Abstract:

Evapotranspiration is an important component of the water cycle, influencing crop growth and grain production. Obtaining long-term, high-resolution evapotranspiration data can help optimize regional water resource allocation. Using Landsat and MODIS remote sensing data, along with ground observation flux data and meteorological data, the SEBS model and ESTARFM model are utilized to derive high spatiotemporal resolution evapotranspiration estimates for the southern red soil area from April to October 2019. The spatiotemporal variation characteristics and influencing factors of the high resolution evapotranspiration are analyzed. The results show that high spatial resolution remote sensing images can achieve higher accuracy in simulating remote sensing-based evapotranspiration. The simulation performance of the SEBS model driven by Landsat remote sensing data is better than that of the SEBS model based on MODIS data. Comparison between the high spatiotemporal resolution evapotranspiration obtained by the ESTARFM model and the measured evapotranspiration using a large aperture scintillometer shows a RMSE of 0.68mm·d⁻¹ and R² of 0.87. The simulated high spatiotemporal resolution daily evapotranspiration from April to October 2019 is spatially correlated with land use types, with evapotranspiration rates as follows: forest > paddy field > other agricultural land. Temporally, evapotranspiration showed an increasing trend from April to August, followed by a gradual decrease from August to October. Temperature is found to be the main meteorological factor affecting evapotranspiration in the study area. The high spatiotemporal resolution evapotranspiration obtained using the SEBS and ESTARFM models shows good agreement with ground measurements. The combination of the SEBS and ESTARFM models can serve as an effective tool for estimating evapotranspiration in southern red soil regions.

Key words: Evapotranspiration, SEBS, ESTARFM, Remote sensing, Spatiotemporal fusion technology