Applicability evaluation of soil moisture constraint algorithms in remote sensing evapotranspiration models

Remotely sensed (RS) evapotranspiration (ET) models have been widely used to estimate ET over large areas. However, a common challenge for these models is the lack of reliable soil moisture (SM) constraints. Due to the lack of reliable, spatiotemporal continuous SM profile data, existing RS-ET models tend to use atmospheric or land surface variables as proxies for SM constraints. Although several proxy algorithms for SM constraints have been developed, few studies have evaluated their performance in ET simulations. To address this gap, we evaluated the applicability of five proxy algorithms for SM constraints (namely, fVPD, fLST, fDT, fZhang, and fdrying) to ET simulations in China using the Penman-Monteith-Leuning (PML) model. These algorithms were evaluated at 14 ChinaFlux sites and 286 basins using flux tower measurements and water balance-based ET estimates, respectively. The results show that among the five algorithms, fdrying performs test at the flux sites, with a median of Kling-Gupta efficiency (KGE) of 0.75. The second-best algorithm is fZhang (KGE = 0.73), followed by fDT (KGE = 0.70), fLST (KGE = 0.68), and fVPD (KGE = 0.65). The performance ranking of the five algorithms at the basin scale is consistent with that at the flux sites. Using the flux site SM measurements as a reference, we further found that the algorithms with better performance in ET simulations also have better SM simulation capabilities. This study highlights the importance of reliable SM constraints in the RS-ET models.

Automated summary

Remotely sensed evapotranspiration (ET) models are commonly used to estimate ET over large areas, but one challenge is the lack of reliable soil moisture (SM) constraints. In this study, five proxy algorithms for SM constraints were evaluated in China, and it was found that the fdrying algorithm performed the best at flux sites.