Artificial Intelligence and Hybrid Optimization Approaches for Potential Evapotranspiration Modeling: A Narrative Review and Future Perspectives

Abstract

Accurate prediction of potential evapotranspiration (PET) is an important factor in irrigation planning, drought monitoring, and sustainable water resource management. This study presents a systematic review of the latest PET modeling studies, focusing on experimental models, artificial intelligence models, and optimized hybrid models. Sixteen studies were analyzed and classified into three main categories: optimized experimental models, standalone AI models, and optimized hybrid AI models. The results show that models such as ANN, RF, XGBoost, LSTM and DNN are able to represent non-linear relationships between climate variables and PET very efficiently. The integration of meta-heuristic optimization algorithms with AI models also contributes to improving accuracy and fine-tuning of transactions. The study recommends focusing in the future on hybrid models, explainable artificial intelligence (XAI), and verification in multiple climates.