Scholarlink

<p>Evaporation forecasting is crucial for effective water resource management, agricultural planning, and climate studies. Traditional methods often rely on empirical equations and historical data, which can limit accuracy and adaptability. Recent advancements in machine learning (ML) offer promising alternatives to enhance forecasting precision. This study utilizes a dataset spanning 17 years from the Beni Haroun Dam (ANBT), incorporating variables such as water level (WL in m), volume (V in Hm³), precipitation (P in mm), mean temperature (T in °C), relative humidity (RH), wind speed (WS), and observed evaporation (E obs in Hm³). Various ML algorithms, including Adaptive Boosting (AdaBoost), Gradient Boosting Regression Trees (GBRT), Extreme Gradient Boosting (XGBoost), Random Forest Regression (RFR) and a hybrid model (Multi Boost-RFR) were applied to predict evaporation rates. The dataset was divided into training (70%) and testing (30%) subsets to evaluate the performance of each model. The evaluation of model performance reveals that the hybrid Multi Boost-RFR model outperformed all other models, achieving the lowest RMSE (0.02) and highest R² (0.98) during training, as well as exceptional testing results with an RMSE of 0.02 and R² of 0.97. Among individual models, RFR, XGBoost, and GBRT demonstrated strong performance, with low RMSE and high R² values, though their testing accuracy was slightly lower than Multi Boost-RFR. AdaBoost, however, was the least effective, with the highest RMSE and lowest R² across both phases. These results demonstrate the superiority of hybrid models in improving evaporation forecasting accuracy, emphasizing the importance of choosing models that align with the specific characteristics of the data for reliable predictions.</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

Evaporation forecasting using different machine learning models in Beni Haroun Dam, Algeria

  • 期刊论文

  • 发表日期：

    2025年1月18日

查看全文
  • Zeghmar Amer,
  • Boutouatou Farah

摘要

Evaporation forecasting is crucial for effective water resource management, agricultural planning, and climate studies. Traditional methods often rely on empirical equations and historical data, which can limit accuracy and adaptability. Recent advancements in machine learning (ML) offer promising alternatives to enhance forecasting precision. This study utilizes a dataset spanning 17 years from the Beni Haroun Dam (ANBT), incorporating variables such as water level (WL in m), volume (V in Hm³), precipitation (P in mm), mean temperature (T in °C), relative humidity (RH), wind speed (WS), and observed evaporation (E obs in Hm³). Various ML algorithms, including Adaptive Boosting (AdaBoost), Gradient Boosting Regression Trees (GBRT), Extreme Gradient Boosting (XGBoost), Random Forest Regression (RFR) and a hybrid model (Multi Boost-RFR) were applied to predict evaporation rates. The dataset was divided into training (70%) and testing (30%) subsets to evaluate the performance of each model. The evaluation of model performance reveals that the hybrid Multi Boost-RFR model outperformed all other models, achieving the lowest RMSE (0.02) and highest R² (0.98) during training, as well as exceptional testing results with an RMSE of 0.02 and R² of 0.97. Among individual models, RFR, XGBoost, and GBRT demonstrated strong performance, with low RMSE and high R² values, though their testing accuracy was slightly lower than Multi Boost-RFR. AdaBoost, however, was the least effective, with the highest RMSE and lowest R² across both phases. These results demonstrate the superiority of hybrid models in improving evaporation forecasting accuracy, emphasizing the importance of choosing models that align with the specific characteristics of the data for reliable predictions.