Open Access

ARTICLE

Groundwater Potential Zone Mapping in Islamabad, Pakistan: An Integrated GIS–AHP and AI Approach

Khlieeq Ul Zaman^1,2,*, Ahmad Saeed³, Muhammad Awais Khan², Hafiz Abdul Basit⁴, Shaharyar², Maria Anum², Rani Ummay Farwa^2,*, Mahmood Iqbal¹, Ali Raza²

1 School of Marine Science and Engineering, Nanjing Normal University, Nanjing, China
2 Department of Earth Sciences, University of Sargodha, Sargodha, Pakistan
3 Institute of Space Science, University of the Punjab, Lahore, Pakistan
4 Department of Earth Sciences, Quaid e Azam University, Islamabad, Pakistan

* Corresponding Authors: Khlieeq Ul Zaman. Email: ; Rani Ummay Farwa. Email:

(This article belongs to the Special Issue: Geospatial Techniques for Precision Agriculture and Water Resources Sustainability)

Revue Internationale de Géomatique 2026, 35, 409-422. https://doi.org/10.32604/rig.2026.083214

Received 31 March 2026; Accepted 14 May 2026; Issue published 02 July 2026

Abstract

Groundwater is the primary buffer against water scarcity in rapidly urbanizing regions, yet its sustainable management is constrained by limited hydrogeological data. This study presents an integrated Geographic Information System (GIS) remote sensing framework strengthened with Artificial Intelligence (AI) to delineate groundwater potential zones (GWPZs) in the Islamabad Capital Territory. Six thematic layers—slope, drainage density, lithology, rainfall, land use/land cover (LULC), and the Normalized Difference Vegetation Index (NDVI)—were derived from SRTM DEM, Sentinel-2 imagery, geological maps, and climate records. Each layer was standardized, reclassified, and weighted using the Analytical Hierarchy Process (AHP). A complete pairwise comparison matrix was constructed and the Consistency Ratio (CR = 0.07 < 0.10) was computed to verify weighting reliability. In parallel, an AI-driven K-Means clustering algorithm was applied to identify natural similarity groups and generate a data-driven groundwater potential map independent of expert bias. Categorical LULC data were one-hot encoded, all continuous variables were min–max normalized, and the optimal cluster number (k = 3) was determined using elbow and silhouette analyses (silhouette score = 0.58). Both approaches revealed consistent spatial patterns: high and very-high groundwater potential zones concentrated across the piedmont plains and alluvial deposits, and low-potential zones dominating the steep, compact formations of the Margalla Hills. Quantitative area statistics show that high and very-high potential zones cover approximately 40% of the study area (~262 km²). Cross-validation between the two methods yielded an Overall Accuracy of 74.3% and a Kappa coefficient of 0.61, indicating substantial agreement. An internal consistency assessment using 30 high-NDVI sample points revealed a significant positive NDVI–GPI correlation (Pearson r = 0.85, p < 0.001; R² = 0.72), confirming spatial coherence of identified recharge zones, though this does not constitute independent external validation. Although the framework does not incorporate subsurface hydraulic properties due to data limitations, the combined GIS–AHP–AI methodology offers a practical and scalable tool for groundwater resource assessment in data-scarce regions.

---

Keywords

---