Measuring water stress in a wheat crop on a spatial scale using airborne thermal and multispectral imagery
Introduction
Understanding crop water stress is important for in-season management of irrigated and rain-fed crops. One method that has long been used in measuring water stress is crop surface temperature (Idso et al., 1981, Jackson et al., 1981, Gonzalez-Dugo et al., 2006). As crops transpire, water loss reduces leaf temperature through evaporative cooling (Allen et al., 1998). In well-watered situations, the leaf surface temperature often becomes much lower than that of the surrounding air. Conversely, water-stressed crops transpire less and crop surface temperature increases, typically rising above the surrounding air temperature (Jackson, 1982). The difference between the crop surface and air temperatures (Ts − Ta) is therefore a good indication of crop water stress.
The first conclusive demonstration that leaf temperature can be cooler than air temperature was given by Ehler (1973). Using thin wire thermocouples to measure leaf temperature, it became evident from his experiments that Ts − Ta was linearly related to vapour pressure deficit (VPD). The VPD is the difference between water vapour pressures of the surrounding air and saturated air at the same temperature. Idso et al. (1981) and Jackson et al. (1981) made use of Ehler's findings in developing a crop water stress index (CWSI) for canopies at full cover, a measure which later received wide acceptance. In their work, canopy temperature was measured by infrared thermometers (IRT) instead of thermocouples.
Distinction needs to be made between individual leaves and the leaves arranged in a crop canopy even though the mechanism of temperature differences is essentially the same (Idso et al., 1967). When an instrument like an IRT provides point measurements, the intent is generally to characterise a whole canopy, which usually comprises multiple point measurements. With the advent of thermal scanners and imagers and their use as hand-held, vehicle-driven, airborne or space-borne instruments, point or small sample information has largely been replaced by the more complete full-sample coverage of canopy temperature.
The theoretical approach to CWSI (Jackson et al., 1981) and its practical applications (Idso et al., 1981, Idso, 1982) require that baselines for non-stressed and fully stressed canopies are defined. Baselines are crop and/or variety specific in relation to Ts − Ta and VPD (Idso, 1982). The concept of the CWSI is confined to full canopies only and does not provide an immediate solution for incomplete cover where bare soil is included. In order to overcome this limitation, Moran et al. (1994) introduced the concept of a vegetation index versus temperature (VIT) trapezoid. This provides a blended surface temperature response comprising a crop and soil temperature mix. A trapezoidal shape is generally formed when Ts − Ta is plotted against vegetation cover, provided sufficient data points are available and adequate spatial variability exists in water status and crop cover. According to Moran et al. (1994) the water deficit index (WDI) of a point is its distance from cold front in proportion to the distance between cold and warm fronts. WDI for any full cover point is equivalent to Idso–Jackson's CWSI at a given VPD.
Clarke (1997) provided a pragmatic approach to the VIT trapezoid by reducing certain conditions on the definition of vertices proposed by Moran et al. (1994) but his proposed approach also relied on field information. The empirical approaches to VIT framework help implementation of the trapezoid models to some extent (Mendez-Barrosa et al., 2008). As the purpose of the VIT trapezoid is to determine the stressed part of the crop at a particular time, Clarke (1997) divided the trapezoid into sections and identified a region of the trapezoid representing a stressed crop.
A limitation of the CWSI is its focus on the acute crop water status and it does not necessarily consider chronic water stress conditions. Similarly, the trapezoid approach uses an instantaneous (acute) measure of crop temperature and, whilst accounting for incomplete cover assumes a linear relationship between soil temperature and crop temperature displacement as soil is covered or uncovered by the crop canopy. For perennial vegetation we assumed that any cover less than unity indicates a chronic stressed condition, but for annual crops this is not necessarily true and in some environments crops never fully cover the ground (O’Connell et al., 2004). We investigate and propose an annual crop version for the trapezoid where vegetative cover is relative to a potential cover predicted at that stage of development through a simple crop model. In this way, both chronic and acute water stress status can be assessed from one graph and appropriate management intervention implemented as needed.
Section snippets
Materials and methods
A field study was established on an experimental farm near Horsham, Victoria, Australia (36.74°S latitude, 142.10°E longitude, elevation 126 m). Crops of wheat (Triticum aestivum L., cv Chara) were sown in two consecutive years. The experiment was set up comprising 48 plots in 2005 and 16 plots in 2006 (Fig. 1). Each plot was of 12 m × 20 m in size. Half of the plots were irrigated and the remaining plots were rain-fed. The crop was sown on the 2nd of June in 2005 and on the 31st of May in 2006. The
Chronic and acute water stress differences
The modified trapezoid of Ts − Ta versus “1 - actual/potential” cover ratio is shown in Fig. 5, Fig. 6. It represents a wide range in crop growth stage from sampling four dates over 2 years (Table 3). It is noted that “potential” was less than “actual” in two cases referring to an early crop stage (21 July) and a late stage (21 November) in 2005. At both stages the overall crop cover itself was very low. Even though the difference between potential and actual was very small in both cases it is
Discussion
The extension of the acute water stress index method of Idso et al. (1981), Idso (1982) and subsequent variants by Moran et al. (1994) and Clarke (1997) to include a measure of chronic water stress progresses toward a unique non-destructive method of assessing crop water stress in a spatial and temporal context. This is particularly important when considering annual crops that do not normally attain full groundcover as is common in much of the semi-arid regions of the world. The spatial aspect
Conclusions
A modified trapezoid water stress model that accounts for chronic and acute water stress is proposed. The input required is largely the same as traditionally used for the existing trapezoid models for acute water stress, but additionally requires potential crop cover that can be derived readily by any suitable crop model. The model is applicable to both irrigated and dryland crops. Our combined stress index might also have application with respect to non-water stresses like nutrient deficiency
Acknowledgements
The authors gratefully acknowledge the financial support for this work through the Victorian Government's Our Rural Landscape initiative. We also thank an anonymous referee for helpful comments on an earlier draft.
References (22)
A soil-adjusted vegetation index (SAVI)
Remote Sens. Environ.
(1988)- et al.
Normalizing the stress-degree-day parameter for environmental variability
Agric. Meteorol.
(1981) Non-water-stressed baselines: a key to measuring and interpret ting plant water stress
Agric. Meteorol.
(1982)- et al.
Estimating crop water deficit using the relation between surface-air temperature and spectral vegetation index
Remote Sens. Environ.
(1994) - et al.
Interception of photosynthetically active radiation and radiation-use efficiency of wheat, field pea and mustard in a semi-arid environment
Field Crops Res.
(2004) - et al.
A simulation model of the wheat crop in response to water and nitrogen supply. I. Model construction
Agric. Syst.
(1996) - et al.
Crop Evapotranspiration
(1998) - Bureau of Meteorology, 2005. Automatic weather stations for agricultural and other applications....
An empirical approach for detecting crop water stress using multispectral airborne sensors
HortTechnology
(1997)- et al.
Estimating soil moisture under low frequency surface irrigation using crop water stress index
J. Irrig. Drain E-ASCE
(2003)
Water stress detection under high frequency sprinkler irrigation with water deficit index
J. Irrig. Drain E-ASCE
Cited by (21)
A robust model for diagnosing water stress of winter wheat by combining UAV multispectral and thermal remote sensing
2024, Agricultural Water ManagementModeling of winter wheat fAPAR by integrating Unmanned Aircraft Vehicle-based optical, structural and thermal measurement
2021, International Journal of Applied Earth Observation and GeoinformationCitation Excerpt :This is not surprising since the CT is not only controlled by biophysical and biochemical properties of plants (Elarab et al. 2015; Seligman et al. 1983; Szlovak et al. 1991) but also influenced by complicated field environment (Kustas and Norman 1999; Patel et al. 2001; Yang et al. 2014). The low accuracy of the CT-only model could be attributed to the fact that the spatial resolution of thermal measurement is relatively low, which could cause more noise in the CT feature in comparing to other types of UAV-based remote sensing features; similar results have been reported in previous similar studies (Abuzar et al. 2009; Maimaitijiang et al. 2017; Matese and Di Gennaro 2018). In our study, the developed model based on the fusion of optical, structural and temperature features obtained the best modeling accuracy (R2 = 0.907, NSE = 0.907, RMSE = 0.041, MAE = 0.034).
High-throughput field phenotyping in dry bean using small unmanned aerial vehicle based multispectral imagery
2018, Computers and Electronics in AgricultureCitation Excerpt :Polania et al. (2016) using hand-held infrared thermometer observed a correlation coefficient of 0.38 between canopy temperature depression and seed yield for a small set of 24 dry bean lines tested under drought conditions. Findings from literature (Moran et al., 1994; Clarke, 1997; Tilling et al., 2007; Abuzar et al., 2009; Polania et al., 2016) combined with ours warrant further investigation of the reliability and applicability of canopy temperature data for determining drought stress response among dry bean breeding materials. New techniques (Deery et al., 2016) applied towards thermal data extraction and interpretation such as pixel frequency distribution may contribute to better high-throughput canopy temperature data for determining potential use in plant breeding.
Water–Nitrogen Colimitation in Grain Crops
2018, Advances in AgronomyCitation Excerpt :Canopy temperature can also be used for monitoring farmers’ paddocks, but this application is not widespread (Jones et al., 2009; Munns et al., 2010). Abuzar et al. (2009) combined thermal imagery and crop cover to account for chronic and acute water stress in wheat. The crop water stress index (CWSI) defined in the early 1980s (Idso, 1982; Jackson et al., 1977, 1981) can be derived from high-resolution aerial thermal imagery, and was related to stomatal conductance of olive trees (Egea et al., 2017) and water content of sugarcane (Veysi et al., 2017).
Spectral monitoring of wheat canopy under uncontrolled conditions for decision making purposes
2016, Computers and Electronics in AgricultureCitation Excerpt :It integrates three CCD with spectral filters, which record in three broad bands of the electromagnetic spectrum; Green in the range 500–600 nm with the peak at 550 nm, red in the range 600–700 nm and the peak at 650 nm, and near-infrared in 750–900 nm with the peak at 830 nm. The MS3100 camera has already been used successfully in agricultural research (Abuzar et al., 2009; Backoulou et al., 2011; Kise and Zhang, 2008). In this case, it was installed on a mobile ground platform (Figs. 1b and 2a) to facilitate picture taking.
A wavelet-artificial intelligence fusion approach (WAIFA) for blending Landsat and MODIS surface temperature
2015, Remote Sensing of EnvironmentCitation Excerpt :Land surface temperature (LST) is one of the main biophysical variables derived from remotely sensed imagery (Anderson et al., 2008). It is widely used for environmental studies such as assessing crop growth (Abuzar, O'Leary, & Fitzgerald, 2009), monitoring vegetation health (Karnieli et al., 2006), agricultural studies (Badeck et al., 2004; Rojas, Vrieling, & Rembold, 2011), estimation of surface energy flux (Muramatsu, Nakayama, & Kaihotsu, 2006), forest burnt area estimation (Ichoku et al., 2003), monitoring evapotranspiration (Anderson, Allen, Morse, & Kustas, 2012; Yang & Wang, 2011) water stress in croplands (Dragutin & Eitzinger, 2007), energy balance (GriendV.D. & Owe, 1993), drought (Karnieli et al., 2010; Sobrino, Gomez, Munoz, & Olioso, 2007) and soil moisture (Hulley, Hook, & Baldridge, 2010). Acquiring satellite images with high temporal and spatial resolutions remains extremely difficult due to satellite technical constrains.