Data mining to aquifer vulnerability assessment

Abstract

The vulnerability map of aquifer pollution is a part of the early warning system to prevent the deterioration of groundwater quality. The weighted index overlay methods are commonly used to map aquifer vulnerability, but they have a series of drawbacks, indicating the need to apply alternative methods that introduce the least number of a priori considerations in the parameters processing and allow a more accurate interpretation of the final results. The purpose of this study is to use data mining techniques for cluster analysis to evaluate the vulnerability of groundwater pollution in the Almondarez Vento Karst Basin in Havana Province, Cuba, and compare the results with those obtained using the RISK method, which is a weighted index overlay method to study karstic aquifers. The variables selected to apply this unsupervised classification technique were: aquifer lithology, topographic slope of the terrain, soil attenuation index to pollutants, fault density per km2 and presence of direct infiltration zones. The cluster analysis achieved greater spatial discrimination and definition of areas with different degrees of vulnerability, demonstrating its high resolution power.