Professional Documents
Culture Documents
net/publication/340081130
CITATIONS READS
110 7,793
6 authors, including:
Zhe Gu Zhiming Qi
Hohai University McGill University
22 PUBLICATIONS 391 CITATIONS 194 PUBLICATIONS 2,864 CITATIONS
All content following this page was uploaded by Zhiming Qi on 25 April 2020.
Abstract: In an effort to improve plant growth and to achieve high yield and/or quality, irrigation scheduling (IS) seeks to provide plants with
Downloaded from ascelibrary.org by Nanjing Forestry University on 03/21/20. Copyright ASCE. For personal use only; all rights reserved.
appropriate quantities of water at appropriate times. To better understand irrigation scheduling’s main processes and principles, its four most
common methods of operation—(1) evapotranspiration and water balance (ET-WB), (2) soil moisture (Θ) status, (3) plant water status, and
(4) models—along with their pros and cons are introduced and compared. Irrigation applications, including software, programs, and asso-
ciated controllers are introduced. Given that some of these methods focus on Θ or plant responses to soil moisture, the determination of target
soil moisture levels, along with estimates (either calculated or measured) of current soil moisture status are key to both scheduling irrigations,
and the precise replenishment of soil moisture to target levels. Accordingly, factors in the soil-crop-atmosphere system affecting soil moisture
must be considered in the scheduling process. As all four types of IS methods focus on soil water content, which serves as a bridge between
irrigation management and crop water requirements for growth, future scheduling methods should focus on the management of soil moisture
based on an advanced understanding of its effects on crop growth either by the integration of existing IS methods or the development of new
models, using intelligent algorithms. Using these approaches, more practical, accurate, and easily adaptable IS applications should be de-
veloped for real-time farming operations. Weather station networks and online data access should be enhanced to better serve these IS
applications. DOI: 10.1061/(ASCE)IR.1943-4774.0001464. © 2020 American Society of Civil Engineers.
Author keywords: Evapotranspiration (ET) and water balance; Soil water status; Plant-based measurements; Model based.
Introduction (40%–80%) (Bazzani 2005; FAO 2001), and water wastage (Walker
1989), the global water scarcity crisis is worsening. Faced with lim-
The roughly 1.55 × 109 ha of arable lands currently (2015) culti- ited arable and irrigated lands, and limited water resources, future
vated in the world represent about 11% of the Earth’s total land agriculture must nonetheless try to feed an increasing population.
surface (World Bank Group 2019). This percentage is predicted This places greater stress on irrigation management to improve water
to reach 13% by 2050 (Alexandratos and Bruinsma 2012). Of these use efficiency (WUE) and productivity (FAO 2001).
arable lands, roughly 17% are subject to any form of irrigation man- The water scarcity also poses new challenges on urban water
agement. However, this relatively small fraction of land equipped utilities, where the need for landscape irrigation will continue to
for irrigation may contribute as much as 30%–40% of the world’s grow with increasing populations (Davis et al. 2009; Mayer et al.
gross agricultural output (Walker 1989). Driven by rising urban and 1999). It was reported that 72% of study participants applied less
industrial competition for limited water resources (Fernández and than 70% of the theoretical irrigation requirement (TIR) to land-
Cuevas 2010; Walker 1989), high agricultural water consumption scape, with about 27% applying only about 10% of TIR; in contrast,
1
about 13% applied excess irrigation (exceeding 130% of TIR), and
Postdoctoral Research Fellow, College of Agricultural Engineering, some even applied more than 300% of TIR (DeOreo et al. 2016). For
Hohai Univ., No. 1 Xikang Rd., Nanjing 210098, China. ORCID:
such landscape irrigation, a well-devised schedule benefited both
https://orcid.org/0000-0002-2483-0880. Email: zhegu2018@hhu.edu.cn
2
Associate Professor, Dept. of Bioresource Engineering, McGill Univ., plant quality and water savings, as an estimated 20%–50% water
Sainte-Anne-de-Bellevue, QC, Canada H9X 3V9 (corresponding author). savings was achieved when using mild to aggressive landscape con-
ORCID: https://orcid.org/0000-0002-8233-165X. Email: zhiming.qi@ servation practices (DeOreo et al. 2016).
mcgill.ca The importance of water in plants arises from its role in support-
3
Graduate Student, Dept. of Bioresource Engineering, McGill Univ., ing photosynthesis, regulating temperature via evaporative cooling,
Sainte-Anne-de-Bellevue, QC, Canada H9X 3V9. Email: rasika maintaining structure through cell turgor pressure (i.e., in maintain-
.burghate@mail.mcgill.ca ing leaf orientation), and transporting nutrients into and throughout
4
Professor, Research Center of Fluid Machinery Engineering and Tech-
the plant, thereby supporting its growth. While irrigation guaran-
nology, Jiangsu Univ., 301 Xuefu Rd., Zhenjiang 212013, China. Email:
shouqiy@ujs.edu.cn tees crop growth in regions where rainfall is insufficient to support
5
Professor, College of Agricultural Engineering and Cooperative crop growth and yield, it must be scheduled properly, lest crops
Innovation Center for Water Safety and Hydro Science, Hohai Univ., suffer water stress at a critical growth stage, or excess water applied
No. 1 Xikang Rd., Nanjing 210098, China. Email: xyjiao@hhu.edu.cn that leads to ponding, waterlogging, runoff and/or deep seepage,
6
Professor, College of Agricultural Engineering, State Key Laboratory thereby leaching applied nutrients and polluting water bodies.
of Hydrology-Water Resources and Hydraulic Engineering, Hohai Univ., Though in regions where soil salinity is of a concern, excessive
No. 1 Xikang Rd., Nanjing 210098, China. Email: xjz481@hhu.edu.cn
water is required to be applied to leach the salts down in the soil
Note. This manuscript was submitted on November 20, 2018; approved
on November 21, 2019; published online on March 20, 2020. Discussion profile. Over the last few decades, a number of methods for sched-
period open until August 20, 2020; separate discussions must be submitted uling and quantifying the required depth of individual irrigation
for individual papers. This paper is part of the Journal of Irrigation and applications have been proposed. The present paper reviews a num-
Drainage Engineering, © ASCE, ISSN 0733-9437. ber of irrigation scheduling (IS) methods effectively employed in
et al. 2017). average, a 43% water saving compared to the time-based irrigation
SmartIrrigation apps are interactive ET-based IS tools that op- approach without a rain sensor (Davis et al. 2009). Only half of the
erate on smartphone platforms. Several apps have been developed theoretical irrigation requirement was applied by the three brands
for different plants, such as citrus, turf, strawberry (Fragaria × of ET controllers (Davis and Dukes 2010).
ananassa Duchesne), cotton (Gossypium hirsutum L.), avocado
(Persea americana Mill.), and vegetables, and is still under devel-
opment for other crops, like blueberry (Vaccinium sect. Cyanococ- Soil-Moisture-Based IS
cus) and soybean. These apps are available on the website (Smart
Irrigation Apps Group 2019). SmartIrrigation apps were first re- Irrigation Timing and Amount
leased in 2013 by Migliaccio et al. (2013) for citrus, turf, and straw-
berry, and further crop apps were later developed. SmartIrrigation The Θ-based IS methods compare monitored Θ (inferred from sen-
apps were developed based on an ET and water balance model, sor measurements) to Θ-based thresholds to trigger irrigation (tim-
using meteorological data, soil parameters, crop phenology, crop ing). The monitored Θ is commonly measured by time domain
coefficients, and irrigation applications to estimate root zone water transmission sensors or reflectometry probes, neutron probes,
deficit (see “ET-WB-Based Approach” section). The apps notify capacitance sensors, granular matrix sensors, etc. (Hedley and Yule
users when irrigation is needed and the amount of water to irrigate. 2009; Migliaccio et al. 2010; Thompson et al. 2007b). Related to
Vellidis et al. (2014, 2015, 2016a) described the development of Θ, soil water tension or soil matric potential (ψm ), measured by
the SmartIrrigation Cotton app, along with the calibration, validation, tensiometers, is another parameter used to quantify soil water avail-
and field evaluation of the model integrated in the app. The Cotton able for plant use.
app was reported to be able to predict soil water content reason- In an effort to benefit crop growth, yield, and quality, Θ-based
ably well and to save water when compared to other IS methods IS methods focus mainly on determining when irrigation should be
(e.g., Checkbook and CWSI methods, discussed in section “Crop applied to maintain root zone Θ within an appropriate range (Viani
Water Stress Index Method”). However, in a wet year, it failed to 2016). Various methods have been proposed to determine irrigation
achieve higher yields and WUE than the simply rainfed treatment timing based on a lower limit or threshold of Θ, θth (Haley and
(Vellidis et al. 2016b). Migliaccio et al. (2016) in introducing the Dukes 2012; Zotarelli et al. 2010), e.g., θth ≤ 0.12 m3 m−3 ; or
development of SmartIrrigation apps for different crops, pointed when soil tension, ψm , reaches a set threshold (Hoppula and Salo
out their limitations: these apps are currently only useful in 2007; Migliaccio et al. 2010; Thompson et al. 2007b), e.g., jψm j >
Florida and Georgia and only for those crops grown in fields 25 kPa. The threshold value varies according to soil properties and
in proximity to meteorological stations. Considering spatial varia- crops, and is determined either through field experiments, where
tion in rainfall, and the finite set of weather stations, it is spatially crop responses to different levels of water stress under various ir-
limited in making accurate schedules. rigation treatments are investigated (Hoppula and Salo 2007;
Besides the SmartIrrigation apps, several IS applications have Migliaccio et al. 2010; Wang et al. 2017), or drawn from existing
been successfully applied. Using a dual crop coefficient approach studies (Haley and Dukes 2012). Therefore, these thresholds are
to estimate ET c , Rosa et al. (2012a, b) developed and tested the optimized for specific locations and crop species due to their re-
SIMDualKc software to simplify the computation and quickly sponses to crop growth, yield, and quality, etc. Thompson et al.
provide ET c information for irrigation scheduling. Employing a (2007b) proposed an alternative to determine threshold values of
short-term water balance that can be adjusted according to mea- ψm , using an indicator of plant water stress, the leaf water potential
sured Θ trends, the Washington Irrigation Scheduling Expert soft- (ψleaf ). A linear regression was established between ψm and ψleaf for
ware (WISE; Leib et al. 2001), allows easy access to ET 0 values conditions ranging from well-watered to nonwatered, and the point
from Washington State’s sixty Public Agriculture Weather System where the relationship broke from a linear relationship was chosen
stations. Thysen and Detlefsen (2006) described an internet imple- as the ψm threshold. Thompson et al. (2007a) also proposed an in
mentation of a previous stand-alone PC-based irrigation decision- situ approach for determining the lower limit in ψm -based irrigation
making program based essentially on an ET and water balance management, by monitoring reductions in Θ as a reflection of crop
method. Chauhan et al. (2013) introduced a web-based DSS, water uptake.
AQUAMAN, to assist Australian peanut (Arachis hypogaea L.) Besides the use of Θ or ψm thresholds, the available soil water
growers to schedule irrigation based on the ET-WB method intro- content threshold (AWCth ) has also been employed in Θ-based IS
duced in FAO-56. Bartlett et al. (2015) developed a smartphone approach, to which the actual available soil water content (AWC, %)
app to extend the usage of the Water Irrigation Scheduling for Ef- is compared:
ficient Application system, an ET-WB-based IS tool developed by
θa − θpwp
Colorado State University. Perea et al. (2017) developed a multi- AWC ¼ 100 · ð2Þ
platform application (IrriFresa) for precision irrigation scheduling θfc − θpwp
sensor calibration and sensor accuracy (Thompson et al. 2007b). Ac- properties including AWC and θfc (Hedley and Yule 2009), which
cordingly, the adoption of a dynamic rooting depth along with dy- were used to determine the triggering threshold (θth or AWCth ).
namic θfc and θpwp values across the rooting depth (Gu et al. 2017a; In contrast to the low spatial resolution of Θ measurements using
Liu et al. 2017) can probably improve irrigation performance. traditional sensors, electromagnetic mapping techniques have the
Under Θ-based irrigation scheduling, the most common way to potential to achieve a high spatial resolution in both Θ and soil
determine the necessary depth of irrigation is to apply a threshold- properties, and therefore serve to map θth to a higher resolution and
based quantity (Migliaccio et al. 2010; Zotarelli et al. 2010), or schedule a more accurate variable rate irrigation (Hedley and Yule
a fixed quantity irrigation at each event using an existing timer/ 2009). The major weakness of an Θ-based IS approach resides in
controller (Haley and Dukes 2012). The threshold-based method the inaccuracy of Θ measurements using sensors (Evett et al. 2011).
interrupts irrigation when the threshold condition no longer exists, Practically, Θ sensors are used to track soil water trends and then
i.e., when jψm j drops below the threshold value (e.g., jψm j < coupled with other IS approaches.
25 kPa), where the sensor measuring the soil water status acts as
a switch to turn on and off the irrigation events (Migliaccio et al.
2010). Similar to an ET-WB-based approach, the irrigation depth Relationships between Θ-Based and ET-WB-Based
determined by Θ-based methods can also represent the quantity of Approaches
water needed to refill the soil profile to θfc (Wang et al. 2017),
Both ET-WB-based and Θ-based methods apply soil water balance
i.e., irrigation ceases when θa reaches θfc (Haley and Dukes 2012;
equations to schedule irrigation, though different current soil water
Hoppula and Salo 2007). However, without considering leaching
estimations and triggering thresholds are employed. Under the ET-
requirements for soil salinity control, recharging the soil up to a
value lower than θfc is recommended to avoid percolation and to WB-based IS method the lower limit of Θ (triggering threshold) is
increase the WUE. When applying irrigation up to the soil’s θfc , soil defined using MAD, while Θ-based approaches use AWCth . How-
properties need to be appropriately estimated. Among the irrigation ever, when AWCth ¼ 1 − MAD, the θth defined in Eq. (3) is equiv-
amount determination methods, applying a fixed amount of irriga- alent to the lowest limit defined in Eq. (1) (Fig. 1). With regard to
tion is effective in alleviating water stress, at the risk of, however, determining the level of θth , the difference between the Θ-based
leading to either deep percolation or insufficient irrigation, result- method and the ET-WB-based method resides in the former using
ing in water and nutrient loss or impaired crop growth. site-specific field experiments to assess the threshold (Thompson
Stirzaker et al. (2017) proposed a novel irrigation timing and et al. 2007a, b), whereas the ET-WB-based method uses a default
depth determination method whereby irrigation was triggered and value drawn from FAO-56 (Allen et al. 1998). For the current Θ
aborted according to the activation/nonactivation of wetting front status, the ET-WB-based method employs an estimated deficit to
θfc , which is calculated through ET estimation and a water balance
equation; whereas the Θ-based method employs Θ sensor measure-
ments. The adoption of different current Θ estimations and trigger-
ing thresholds in irrigation timing is summarized in Table 1. With
fc
these different combinations of current Θ estimates and triggering
So il wat er con ten t ( )
RAW = AW × MAD
thresholds, different irrigation timing methods have been applied in
a
practice. For example, using a soil water balance equation to predict
AW current θa , irrigation was triggered when the calculated θa dropped
th below a preset threshold (Haley and Dukes 2012), making it easier
AWC
AWC th to conduct irrigation scheduling as Θ sensors and soil θfc proper-
ties were not necessary in this case. However, under the effect
pwp of capillary rise from a shallow water table and/or lateral flow,
0 soil-water-balance-predicted Θ may underestimate θa (Hedley and
Yule 2009). Hedley and Yule (2009) applied an MAD value of
Fig. 1. Similarity and differences in defining the soil moisture content 0.55 to define AWCth , such that irrigation was triggered when
threshold (θth ) between ET-based methods and AWC threshold methods. θa reached this threshold (Table 1). Evett et al. (2011) investigated
AW = available water; AWC = available soil water content; AWCth = whether Θ sensors were sufficiently accurate to conduct IS using a
available soil water content threshold; MAD = management allowed MAD-defined threshold and found that they all required soil-
depletion; RAW = readily available water; θa = actual volumetric soil
specific calibration except conventional time domain reflectome-
water content; θfc = volumetric soil moisture at field capacity; θpwp =
try probes.
volumetric soil moisture at permanent wilting point; and θth = soil moist-
To conclude, ET-WB-based and Θ-based IS methods are tightly
ure threshold.
connected in the way they focus on Θ status.
Θ-Based IS Applications indices. Plant-based IS methods are based on the relationship be-
tween crop water stress and soil water deficit. Accordingly, they
Developed by the United States Department of Agriculture Agricul-
can be used to define the optimum Θ level for crop growth. As
tural Research Service (USDA-ARS) and University of Georgia on
the basis of 25 years of data on conservation-minded irrigation man- the sensitivity to water deficit varies among different plant species,
agement (USDA-ARS 2016), Irrigator Pro is an expert system de- plant tissues, and crop phenological stages, a wide range of plant-
signed to provide irrigation scheduling recommendations based on based stress measurements have been proposed as being germane to
soil measurements like soil ψm and soil temperature. Limited to three irrigation scheduling. Reviewing plant-based measurements for ir-
crops—peanuts (Arachis hypogaea L.), corn, and cotton—Irrigator rigation scheduling, Jones (2004) found two principal categories:
Pro has been evaluated and applied for the implementation of var- 1. plant-water-status-based—direct measurements of leaf/xylem/
iable rate irrigation (VRI) (Bauer et al. 2008; Lamb et al. 2015; Stone stem water potential and indirect measurements of leaf thick-
et al. 2015; Vellidis et al. 2013). The system seeks to maintain soil ness, stem and fruit diameter variation, and turgor pressure
temperatures and Θ within their optimum ranges according to (Padilla-Díaz et al. 2016); and
principles described on the website of the USDA-ARS National Pea- 2. those based on plant physiology—measurements of sap flow,
nut Research Laboratory (USDA-ARS, National Peanut Research xylem cavitation, stomatal conductance, and thermal sensing.
Laboratory 2019a, b). Employing a largely Θ-based method, Irriga- An efficient plant-based irrigation program depends on whether
tor Pro triggers irrigation in one or more of these situations: whatever measurement is made is sensitive enough to properly as-
1. dry soil, namely when AWCðprobeÞ drops below zero, where sess the water deficit for the specific plant/crop (Jones 2004). For
the AWCðprobeÞ is calculated as the difference between θa — example, plants with strong endogenous control systems maintain a
estimated by Θ probe readings—and a baseline Θ (e.g., set stable leaf water status over a wide range of evaporative demand or
to when soil jψm j equals to 40 kPa); soil water supplies; therefore, one cannot use plant water status as
2. heat stress, when the maximum soil temperature since the last the index for irrigation scheduling. These include cowpea [Vigna
watering event (rain or irrigation) reaches the allowed maxi- unguiculata (L.) Walp.], corn, and poplar (Populus L.). Since leaf/
mum; and/or xylem/stem water potential are not well adapted for automation of
3. when a specific crop growth stage is reached, irrigation is applied irrigation scheduling due to difficulties in real-time measurement,
until the AWCðprobeÞ reaches its maximum for the field’s soil type. stem diameter variation (SDV), sap flow and canopy temperature
When soil moisture probe readings are not available, the sensing are more often applied in automatic irrigation systems.
AWCðprobeÞ can alternatively be calculated using a checkbook While SDV and sap flow are mostly used for commercial orchard
method based on rainfall, irrigation, and expected water use. Though plants, e.g., grapevine (Vitis L.), apple (Malus domestica Borkh.),
Irrigator Pro provides no specific recommendations for the quantity pear (Pyrus communis auct. iber.), and olive (Olea europaea L.)
of irrigation to be applied, it is able to predict the timing when irri- trees (Steppe et al. 2008; Fernández et al. 2008), thermal sensing
gation is needed through a criteria that the AWCðprobeÞ drops below technique are more commonly used for herbaceous crops and trees,
0.0 in the next 7 days. e.g., soybean, cotton, grain sorghum [Sorghum bicolor (L.) Moench]
Based on a Θ-based scheduling method Gutiérrez et al. (2014), (O’Shaughnessy and Evett 2010; O’Shaughnessy et al. 2011, 2012),
Nemali and van Iersel (2006) developed soil moisture sensor (SMS) corn (Evett et al. 1996; DeJonge et al. 2015), bermudagrass [Cyn-
controllers to maintain a certain Θ. For such systems, the installa- odon dactylon (L.) Pers.] (Emekli et al. 2007), apple trees (Osroosh
tion of a wireless sensor network is usually necessary for field-scale et al. 2015, 2016), and wine grapes (Bellvert et al. 2016), etc. Re-
irrigation management (Haule and Michael 2014; Vellidis et al. viewing several SDV-derived IS approaches, including maximum
2008). The performance of several brands of SMS controllers avail- daily stem shrinkage/growth rate, Fernández and Cuevas (2010)
able on the market (e.g., Acclima, Rain Bird, Irrometer, Water pointed out that the effectiveness of an SDV-derived index must be
Watcher, etc.) relies on their accuracy in measuring volumetric Θ evaluated, because many factors other than water status may affect
(Cardenas-Lailhacar and Dukes 2010). The evaluation of well- the index, e.g., growth patterns, crop load, plant age, and field man-
programed ET controllers and SMS controllers showed a reduction agement practices. Besides, SDV measurements tend to be affected
in water application compared to time-based treatments in which by noises from raindrops and small animals. Expert supervision is
irrigation was set at a fixed time interval (McCready and Dukes routinely required, which compounds the limited applicability of
2011; McCready et al. 2009). SDV-derived indices for automating irrigation.
Given the recent development of thermal sensing using infrared
thermometers and other thermography techniques, canopy temper-
Plant-Based IS ature (T °c ) has recently been applied as a key factor for irrigation
scheduling. Commonly used canopy-temperature-based methods
Instead of using the Θ threshold to schedule irrigation in Θ-based include the temperature-time-threshold (TTT), crop water stress in-
IS methods, plant-based IS methods use plant-water-status-related dex (CWSI), and temperature stress day (TSD). One of the major
is required, with the diurnal T °c dynamics embedded into the meth- through an evaluation of the CWSI under different irrigation treat-
ods (O’Shaughnessy and Evett 2010; O’Shaughnessy et al. 2011, ments, and the baseline temperature was acquired through experi-
2012; Osroosh et al. 2016; Peters and Evett 2008; Taghvaeian ments with fully stressed and nonstressed treatments (Emekli et al.
et al. 2014). 2007; Gontia and Tiwari 2008). The CWSI threshold was set by
Under a plant-based approach, Steppe et al. (2008) derived the users and can serve to trigger irrigations using thermal-sensed T °c .
desired depth of irrigation from the time integration of the sap flow; The accuracy of baselines is the key to this method, especially for
however, for most cases the calculated irrigation depth is not im- D2 (Idso 1982). Some studies suggest that D2 changed at different
portant as the control system will not terminate the irrigation until crop growth stages. For example, preheading and postheading
the measured plant stress index returns to the desired range (Jones baselines for corn varied and needed to be calibrated separately
2004; Osroosh et al. 2015). To account for a possible lag in the (Alderfasi and Nielsen 2001; Gontia and Tiwari 2008). Three meth-
physiological response for some crops, a fixed depth of irrigation ods were found in the literature to obtain baseline temperatures for
can be applied (Osroosh et al. 2015). determining CWSI: Idso et al. (1981) and Idso (1982) defined the
baselines (D1 and D2 ) using an empirical linear relationship be-
Temperature-Time-Threshold Method tween T °c − T °a and VPD, and Jackson et al. (1988) related these
parameters through a one-layer canopy energy balance model,
The TTT method triggers irrigation only when T °c exceeds a crop- while Alves and Pereira (2000) modified this canopy energy model
specific temperature threshold (T °th ) for greater than a predeter- to avoid the evaluation of crop surface resistances. Yuan et al.
mined time period (time threshold, tth ) within 1 day (Wanjura et al. (2004) evaluated those three methods and concluded that baselines
1995, 2004, 2006). The amount of water is applied according to derived through the method of Alves and Pereira (2000) were the
local daily ET. For instance, Evett et al. (1996) proposed that most practical, and those defined by Jackson et al. (1988) the most
30 mm of water, which euqualed to three times of the peak daily reasonable. However, both Alves and Pereira (2000) and Jackson
ET (3 × 10 mm ¼ 30 mm), should be applied to corn when the et al. (1988) required the measurement of more components
daily accumulated time for the canopy temperature above 28°C (e.g., net radiation and ground heat flux) than Idso et al. (1981) and
(T °th ¼ 28°C) exceeded 240 min (tth ¼ 240 min) on the 3rd day Idso (1982) did, making irrigation scheduling more complex. Even
after the previous irrigation. The TTT method has been evaluated though the empirical Idso definition of CWSI baselines is not as
on a number of occasions (DeJonge et al. 2015; Osroosh et al. accurate as that of the other two definitions, it had been commonly
2016; O’Shaughnessy and Evett 2010; Peters and Evett 2008), adopted by the irrigation community. As leaf temperature fluctuates
and was reported to show promise for auto-irrigation scheduling, rapidly with changes in radiation, windspeed, and air temperature,
given appropriate T °th and tth values determined in previous studies. the baseline temperature in CWSI [Eq. (4)] is affected by the envi-
The TTT method is easy to apply because it only needs the T °th and ronmental variations as well. In areas under more humid environ-
tth , e.g., T °th ¼ 22.2°C and tth ¼ 225 min for drip-irrigated apple ment (lower VPD) and lower levels of radiation, the sensitivity of
trees in the state of Washington (Osroosh et al. 2016); T °th ¼ baselines in CWSI calculations is reduced and thereby the reliabil-
28°C and tth ¼ 452 min for cotton in Texas (O’Shaughnessy and ity of CWSI method decreases. To cope with the low sensitivity of
Evett 2010). However, it can be inaccurate because canopy temper- baselines, Jones (1999) suggested to measure maximum and mini-
ature can be affected by the ambient temperature, e.g., T °c can be mum T °c using infrared thermometers from a dry (nontranspiring,
high on a hot day even if the crop is well watered. Moreover, the by covering the leaf with petroleum jelly) and wet real leaf surface,
TTT method only consider canopy temperature threshold and the respectively.
time when the threshold was exceeded, but no attention was paid to A CWSI time threshold (CWSI-TT) method was proposed by
the extent to which T °th is exceeded (DeJonge et al. 2015). More- O’Shaughnessy et al. (2012) to overcome the limit that the CWSI
over, the irrigation amount determined by peak daily ET is inac- has to be measured near solar noon or soon after, and under cloud-
curate and may cause deep seepage. free conditions. Similar to the TTT method, the CWSI-TT method
triggers irrigation when the observed CWSI value exceeds the
CWSI threshold for longer than the time threshold. Compared to
Crop Water Stress Index Method
manual irrigation based on weekly neutron probe readings, the
Proposed by Idso et al. (1981) and Jackson et al. (1981), the crop CWSI-TT method proved to be more effective on scheduling irri-
water stress index (CWSI) method schedules irrigation on the basis gation for grain sorghum (O’Shaughnessy et al. 2012). However,
of the crop’s water stress based on the T °c and atmospheric vapor O’Shaughnessy et al. (2012) pointed out that thermal-based indices
pressure deficit (VPD). The CWSI is calculated as (either CWSI or CWSI-TT) monitored using radiometric sensors
may generate false irrigation triggers in the early season when
ðT °c − T °a Þ − D2 the soil temperature affected T °c measurement. Osroosh et al. (2015)
CWSI ¼ ð4Þ
D1 − D2 accordingly developed a CWSI dynamic threshold (CWSI-DT),
With the help of thermal imaging or infrared thermography tech- CROPWAT, AquaCrop, and RZWQM2 models are highlighted in
niques, the CWSI method has been used to monitor spatial and tem- this review. These models, after appropriate calibration, can accu-
poral crop water stress (Bellvert et al. 2016; O’Shaughnessy et al. rately simulate crop responses to variable atmospheric and soil con-
2011) and then to schedule a variable irrigation in different regions ditions and management scenarios, e.g., the Root Zone Water
of the field. The fusion of thermal and visible imaging improved the Quality Model (RZWQM2) model (Ahuja et al. 2000; Ma et al.
accuracy of CWSI by accounting for soil background effects on mea- 2012a, b). As these models contribute to investigating crop re-
sured temperature (Möller et al. 2006). The 2D-CWSI developed by sponses to different irrigation management regimes, they in return
Tilling et al. (2007) based on the vegetation index-temperature (VIT) benefit to irrigation scheduling. However, process-based models
trapezoid (Moran et al. 1994) allowed the application of CWSI to are commonly used for irrigation planning but not real-time deci-
partially vegetated fields. sion-making. To fill this gap, Gu et al. (2017a) developed a soft-
ware, IrrSch, to determine irrigation timing and amount based on
Temperature Stress Day Method RZWQM2 simulated plant water stress and soil water regime. This
software, along with irrigation control hardware, has been adopted
The TSD method is based on the difference between the stressed
in an on-going field experiment to test its feasibility in real-time
and nonstressed T °c of a given crop (Clawson and Blad 1982;
irrigation scheduling for cotton (Chen et al. 2019).
Gardner et al. 1981a, b). Even though it is more easily affected by
Irrigation events modelled in SWAT can be scheduled either:
some environmental factors (e.g., humidity) (Clawson et al. 1989),
1. by the user—by providing inputs for application dates, amounts,
given their simplicity, TSD indices have been applied in a number
and application efficiencies; or
of recent studies (DeJonge et al. 2015; Taghvaeian et al. 2014).
2. by an automatic irrigation module (Arnold et al. 1998; Neitsch
Taghvaeian et al. (2014) evaluated the simple TSD index, the
et al. 2011), whereby irrigations are triggered by either a water
degrees above non-stressed (DANS), which represented the differ-
stress threshold or a soil water deficit threshold fixed by the user.
ence between stressed and nonstressed canopy temperatures. When
The quantity of water applied is calculated to replenish the root
applied to sunflowers (Helianthus annuus L.), they found DANS to
zone soil to θfc (though a lesser amount is suggested to avoid ex-
be strongly correlated with several crop parameters, including the
cessive percolation if rainfall occurs in the following days) or a
fraction of intercepted photosynthetically active radiation (fIPAR),
fraction thereof for deficit irrigation. The SWAT model calculates
leaf area index (LAI), leaf water potential, and root growth. Accord-
water stress by comparing actual and potential plant transpiration
ingly, it was suggested that a simple DANS index, based solely on
(T a , T pot , respectively). Transpiration is either simulated by the
T °c and estimated by a simple subtraction, might be used as effec-
Penman-Monteith method or derived as a linear function of poten-
tively as CWSI in monitoring of plant water status and therefore
tial evapotranspiration (ET pot ) and LAI (Neitsch et al. 2011; Sun
in scheduling irrigations for sunflower in arid/semiarid regions.
and Ren 2014). Both Sun and Ren (2014) and Maier and Dietrich
DeJonge et al. (2015) evaluated six canopy-temperature-based
(2016) found that, when used for irrigation scheduling, the SWAT
water stress indices, including three previously proposed (CWSI,
model consistently overestimated the amount of irrigation required.
DANS, and TTT), and three newly introduced: degrees above critical
Therefore, the model’s accuracy in automatic irrigation control
temperature (DACT; critical temperature is equivalent to temperature
needs to be improved.
threshold defined in TTT method), integrated DANS (IDANS), and
Developed by the Land and Water Development Division of
integrated DACT (IDACT). Their experiment showed that the simple
FAO, CROPWAT is a computer program to calculate crop water
indices, DANS and DACT, and their integrated surrogates generated
and irrigation requirements from climatic and crop data (Smith
similar representations of water status as CWSI, but with fewer
1992; Savva and Frenken 2002). CROPWAT uses what is essen-
parameters. Given its simplicity, the TSD method is likely to be
tially the ET-WB method, combined with the ability to predict
adopted by farmers in irrigation scheduling for some crops in some
yield reduction due to water stress (Doorenbos and Kassam 1979).
regions.
To schedule irrigations, CROPWAT requires climatic data (tem-
perature, relative humidity, wind speed, sunshine hours, rainfall),
Model-Based IS crop data (planting date, K c curve, rooting depth at different
growth stages, the allowable Θ depletion level, and the yield re-
The model-based IS methods reviewed are limited to those in- sponse factor K y ), along with soil data (total available soil water
tended for determining irrigation timing and amount for a specific content, initial soil water depletion). Irrigations under CROPWAT
field, rather than the determination of the distribution of water re- can be triggered by a Θ threshold, at fixed intervals, by a predeter-
sources for several fields and/or crops (Alvarez et al. 2004). Among mined depleted amount of water, or by a reduction in ET. The
these model-based IS methods, the utilization of process-based mod- water depth applied is the amount to bring Θ back to θfc or to
els and regression models are summarized. Irrigations are scheduled a fixed level below or above θfc ; otherwise, the replenishment
using simulated results from a process-based model. For regression point is at the user’s discretion. In the latest version of CROPWAT
irrigation events by manually specifying the time and depth of each levels for a crop (Linker and Kisekka 2017), etc. Research on real-
application, or by the model automatically developing a schedule. time irrigation scheduling based on model outputs is less common
In the latter case, irrigations are scheduled either at a fixed time at present. Thorp et al. (2017) used the CSM-CROPGRO-Cotton
interval and depth, or by a fixed percentage of allowable water model to assist in-season irrigation scheduling by estimating
depletion of the root zone. Using the AquaCrop model and drawing ET. Gu et al. (2017a) developed an irrigation scheduling software
on a long series of historical climate data, Geerts et al. (2010) opti- based on RZWQM2. Water-stress-based methods, such as those
mized irrigation frequency during sensitive crop growth stages employed in the SWAT model (Neitsch et al. 2011; Sun and Ren
based on indicative crop development traits at the stage’s onset. 2014) and the RZWQM2 model (Saseendran et al. 2014, 2015; Gu
A simplified readable irrigation chart was subsequently generated et al. 2017a), are generally more reasonable and practical once the
for farmers. However, this simplified irrigation chart only suggests model is assured of an accurate prediction in plant water stress after
the irrigation frequency during the critical crop growth stage, with- calibration and validation. However, without links to precise weather
out irrigation amount information or adjustment due to atmospheric forecasts, these models, when involved in real-time IS, may derive an
conditions of given years. erroneous timing and quantity of irrigation for a field, especially
Developed by USDA-ARS, RZWQM2 is a one-dimensional when a large rainfall events occurs soon after irrigation. Once the
model with emphasis on management effects on hydrologic cycle, present weather data are recorded, the crop water status need to be
water quality, and crop production (Ahuja et al. 2000). Equipped updated.
with the Decision Support System for Agrotechnology Transfer
(DSSAT) model (Jones et al. 2003) and the Simultaneous Heat and
Water (SHAW) model (Flerchinger and Saxton 1989), RZWQM2 Regression Models
simulates crop growth on a daily basis under different field man- Besides process-based models for IS, some scholars have recently
agement practices and specific soil-plant-atmosphere systems. The proposed methods using regression models. Lopes et al. (2016)
RZWQM2 model provides several IS methods: by fixed interval, transformed the irrigation planning problem into an optimal control
triggered by a root zone depletion (MAD), or triggered by a thresh- problem for the object of minimizing the total volume of irrigation
old of the ratio of actual ET to potential ET. By obtaining field- during the entire growing season. The optimal control problem was
observed climate data and online forecasting data using a software subjected to a soil water balance function, minimum Θ constraint,
they developed, Gu et al. (2017a, b) and Liu et al. (2017) applied and maximum irrigation amount for a single event. However, this
the RZWQM2 model to simulate the predicted crop water stress, algorithm required model-derived whole-year weather data, which
and to judge whether irrigation was required on a daily basis. The are unpredictable and may lead to the failure of the proposed
irrigation depth was calculated to return the current soil water con- method. To overcome this problem, Lopes et al. (2016) further im-
tent to θfc . In this way, irrigations can be scheduled in a real-time proved the irrigation planning by applying model predictive control
sense as long as the real-time atmospheric data are available. The (MPC) techniques, where the optimal control problem was fre-
RZWQM2 model also integrates fixed intervals, specified dates, or quently recomputed at each interval by taking into account mea-
an option dependent on soil water depletion in the root zone in its sured system variables. The MPC method, also known as receding
irrigation module. horizon control (RHC), was applied by Park et al. (2009) to the
Other models reportedly used for irrigation scheduling pur- problem of irrigation management, by incorporating sensor mea-
poses include DAISY, a one-dimensional mechanistic soil-plant- surements (feedback), predictive models, and optimization algo-
atmosphere system model (Abrahamsen and Hansen 2000; Seidel rithms to maintain Θ levels. An MPC algorithm was also tested
et al. 2016) and SIMETAW# (SIMulation of EvapoTranspiration of by Saleem et al. (2013) for real-time IS based on soil water balance
Applied Water), a model capable of assessing climate change im- and a given Θ range; however, the system dynamic function was
pacts on future irrigation demand, thereby facilitating user develop- based on a simple water balance equation and ET c and precipita-
ment of adaptation strategies (Mancosu et al. 2016). Seidel et al. tion were assumed to be directly available. To solve this problem,
(2016) used a partially calibrated DAISY model to schedule irri- Delgoda et al. (2016) combined an MPC algorithm with the Aqua-
gation: when the soil ψm reached a fixed threshold, an irrigation Crop model, where the system dynamic function was given by a
of fixed depth was triggered. A web-based user interface was de- linear first-order time-series model based on system identification
veloped to run the modified van Genuchten model (to convert the from AquaCrop. The AquaCrop model also provided feedback on
soil water tension data to volumetric water content) and to derive the root zone Θ deficit during the MPC’s optimization process.
irrigation scheduling recommendations. By integrating the web- A predominant advantage of using MPC/RHC algorithms in IS
based interface with a wireless Θ sensing array (UGA SSA) and is its ability to be programmed into a controller and thereby be used
a VRI-enabled center pivot irrigation system, a precise automated in real-time applications. Park et al. (2009) executed the RHC al-
VRI system was achieved (Liakos et al. 2015; Liang et al. 2016; gorithm in a real-time manner on a field computer through collect-
Vellidis et al. 2016c). ing real-time soil moisture, temperature, and meteorological data,
learning technologies) to calculate appropriate irrigation decisions irrigation depth. However, the other three types of methods require
considering the weather, crop, and soil characteristics, along with specific preparations prior to their application: Θ- and plant-based
field measurements from soil sensors. Linker et al. (2016) used methods need sensors and monitoring systems to be installed;
both the crop model AquaCrop and a nonlinear constrained opti- thresholds to trigger the irrigation need to be researched and vali-
mization to investigate the highest yield achievable for any given dated; while for models, calibration via previous field experiment
water quantity through optimizing the water-depletion level and re- may be required. Installing sensors and a monitoring system in the
plenishment level implemented at five different growth stages. With field is laborious for farmers, so they might prefer to equip them-
a well-calibrated RZWQM2 model to assess the derived solutions, selves with infrared thermometers on a portable irrigation system
Nguyen et al. (2017) applied an ant colony algorithm to search for (O’Shaughnessy et al. 2012). Plant-based methods may require
an optimal irrigation schedule to maximize the net return of a sin- expert supervision as the monitored data can be a puzzle to farmers,
gle crop. making irrigation less reliable (Fernández and Cuevas 2010). Model-
However, for real-time IS, regression methods using either
based methods are easier for the users, given a developed software or
optimization or AI algorithms must be validated since they are
procedure based on a well-calibrated model (George et al. 2000; Gu
commonly trained using the historical weather data, which may
et al. 2017a).
not actually possess the same trend as the current weather data.
An IS approach must be appropriate to the specific conditions at
The effect of the uncertainty of predicted weather data should also
hand. For an orchard, a plant-based method measuring either the
be evaluated. The performance of regression models should be
stem diameter or sap flow can be used, whereas an ET-WB-based
compared to that of other methods, e.g., Θ-based, plant-based,
method or a model-based method would prove inappropriate, given
or process-based models, so that the effectiveness could be evalu-
that the tree species may not likely be available in the model or
ated. Development of those softwares and controllers that inte-
grated with regression models/algorithms is challenging but is the suggested K c and duration of crop stages are not appropriate
driven by the commercial value of real-time IS. Fortunately, with in the FAO document for an individual field. For areas where it
the development of AI hardware, e.g., Raspberry Pi that is able to is difficult or impossible to access weather data, an ET-WB-based
run complex mathematical algorithms, those regression IS methods method or a model-based method are not achievable. The main re-
can be implemented in real time with low cost. quirements and constraints for the different methods are listed
briefly in Table 2.
It is worth noting that both the ET-WB-based and model-based
Concluding Remarks and Future Perspectives methods need only onsite weather data when operating, without
further measured information on the Θ and plant growth, because
the ET-WB-based methods assume full irrigation and for the model-
Advantages and Disadvantages of the IS Approaches based methods Θ and plant growth are predicted by the model.
The four types of IS methods each have their advantages and pit- This is similar to an open loop control, under which errors between
falls, being effective in some conditions but presenting less than the calculated or simulated parameters may accumulate during the
ideal results in some fields when poorly suited or operated. Osroosh scheduling process and thus result in delayed or early onset irriga-
et al. (2016) compared seven different irrigation methods for apple tion. The Θ- and plant-based methods both focus on measured
orchards (e.g., ET-WB-based, conventional, Θ-based methods, and real-time conditions of the soil or plant, which mimics a closed loop
plant-based methods) based on the total irrigation water applied, control. In this way, the scheduling effectiveness depends strongly on
Table 2. Requirements and constraints of the four types of irrigation scheduling methods
Methods Requirements Constraints
ET-WB-based Weather data; K c curve of the crop; suggested MAD Limited to the crops for which K c and MAD are provided in FAO
and estimated soil properties document; cumulative error may occur during the scheduling process
Θ-based Sensors and monitoring system; validated trigger Depends on the accuracy of measured soil moisture and the validity of
threshold(s) the threshold; sensor measurements may vary and not be representative
of the full field conditions due to field variability
Plant-based Sensors and monitoring system; validated trigger Noise in sensitive measured data (e.g., SDV) may make scheduling
threshold(s)/conditions impossible; depends on the sensitivity of the measured parameter and
the validity of the trigger thresholds or conditions
Model-based Calibrated model via previous field experiment data; Depends on the calibration quality and model accuracy; cumulative
weather data; developed software or procedure error may occur in real-time scheduling
themselves [Checkbook, SmartIrrigation App and the University of over the growing season. To better schedule irrigation, the factors
Georgia Smart Sensor Array (UGA-SSA; Vellidis et al. 2013)] and mentioned above should be considered in setting the target level,
to a strictly rainfed treatment. They found that in a wet year the and a reliable and precise estimation of θa is needed. Water stress
scheduling methods resulted in a significantly lower yield and methods, which integrate the primary factors stated above, re-
water use efficiency than the rainfed treatment. This study indicates present advanced methods in terms of determining appropriate tar-
that for cotton a slight water stress during certain phenological get Θ values. Using a more precise, or at least well-calibrated
stages stimulated its growth. Similarly, in Irrigator Pro for cotton model, plant water stress and real-time Θ can be estimated accu-
irrigation, once the first cracked boll stage occurred, irrigation was rately, though this accuracy could be enhanced when a few mea-
applied to return the soil to its maximum AWCðprobeÞ , and then no surements could serve as feedback. When determining the depth of
more irrigation was applied for the rest of the season, promoting irrigation in Step 3, weather forecasts are also crucial factors to be
boll opening and preventing boll rot and attending yield and quality considered to avoid run off and/or deep seepage after an ensuing
loss (USDA-ARS, National Peanut Research Laboratory 2019a). precipitation event. Moreover, the forecasted rainfall data should be
Delving deep into the four types of IS methods, it is found that updatable. McCready et al. (2009) found that a high setting of sat-
they all focus on the Θ, which functions as a bridge/connection uration allowance for the Rain Bird ET Manager controller, that
between irrigation management and crop water required for growth. allows for rainfall to be stored in the root zone by not irrigating
In other words, for the purpose of improving crop growth and sav- fully to field capacity, resulted in low soil-water content during
ing water, irrigation management methods are intended to adjust Θ periods of low rainfall and therefore affected turf quality. A pre-
to a certain level where crop growth can be guaranteed with im- cisely forecasted rainfall in a few days following the irrigation
proved irrigation water use efficiency. To achieve this, the follow- event makes the irrigation decision more efficient because the fore-
ing three points need to be implemented: casted rainfall will be deducted to avoid deep percolation when this
1. Determine a range of Θ levels (target level) that promote crop rainfall event occurs.
growth throughout the entire growing season. Different target Although the IS approaches discussed above have been widely
levels should be determined for different growth stages by used, several practical factors, including the ease of use, economic,
applying various irrigation treatments in the field before employ- governmental, and water availability constraints or salt leaching re-
ing the Θ-based or plant-based scheduling methods (Padilla- quirements, were not considered when developing those systems.
Díaz et al. 2016; Wang et al. 2017). For example, when using For instance, as salt may either rise from deeper soil layers or origi-
water stress indices, the lower Θ limit is defined as where water nate in poor-quality irrigation water, a leaching irrigation can be
stress occurs, while the upper Θ might be defined as θfc . scheduled during or between crop growing seasons. Alternatively,
2. Estimate actual Θ correctly and precisely, using values inferred a coefficient for salinity leaching requirement Lr (Hoffman et al.
from sensor measurement in a Θ-based method, through plant 1990) can be applied in calculating the required depth for each ir-
measurements when using plant-based methods, or estimated rigation event to leach salts beyond the root zone.
through an ET-WB method (considering the weather and soil
characteristics) or other models. A number of factors may contri-
Conclusions and Perspectives in IS
bute to a more precise and reliable measurement or estimation of
Θ, e.g., weather, soil, crop, and their interactions. These are con- Irrigation scheduling methods for better crop growth were re-
sidered in advanced agricultural system models, e.g., RZWQM2. viewed. These methods are summarized and categorized into four
When using plant measurements to represent Θ, considering more types: ET-WB-, Θ-, plant-, and model-based methods. The four
factors such as crop growth stage would benefit scheduling types of methods are all able to successfully schedule irrigations
(Bellvert et al. 2016). when properly operated. With more applications needed for IS,
3. Apply the exact amount of water to return Θ from its current tools (software, programs, and controllers) have been developed
status to a target level. To achieve this, irrigation frequency to facilitate the decision process.
and quantity should be coordinated so the crop is neither under However, today’s IS methods must reach a compromise between
water stress nor overirrigated (resulting in runoff, deep seepage practicability, accuracy, and universalizability. The ET-WB method
or unnecessary drainage) (McCready and Dukes 2011). Irriga- can be widely used regardless of the field conditions (e.g., soil and
tion frequency will be higher when irrigation quantity at each crop growth), as long as the weather data, K c , and MAD are avail-
event is decreased. able (soil water capacity is not even necessary). This method, while
In general, to schedule an appropriate irrigation, factors includ- practicable with the guidelines provided by FAO-56, is not very
ing the weather, soil properties, and crop growth should be consid- accurate under site-specific conditions, especially when the error
ered in both Steps 1 and 2. The ET-WB method uses MAD and soil of daily Θ estimations cumulates. Θ- and plant-based methods
properties to define the target Θ level and estimates θa by the water are popular and accurate when the measurement is applicable
balance, where the target Θ is a general value provided by FAO and reliable, but they are relatively less practical because of the
crop growth responses to water; thus, studies on crop phenology T °c = canopy temperature;
and responses to various factors in soil-crop-atmosphere system T °th = canopy temperature threshold;
should be enhanced. Given the importance of Θ in IS (see section tth = time threshold;
“Factors considered in IS”), the development of more effective VPD = atmospheric vapor pressure deficit;
scheduling methods should focus on the control of Θ, based on
Θ = soil moisture;
an advanced understanding of its effect on crop growth. With a pre-
θa = actual volumetric soil water content;
cise estimation of real-time Θ and well-defined Θ/water stress
threshold by the model, as well as weather forecast data and mea- θfc = volumetric soil moisture at field capacity;
sured feedback if available, the process-based model can yield a θpwp = volumetric soil moisture at permanent wilting point;
better irrigation regime in terms of water savings and favorable θth = soil moisture threshold;
plant growth. In viewing the development and application of intel- ψm = soil matric potential; and
ligent algorithms, agricultural soil-crop-atmosphere system model- ψleaf = leaf water potential.
ing might achieve an evolution over the conventional process-based
models and result in a more effective irrigation scheduling. Those
approaches based on regression models should be pressed forward References
to integrate with software or decision support systems for the pur-
pose of real-time operation. Last but not least, less expensive equip- Abrahamsen, P., and S. Hansen. 2000. “Daisy: An open soil–crop–
ment (e.g., sensors and weather station) need to be developed so as atmosphere system model.” Environ. Modell. Software 15 (3): 313–330.
to be more affordable for farmers. Moreover, the accessibility of https://doi.org/10.1016/S1364-8152(00)00003-7.
Ahuja, L., K. Rojas, J. Hanson, M. Shaffer, and L. Ma. 2000. Root zone
online historical and forecast weather data need to be improved,
water quality model. Highlands Ranch, CO: Water Resources
especially in developing countries. Publications.
Alderfasi, A. A., and D. C. Nielsen. 2001. “Use of crop water stress index
for monitoring water status and scheduling irrigation in wheat.” Agric.
Acknowledgments Water Manage. 47 (1): 69–75. https://doi.org/10.1016/S0378-3774(00)
00096-2.
This research was supported by the National Key Research and De- Alexandratos, N., and J. Bruinsma. 2012. World agriculture towards 2030/
velopment Program of China (No. 2017YFC0403303), the Natural 2050: The 2012 revision. Rome: Food and Agriculture Organization.
Science and Engineering Research Council of Canada-Discovery Allen, R. G., L. S. Pereira, D. Raes, and M. Smith. 1998. Crop
program (RGPIN 04784-14), the Fundamental Research Funds evapotranspiration-guidelines for computing crop water requirements—
FAO irrigation and drainage paper 56. Rome: Food and Agriculture
for the Central Universities (2019B18214), the National Natural
Organization.
Science Foundation of China (51909064), the Project funded by Alvarez, J. F. O., J. A. de Juan Valero, J. M. T. Martín-Benito, and E. L.
China Postdoctoral Science Foundation (2019M651684), and Mata. 2004. “Mopeco: An economic optimization model for irrigation
The Belt and Road Special Foundation of the State Key Laboratory water management.” Irrig. Sci. 23 (2): 61–75. https://doi.org/10.1007
of Hydrology-Water Resources and Hydraulic Engineering /s00271-004-0094-x.
(2019490411). Many thanks to China Scholarship Council (CSC) Alves, I., and L. S. Pereira. 2000. “Non-water-stressed baselines for irri-
program that financially supported Zhe’s visiting study at McGill gation scheduling with infrared thermometers: A new approach.” Irrig.
University (Canada) that promotes the production of this review Sci. 19 (2): 101–106. https://doi.org/10.1007/s002710050007.
paper, and appreciation to Dr. Georges T. Dodds who proofread Arnold, J. G., R. Srinivasan, R. S. Muttiah, and J. R. Williams. 1998.
this review. We also wish to thank the editor and anonymous re- “Large area hydrologic modeling and assessment. Part I: Model devel-
opment.” J. Am. Water Resour. Assoc. 34 (1): 73–89. https://doi.org/10
viewers for their helpful comments and suggestions.
.1111/j.1752-1688.1998.tb05961.x.
Augustin, L. K., A. H. Yagoob, W. K. Kirui, and Y. Peiling. 2015. “Optimal
irrigation scheduling for summer maize crop: Based on GIS and CROP-
Notation WAT application in Hetao district; Inner Mongolia autonomous region,
China.” J. Biol. Agric. Healthcare 5 (18): 95–102.
The following symbols are used in this paper: Bartlett, A., A. Andales, M. Arabi, and T. Bauder. 2015. “A smartphone app
AWC = actual available soil water content; to extend use of a cloud-based irrigation scheduling tool.” Comput. Elec-
AWCth = available soil water content threshold; tron. Agric. 111 (Feb): 127–130. https://doi.org/10.1016/j.compag.2014
.12.021.
AWCðprobeÞ = difference between measured and baseline soil
Bauer, P. J., K. C. Stone, W. Bussche, J. A. Millen, D. E. Evans, and E. E.
moisture (in Irrigator Pro); Strickland. 2008. “Site-specific irrigation of peanuts on a coastal plain
Dr = root zone depth; field.” In Vol. 222 of Proc., 9th Int. Conf. on Precision Agriculture.
D1 = maximum crop water stress baseline; Princeton, NJ: CiteSeerX.
age. 97 (5): 666–672. https://doi.org/10.1016/j.agwat.2009.12.009. Flerchinger, G. N., and K. E. Saxton. 1989. “Simultaneous heat and water
Chauhan, Y. S., G. C. Wright, D. Holzworth, R. C. Rachaputi, and J. O. model of a freezing snow-residue-soil system: I. Theory and develop-
Payero. 2013. “Aquaman: A web-based decision support system for ment.” Trans. ASABE 32 (2): 565–571. https://doi.org/10.13031/2013
irrigation scheduling in peanuts.” Irrig. Sci. 31 (3): 271–283. https://doi .31040.
.org/10.1007/s00271-011-0296-y. Gardner, B., B. Blad, D. Garrity, and D. Watts. 1981a. “Relationships be-
Chen, X., Z. Qi, D. Gui, Z. Gu, L. Ma, F. Zeng, L. Li, and M. W. Sima. tween crop temperature, grain yield, evapotranspiration and phenologi-
2019. “A model-based decision support system for irrigation scheduling cal development in two hybrids of moisture stressed sorghum.” Irrig.
to improve water productivity.” Agronomy 9 (11): 686. https://doi.org Sci. 2 (4): 213–224. https://doi.org/10.1007/BF00258375.
/10.3390/agronomy9110686. Gardner, B., B. Blad, and D. Watts. 1981b. “Plant and air temperatures in
Clawson, K., R. Jackson, and P. Pinter. 1989. “Evaluating plant water stress differentially irrigated corn.” Agric. Meteorol. 25: 207–217. https://doi
with canopy temperature differences.” Agron. J. 81 (6): 858–863. .org/10.1016/0002-1571(81)90073-X.
https://doi.org/10.2134/agronj1989.00021962008100060004x. Geerts, S., D. Raes, and M. Garcia. 2010. “Using AquaCrop to derive defi-
Clawson, K. L., and B. L. Blad. 1982. “Infrared thermometry for sched- cit irrigation schedules.” Agric. Water Manage. 98 (1): 213–216. https://
uling irrigation of corn.” Agron. J. 74 (2): 311–316. https://doi.org/10 doi.org/10.1016/j.agwat.2010.07.003.
.2134/agronj1982.00021962007400020013x.
George, B. A., S. A. Shende, and N. S. Raghuwanshi. 2000. “Development
Davis, S., and M. Dukes. 2010. “Irrigation scheduling performance by
and testing of an irrigation scheduling model.” Agric. Water Manage.
evapotranspiration-based controllers.” Agric. Water Manage. 98 (1):
46 (2): 121–136. https://doi.org/10.1016/S0378-3774(00)00083-4.
19–28. https://doi.org/10.1016/j.agwat.2010.07.006.
Glenn, E. P., A. R. Huete, P. L. Nagler, K. K. Hirschboeck, and P. Brown.
Davis, S., M. Dukes, and G. Miller. 2009. “Landscape irrigation by
2007. “Integrating remote sensing and ground methods to estimate
evapotranspiration-based irrigation controllers under dry conditions
evapotranspiration.” Crit. Rev. Plant Sci. 26 (3): 139–168. https://doi
in Southwest Florida.” Agric. Water Manage. 96 (12): 1828–1836.
.org/10.1080/07352680701402503.
https://doi.org/10.1016/j.agwat.2009.08.005.
Gnecchi, J. A. G., A. M. Patiño, F. L. Paramo, A. D. C. T. Anguiano, and
DeJonge, K. C., S. Taghvaeian, T. J. Trout, and L. H. Comas. 2015. “Com-
D. L. Espinoza. 2015. “Investigation of wetting front propagation dy-
parison of canopy temperature-based water stress indices for maize.”
namics using soil impedance measurements: Implications for modelling
Agric. Water Manage. 156 (Jul): 51–62. https://doi.org/10.1016/j.agwat
and irrigation scheduling.” Water Resour. Manage. 29 (1): 197–210.
.2015.03.023.
https://doi.org/10.1007/s11269-014-0835-4.
Delgoda, D., H. Malano, S. K. Saleem, and M. N. Halgamuge. 2016.
“Irrigation control based on model predictive control (MPC): Formu- Gontia, N., and K. Tiwari. 2008. “Development of crop water stress index
lation of theory and validation using weather forecast data and AQUA- of wheat crop for scheduling irrigation using infrared thermometry.”
CROP model.” Environ. Modell. Software 78 (Apr): 40–53. https://doi Agric. Water Manage. 95 (10): 1144–1152. https://doi.org/10.1016/j
.org/10.1016/j.envsoft.2015.12.012. .agwat.2008.04.017.
DeOreo, W. B., P. Mayer, B. Dziegielewski, and J. Kiefer. 2016. Residential Gowda, P. H., J. L. Chavez, P. D. Colaizzi, S. R. Evett, T. A. Howell, and
end uses of water, version 2: Executive report. Denver: Water Research J. A. Tolk. 2008. “ET mapping for agricultural water management:
Foundation. Present status and challenges.” Irrig. Sci. 26 (3): 223–237. https://doi
Devitt, D., K. Carstensen, and R. Morris. 2008. “Residential water savings .org/10.1007/s00271-007-0088-6.
associated with satellite-based ET irrigation controllers.” J. Irrig. Drain. Grabow, G., I. Ghali, R. Huffman, G. Miller, D. Bowman, and A. Vasanth.
Eng. 134 (1): 74–82. https://doi.org/10.1061/(ASCE)0733-9437(2008) 2013. “Water application efficiency and adequacy of ET-based and soil
134:1(74). moisture–based irrigation controllers for turfgrass irrigation.” J. Irrig.
Doorenbos, J., and A. Kassam. 1979. Yield response to water—Irrigation Drain. Eng. 139 (2): 113–123. https://doi.org/10.1061/(ASCE)IR
and drainage paper 33. Rome: Food and Agriculture Organization. .1943-4774.0000528.
Dukes, M. 2012. “Water conservation potential of landscape irrigation Gu, Z., Z. Qi, L. Ma, D. Gui, J. Xu, Q. Fang, S. Yuan, and G. Feng. 2017a.
smart controllers.” Trans. ASABE 55 (2): 563–569. https://doi.org/10 “Development of an irrigation scheduling software based on model pre-
.13031/2013.41391. dicted crop water stress.” Comput. Electron. Agric. 143 (Dec): 208–221.
Emekli, Y., R. Bastug, D. Buyuktas, and N. Y. Emekli. 2007. “Evaluation of https://doi.org/10.1016/j.compag.2017.10.023.
a crop water stress index for irrigation scheduling of bermudagrass.” Gu, Z., Z. Qi, L. Ma, and S. Yuan. 2017b. “Water stress based deficit irri-
Agric. Water Manage. 90 (3): 205–212. https://doi.org/10.1016/j.agwat gation scheduling using RZWQM2 model for maize in Colorado.”
.2007.03.008. In Proc., 2017 ASABE Annual Int. Meeting, St. Joseph, MI: American
Evett, S. R., T. A. Howell, A. D. Schneider, D. R. Upchurch, and D. F. Society of Agricultural and Biological Engineers.
Wanjura. 1996. “Canopy temperature based automatic irrigation con- Gutiérrez, J., J. F. Villa-Medina, A. Nieto-Garibay, and M. Á. Porta-
trol.” In Proc., Int. Conf. on Evapotranspiration and Irrigation Sched- Gándara. 2014. “Automated irrigation system using a wireless sensor
uling, 207–213. St. Joseph, MI: American Society of Agricultural network and GPRS module.” IEEE Trans. Instrum. Meas. 63 (1):
Engineers. 166–176. https://doi.org/10.1109/TIM.2013.2276487.
Evett, S. R., R. C. Schwartz, N. T. Mazahrih, M. A. Jitan, and I. M. Shaqir. Haley, M. B., and M. D. Dukes. 2012. “Validation of landscape irrigation
2011. “Soil water sensors for irrigation scheduling: Can they deliver a reduction with soil moisture sensor irrigation controllers.” J. Irrig. Drain.
management allowed depletion?” Acta Hort. 888 (888): 231–237. Eng. 138 (2): 135–144. https://doi.org/10.1061/(ASCE)IR.1943-4774
https://doi.org/10.17660/ActaHortic.2011.888.26. .0000391.
effectively schedules and controls center pivot irrigation of cotton.” Smart Irrigation Apps Group. 2019. “Smart irrigation apps.” Accessed
Agric. Water Manage. 97 (9): 1310–1316. https://doi.org/10.1016/j August 23, 2019. http://smartirrigationapps.org/.
.agwat.2010.03.012. Smith, M. 1992. Vol. 46 of CROPWAT: A computer program for irrigation
O’Shaughnessy, S. A., S. R. Evett, P. D. Colaizzi, and T. A. Howell. 2012. planning and management. Rome: Food and Agriculture Organization.
“A crop water stress index and time threshold for automatic irrigation Smith, M., D. Kivumbi, and L. Heng. 2002. “Use of the FAO CROPWAT
scheduling of grain sorghum.” Agric. Water Manage. 107 (May): model in deficit irrigation studies.” In Deficit irrigation practices,
122–132. https://doi.org/10.1016/j.agwat.2012.01.018. 17–27. Rome: Food and Agriculture Organization.
Osroosh, Y., R. T. Peters, C. S. Campbell, and Q. Zhang. 2015. “Automatic Steduto, P., T. C. Hsiao, D. Raes, and E. Fereres. 2009. “Aquacrop—The
irrigation scheduling of apple trees using theoretical crop water stress FAO crop model to simulate yield response to water: I. Concepts and
index with an innovative dynamic threshold.” Comput. Electron. Agric. underlying principles.” Agron. J. 101 (3): 426–437. https://doi.org/10
118 (Oct): 193–203. https://doi.org/10.1016/j.compag.2015.09.006. .2134/agronj2008.0139s.
Osroosh, Y., R. T. Peters, C. S. Campbell, and Q. Zhang. 2016. “Compari- Steppe, K., D. J. De Pauw, and R. Lemeur. 2008. “A step towards new
son of irrigation automation algorithms for drip-irrigated apple trees.” irrigation scheduling strategies using plant-based measurements and
Comput. Electron. Agric. 128 (Oct): 87–99. https://doi.org/10.1016/j
mathematical modelling.” Irrig. Sci. 26 (6): 505–517. https://doi.org/10
.compag.2016.08.013.
.1007/s00271-008-0111-6.
Padilla-Díaz, C., C. Rodriguez-Dominguez, V. Hernandez-Santana, A.
Stirzaker, R. J., T. C. Maeko, J. G. Annandale, J. M. Steyn, G. T. Adhanom,
Perez-Martin, and J. Fernández. 2016. “Scheduling regulated deficit ir-
and T. Mpuisang. 2017. “Scheduling irrigation from wetting front
rigation in a hedgerow olive orchard from leaf turgor pressure related
depth.” Agric. Water Manage. 179 (Jan): 306–313. https://doi.org/10
measurements.” Agric. Water Manage. 164 (Part 1): 28–37. https://doi
.1016/j.agwat.2016.06.024.
.org/10.1016/j.agwat.2015.08.002.
Stone, K., P. Bauer, W. Busscher, J. Millen, D. Evans, and E. Strickland.
Park, Y., J. S. Shamma, and T. C. Harmon. 2009. “A Receding Horizon
2015. “Variable-rate irrigation management using an expert system in
Control algorithm for adaptive management of soil moisture and chemi-
the eastern coastal plain.” Irrig. Sci. 33 (3): 167–175. https://doi.org/10
cal levels during irrigation.” Environ. Modell. Software 24 (9): 1112–
.1007/s00271-014-0457-x.
1121. https://doi.org/10.1016/j.envsoft.2009.02.008.
Sun, C., and L. Ren. 2014. “Assessing crop yield and crop water produc-
Perea, R. G., I. F. García, M. M. Arroyo, J. R. Díaz, E. C. Poyato, and
tivity and optimizing irrigation scheduling of winter wheat and summer
P. Montesinos. 2017. “Multiplatform application for precision irrigation
scheduling in strawberries.” Agric. Water Manage. 183 (Mar): 194–201. maize in the Haihe plain using SWAT model.” Hydrol. Process. 28 (4):
https://doi.org/10.1016/j.agwat.2016.07.017. 2478–2498. https://doi.org/10.1002/hyp.9759.
Pereira, L. S., R. G. Allen, M. Smith, and D. Raes. 2015. “Crop evapotran- Taghvaeian, S., L. Comas, K. C. DeJonge, and T. J. Trout. 2014. “Conven-
spiration estimation with FAO56: Past and future.” Agric. Water Man- tional and simplified canopy temperature indices predict water stress in
age. 147 (Jan): 4–20. https://doi.org/10.1016/j.agwat.2014.07.031. sunflower.” Agri. Water Manage. 144 (Oct): 69–80. https://doi.org/10
Peters, R. T., and S. R. Evett. 2008. “Automation of a center pivot using the .1016/j.agwat.2014.06.003.
temperature-time-threshold method of irrigation scheduling.” J. Irrig. Thompson, R., M. Gallardo, L. Valdez, and M. Fernández. 2007a. “Deter-
Drain. Eng. 134 (3): 286–291. https://doi.org/10.1061/(ASCE)0733 mination of lower limits for irrigation management using in situ assess-
-9437(2008)134:3(286). ments of apparent crop water uptake made with volumetric soil water
Raes, D., P. Steduto, T. C. Hsiao, and E. Fereres. 2009. “Aquacrop—The content sensors.” Agric. Water Manage. 92 (1): 13–28. https://doi.org
FAO crop model to simulate yield response to water: II. Main algorithms /10.1016/j.agwat.2007.04.009.
and software description.” Agron. J. 101 (3): 438–447. https://doi.org/10 Thompson, R., M. Gallardo, L. Valdez, and M. Fernández. 2007b. “Using
.2134/agronj2008.0140s. plant water status to define threshold values for irrigation management
Rosa, R. D., P. Paredes, G. C. Rodrigues, I. Alves, R. M. Fernando, L. S. of vegetable crops using soil moisture sensors.” Agric. Water Manage.
Pereira, and R. G. Allen. 2012a. “Implementing the dual crop coeffi- 88 (1): 147–158. https://doi.org/10.1016/j.agwat.2006.10.007.
cient approach in interactive software. 1. Background and computa- Thorp, K. R., D. J. Hunsaker, K. F. Bronson, P. Andrade-Sanchez, and
tional strategy.” Agric. Water Manage. 103 (Jan): 8–24. https://doi E. M. Barnes. 2017. “Cotton irrigation scheduling using a crop growth
.org/10.1016/j.agwat.2011.10.013. model and FAO-56 methods: Field and simulation studies.” Trans.
Rosa, R. D., P. Paredes, G. C. Rodrigues, R. M. Fernando, I. Alves, L. S. ASABE 60 (6): 2023–2039. https://doi.org/10.13031/trans.12323.
Pereira, and R. G. Allen. 2012b. “Implementing the dual crop coeffi- Thysen, I., and N. K. Detlefsen. 2006. “Online decision support for irriga-
cient approach in interactive software: 2. Model testing.” Agric. Water tion for farmers.” Agric. Water Manage. 86 (3): 269–276. https://doi.org
Manage. 103 (Jan): 62–77. https://doi.org/10.1016/j.agwat.2011.10 /10.1016/j.agwat.2006.05.016.
.018. Tilling, A. K., G. J. O’Leary, J. G. Ferwerda, S. D. Jones, G. J. Fitzgerald,
Saleem, S. K., D. K. Delgoda, S. K. Ooi, K. B. Dassanayake, L. Liu, D. Rodriguez, and R. Belford. 2007. “Remote sensing of nitrogen and
M. N. Halgamuge, and H. Malano. 2013. “Model predictive control water stress in wheat.” Field Crop. Res. 104 (1): 77–85. https://doi.org
for real-time irrigation scheduling.” IFAC Proc. Volumes 46 (18): /10.1016/j.fcr.2007.03.023.
299–304. https://doi.org/10.3182/20130828-2-SF-3019.00062. Tsang, S., and C. Jim. 2016. “Applying artificial intelligence modeling to
Saseendran, S., L. Ahuja, L. Ma, D. Nielsen, T. Trout, A. Andales, J. optimize green roof irrigation.” Energy Build. 127 (Sep): 360–369.
Chávez, and J. Ham. 2014. “Enhancing the water stress factors for https://doi.org/10.1016/j.enbuild.2016.06.005.
sessment of a smartphone application for irrigation scheduling in cot- Wanjura, D., D. Upchurch, and J. Mahan. 2004. “Establishing differential
irrigation levels using temperature-time thresholds.” Appl. Eng. Agric.
ton.” Comput. Electron. Agric. 127 (Sep): 249–259. https://doi.org/10
20 (2): 201. https://doi.org/10.13031/2013.15892.
.1016/j.compag.2016.06.021.
Wanjura, D., D. Upchurch, and J. Mahan. 2006. “Behavior of temperature-
Vellidis, G., V. Liakos, C. Perry, W. Porter, M. Tucker, S. Boyd, M.
based water stress indicators in biotic-controlled irrigation.” Irrig. Sci.
Huffman, and B. Robertson. 2016b. “Irrigation scheduling for cotton
24 (4): 223–232. https://doi.org/10.1007/s00271-005-0021-9.
using soil moisture sensors, smartphone apps, and traditional methods.”
World Bank Group. 2019. “World bank open data.” Accessed August 23,
In Vol. 16779 of Proc., 2016 Beltwide Cotton Conf., 772–780. New
2019. https://data.worldbank.org.
Orleans: National Cotton Council. Xu, J., W. Bai, Y. Li, H. Wang, S. Yang, and Z. Wei. 2019. “Modeling rice
Vellidis, G., V. Liakos, W. Porter, M. Tucker, and X. Liang. 2016c. A dy- development and field water balance using AquaCrop model under
namic variable rate irrigation control system. St. Louis: Academic drying-wetting cycle condition in eastern China.” Agric. Water Manage.
Publishers. 213 (Mar): 289–297. https://doi.org/10.1016/j.agwat.2018.10.028.
Vellidis, G., V. Liakos, M. Tucker, C. Perry, J. Andreis, C. Fraisse, and K. Yang, G., L. Liu, P. Guo, and M. Li. 2017. “A flexible decision support
Migliaccio. 2015. “A smartphone app for precision irrigation schedul- system for irrigation scheduling in an irrigation district in China.” Agric.
ing in cotton.” In Precision agriculture’15, 701–708. Wageningen, Water Manage. 179 (Jan): 378–389. https://doi.org/10.1016/j.agwat
Nertherlands: Wageningen Academic Publishers. .2016.07.019.
Vellidis, G., M. Tucker, C. Perry, C. Kvien, and C. Bednarz. 2008. “A real- Yuan, G., Y. Luo, X. Sun, and D. Tang. 2004. “Evaluation of a crop water
time wireless smart sensor array for scheduling irrigation.” Comput. stress index for detecting water stress in winter wheat in the North
Electron. Agric. 61 (1): 44–50. https://doi.org/10.1016/j.compag.2007 China plain.” Agric. Water Manage. 64 (1): 29–40. https://doi.org/10
.05.009. .1016/S0378-3774(03)00193-8.
Vellidis, G., M. Tucker, C. Perry, D. Reckford, C. Butts, H. Henry, V. Zotarelli, L., M. D. Dukes, J. M. S. Scholberg, K. Femminella, and R.
Liakos, R. Hill, and W. Edwards. 2013. “A soil moisture sensor-based Munoz-Carpena. 2010. “Irrigation scheduling for green bell peppers
variable rate irrigation scheduling system.” In Precision agriculture’13, using capacitance soil moisture sensors.” J. Irrig. Drain. Eng. 137 (2):
713–720. New York: Springer. 73–81. https://doi.org/10.1061/(ASCE)IR.1943-4774.0000281.