Groundwater potential mapping using machine learning approach in West Java, Indonesia

Groundwater availability is a challenge as it is utilized for vital sectors such as agricultural sector, human consumption, and industrial sector. Therefore, water resource mapping is needed to be performed to maintain water resource sustainability. This research aims to investigate groundwater potential in West Java, Indonesia using supervised machine learning (ML) methods namely Random Forest (RF), Support Vector Machine (SVM), and Artificial Neural Network (ANN). Several groundwater conditioning factors were used in this research such as Topographic Wetness Index (TWI), Normalized Difference Vegetation Index (NDVI), lithology, geomorphology, land use land cover (LULC), soil type, and land system. The groundwater potential prediction model was validated using the groundwater potential map and well locations obtained from the Ministry of Energy and Mineral Resources and the Ministry of Public Works and Public Housing of Republic of Indonesia, respectively. The results show that the highest overall accuracy was achieved using RF method (0.8). We found that the land system was the highest contributor to groundwater potential mapping (25%), followed by lithology (16%), NDVI (15%), geomorphology and TWI (14% each), and LULC and soil type (8% each). More than 50% of the West Java Province region exhibited groundwater potential in very low and low classes, while the high and very high classes of groundwater potential were only less than 16%. Ground geoelectric measurements were conducted in sample areas in Bandung City and Sukabumi District, representing very high and very low groundwater potentials, respectively. This study emphasizes the critical need to implement measures that ensure the sustainability of water resources and prevent mismanagement of groundwater extraction, particularly in West Java.