Remote sensing based continuous estimation of regional evapotranspiration by improved SEBS model

Abstract

Remote sensing (RS) has been considered as the most promising tool for evapotranspiration (ET) estimation at regional scale. However, large errors implied in the process of extrapolating instantaneous latent heat flux derived at satellite over-passing time to daily ET inevitably constrains the application of RS models. In this study, we modified Surface Energy Balance System (SEBS) model by replacing the instantaneous inputs with daily representative parameters to estimate daily ET directly. A further strategy was added to the model for estimating ET during cloud-contaminate period using moving window averaged Bowen ratio. One merit of the improved model is that the calculation of daily ET can be avoided by means of instantaneous input from ground observations is avoided, which is insufficient at regional scale from meteorological stations. The second merit is the model circumvents the scaling up process implied in the traditional methods. Another merit is that the cloud-free constrain of ET estimation based on RS data is circumvented through a gap filling approach, which makes continuous ET estimation possible. For the purpose of model performance evaluation, the model was tested at the Weishan flux site in the North China Plain from 2006 to 2007. Two-year continuous simulation results show that the model has a good performance for daily ET estimation with a deterministic coefficient of 0.61 and a bias of 3%. Then the model was applied to the 5711 km² Weishan Irrigation District at 1-km spatial resolution.