Estimating evapotranspiration using remotely sensed solar-induced fluorescence measurements

Terrestrial evapotranspiration (ET) is a critical component of the land surface and its accurate estimation is crucial in understanding the global water and energy budgets. Current large-scale ET estimation models depend on remotely sensed data products of meteorological conditions and vegetation phenology, but high uncertainty remains to correctly represent ecological factors, which are regulated by biochemistry. In recent years, solar-induced fluorescence (SIF), a proxy for photosynthesis, has been extensively used in different ecological research and is conceived to be of great potential for ET estimation to constrain the transpiration flux, which is the significant component of ET. So far, most SIF-based ET estimation methods are achieved via empirical methods, which lack a solid physical foundation. In this work, we estimated ET by combining SIF and meteorological variables with Fick's law and an optimal stomatal behavior model and validated the model across different eddy-covariance flux stations around the globe. The model showed overall good performance across different ecosystems and when compared with other remote sensing-based ET models, such as the Priestley Taylor-JPL (PT-JPL) model, the Moderate Resolution Imaging Spectroradiometer (MODIS)-ET products, and a simple empirical linear SIF-ET model, the RMSE of estimated ET of the model proposed in this paper demonstrated better performance. The model we proposed here can be potentially extended for a reliable ET estimation at a global scale.