Environment & Ecosystem Science (EES)

Accurate estimation of reference evapotranspiration (ETo) is essential for determining crop water requirements and optimizing irrigation management. While the FAO-56 Penman-Monteith (FAO-56 PM) method remains the global standard for ETo estimation, simpler alternative methods persist due to their lower data requirements. This study developed and evaluated multiple linear regression models for ETo prediction across three stations in agriculture-dominated northwestern Bangladesh (Rajshahi, Bogra, and Rangpur) using meteorological data from 1979 to 2022. The analysis revealed solar radiation as the strongest predictor of ETo (Kendall’s τ > 0.74), followed by maximum temperature (τ > 0.68), while sunshine hours and relative humidity showed weaker correlations (τ < 0.1). Among the three developed models, Model 3—incorporating average temperature, relative humidity, solar radiation, and wind speed—achieved superior performance (R² > 0.97, RMSE < 0.024) compared to the other models (R² > 0.93, RMSE < 0.07). These robust regression models provide valuable tools for agricultural water management in the study regions, enabling precise irrigation scheduling, improved yield forecasting, and climate-resilient policy development.