Employing GIS techniques and unsupervised learning to delineate groundwater recharge potential: A case study in the karst region of northern Puerto Rico

Severe drought conditions, along with excessive water extraction, has imposed huge stress on groundwater resources in many regions across the world. Knowledge of potential recharge zones can provide authorities valuable data regarding groundwater resource management, land development, or environmental protection. This study evaluates the feasibility of using geographical information system (GIS) data and unsupervised learning, along with high-resolution World-View satellite imagery to determine potential recharge areas in the karst region of northern Puerto Rico. Groundwater recharge parameters, such as geology, precipitation, lineament density, drainage density, topographic wetness index, slope, land use/cover and sinkhole density were generated as GIS layers and analyzed for groundwater recharge potential, employing principal component analysis in ArcGIS Pro and Environment for Visualizing Images (ENVI). The map generated categorizes groundwater potential zones into four categories: high, moderate, low, and very low. Results revealed that the study area shows a 76% moderate-to-high groundwater recharge capability in the study area. Even though this methodology was implemented as a case study, it can certainly be extrapolated to other regions and can provide critical information regarding sustainable groundwater resource management.