Bayesian Model Averaging for Estimating Evapotranspiration and Water Footprint in Wheat Cultivation
نویسندگان
1 Irrigation & Reclamation Engrg. Dept. University of Tehran Karaj, Iran.
2 Associate Prof., Irrigation & Reclamation Engrg. Dept. University of Tehran Karaj, Iran.
doi
10.48308/pae.2026.242112.1126چکیده
Water resource management is of utmost importance in arid and semi-arid regions. The incorporation of the water footprint (WF) concept, which connects various water-consuming sectors in crop production, serves as a practical tool for water sector policies. The accuracy of three crop models (CropSyst, DSSAT, and SSM-Wheat) in estimating evapotranspiration (ET) was compared with the FAO Penman-Monteith (FAO56) reference model. Subsequently, the Bayesian model averaging (BMA) approach was employed to integrate the models. The application of the BMA approach resulted in a reduction of the Normalized Root Mean Square Error (NRMSE) in comparison to the individual models. Moreover, the coefficient of determination (R2), Nash-Sutcliffe efficiency (EF), and Kling-Gupta efficiency (KGE) achieved values of 99%, 0.99, and 0.96, respectively. In the subsequent step, the WF was calculated based on the yield and evapotranspiration values. The findings revealed that the green WF exceeded the blue WF in most fields, primarily due to sufficient rainfall in the area during the growth period, which allowed the plants to utilize soil moisture. Consequently, the pressure on soil moisture (effective rainfall) surpassed that on blue water. The objective of this study was to calculate the WF of wheat in Gorgan, Iran and the results highlight the requirement of effective crop management strategies to achieve a balance in water consumption, thereby minimizing the blue WF and maximizing yield. For instance, modifying the planting date to align with rainfall during the growth period can significantly reduce the blue WF.