Evaluating the impact of grid cell properties in spatial discretization of groundwater model for a tropical karst catchment in Rote Island, Indonesia

Abstract

To assess the effect of three grid cell properties (size, mean slope of the surface and distance between centre of grid and observation well) on groundwater models' performances, a tropical karst catchment characterized by monsoonal season in Rote Island, Indonesia was selected. Here, MODFLOW was used to develop models with five different spatial discretization schemes: 10 ×10 m, 20 ×20 m, 30×30 m, 40×40 m and 50×50 m. Using parameter estimation method, hydraulic conductivity and specific yield values over a selection of pilot points were estimated. The trends of the performances were calculated at each observation well in order to recommend the most appropriate location for observation well placement in terms of topographical characteristic. It is confirmed that the deterioration of model performance is mainly controlled by the increase of distance between well and centre of the cell, and the mean slope of the surface. Results reveal that model performance increases substantially for areas of low slope (<3%) and medium slope (3-10%) for a smaller grid cell size. Therefore, to improve model performance, it is recommended that the observations wells are placed in areas of low and medium slopes.