A gamma mixture model-based approach for the estimation of natural background levels of ^NO3- – N in groundwater

The first stage in determining the chemical status of a groundwater body in an aquifer system is to determine natural background levels (NBLs). The various sources of NO3-–N in an environment as well as the interaction of natural and anthropogenic processes, present considerable obstacles in determining NBLs. Another constraint on NBL estimation is choosing the right statistical technique. In this paper, the NO3-–N levels of groundwater and the high-risk zones in the Densu Basin were evaluated. The evaluation was done using the Gamma mixture probability distribution and the iterative outlier removal technique. We also considered the strengths and weaknesses of these two models by assuming the NO3-–N concentration is coming from a single source. The Gamma mixture model was used to identify the sub-populations in the NO3-–N data set and also, estimate the optimal parameters for the hidden clusters. The initial component with the lower NO3-–N concentration was considered as the NBL. This was measured at 2.56 ± 2.56 mg/L, whereas considering a single source the iterative technique recorded the NBL at 5.6 ± 5.3 mg/L. Assuming the groundwater contamination is from a single source, then the iterative method introduces an error of 3.1 ± 2.8 mg/L in the NBL estimation. The result suggests that the Gamma mixture model is more robust in estimating pollution with multiple sources (that is, natural and human-induced sources).