Nitrate (NO3-) pollution of groundwater is a global concern in agricultural areas. To gain a comprehensive understanding of the sources and destiny of nitrate in soil and groundwater within intensive agricultural areas, this study employed a combination of chemical indicators, dual isotopes of nitrate (δ15N-NO3- and δ18O-NO3-), random forest model, and Bayesian stable isotope mixing model (MixSIAR). These approaches were utilized to examine the spatial distribution of NO3- in soil profiles and groundwater, identify key variables influencing groundwater nitrate concentration, and quantify the sources contribution at various depths of the vadose zone and groundwater with different nitrate concentrations. The results showed that the nitrate accumulation in the cropland and kiwifruit orchard at depths of 0-400 cm increased, leading to subsequent leaching of nitrate into deeper vadose zones and ultimately groundwater. The mean concentration of nitrate in groundwater was 91.89 mg/L, and 52.94% of the samples exceeded the recommended grade III value (88.57 mg/L) according to national standards. The results of the random forest model suggested that the main variables affecting the nitrate concentration in groundwater were well depth (16.6%), dissolved oxygen (11.6%), and soil nitrate (10.4%). The MixSIAR results revealed that nitrate sources vary at different soil depths, which was caused by the biogeochemical process of nitrate. In addition, the highest contribution of nitrate in groundwater, both with high and low concentrations, was found to be soil nitrogen (SN), accounting for 56.0% and 63.0%, respectively, followed by chemical fertilizer (CF) and manure and sewage (M&S). Through the identification of NO3- pollution sources, this study can take targeted measures to ensure the safety of groundwater in intensive agricultural areas.
Keywords: Bayesian isotope mixing model; Groundwater contamination; Nitrate accumulation; Random forest model; Stable isotope.
Copyright © 2023 Elsevier Inc. All rights reserved.