Paddy field irrigation forecast based on a satellite data driven energy water balance model

This study proposes a satellite-based irrigation forecast approach for improve irrigation water management in paddy fields. This method integrates remotely sensed data with a distributed hydrological energy water balance model (FEST–EWB–paddy), supplemented by weather forecasts, to accurately forecast irrigation requirements. This model develops different structural combinations of water and energy balance equations on the basis of relative changes between field water status and crop growth. The land surface temperature (LST) simulated by the calibrated model achieved an absolute mean bias error (AMBE) of 1.80 °C and a root mean square error (RMSE) of 2.06 °C compared with satellite-based LST data. The average AMBE and RMSE between the simulated and observed water levels were 18 mm and 21 mm, respectively. Weather forecast accuracy generally declined with longer forecast periods but remained within acceptable ranges for irrigation forecasting. When an alternate wetting and drying (AWD) irrigation strategy was applied, compared with traditional practices, the FEST–EWB–paddy model maintained a relatively low water level in paddy fields, resulting in a 36 % reduction in irrigation water use and a 14 % decrease in irrigation frequency. The average values of the irrigation forecast depth error, timing error, and count error were 12 mm, 0.18, and 0.41, respectively. Most of the forecast inaccuracies occurred after Day 3 of the forecast horizon, while the highest error rate was observed on Day 5. Errors in weather forecasts had a minimal effect on irrigation forecasts accuracy, underscoring the model’s reliability in practical applications.