Groundwater Nitrate Source Apportionment in the Pineland Sand Region Using Bayesian Modeling

Groundwater nitrate contamination imposes a safety concern. Mitigating groundwater nitrate levels requires a quantitative assessment of source contributions, which is also challenging to estimate considering the data uncertainty and the agriculture environmental complexity. The Pineland Sand Region in north central Minnesota, United States, features sandy soils and has a mixed land-use landscape of agriculture and natural background, with part of the area located within the White Earth Nation Reservation region. We analyzed the land-use patterns and identified nitrate contaminant sources in nature, wetlands, row crops, pasture, and urban areas. To estimate the nitrate source apportionment and its uncertainty in this study area, we first characterized these sources using calcium, sulfate, chloride, and sodium, and then applied a Bayesian mass-balance mixing model using a Monte Carlo Markov Chain approach in 164 groundwater wells and further obtained the probability distributions of the fractional contributions from those specific sources. The results show that nature and wetlands source fractions dominate sites with negligible agriculture impacts, while row crop and pasture sources are estimated with higher contaminant contributions in sites subject to more intensive agriculture management. We also discussed a few specific sites’ contaminant source fractions and suggest a future source characterization scheme to advance groundwater nitrate understanding in the study area.