| Popis: |
The objective is to propose an approach to care the groundwater quality from anthropogenic threats with minimum funds or poor data, where statistical methods such as popular principal components analysis and K-means, afford non-significant results. It is a frequent dilemma in developing countries. To overcome it, data mining (DM) techniques were applied to evaluate hidden patterns between 15 hydrogeochemical parameters from 29 production wells and the DRASTIC vulnerability index (DVI), to identify the specific parameters related to the threat, even natural or anthropogenic. The DM classifiers afforded four wells’ clusters, located in correspondence to their DVI-scaled areas in the map. The DM informational and differential weights, with the interaction and multi tests procedures, pointed the key water quality parameters as reliable forecasters and no need for others. The approach would be useful to foresee warning criteria for policy-makers, saving funds in the groundwater quality control analysis. The groundwater quality was adequate, but high and moderate DVI values areas need prevention. DM identified critical physicochemical parameters as predictors for the aquifer vulnerability (Mg2+, HCO3−, Cl−, K+, Na+ concentrations, electric conductivity, and total dissolved solids characterize highest DVIs). The main surface threats were the urban and industrial activities in the center, agriculture along the flanks, and cheese manufacturing in the north. |
