Utilizing the concept of risk in calibrating the vulnerability of coastal-alluvial aquifers based on machine learning methods

MF Baensaf and M Rafati and HK Moghaddam, INTERNATIONAL JOURNAL OF ENVIRONMENTAL SCIENCE AND TECHNOLOGY, 22, 16747-16762 (2025).

DOI: 10.1007/s13762-025-06729-2

This study aims to calibrate the vulnerability index of coastal-alluvial aquifers by utilizing the concept of exploitation risk, with consideration of the aquifer's inherent vulnerability. The risk concept was defined based on three parameters: nitrate concentration, water resource exploitation density, and land use changes. These factors were integrated using fuzzy logic, which applies gradual membership values rather than strict thresholds to reflect uncertainty and variation in environmental conditions. The results showed that the proposed risk- based calibration significantly improved the correlation between vulnerability indices and observed pollution indicators, increasing the DRASTIC-risk correlation from 0.35 to 0.75 and the GALDIT-risk correlation from 0.29 to 0.74. These improvements demonstrate a stronger alignment between modeled vulnerability and actual environmental stressors, supporting more reliable aquifer risk assessments. The spatial analysis revealed that the northern part of the aquifer, influenced by coastal dynamics, was best represented by the GALDIT index (11% of the area), while the remaining 89%-mainly affected by agricultural activities-was better captured by the DRASTIC index. Overall, the integration of risk-based calibration with artificial intelligence offers a practical framework for enhancing groundwater vulnerability assessments, supporting more effective resource management and planning under increasing anthropogenic and climatic pressures. This study highlights the practical application of hybrid vulnerability indices for sustainable groundwater management under increasing anthropogenic pressures.

Return to Publications page