Predicting Evapotranspiration in the Semi-Arid Region of Indore Using AI Models

Indore is facing increasing water scarcity and inefficient irrigation is a major contributor to it. Hence, this research explores the use of Artificial Intelligence models like Artificial Neural Networks (ANNs) and Light Gradient Boosting Machine (LGBM) in predicting reference evapotranspiration (ET0) using limited and sufficient data for Indore’s semi-arid climate. In places where water is scarce, accurate prediction of ET0 plays a vital role in efficient irrigation planning. Based on historical meteorological data from 1985 to 2022, the models were trained and tested, with ANN generally outperforming LGBM especially when an extensive set of input variables was used. Furthermore, wind speed and net radiation were found to be crucial factors for ET0 estimation. Nonetheless, even though the accuracy of ANN was higher than that of LGBM, its computational efficiency was higher and it proved to be more useful in certain scenarios where data is limited.