Characterization and sensitivity analysis of parameters for delineating groundwater potential zones in the Phakphokthum Watershed, Ilam, Eastern Nepal

Abstract

Groundwater potential (GWP) mapping is an important technique for the sustainable exploitation and management of water resources. Although GWP using Multi Criteria Decision Analysis (MCDA) is commonly assessed using multiple thematic layers, the sensitivity of individual parameters to the final GWP map remains poorly evaluated. This study assesses the sensitivity of nine thematic layers including slope, geology, Normalized Difference Vegetation Index (NDVI), Topographic Position Index (TPI), aspect, hydrogeomorphology, land use/land cover (LULC), drainage density (DD), and lineament density (LD) used to delineate GWP zones in the Phakphokthum watershed, Ilam, eastern Nepal. The GWP map was generated using the Analytic Hierarchy Process (AHP) within a Geographic Information System (GIS) environment, following the derivation of criteria weights for each thematic layer and its respective subclasses. The GWP map of the area, indicated 20% area as high potential zone and 55% as low potential zone and is validated against the spring inventory using the ROC/AUC method with an accuracy of 70.3 %. Sensitivity analysis was performed using a one-parameter-removal approach, in which each parameter was systematically excluded and the AHP recalculated using the remaining parameters, yielding nine distinct scenarios. The influence of each parameter on groundwater occurrence was evaluated through validation against 85 spring locations using the receiver operating characteristic–area under curve (ROC-AUC) method, with prediction accuracies ranging from 63.1% to 73.1%. TPI and LD were identified as the most influential parameters controlling groundwater occurrence in the study area. These parameters are considered critical and are recommended for the delineation of groundwater potential zone in similar geographic settings. The findings also highlight the importance of sensitivity analysis as a complementary approach to AHP for identifying influential paramenters, evaluating model robustness, and improving the reliability of GWP maps.