Scholarlink

<p>The accelerating degradation of groundwater resources in arid and semi-arid regions demands a fundamental transformation in hydrogeological assessment and water governance. This study introduces an advanced spatial artificial intelligence framework that combines Maximum Entropy (ME), Random Forest (RF), and Extreme Gradient Boosting (XGBoost) algorithms. It then integrates the framework with high-resolution geo-environmental datasets to delineate <i>aflaj</i> groundwater potential zones in <i>wilayat</i> (district) Sohar, Oman. The model incorporates 16 conditioning variables, including rainfall, elevation, geology, fault density, and land use, to achieve a high predictive fidelity, with AUC of 81.1%, 82.8%, and 84.1% for the ME, RF, and XGBoost models, respectively. An ROC analysis validated these results against actual groundwater data. Sensitivity diagnostics and jackknife test identified rainfall and geological structure as the primary drivers of hydrological dynamics. The ME model excelled in spatial generalization in presence-only data contexts and correctly identified 21% of high-potential zones. This integrative approach transcends conventional groundwater mapping by embedding machine learning within a geospatial decision-support architecture, thereby enabling precise targeting of <i>aflaj</i> restoration zones. The resulting maps offer actionable intelligence for sustainable water resource planning, agroecological resilience, and cultural heritage conservation. By operationalizing spatial AI in a data-scarce, ecologically fragile landscape, this study sets a replicable benchmark for groundwater governance under climate stress, with implications on regional policy, transboundary aquifer management, and global water security frameworks.</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

Harnessing the power of machine learning to unveil groundwater potential in semi-arid environments

  • 期刊论文

  • 发表日期：

    2026年3月23日

查看全文
  • Khalifa M. Al-Kindi,
  • Yunis Bani-Oraba

摘要

The accelerating degradation of groundwater resources in arid and semi-arid regions demands a fundamental transformation in hydrogeological assessment and water governance. This study introduces an advanced spatial artificial intelligence framework that combines Maximum Entropy (ME), Random Forest (RF), and Extreme Gradient Boosting (XGBoost) algorithms. It then integrates the framework with high-resolution geo-environmental datasets to delineate aflaj groundwater potential zones in wilayat (district) Sohar, Oman. The model incorporates 16 conditioning variables, including rainfall, elevation, geology, fault density, and land use, to achieve a high predictive fidelity, with AUC of 81.1%, 82.8%, and 84.1% for the ME, RF, and XGBoost models, respectively. An ROC analysis validated these results against actual groundwater data. Sensitivity diagnostics and jackknife test identified rainfall and geological structure as the primary drivers of hydrological dynamics. The ME model excelled in spatial generalization in presence-only data contexts and correctly identified 21% of high-potential zones. This integrative approach transcends conventional groundwater mapping by embedding machine learning within a geospatial decision-support architecture, thereby enabling precise targeting of aflaj restoration zones. The resulting maps offer actionable intelligence for sustainable water resource planning, agroecological resilience, and cultural heritage conservation. By operationalizing spatial AI in a data-scarce, ecologically fragile landscape, this study sets a replicable benchmark for groundwater governance under climate stress, with implications on regional policy, transboundary aquifer management, and global water security frameworks.