Comparison of evapotranspiration upscaling methods from instantaneous to daytime scale for tea and wheat in southeast China

Abstract

Accurately converting instantaneous evapotranspiration (λETi) at satellite over-passing time into daily evapotranspiration (λETd) is a key issue of applying remotely sensed data to estimate regional evapotranspiration (λET) from remote sensing satellites, which plays an important role for effective water resource management. The scaling methods that take advantage of the relationship between λET and other environmental factors and can be used to convert λETi into λETd. In this study, five scaling methods of converting λETi into λETd, including the evaporative fraction method (Eva-f method), revised evaporative fraction method (R-Eva-f method), crop coefficient method (Kc-ET0 method), revised crop coefficient method (R-Kc-ET0 method) and direct canopy resistance method (Direct-rc method), were evaluated based on the detailed meteorological data measured from 2016 to 2018 in a tea field and 2018 to 2020 in a wheat field in southeast China. The estimated λETd was compared with the measured λETd by the Bowen ratio energy balance (BREB) method. The results indicated that the Eva-f and R-Eva-f methods with the mean root mean square error (RMSE) and coefficient of efficiency (ε) equaled 9.02 W m−2 and 0.92; 12.06 W m−2 and 0.89, respectively, were superior to the Kc-ET0, R-Kc-ET0 and Direct-rc method; the Kc-ET0 method with the mean RMSE and ε equaled 20.62 W m−2 and 0.79 was also a good option for simulating the λETd of tea and wheat; while the R-Kc-ET0 method simulated the λETd well for the wheat with mean RMSE and ε equaled 36.29 W m−2 and 0.71, but significantly overestimated the tea λETd with the mean RMSE and ε values of 39.61 W m−2 and 0.59 for tea; the Direct-rc method overestimated λETd of tea and wheat for the most of intervals with the mean RMSE and ε of 39.58 W m−2 and 0.62, and was not recommended to use in the present study areas.