and turfgrass quality (P = 0.082).

Fig. 3. Winter 2006–2007 cumulative and daily water applied and daily rainfall (1 Fig. 4. Spring 2007 cumulative and daily water applied and daily rainfall (27
December–26 February). February–31 May).
 S.L. Davis et al. / Agricultural Water Management 96 (2009) 1828–1836
Both ET controllers applied less water than RTIME unlike any
other time of year. The ET controller treatments showed the
potential to save over 50% of water applied during the cooler
months of the year. These results suggest that the time schedule
can be greatly reduced in this part of the year.
3.3. Spring 2007
In spring 2007, irrigation application ranged from 244 mm by
RTIME to 445 mm by the Weathermatic controller (Fig. 4). Since
the ET controllers were given the option of irrigating any day of the
week, these devices irrigated a smaller amount per event, but more
frequently than the time-based treatments (Table 4). The Weath-
ermatic controller irrigated almost every day, averaging 6.2 events
per week, and the Toro controller irrigated almost three times as
often, averaging 4.5 events per week. Weekly water applications
were not necessarily less by the ET controllers relative to TIME.
Weekly water applications were similar by the Weathermatic
controller, averaging 32 mm/wk, the Toro controller, averaging
30 mm/wk, and TIME, averaging 29 mm/wk. Each of these
treatments applied more weekly irrigation than RTIME that
averaged 17 mm/wk (P < 0.0001). The ET controllers had similar
weekly application rates and were higher than RTIME (P < 0.0001).
Irrigation events occurred every day for the Weathermatic
controller except when a daily event was bypassed by the rain
sensor integral to that controller (Table 4).
All treatments maintained similar turfgrass quality ratings well
above the minimally acceptable level; averages ranging from 6.1 to
6.4, were not statistically different (Table 5). Turfgrass quality was
not correlated with irrigation application (P = 0.745). Despite the
reduced watering by RTIME, the reduced time-based schedule still
had a turfgrass quality rating similar to other treatments (6.1).
Rainfall totaled 109 mm over this period. The time-based
schedules, TIME and RTIME, applied irrigation during every
scheduled event for the months of March and May (Fig. 4). Each
rain event occurring in March was not substantial enough to
trigger the rain sensor to bypass irrigation and there was no rainfall
in May. Irrigation savings by the ET controller treatments were
based purely on their ability to match irrigation application with
environmental demand and not affected by the variability of the
rain sensor during these two months.
Water savings by all treatments compared to the TIME WORS
treatment ranged from 9% by the Weathermatic controller to 50%
by RTIME (Table 5). Average weekly water application for the TIME
WORS treatment was higher than the Toro controller (P = 0.0009),
but was not different from the Weathermatic controller at 35 mm/
wk versus 32 mm/wk (P = 0.165; Table 4). The time-based
treatments, TIME and RTIME had lower weekly application rates
(P < 0.0001) than TIME WORS.
3.4. Summer 2007
Rainfall was more frequent during the summer of 2007, totaling
446 mm, but was still below the historical average. Irrigation
ranged from 228 mm by RTIME to 425 mm by TIME (Fig. 5). The
ETwater controller irrigated every day because it was programmed
with a 25% allowable depletion instead of 50% originally
programmed causing the controller to irrigate when 25% of the
water was calculated to have left the root zone. This controller also
would not recognize a rain sensor despite repeated attempts with
customer service to repair. A power outage occurring on 8 June
2007 caused the equipment associated with the Weathermatic
controller to discontinue taking measurements to calculate ETo.
Since the Weathermatic controller did not operate based on an ET
schedule during this time, data for this controller were removed for
this period.
1833
Fig. 5. Summer 2007 cumulative and daily water applied and daily rainfall (1 June–
31 August).
The ET controllers, Toro and ETwater, irrigated less depth per
event, but applied irrigation more frequently as can be seen by the
reduced slopes (compared to TIME WORS) for these treatments
(Fig. 5); however, average weekly irrigation applied by the ET
controllers was not statistically different (P = 0.802), 26 mm/wk by
Toro and 24 mm/wk by ETwater, but was greater than RTIME
(16 mm/wk; P < 0.0001; Table 4). The TIME schedule resulted in
nearly twice the weekly irrigation rate compared to RTIME
(P < 0.0001), 25% higher irrigation rate than the ETwater controller
(P = 0.001), and a 15% greater rate than the Toro controller
(P = 0.046). Turfgrass quality ratings were not different across
treatments (P = 0.933) and remained above the minimally
acceptable levels. Also, turfgrass quality was not correlated with
water application (P = 0.591).
Water savings by all treatments compared to the TIME WORS
treatment (Table 5) ranged from 31% for TIME to 63% for RTIME.
Savings for the ET controller treatments fell between the other
treatments by saving 41% and 45%, respectively. The average
weekly water application for TIME WORS was 44 mm/wk and was
higher than all treatments (P < 0.0001).
3.5. Fall 2007
Irrigation ranged from 209 mm for the Toro controller to
427 mm for TIME in fall 2007 (Fig. 6). Rainfall during this period
totaled 264 mm with only a few small events later in the season.
Average weekly water application (Table 4) for the ETwater
controller was 18 mm/wk and was not different compared to the
other ET controller treatments, Weathermatic (20 mm/wk;
P = 0.552) and Toro (15 mm/wk; P = 0.149), as well as RTIME
(18 mm/wk; P = 1.000). The Toro controller had a lower weekly
irrigation rate compared to the Weathermatic controller
(P = 0.001). The Toro applied less events per week than the other
ET controllers, averaging 1.7 events compared to 5.2 events for the
Weathermatic and 4.6 for the ETwater. Irrigation scheduling by the
Toro controller was similar to the RTIME schedule; both
treatments applied similar average depths for 1.7 events per
week. All treatments had lower weekly application rates than TIME
(31 mm/wk; P < 0.0001). Turfgrass quality was similar across all
treatments and higher than the minimally acceptable value of 5,
1834 S.L. Davis et al. / Agricultural Water Management 96 (2009) 1828–1836
Fig. 6. Fall 2007 cumulative and daily water applied and daily rainfall (1
September–30 November).
ranging from 6.4 to 7.1 (Table 5); quality was not different between
treatments (P = 0.170). Similar to all other experimental periods,
turfgrass quality was not correlated with irrigation depth
(P = 0.178).
The weather monitor for the Weathermatic controller measures
temperature and senses rainfall using an expanding disk rain
sensor. This monitor was found inverted and hanging from its
mounted location on 9 October 2007. The length of time prior to
this date that the monitor was damaged was unknown. Irrigation
occurred despite rain events due to the misalignment of the
weather monitor.
The Weathermatic controller saved 43% compared to TIME
WORS while the Toro and ETwater controllers saved 59% and 50%,
respectively (Table 5). Both TIME and RTIME also showed water
savings of 15 and 50%, respectively. Savings from the Weath-
ermatic controller would have been slightly higher had the
weather monitor been aligned appropriately for at least seven rain
events that exceeded the 6 mm threshold of the sensor.
4. Discussion
All treatments applied less water compared to cumulative
irrigation for TIME WORS (Table 5). Average potential water
savings using a rain sensor at a 6 mm threshold was 21% over the
entire study period. Rainfall was less than the historical average
resulting in dry conditions (Fig. 1). These savings occurred despite
dry conditions due to a schedule of only two irrigation events per
week. There was a high probability of rainfall events greater than
6 mm occurring within each period to cause at least one of the
irrigation events to bypass during random weeks, resulting in
water savings. RTIME averaged 53% savings for the study period
due to the reduced runtimes and bypassing by the rain sensor.
When operating properly, all ET controller treatments exhibited
considerable savings according to statistical differences compared
to TIME WORS for every time period except spring 2007 (Table 5).
Differences were not significant in spring 2007 because the time-
based treatments were developed considering historical effective
rainfall. However, the spring 2007 period experienced only half of
historical normal rainfall, increasing irrigation demand. Even
though more irrigation occurred compared to the time-based
treatments, the ET controllers were reacting to the plant water
needs based on real-time conditions and not historical needs.
Despite this fact, it is clear that the ET controllers could have been
programmed to apply less water (e.g. some fraction of ET) and turf
quality would not have suffered as evidenced by application of half
as much water by RTIME with damaging turf quality.
The ET controllers schedule irrigation so that the plant material
being irrigated is always well-watered. However, it is acceptable
for turfgrass to have some stress from deficit irrigation strategies
and still be able to recover quickly without loss of quality. Deficit
irrigation occurred during the spring 2007 period for the time-
based treatments due to the lack of rainfall compared to the
historical average (Fig. 4). The ET controllers responded to the
increased climatic demand, keeping the turfgrass healthy under
well-watered conditions, while still producing some water savings
over the same period. Programming these controllers to allow
some deficit conditions could potentially result in both acceptable
turfgrass quality and higher water savings assuming good
irrigation distribution uniformity.
Overall, water savings by these controllers were relatively
similar despite the different approaches to irrigation scheduling by
the manufacturers. The Weathermatic controller averaged 35%
savings for the entire study period compared to the TIME WORS
treatment. Average savings during 2 d/wk irrigation, fall 2006 and
winter 2006–2007, were 44%. This could be attributed to less
cumulative irrigation application over the winter months due to
more accurate estimation of water need for the period (Fig. 3).
Savings for 2007 periods averaged 26% and was lower than that of
other time periods due to dry spring conditions. Savings from the
Weathermatic controller would have been slightly higher had the
weather monitor been aligned appropriately for at least seven rain
events that exceed the 6 mm threshold of the sensor.
The Toro controller showed considerable savings during both
years averaging 50% for the fall 2006 and winter 2006–2007
periods and 38% for the 2007 periods, averaging 43% overall.
Average savings were similar to the Weathermatic controller,
rarely statistically different, and savings were less for the 2007
periods due to increased water demand for spring. The ETwater
controller resulted in 47% savings for the last two periods, also
similar to other ET controller savings. Water savings using ET
controllers was not at the expense of turfgrass quality (Table 4).
The ET controller treatments consistently applied less than
TIME for the first two periods with only the exception of the
incorrect time-based irrigation schedule in October. The ET
controller treatments always applied less than TIME over the last
three periods except for the following three months: 44% more for
April (Weathermatic), 7% more for May (Weathermatic), and 5%
more for July (Toro).
The Weathermatic and Toro controllers resulted in more water
savings for fall 2007 compared to fall 2006 (Table 5). It was likely
that water savings resulted from these treatments because
watering restrictions were removed and the controllers were
more efficient at scheduling irrigation and accounting for rainfall
given more opportunities to irrigate each week. More savings were
also possible for fall 2007 due to increasing the net irrigation
requirement replacement for the time-based schedules from 60%
to 100% after winter 2006–2007.
Haley et al. (2007) found that homes in Central Florida used an
average of 149 mm/month when their time clocks were not
adjusted over the year. Compared to this benchmark, fall 2006 and
winter 2006–2007 savings for the ET controller treatments were
60% and 71% while the time-based treatments, TIME, RTIME, and
TIME WORS, saved 47%, 63%, and 29%, respectively. During the last
three periods, all treatments irrigated less than 149 mm/month
except for TIME WORS (29% increase). Savings ranged from 6% by
TIME to 46% by RTIME. ET controller savings were 20%, 26%, and
 S.L. Davis et al. / Agricultural Water Management 96 (2009) 1828–1836
30% for the Weathermatic, Toro, and ETwater controllers,
respectively.
The 2006 periods, fall and winter, had time-based treatments
that were very conservative as well as ET controllers that were
scheduled with almost 100% efficiency. Despite scheduling
conservatively, the ET controllers still exhibited water savings,
with maximum savings during winter 2006–2007, and above
acceptable turfgrass quality (Table 5). Fall 2006 had more rainfall
distributed over the treatment period than fall 2007 causing more
rainfall to be effectively stored in the root zone and less irrigation
to be required. This difference produced similar savings between
the periods despite the alternative settings.
There are financial benefits to using ET controllers compared to
TIME WORS. Assuming an average irrigated landscape size of
500 m2 and a conservative flat rate of $ 4.00 per 3786 l of water
used, the cost for TIME WORS was $ 1144 over the entire study
period. Comparatively, the average cost for the ET controller
treatments was $ 697 over the study period. The potential water
savings by the ET controllers could save $ 28 per month, on
average, for a typical homeowner. If the average cost of
implementing one of the ET controller technologies is $ 500, then
the payback period would be about 18 months.
All treatments maintained acceptable turfgrass quality despite
conservative settings, but it is unlikely that all treatments could
have maintained above acceptable quality during the rest of the
study. The time-based treatments had lower weekly water
application and higher water savings during spring 2007 even
after adjusting these schedules to apply 40% more water than the
previous periods. It is possible that allowing the conservative
schedule to continue through spring 2007 would have decreased
turfgrass quality from prolonged water stress. Though ET
controllers could be programmed incorrectly by homeowners,
such as using 100% efficiency, the controllers would continually
respond to changes in climatic demand whereas the time-based
treatments could not adjust.
5. Conclusions
The TIME treatment developed from 100% replacement of the
net irrigation requirement, consistently applied more cumulative
irrigation compared to the ET controller treatments. The RTIME
schedule applied the least amount of water in all periods except
winter 2006–2007 and fall 2007. Turfgrass quality remained above
the minimally acceptable level for both of these time-based
treatments and there were no statistical differences between the
ratings among treatments. As a result, 60% replacement of net
irrigation requirements is appropriate for effective water applica-
tion with good irrigation distribution uniformity and weather
similar to the historical average.
The ET controllers averaged 43% water savings compared to a
time-based treatment without a rain sensor and were about twice
as effective at reducing irrigation compared to a rain sensor alone.
Turfgrass quality remained above minimally acceptable despite
water savings and the dry conditions compared to the historical
average. The controllers adjusted their irrigation schedules to the
climatic demand effectively, with maximum savings of 60% during
the winter 2006–2007 period and minimum savings of 9% during
spring 2007 when demand was highest.
The RTIME treatment resulted in similar savings as ET
controllers. Thus, as has been shown in previous research in
Florida, changing time clock settings throughout the year can
result in substantial irrigation savings. Fall 2006 and winter 2006–
2007 were scheduled for only 36% replacement of net irrigation
requirement for the reduced time-based treatment, but still
irrigated more in the winter compared to the ET controller
treatments. This result indicates that the ET controllers were
1835
effective at reducing irrigation application during low climatic
demand. Time-based treatments were developed from the net
irrigation requirement for the area resulting in less water applied
than if scheduled without using historical ET and effective rainfall.
However, time-based schedules do not fluctuate with changing
weather conditions and many homeowners do not adjust time
clock irrigation schedules on a regular basis. Thus, the ET
controllers tested here have shown that they can adjust irrigation
in response to climatic demand in Florida. Actual water
conservation potential of these controllers in landscapes will
depend on irrigator habits and preferences. These controllers need
to be evaluated under ‘‘real world’’ conditions to verify water
savings.
Acknowledgments
The authors would like to acknowledge the following funding
agencies for their support of this research: Hillsborough County
Water Resource Services, Florida Department of Agricultural and
Consumer Services, Florida Nursery Growers and Landscape
Association, and Florida Agricultural Experiment Station. The
authors would also like to thank the following individuals for their
efforts in making this project possible: David Crockett, Larry Miller,
Daniel Preston, Sudeep Vyapari, Sydney Park Brown, Amy Shober,
Melissa Baum Haley, Mary Shedd McCready, and Gitta Shurberg.
References
Allen, R.G., Pereira, L.S., Raes, D., Smith, M., 1998. Crop evapotranspiration: Guide-
lines for computing crop requirements. Irrigation and Drainage Paper No. 56,
FAO, Rome, Italy, 300 pp.
ASCE-EWRI, 2005. The ASCE standardized reference evapotranspiration equation.
Technical Committee Report to the Environmental and Water Resources Insti-
tute of the American Society of Civil Engineers from the Task Committee on
Standardization of Reference Evapotranspiration. ASCE-EWRI, 1801 Alexander
Bell Drive, Reston, VA 20191-4400, 173 pp.
Aquacraft Inc., 2002. Performance evaluation of WeatherTRAK irrigation controllers
in Colorado. Available at: http://www.aquacraft.com/Download_Reports/
WeatherTRAK_2001_Study_Report.pdf (Accessed 1 July 2009) 22 pp.
Aquacraft Inc., 2003. Report on Performance of ET Based Irrigation Controller:
Analysis of operation of WeatherTRAK controller in field conditions during
2002. Available at: http://www.aquacraft.com/Download_Reports/
WthrTrk_2002_Study_Report.pdf (Accessed 1 July 2009) 31 pp.
Bamezai, A., 2004. LADWP weather-based irrigation controller pilot study. Available
at: http://www.cuwcc.org/uploads/product/LADWP-Irrigation-Controller-
Pilot-Study.pdf (Accessed 31 October 2005) 40 pp.
Black, R.J., Ruppert, K., 1998. Your Florida Landscape: A Complete Guide to Planting
and Maintenance. UF-IFAS publication. University of Florida Press, Gainesville,
FL, p. 241.
Buss, E.A., 1993. Southern chinch bug management on St. Augustinegrass. ENY-325,
Institute of Food and Agricultural Sciences. University of Florida, Gainesville, FL.
Available at: http://edis.ifas.ufl.edu/LH036 (Accessed 1 July 2009) 5 pp.
Devitt, D.A., Carstensen, K., Morris, R.L., 2008. Residential water savings associated
with satellite-based ET irrigation controllers. Journal of Irrigation and Drainage
Engineering. 134, 74–82.
Diamond, R.A., 2003. Project review of Irvine ET controller residential runoff
reduction study. Irvine Ranch Water District. Available at: http://www.irriga-
tion.org/swat/images/irvine_runoff_reduction.pdf (Accessed 1 November
2005) 4 pp.
Dukes, M.D., Haman, D.Z., 2002. Operation of residential irrigation controllers.
CIR1421, Institute of Food and Agricultural Sciences, University of Florida,
Gainesville, FL. Available at: http://edis.ifas.ufl.edu/AE220 (Accessed 1 July
2009) 10 pp.
Haley, M.B., Dukes, M.D., Miller, G.L., 2007. Residential irrigation water use in
Central Florida. Journal of Irrigation and Drainage Engineering 133 (5), 427–
434.
Hunt, T., Lessick, D., Berg, J., Wiedmann, J., 2001. Residential weather-based
irrigation scheduling: Evidence from the Irvine ‘‘ET Controller’’ study. Available
at: http://www.irrigation.org/swat/images/irvine.pdf (Accessed 30 October
2005) 53 pp.
Mayer, P.W., DeOreo, W.B., Opitz, E.M., Kiefer, J.C., Davis, W.Y., Dziegielewski, B.,
Nelson, J.O., 1999. Residential End Uses of Water. American Water Works
Association Research Foundation, Denver, CO, 310 pp.
NOAA (National Oceanic and Atmospheric Administration), 2005. Monthly pre-
cipitation 1975–2005 for Parrish, FL. Available at: http://cdo.ncdc.noaa.gov/pls/
plclimprod/poemain.cdobystn?dataset=DS3220&StnList=086880NNNNN
(Accessed 1 November 2006).
1836 S.L. Davis et al. / Agricultural Water Management 96 (2009) 1828–1836

Riley, M., 2005. The cutting edge of residential smart irrigation technology. Cali-
fornia Landscaping. July/August, pp. 19–26.
Sartain, J.B., 1991. General recommendations for fertilization of turfgrasses on
Florida soils. SL21, Institute of Food and Agricultural Sciences, University of
Florida, Gainesville, FL. Available at: http://edis.ifas.ufl.edu/LH014 (Accessed 1
July 2009) 7 pp.
Shearman. R.C., Morris, K.N., 1998. NTEP Turfgrass Evaluation Workbook. NTEP
Turfgrass Evaluation Workshop, 17 October 1998, Beltsville, MD, 5 pp.
Solley, W.B., Pierce, R.R., Perlman, H.A., 1998. Estimated use of water in the United
States in 1995. United States Geological Survey Circular 1200, p. 78.
United States Census Bureau, 2005. Population estimates. Washington, DC. Available
at: http://www.census.gov/popest/estimates.php (Accessed 19 January 2006).
Wong, F., Harivandi, M.A., Hartin, J., 2005. UC IPM Pest Management Guidelines:
Turfgrass. University of California Agriculture and Natural Resources Publica-
tion 3365-T, Davis, CA Available at: http://www.ipm.ucdavis.edu/PMG/
r785100311.html (Accessed 15 March 2008) p. 2.