EUVIMEDEUVIMED entdecken
Crossref

Artificial Intelligence-Driven Integrated Assessment, Prediction, and Health-Risk Analysis of Groundwater Quality in Solapur District, Maharashtra: A Multi-Model Ensemble and SHAP Explainability Framework

Mustaq Ahmad Shaikh, Farjana Birajdar

International Journal of Advanced Geosciences · 2026

DOI: 10.14419/bt1xbt29 ↗

Kurz gefasst

Fachartikel in der Zeitschrift International Journal of Advanced Geosciences, 2026. Verfasst von Mustaq Ahmad Shaikh und 1 weiteren Autor:innen.

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Groundwater quality deterioration across the semi-arid Deccan Basalt terrain of Solapur District, Maharashtra poses an escalating crisis for over 4.3 million residents dependent on hard-rock aquifer systems. This study presents the most comprehensive Artificial Intelligence-based integrated framework for groundwater quality assessment, spatiotemporal prediction, and public health risk quantification yet conducted for this region. Leveraging a robust dataset of 534 observations from 89 GSDA-monitored wells across 11 talukas (2022–2024, bi-annual pre- and post-monsoon campaigns), we developed and validated a weighted stacking ensemble model combining Random Forest (RF), XGBoost, Deep Artificial Neural Network (ANN), and Long Short-Term Memory (LSTM) networks. The proposed ensemble achieves outstanding predictive performance: R²=0.967, RMSE=2.43 WQI units, MAE=1.92, and NSE=0.964 on a 30% holdout test set (n=160), outperforming all nine individual benchmark models and reducing RMSE by 73% versus Multiple Linear Regression. Water Quality Index analysis reveals that Akkalkot, Mangalvedhe, and Sangola talukas are classified as "Very Poor" (WQI) and "Critical" on the Health Risk Index, with fluoride reaching 2.82 mg/L (188% of BIS Maximum Permissible Limit), nitrate 84.2 mg/L (187% of BIS MPL), and TDS 1382 mg/L — creating compound health risks including dental and skeletal fluorosis, methaemoglobinaemia, and enhanced arsenic bioavail-ability. SHAP (SHapley Additive exPlanations) analysis identifies fluoride (24.2%), nitrate (20.2%), and TDS (17.6%) as dominant WQI predictors. A composite Health Risk Index (HRI) framework targeting three contaminants (F⁻, NO₃⁻, As) identifies three talukas requiring urgent defluoridation and denitrification infrastructure. Seasonal post-monsoon WQI improvement of 5.6–7.1 WQI units (p < 0.001) is documented in Atal Bhujal Yojana recharge intervention zones. The framework is operationally integrated with GSDA Solapur Division's Early Warning Indicator (EWI) system and Maharashtra's District Water Security Plan, enabling real-time contamination alerts, remediation prioritisation, and climate-adaptive water governance.

Zitieren

APA

Shaikh, M. A., & Birajdar, F. (2026). Artificial Intelligence-Driven Integrated Assessment, Prediction, and Health-Risk Analysis of Groundwater Quality in Solapur District, Maharashtra: A Multi-Model Ensemble and SHAP Explainability Framework. International Journal of Advanced Geosciences. https://doi.org/10.14419/bt1xbt29

Vancouver

Shaikh MA, Birajdar F. Artificial Intelligence-Driven Integrated Assessment, Prediction, and Health-Risk Analysis of Groundwater Quality in Solapur District, Maharashtra: A Multi-Model Ensemble and SHAP Explainability Framework. International Journal of Advanced Geosciences. 2026. doi:10.14419/bt1xbt29

BibTeX
@article{Shaikh2026,
  author = {Shaikh, Mustaq Ahmad and Birajdar, Farjana},
  title = {Artificial Intelligence-Driven Integrated Assessment, Prediction, and Health-Risk Analysis of Groundwater Quality in Solapur District, Maharashtra: A Multi-Model Ensemble and SHAP Explainability Framework},
  journal = {International Journal of Advanced Geosciences},
  year = {2026},
  doi = {10.14419/bt1xbt29}
}
Kostenlos starten →