Comparative evaluation of GIS-based models for mapping aquifer vulnerability in hard-rock terrains

Aquifer vulnerability assessment plays a vital role in identifying the areas/regions which are highly susceptible to groundwater contamination. The main intent of this study is to modify original DRASTIC and DRASTIC-P models by adding two exogenous factors [land use/land cover' (LU) and lineament density' (LD)] which significantly influence groundwater contamination. The performances of original DRASTIC and DRASTIC-P (agricultural DRASTIC) models were compared with those of six modified forms of these models, viz., DRASTIC-LD', DRASTIC-LU', DRASTIC-LDLU', DRASTIC-P-LD', DRASTIC-P-LU' and DRASTIC-P-LDLU'. The results of these models were validated with two scientifically sound and pragmatic approaches: (i) using the single water-quality parameter as a source of groundwater contamination (Approach I), and (ii) using multi-water quality parameters causing groundwater contamination (Approach II). Moreover, the sensitivity of these models was analyzed to identify most influential parameters in each case. The results revealed that irrespective of the models employed more than 50% of the studyarea falls under High' and Very High' vulnerability zones. The Approach I validation results indicated that the DRASTIC-P-LDLU' model performs the best with an accuracy of 61% and 68% with respect to nitrate and chloride concentrations, respectively, followed by the DRASTIC-LDLU model (respective accuracy=59% and 61%). The results of model validation using Approach II also confirmed that among the eight models, the Specific' aquifer vulnerability predicted by the DRASTIC-P-LDLU' model (accuracy =30%) is reasonably more accurate than DRASTIC-LDLU (accuracy=29.7%) and DRASTIC-P-LU (accuracy=29.6%) models. Hence, it is recommended to assess Specific' aquifer vulnerability instead of Intrinsic' aquifer vulnerability. The results of the model sensitivity analyses also indicated that lineament density' and land use/land cover' are the most significant parameters for vulnerability assessment.