Modeling groundwater nitrate concentrations using spatial and non-spatial regression models in a semi-arid environment

Abstract

Nitrate nitrogen (NO3−-N) from agricultural activities and in industrial wastewater has become the main source of groundwater pollution, which has raised widespread concerns, particularly in arid and semi-arid river basins with little water that meets relevant standards. This study aimed to investigate the performance of spatial and non-spatial regression models in modeling nitrate pollution in a semi-intensive farming region of Iran. To perform the modeling of the groundwater's NO3−-N concentration, both natural and anthropogenic factors affecting groundwater NO3−-N were selected. The results of Moran's I test showed that groundwater nitrate concentration had a significant spatial dependence on the density of wells, distance from streams, total annual precipitation, and distance from roads in the study area. This study provided a way to estimate nitrate pollution using both natural and anthropogenic factors in arid and semi-arid areas where only a few factors are available. Spatial regression methods with spatial correlation structures are effective tools to support spatial decision-making in water pollution control.

Keywords

• Groundwater
• Nitrate
• Natural and anthropogenic factors
• Spatial autoregression models
• Spatial autocorrelation