نوع مقاله : مقاله پژوهشی
عنوان مقاله English
نویسندگان English
Faced with growing challenges from climate change and rising water demand, accurate prediction of reference evapotranspiration (ET0) has become a key component of sustainable water resource management. Traditional models like FAO-Penman-Monteith, although scientifically reliable, often lose accuracy in areas with limited meteorological data. This study aims to improve ET0 prediction by combining MODIS remote sensing data with advanced machine learning algorithms. Two models were developed: the basic Extreme Gradient Boosting (XGB) and an enhanced version called Hybrid XGB (HXGB), which offers better generalization. Meteorological and satellite data from Ramsar and Bandar Anzali stations (2001–2023) were used for training and evaluation. Results showed that HXGB outperformed the base model at both stations. In Bandar Anzali, Scenario 8 (using both meteorological data and satellite-based ETMODIS) reduced RMSE to 0.11 mm/day. In Ramsar, the same scenario achieved an RMSE of 0.19 mm/day. This data fusion approach increased the models sensitivity to spatial and temporal variations, significantly improving prediction accuracy.
کلیدواژهها English