Scholarlink

<p>Hydrological and irrigation planning rely on Reference Evapotranspiration (ET₀), often calculated using the Penman-Monteith (PM) equation, which requires multiple inputs. In our proposed model, Machine Learning (ML) techniques are used for the appropriate estimation of ET₀ from Solar Radiation (SR) data utilizing Extreme Gradient Boosting (XGB), Multilayer Perceptron (MLP), Support Vector Regression (SVR), and Random Forest (RF). Performance evaluation has been conducted using Mean Absolute Error (MAE), Root Mean Square Error (RMSE), Coefficient of Determination (R²), and Nash-Sutcliffe Efficiency (NSE). To ensure robustness across varying datasets, we propose a Weighted Average Ensemble (WAE) method, which outperforms other ensemble-based models and other approaches, including Deep Neural Networks (DNN), as identified in recent studies. The proposed model offers a promising solution for estimating ET₀ in areas with limited meteorological data, particularly when SR is the only available parameter.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Single Meteorological Parameter-Based Ensemble Modeling for Reference Evapotranspiration in California’s Weather Stations

  • 期刊论文

  • 发表日期：

    2025年7月24日

查看全文
  • Abhishek Patel,
  • Syed Taqi Ali

摘要

Hydrological and irrigation planning rely on Reference Evapotranspiration (ET₀), often calculated using the Penman-Monteith (PM) equation, which requires multiple inputs. In our proposed model, Machine Learning (ML) techniques are used for the appropriate estimation of ET₀ from Solar Radiation (SR) data utilizing Extreme Gradient Boosting (XGB), Multilayer Perceptron (MLP), Support Vector Regression (SVR), and Random Forest (RF). Performance evaluation has been conducted using Mean Absolute Error (MAE), Root Mean Square Error (RMSE), Coefficient of Determination (R²), and Nash-Sutcliffe Efficiency (NSE). To ensure robustness across varying datasets, we propose a Weighted Average Ensemble (WAE) method, which outperforms other ensemble-based models and other approaches, including Deep Neural Networks (DNN), as identified in recent studies. The proposed model offers a promising solution for estimating ET₀ in areas with limited meteorological data, particularly when SR is the only available parameter.