A Comprehensive Review on Applications of Machine Learning Algorithms for Evapotranspiration Estimation

Abstract

The measurement of evapotranspiration constitutes a critical component in irrigation scheduling. Evapotranspiration refers to the combined loss of water from plant surfaces and soil. Evaporative parameters are extensively applied in the analysis of water balance, the management of water resources, and the design of irrigation systems, as well as in the estimation of plant growth and height. It can be quantified through a range of methods incorporating different variables. Evapotranspiration is influenced by climatic variability, and given the considerable geographical heterogeneity of climate conditions, many previously developed systems have not incorporated the full range of available meteorological data, resulting in models that are not sufficiently robust. The goal of this work is to examine earlier research on the use of AI models in ET modeling under various models. The study's findings showed that while AI-based methods provide interesting substitutes because of their capacity to represent intricate nonlinear interactions, conventional models such as the Penman–Monteith (PM) model necessitate a large amount of input data. The need for standardized input configurations, improved pre-processing methods, and the integration of hydrological and remote sensing data is highlighted by the difficulties AI models face, despite their potential, including overfitting, interpretability, inconsistent input variable selection, and lack of integration with physical ET processes.