Enhanced Estimation of Reference Evapotranspiration (ET₀) through Multiple Linear Regression Approaches in Semi-Arid Regions of Madhya

Reference evapotranspiration (ET₀) at the Ashok Nagar station was estimated using multiple linear regression (MLR) techniques. A 30-year dataset (1994–2023) comprising climatological parameters such as maximum and minimum air temperatures, mean relative humidity, wind speed, and solar radiation for the Ashok Nagar district, Madhya Pradesh, India, was utilized. Observed ET₀ values, computed using the FAO-56 Penman–Monteith equation, served as the dependent variable for model development in SPSS v21. Three MLR models were formulated with different combinations of independent variables: Model-1 included maximum and minimum temperatures, Model-2 incorporated maximum and minimum temperatures along with solar radiation, and Model-3 combined maximum and minimum temperatures, wind speed, and relative humidity. Performance evaluation was conducted by comparing the models with the Penman–Monteith standard using statistical indices such as the correlation coefficient (R), coefficient of determination (R²), mean absolute error (MAE), and root mean square error (RMSE). Model-3 exhibited superior predictive performance, with R, R², MAE, and RMSE values of 0.976, 0.952, 0.366, and 0.477 for the calibration dataset (70%), and 0.978, 0.957, 0.449, and 0.584 for the validation dataset (30%). The strong correlation and low error metrics confirm that Model-3 is the most reliable approach for ET₀ estimation in the semi-arid conditions of this region.