Development of multiple linear regression models for annual reference evapotranspiration estimation under limited data conditions

Ahmadpari, Hedieh; Khaustov, Vitaly; Amini, Ata

doi:10.22098/mmws.2025.18810.1726

Development of multiple linear regression models for annual reference evapotranspiration estimation under limited data conditions

مقالات آماده انتشار

نوع مقاله : Special Issue: New Approaches to Water and Soil Management and Modeling

نویسندگان

¹ Ph.D. Candidate, Hydrology of Land, Water Resources, Hydrochemistry, Russian State Hydrometeorological University, Saint Petersburg, Russia

² Candidate of Technical Sciences, Associate Professor at the Department of Engineering Hydrology of the RSHU, Saint Petersburg, Russia

³ Professor, Soil Conservation and Watershed Management Research Department, Kurdistan Agricultural and Natural Resources Research and Education Center, AREEO, Sanandaj, Iran

10.22098/mmws.2025.18810.1726

چکیده

Accurate estimation of reference evapotranspiration (ET₀) is essential for agricultural water management, particularly in regions with limited data availability. The aim of this study was to evaluate multiple linear regression (MLR) models to estimate ET₀ at the annual scale. Meteorological data from the Kuhdasht synoptic station, Iran for a 25-year period (1998–2022) were used. ET₀ was calculated using the FAO-56 Penman-Monteith method implemented through the CROPWAT 8.0 software. A total of 31 MLR models were developed using the Regression option from the Analysis ToolPak of Microsoft Excel 2019 to quantify the relationship between ET₀ and climatic variables. Seven statistical indices were used to evaluate the performance of the MLR models in estimating ET₀. Results showed that 16 models achieved very high accuracy, with coefficients of determination (R²) greater than 0.92. Among single-variable models, wind speed ing up to 92% of ET₀ variability. Several two-variable models achieved R² = 0.92–0.96, and most three-variable models reached R² = 0.93–0.97. Four-variable models also performed strongly (R² ≈ 0.95–0.97), while the five-variable model yielded R² ≈ 0.97, similar to simpler models. Wind speed emerged as the most influential factor, highlighting that well-chosen two- or three-variable models can estimate ET₀ as effectively as more complex alternatives.

کلیدواژه‌ها

موضوعات