Sensitivity of METRIC-based tree crop evapotranspiration estimation to meteorology, land surface parameters and domain size
Introduction
Evapotranspiration (ET) is an important component of the hydrological cycle and surface energy balance that links water, energy, and carbon cycles (Campbell and Norman, 1998, Jung et al., 2010, Zhang et al., 2019). It is closely related to crop water use in the context of agricultural irrigation management (Rana et al., 2005, Williams and Ayars, 2005, French et al., 2015, He et al., 2017, Xue et al., 2020). Accurate estimation of crop ET can improve crop water use efficiency and rationally allocate water resources (Qiu et al., 2019, Han et al., 2021, Wang et al., 2021), especially for water scarce regions. Compared with ground-based methods to measure ET, such as eddy covariance (Twine et al., 2000, Gu et al., 2012), lysimeters (Allen et al., 1991), and surface renewal techniques (McElrone et al., 2013, Shapland et al., 2014), remote sensing-based ET models have a great advantage of estimating and mapping ET in the temporal and spatial patterns. It would be providing a scientific guidance for improving water resource management.
Mapping evapotranspiration at high resolution with internalized calibration (METRIC) was proposed by Allen et al. (2007a) based on the theoretical and computational basis of surface energy balance algorithms for land (SEBAL, Bastiaanssen et al., 1998) for estimating ET by incorporating satellite observations and a few of high quality weather data. METRIC model has its significant advantage over SEBAL in that using the weather station-based reference ET (ETr) to serve as a real ground reference on the satellite image-based actual ET (Allen et al., 2007a). The ETr was used to build surface energy balance conditions at the cold and hot pixels for a linear relationship between land surface temperature (LST) and near-surface temperature difference (dT) in sensible heat flux (H) estimation. Also, METRIC model applies fraction of reference ET (ETrF) as a real-time crop coefficient for extrapolating from instantaneous ET (satellite overpassing time) to daily ET (satellite overpassing date) (Allen et al., 2007a, Allen et al., 2011). Contrasting with other remote sensing ET models, such as the LST-based Atmosphere Land Exchange Inverse (ALEXI) model at continental scales (Anderson et al., 2007) and Simplified Surface Energy Balance (SSEB) model with MODIS images (Senay et al., 2007), METRIC model was initially designed for regions smaller than a few hundred kilometers in scale and produced high accurate ET maps at high spatial resolution (Allen et al., 2007a). For the relatively complex terrain and weather conditions, it was suggested to divide the satellite image into several parts and to calculate ET separately since the weather of study area need to be relatively uniform for cold and hot pixels selection in the METRIC model assumes (Allen et al., 2014).
The METRIC model is a feasible tool for estimating and mapping ET in temporal-spatial variability. It has been widely used for water consumption from natural vegetation to agricultural crops, determination of crop coefficients, irrigation schedules, and water rights accounting in Idaho, California, and New Mexico (Allen et al., 2007b). Case studies concluded that errors for seasonal ET estimates were less than 4% in the growing seasons and errors for daily ET estimates were about 13%−20% for irrigated meadow and sugar beet crop (Allen et al., 2007b). French et al. (2015) found METRIC model was used to provide good estimates of daily ET over a cotton study site, and they suggested METRIC model was preferred when sufficient ancillary data were not available. He et al. (2017) showed that monthly ET estimates had a 9.68% relative mean difference between METRIC-based versus observed ET over an almond orchard. Ortega-Salazar et al. (2021) found that the estimated errors within the METRIC model for daily ET estimates were only between 4% and 6% of the measurements and the root mean square error (RMSE) was 0.42 mm/day for a drip-irrigated olive orchard.
Sensitivity analysis is very useful to identify the sensitive variables in a model and the suitable study size for improving model accuracy (Zhang et al., 2017, Long et al., 2011). Valipour et al. (2020) analyzed the effect of meteorological variables on the reference ET (ETo). They found that wind speed was the most important variable driving the trends of ETo in most regions. Anwar et al. (2021) examined the sensitivity of the simulated summer potential ET (PET) of tropical Africa to LAI and vegetation-runoff systems using the RegCM4 model. Other studies focused on the sensitivity of actual ET estimates to bias in calibration and LST of METRIC model (Morton et al., 2013, Kilic et al., 2016), selection of cold and hot pixels of the triangle, SEBAL and METRIC models (Long and Singh, 2013), and meteorological factors of Penman method at a region scale (Yang et al., 2019; Biazar et al., 2019). Allen et al. (2013) performed a sensitivity analysis on METRIC model in mountainous areas. They showed that increasing aerodynamic roughness increased convective heat transfer and decreased ET estimation. French et al. (2015) revealed that the sensitivity of METRIC model was related to the season. The relatively great sensitivity showed in the early growth of cotton at a fine spatial resolution.
However, tree crops have complex canopy structure and physiology which cause the energy exchange processes more complicated (Jin et al., 2018). The sensitivity analysis of tree crop ET has received less attention compared to agricultural row crops. We are aware of no studies focusing on the quantifying errors of METRIC-based daily ET estimates over orchards during the entire growth period. Also, it is not clear how different study sizes impact on daily ET estimates in the METRIC model. In this study, we performed a comprehensive sensitivity analysis of tree crop daily ET estimates by using METRIC model on two types of commercial orchards throughout the growth period. The objective was to quantify the sensitivity of daily ET estimates to errors in various meteorological variables and land surface parameters. In addition, appropriate study size could improve the estimated accuracy and it is crucial for correctly selecting cold and hot pixels in the METRIC model. Therefore, we also investigated the impact of different study sizes on daily ET estimates of an almond and three pistachio orchards in the San Joaquin Valley, California, USA.
Section snippets
Study area
This study focused on two types of commercial mature orchards located in south San Joaquin Valley, California, US (Fig. 1), including one almond orchard (He et al., 2017) and three pistachio orchards (Jin et al., 2018). The study site has a typical Meditterranean climate of a hot and dry summer and a mild and wet winter. The almond trees were planted in 1999 with a well-watered management of drip and fanjet during the period 2009–2012. The canopy structure was homogeneous with the summer NDVI
Sensitivity analysis to meteorological variables
The METRIC-based daily ET estimates of both almond and pistachio trees were more sensitive to instantaneous ETo compared to Ta, VP and WS throughout the entire growing season (Fig. 2). Results indicated that an error of 5% in instantaneous ETo can result in a mean increase of 6.80 ± 4.41% (~0.31 ± 0.21 mm/day) in the daily ET of almond trees, corresponding to the changes of 36.39 ± 10.84 W/m2 in H estimates and 0.05 ± 0.03 in EToF. The change of daily ET in the pistachio sites was smaller than
Meteorological variables
It is worth noting that instantaneous ETo was a key variable in the METRIC model since it was used to internally calibrate to compute H and reference ET fraction (EToF) which is similar to the crop coefficient. When errors varying from − 50 to 50% were added to instantaneous ETo, H estimates decreased but EToF increased, thereby resulting in an increase in daily ET of both almond and pistachio trees. In the Fig. 3, the shape of the curves for almond trees was different from that for pistachio
Conclusions
In this study, we quantified the sensitivity of METRIC-based daily ET estimates of almond and pistachio trees to weather variables and land surface parameters, with the introduced errors up to ± 50% (∆=5%) or ± 5 K (∆=0.5 K). We also analyzed the influence of different study sizes on daily ET estimates throughout the entire growing season.
For the sensitivity to meteorological variables, our results showed that instantaneous ETo played an important role in determining METRIC-based daily ET
Declaration of Competing Interest
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.
Acknowledgements
This work is supported in part by the National Natural Science Foundation of China under Grant 41971300 and Grant 61901278; in part by the Key Project of Department of Education of Guangdong Province under Grant 2020ZDZX3045; in part by the Guangdong Basic and Applied Basic Research Foundation under Grant 2022A1515011290; in part by the Natural Science Foundation of Guangdong Province under Grant 2021A1515011413; in part by Shenzhen Scientific Research and Development Funding Program under
References (43)
- et al.
A remote sensing surface energy balance algorithm for land (SEBAL): 1. Formulation
J. Hydrol.
(1998) - et al.
Estimation of the netradiation using MODIS (Moderate Resolution Imaging Spectroradiometer) data forclear sky days
Remote Sens. Environ.
(2005) - et al.
Estimation of net radiationfrom the MODIS data under all sky conditions: Southern Great Plains case study
Remote Sens. Environ.
(2010) - et al.
Remote sensing of evapotranspiration over cotton using the TSEB and METRIC energy balance models
Remote Sens. Environ.
(2015) - et al.
The fundamental equation of eddy covariance and its application in flux measurements
Agr. For. Meteorol.
(2012) - et al.
Crop evapotranspiration prediction by considering dynamic change of crop coefficient and the precipitation effect in back-propagation neural network model
J. Hydrol.
(2021) - et al.
Spatially variable evapotranspiration over salt affected pistachio orchards analyzed with satellite remote sensing estimates
Agr. For. Meteorol.
(2018) - et al.
Sensitivity of evapotranspiration retrievals from the METRIC processing algorithm to improved radiometric resolution of Landsat 8 thermal data and to calibration bias in Landsat 7 and 8 surface temperature
Remote Sens. Environ.
(2016) - et al.
Performance of the METRICmodel for mapping energy balance components and actual evapotranspiration overa superintensive drip-irrigated olive orchard
Agr.Water Manag.
(2021) - et al.
Satellite-based evapotranspiration of a super-intensive olive orchard: application of METRIC algorithms
Biosyst. Eng.
(2014)
Evapotranspiration estimation using a modified Priestley-Taylor model in a rice-wheat rotation system
Agr. Water Manag.
Measurement and modelling of evapotranspiration of irrigated citrus orchard under Mediterranean conditions
Agr. For. Meteorol.
Modelling surface energy fluxes over maize using a two-source patch model and radiometric soil and canopy temperature observations
Remote Sens. Environ.
Thermocouple frequency response compensation leads to convergence of the surface renewal alpha calibration
Agr. For. Meteorol.
Correcting eddy-covariance flux underestimates over a grassland
Agr. For. Meteorol.
Evaluation of crop coefficient and evapotranspiration data for sugar beets from landsat surface reflectances using micrometeorological measurements and weighing lysimetry
Agr. Water Manag.
Grapevine water use and the crop coefficient are linear functions of the shaded area measured beneath the canopy
Agr. For. Meteorol.
Evaluation of remote sensing-based evapotranspiration models against surface renewal in almonds, tomatoes and maize
Agr. Water Manag.
Sensitivity of potential evapotranspiration to meteorological factors and their elevational gradients in the Qilian Mountains, northwestern China
J. Hydrol.
Lysimeters for Evapotranspiration and Environmental Measurements
Satellite-based energy balance for mapping evapotranspiration with internalized calibration (METRIC)—Model
J. Irrig. Drain. Eng.
Cited by (3)
Linking Transpiration to Reef Nitrogen Supply on a Tropical Coral Island
2024, Journal of Geophysical Research: Biogeosciences