New hybrid evolutionary algorithm for optimizing index-based groundwater vulnerability assessment method

Abstract Limited hydrogeological data accessibility leads scholars to improve the robustness of present qualitative groundwater vulnerability assessment methods using mathematical techniques. In the present study, we implemented three GIS-based groundwater vulnerability assessment indices, namely DRASTIC (Depth to water table, net Recharge, Aquifer media, Soil media, Topography, Impact of vadose zone, and hydraulic Conductivity), SINTACS (Soggicenza, Infiltrazione, Non saturo, Tipologia della copertura, Acquifero, Conducibilita, and Superficie topografica), and GODS (Groundwater confinement, Overlying strata, Depth to groundwater, and Soil media) to assess the groundwater vulnerability levels. Although DRASTIC results showed better performance with respect to the nitrate concentration data from 50 observation wells in the study site, the index is still unreliable due to its inherent drawbacks, including subjectivity. Hybrid PSO-GA method is a successful optimization algorithm gathering the advantages of Particle Swarm Optimization (PSO) and Genetic Algorithm (GA) while avoiding their shortcomings. The DRASTIC weighting system is optimized using PSO-GA optimization algorithm. Also, Step-wise Weight Assessment Ratio Analysis (SWARA) as a Multi-Attribute Decision Making (MADM) method is applied for changing ranges of DRASTIC rates and weights. The vulnerability indices obtained from SWARA-SWARA, DRASTIC-PSO-GA, and SWARA-PSO-GA frameworks are evaluated and compared with generic DRASTIC regarding the nitrate concentration dataset by employing Area Under the ROC Curve (AUC) and Grey relational analysis methods. Results show a noticeable improvement of correlation between indices and observed nitrate concentration after modifications and optimizations. The new hybrid SWARA-PSO-GA framework is the most effective framework in assessing the vulnerability of the present study area.