A Novel Remote Sensing Data–Driven One-Step Framework for Large-Scale Actual Evapotranspiration Estimation

Accurate estimation of actual evapotranspiration (ETa) is crucial for sustainable water management and efficient irrigation planning. This study presents a novel remote sensing data–driven one-step simulation–optimization framework designed for large-scale, multi-crop ETa estimation. Unlike conventional methods that rely on linear regression or iterative calibration for single crops, this framework simultaneously optimizes model parameters in a single step, capturing their nonlinear dynamics effectively. Daily ETa was estimated using remote sensing inputs, including NDVI, land surface temperature (LST), surface albedo (αs), alongside available meteorological data. The model was tested under two scenarios with different training and testing datasets. Scenario I, with limited data, achieved test R² = 0.8195 and all-data R² = 0.8152, closely matching Scenario II (test R² = 0.8357, all-data R² = 0.8162), demonstrating that high accuracy can be maintained even with reduced data availability. Optimal parameters were consistent across scenarios (a ≈ 0.315, b ≈ −0.0012), and seasonal performance remained stable. These results highlight the framework’s robustness, generalization capability, and practical applicability for ETa estimation in heterogeneous landscapes, emphasizing the power of remote sensing data–driven optimization.