Groundwater Vulnerability Zoning to Nitrate Based on DRASTIC-LU Parameters and Data-driven Models

Abstract

Assessment of groundwater vulnerability to prioritize these resources from the perspective of exploitation, management, and control of pollution in different areas is important. The purpose of this study is to estimate groundwater vulnerability (nitrate concentration) of the Birjand plain aquifer based on DRASTIC-LU parameters and spatial and non-spatial data-driven methods. The novelty of this study is to present new combination approaches to determine the effective DRASTIC-LU parameters in groundwater vulnerability zoning (Birjand plain aquifer). In this regard, geographically weighted regression (GWR) with exponential and bi-square kernels and artificial neural network (ANN) were combined with a binary particle swarm optimization algorithm (BPSO). The best value of the fitness function (1-R2) for ANN, GWR with exponential kernel, and GWR with bi-square kernel was obtained 0.1060, 0.07453, and 0.0069, respectively, Which indicates higher compatibility of the bi-square kernel than the other methods. It was also found that the DRASTIC-LU parameters have a significant effect on the rate of groundwater vulnerability and estimation of the nitrate concentration in the study area.