Groundwater vulnerability mapping using the modified DRASTIC model: the metaheuristic algorithm approach

• Published in: Environmental monitoring and assessment Vol. 193; no. 1; p. 25
• Main Authors: L, Balaji, R, Saravanan, K, Saravanan, N.A, Sreemanthrarupini
• Format: Journal Article
• Language: English
• Published: Cham Springer International Publishing 03.01.2021; Springer Nature B.V
• Subjects: Algorithms; Atmospheric Protection/Air Quality Control/Air Pollution; Computer applications; Correlation coefficient; Correlation coefficients; Earth and Environmental Science; Ecology; Ecotoxicology; Environment; Environmental Management; Environmental Monitoring; Environmental science; Groundwater; Groundwater resources; Heuristic methods; Invasive plants; Machine learning; Mapping; Model accuracy; Models, Theoretical; Monitoring/Environmental Analysis; Nitrates - analysis; Optimization; Particle swarm optimization; Ranking; Resource management; Vulnerability; Water resources; Weight; Groundwater vulnerability; Wilcoxon test; Metaheuristic algorithm; TOPSIS; DRASTIC model
• ISSN: 0167-6369; 1573-2959
• DOI: 10.1007/s10661-020-08787-0

Vulnerability assessment and mapping is a significant tool for sustainable management of the precious natural groundwater resources. DRASTIC is an extensively used index model to map groundwater vulnerable zones. However, the original DRASTIC model rates and weights used in most of the research depict the poor correlation between nitrate concentration and groundwater vulnerability index. Wilcoxon test and five population-based metaheuristic (MH) algorithms, namely, firefly algorithm (FA), invasive weed optimization (IWO), teaching learning-based optimization (TLBO), shuffled frog leaping algorithm (SFLA), and particle swarm optimization (PSO), were used to optimize the rates and weights of the DRASTIC model to improve its accuracy. The performance of all the employed metaheuristic algorithms converges to a global optimal solution at different iterations, and to choose the best algorithm for DRASTIC weights optimization, a ranking methodology was proposed. The algorithms were ranked by calculating the relative closeness of alternatives with computational speed and the number of iterations as attributes in the TOPSIS method. This study identifies FA as the outperforming algorithm among the employed for this specified weight optimization problem based on ranking. The result of the optimization model proposed depicts significant improvement in the correlation coefficient between the groundwater vulnerability index and nitrate concentration from 0.0545 for the original DRASTIC model to 0.7247 for the Wilcoxon-MH- DRASTIC. Hence, this ranking approach can be adopted when global optimal solution is found by all employed algorithms in DRASTIC weight optimization.