Estimating how much water is lost by crops to the atmosphere is a cornerstone of efficient irrigation planning and water resource management. Many semiarid regions lack reliable weather data to calculate what is known as reference evapotranspiration accurately. The study “Comparison of Machine Learning and Empirical Methods for Data-Efficient Estimation of Reference Evapotranspiration in a Semiarid Region” looks at how this gap might be addressed. Researchers Murat Pinarlik, Burak Bostancioglu, Adebayo J. Adeloye, and Bulent Selek explore whether newer machine learning techniques can work effectively with limited data, alongside or in place of traditional empirical formulas. At its core, the study aims to show that machine learning models using fewer meteorological inputs can still produce ET0 estimates that are competitive with long-established methods.
To test this idea, the authors compared several commonly used empirical approaches with a set of machine learning models, evaluating how each performed when data inputs were intentionally reduced. They examined accuracy, reliability, and how sensitive each method was to missing information, offering insight into the balance between simplicity and predictive power. The results point toward promising flexibility in machine learning approaches, especially in data-scarce settings. For those involved in irrigation and water management, the study highlights practical opportunities to adapt ET0 estimation methods based on available data. Learn more about how this research can improve decision-making without requiring extensive new monitoring systems in the Journal of Irrigation and Drainage Engineering at https://ascelibrary.org/doi/10.1061/JIDEDH.IRENG-10693. The abstract is below.
Abstract
Accurate estimation of reference evapotranspiration (ET0) is critical for efficient irrigation planning and water resource management, particularly in semiarid regions where meteorological data are limited. This study compares the performance of empirical models (Hargreaves, Thornthwaite, and Blaney–Criddle) and machine learning (ML) algorithms, i.e., support vector regression (SVR), random forest, decision tree, linear regression, and artificial neural networks, for estimating ET0 in the Çekerek subbasin of Türkiye. Each method was evaluated under multiple input scenarios simulating various data-scarce conditions. Results show that ML models significantly outperformed empirical equations, with SVR achieving the best performance (𝑅=0.997; RMSE=0.145 mm/day; NSE=0.991) even with reduced input parameters. The superior performance of ML approaches is attributed to their ability to capture nonlinear relationships among meteorological variables and to maintain robustness under reduced-input scenarios, unlike empirical equations that are structurally limited. Among empirical approaches, the Hargreaves method yielded the most consistent results but remained sensitive to seasonal variation. The findings demonstrate the potential of ML-based models to provide accurate, data-efficient alternatives for ET0 estimation in operational water resource applications, especially in regions with sparse climate monitoring networks.
Learn more about how machine learning models improve limited-data evaporation estimates without requiring new monitoring systems at https://ascelibrary.org/doi/10.1061/JIDEDH.IRENG-10693.
