Scholarlink

<p>Mapping the artificial groundwater recharge sites with greater precision is always a challenging task.This research introduces an artificial neural network (ANN) model and performs a comparative assessment against three different machine learning techniques: Random Forest (RF), eXtreme Gradient Boosting (XGBoost) and Support Vector Machine (SVM). All the contributing features for selection of artificial groundwater recharge site in the chosen study area are mainly soil type, rainfall, normalized difference vegetation index (NDVI), elevation, drainage density, landuse and landcover (LULC), slope, geology, geomorphology, lineament density and groundwater fluctuation map. All the required information pertaining to these selected features have been gathered from remote sensing data, field visit data observations, and from different organizations. The performance of the models is evaluated using statistical tools namely, coefficient of determination (R<sup>2</sup>), root mean squared error (RMSE), mean square error (MSE) and mean absolute error (MAE). Results from the study demonstrates that Random Forest machine learning algorithm out performs in selection of suitable groundwater recharge location with exhibiting the highest accuracy measured in terms of coefficient of determination(R<sup>2)</sup>), i.e. R<sup>2</sup> = 0.9275in opposition to SVM(R<sup>2</sup> = 0.9015), ANN(R<sup>2</sup> = 0.8621) and XGBoost(R<sup>2</sup> = 0.8941). Subsequently, the output from the models is validated through entropy weight water quality index (EWQI). Validation using EWQI approach leverages entropy-based parameter weighting to ensure accurate and objective assessment, also offer a reliable basis for assessing model performance in identifying artificial ground water recharge zones. The outcome of this study will aid in sustainable groundwater management.</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

Identification of Artificial Groundwater Recharge Zones in Municipal Area of Coal Capital of India: An Integrated GIS and Machine Learning Based Approach

  • 期刊论文

  • 发表日期：

    2025年9月23日

查看全文
  • Aman Kumar,
  • Rohit Sharma,
  • Rajesh Kumar Paswan

摘要

Mapping the artificial groundwater recharge sites with greater precision is always a challenging task.This research introduces an artificial neural network (ANN) model and performs a comparative assessment against three different machine learning techniques: Random Forest (RF), eXtreme Gradient Boosting (XGBoost) and Support Vector Machine (SVM). All the contributing features for selection of artificial groundwater recharge site in the chosen study area are mainly soil type, rainfall, normalized difference vegetation index (NDVI), elevation, drainage density, landuse and landcover (LULC), slope, geology, geomorphology, lineament density and groundwater fluctuation map. All the required information pertaining to these selected features have been gathered from remote sensing data, field visit data observations, and from different organizations. The performance of the models is evaluated using statistical tools namely, coefficient of determination (R2), root mean squared error (RMSE), mean square error (MSE) and mean absolute error (MAE). Results from the study demonstrates that Random Forest machine learning algorithm out performs in selection of suitable groundwater recharge location with exhibiting the highest accuracy measured in terms of coefficient of determination(R2)), i.e. R2 = 0.9275in opposition to SVM(R2 = 0.9015), ANN(R2 = 0.8621) and XGBoost(R2 = 0.8941). Subsequently, the output from the models is validated through entropy weight water quality index (EWQI). Validation using EWQI approach leverages entropy-based parameter weighting to ensure accurate and objective assessment, also offer a reliable basis for assessing model performance in identifying artificial ground water recharge zones. The outcome of this study will aid in sustainable groundwater management.