Improved Evapotranspiration Estimation in Coarse-Textured Soils Using a Comprehensive Water Balance Model

Advances in soil moisture monitoring techniques and sensor networks have made data assimilation a powerful approach for estimating evapotranspiration (ET). The commonly used simple water balance (SWB) model provides reliable ET estimates within data assimilation frameworks; however, it often neglects the influence of ET on vertical fluxes. While this assumption may be valid during drying periods in low-drainage soils, it is less appropriate for soils with high hydraulic conductivity. This study introduces a comprehensive water balance (CWB) model that explicitly accounts for ET-driven percolation. The model captures the effect of ET on vertical fluxes by comparing soil water depletion with and without ET, thereby highlighting the role of root water uptake (RWU) in controlling percolation. The CWB model, coupled with an ensemble Kalman filter, predicts daily ET using soil moisture sensor data across different soil types. Within this framework, RWU is treated as the observable variable for state updating. Model performance was evaluated against the conventional SWB model under varying drainage conditions. For loamy sand, the CWB model was independently validated using weighing lysimeter measurements. Results demonstrate that the CWB model outperforms the SWB model, particularly in coarse-textured soils, reducing ET estimation error by up to 45% and achieving higher accuracy (NSE = 0.918 vs. 0.727). Sensitivity analyses incorporating sensor uncertainty show that fine-textured soils exhibit lower sensitivity to measurement errors, resulting in more robust ET estimates. These findings underscore the importance of incorporating vertical flux effects to avoid ET underestimation, especially in highly permeable soils. 

Keywords: Water balance model, vertical flux, evapotranspiration, Ensemble Kalman filter, root water uptake.