DATA MINING APPLICATION TO AQUIFERS VULNERABILITY ASSESSMENT IN CUYAGUATEJE AND COSTERA SUR BASINS OF PINAR DEL RÍO, CUBA.

Aplicação da mineração de dados na avaliação da vulnerabilidade dos aquíferos nas bacias Cuyaguateje e Costera Sur de Pinar del Rio, Cuba

Abstract

Groundwater accounts for more than 97% of the total freshwater availability on Earth and obviously needs protection. It is necessary to prevent quality groundwater deterioration and aquifer intrinsic vulnerability maps are part of an early warning system. Weighted index overlay methods are commonly used to map aquifer vulnerability. However, these methods have disadvantages that make it convenient to apply alternative methods to obtain more accurate results. The research objective was evaluate groundwater contamination vulnerability in Cuyaguateje and Costera Sur basins of Pinar del Río, Cuba, using data mining techniques (cluster analysis and decision tree), and compare the results with those obtained by the RISK method, which is a weighted index superposition method. The variables used were: land topographic slope, aquifer lithology, soil attenuation index to pollutants and sinkholes density per square kilometer. The unsupervised an supervised grouping analysis was able to better discriminate the areas with different degrees of vulnerability, demonstrating greater resolving power.

Keywords: Aquifer intrinsic vulnerability. Groundwater. Cluster analysis. Decision tree.