<![CDATA[The estimation of Potential Evapotranspiration (ETp) is crucial for water distribution planning andcropping patterns. Generally, ETp calculation is obtained from empirical models such as the PenmanMonteith (PM) model recommended by the Food and Agriculture Organization (FAO). However,implementing this model requires numerous weather variables and adequate weather data availability. Thisresearch aims to develop an ETp estimation model using the Random Forest (RF) algorithm The weathervariables used in this research as inputs for ETp modeling are solar radiation (Rs); air temperature (T); airhumidity (RH); and a combination of Rs and T. Weather variable data were obtained from an automaticweather station (AWS) in the Tungkub Irrigation Area, Bali. The research results indicate that the weathervariable Rs is the best estimation model input, while the weather variable RH is the weakest. In modelcalibration, three evaluation metrics were used to assess model performance, R2, MSE, and RMSE.Meanwhile, for model validation, three techniques were employed, prediction error plot, residuals plot,and k-fold cross-validation. The research results indicate that the average ETp estimation value with thescenario of input Rs using the RF algorithm in the Tungkub Irrigation Area is 0,14 mm/hour (R2 = 1,00,MSE = 0,00, RMSE = 0,01). Meanwhile, the average ETp PM value is 0,15 mm/hour. The scenario ofinput Rs using the RF algorithm shows estimation values close to the PM ETp value.]]>
Copyrights © 2025
