Resistance optimization and applicability of the TSEB model in desert–oasis areas

Abstract

Evapotranspiration (ET) is crucial for the water and carbon cycles of ecosystems, and accurate characterization of ET processes is essential for understanding regional water cycle mechanisms and ecohydrological dynamics. However, limitations in observational data and conventional modeling frameworks continue to constrain the accuracy of ET estimation over complex desert–oasis surfaces. In this study, the Enhanced Spatial and Temporal Adaptive Reflectance Fusion Model (ESTARFM) was integrated with the Two-Source Energy Balance (TSEB) model to simulate ET at an 8-day temporal and 30 m spatial resolution in a typical desert–oasis region. The interpolated meteorological station data and MODIS-Landsat fusion data were employed as the model inputs, and an additional heat transfer resistance formulation (KB) combined with an improved canopy/soil resistance equation was incorporated to optimize the TSEB model. Model parameters were further refined using the Monte Carlo method, and the model was validated against eddy covariance observations. The results demonstrated that the introduction of additional heat transfer resistance and improved canopy/soil resistance equations substantially reduced the overestimation and underestimation of latent heat flux (LE) and sensible heat flux (H) in the original model. The TSEB-MH model produced the most accurate sensible heat flux estimates (RMSE = 22.18 W m², R² = 0.64), reducing RMSE by 49.9 %, 20.1 %, and 34.0 % relative to TSEB, TSEB-KB, and TSEB-CH, respectively (RMSE = 44.30/27.76/33.60 W m^2; R² = 0.49/0.51/0.60). Meanwhile, R² increased by 0.15, 0.13, and 0.04 compared with TSEB, TSEB-KB, and TSEB-CH, respectively. For latent heat flux, the simulation accuracies of the TSEB, TSEB-KB, TSEB-CH, and TSEB-MH models were RMSE = 62.36/53.13/53.32/54.8 W/m² and R² = 0.62/0.70/0.72/0.69, with the TSEB-CH model performing best. The comparison of multi-year cumulative ET from 2013 to 2022 estimated by the TSEB-CH and TSEB-MH models and two ET products indicated that the TSEB-MH and TSEB-CH models provided higher spatial resolution and accuracy in ET estimation, particularly in arid and topographically complex areas. This study enhances the applicability of the TSEB model for energy flux estimation across desert–oasis surfaces and provides technical support for desertification control, ecological restoration, and water resource management in arid regions.