Assessing the impacts of uncertainties from satellite Leaf Area Index on water use monitoring in California vineyards

Satellite-estimated Evapotranspiration (ET) directly quantifies plant water consumption and supports irrigation scheduling, a critical part of water conservation and agricultural sustainability in California where challenges from increasingly severe droughts are ever present. Many remote-sensing-based ET models use satellite-derived Leaf Area Index (LAI) as a key input to control energy and water flux partitioning between soil/substrate and plant canopy. However, LAI estimation from satellites is highly uncertain, while its impact on ET modeling and the partitioning between plant transpiration (T) and soil evaporation (E) remains poorly understood. Here, we evaluated satellite LAI estimation methods with ground measurements in four California vineyards and assessed the impacts of LAI uncertainties on ET simulation and E/T partitioning in the thermal-based Two-Source Energy Balance (TSEB) model. We found that the Sentinel-2 Level-2 Prototype Processor and two Landsat-based MODIS-consistent LAI algorithms predicted low to medium LAI well but underestimated high LAI by up to 50% (RMSE ~ 1 1.3). Regression models trained with ground measurements substantially reduced LAI errors (RMSE ~ 0.3 0.6). TSEB ET was more sensitive to positive biases in LAI where a +50% error in LAI resulted in up to 50% deviation in ET, but a -50% error in LAI led to only a 10% change in ET. However, even when ET changes were minimal under negative LAI biases, the impact on E and T was significant. A 50% reduction in LAI caused up to 50% deviations in E and T due to a divergent response of soil and plant water loss to LAI. These findings call for careful consideration of LAI uncertainties in ET modeling and particularly in E/T partitioning.