TY - JOUR
T1 - Evapotranspiration estimates from an energy-water-balance model calibrated on satellite land surface temperature over the Heihe basin
AU - Paciolla, Nicola
AU - Corbari, Chiara
AU - Hu, Guangcheng
AU - Zheng, Chaolei
AU - Menenti, Massimo
AU - Jia, Li
AU - Mancini, Marco
PY - 2021
Y1 - 2021
N2 - A distributed hydrological energy-water-balance model (FEST-EWB) is calibrated over the Heihe Basin, a mainly desertic basin in China, employing remotely-sensed Land Surface Temperature (LST) (MODIS, 1-km resolution) as calibration variable. This approach overcomes the problem of model parameters characterization, which are usually difficult to define especially over large basins, allowing a pixel-by-pixel calibration, preserving the spatial heterogeneity. Hence, the spatial distribution of the modelled LST, but also of soil moisture (SM) and evapotranspiration (ET) is improved. The accuracy of the calibration process is documented through common statistical indexes. The modelled ET is compared locally against two eddy covariance stations in the agricultural area, while distributively against the ET estimates of the ETMonitor model and some global re-analysis products (ERA-Interim, GLDAS2, GLEAM and MERRA-2). Calibration and validation performed in this study prove that a considerable model accuracy is attainable even in extremely arid environments. An average LST bias of 2.6 °C is obtained over the basin. A good adaptation of FEST-EWB is also obtained against eddy-covariance stations ET with a little bias around −1 mm/d. On the other hand, the reanalysis products display a much worse performance, with higher absolute biases (around −3.5 mm/d), although with high variability among the models.
AB - A distributed hydrological energy-water-balance model (FEST-EWB) is calibrated over the Heihe Basin, a mainly desertic basin in China, employing remotely-sensed Land Surface Temperature (LST) (MODIS, 1-km resolution) as calibration variable. This approach overcomes the problem of model parameters characterization, which are usually difficult to define especially over large basins, allowing a pixel-by-pixel calibration, preserving the spatial heterogeneity. Hence, the spatial distribution of the modelled LST, but also of soil moisture (SM) and evapotranspiration (ET) is improved. The accuracy of the calibration process is documented through common statistical indexes. The modelled ET is compared locally against two eddy covariance stations in the agricultural area, while distributively against the ET estimates of the ETMonitor model and some global re-analysis products (ERA-Interim, GLDAS2, GLEAM and MERRA-2). Calibration and validation performed in this study prove that a considerable model accuracy is attainable even in extremely arid environments. An average LST bias of 2.6 °C is obtained over the basin. A good adaptation of FEST-EWB is also obtained against eddy-covariance stations ET with a little bias around −1 mm/d. On the other hand, the reanalysis products display a much worse performance, with higher absolute biases (around −3.5 mm/d), although with high variability among the models.
KW - Distributed calibration
KW - Eddy covariance stations
KW - Evapotranspiration
KW - Land surface temperature
KW - Meteorological reanalysis
KW - Remote sensing
UR - http://www.scopus.com/inward/record.url?scp=85100766746&partnerID=8YFLogxK
U2 - 10.1016/j.jaridenv.2021.104466
DO - 10.1016/j.jaridenv.2021.104466
M3 - Article
AN - SCOPUS:85100766746
SN - 0140-1963
VL - 188
SP - 1
EP - 13
JO - Journal of Arid Environments
JF - Journal of Arid Environments
M1 - 104466
ER -