Using a simple 2D steady-state saturated flow and reactive transport model to elucidate denitrification patterns in a hillslope aquifer

Abstract

In the last 50 years, agricultural intensification has resulted in increasing nutrient losses that threaten the health of the lakes on the volcanic plateau of New Zealand’s North Island. As part of our efforts to understand the transport and transformations of nitrogen in this landscape, the 2D vertical groundwater transport model AquiferSim 2DV was used to simulate water flow, nitrate transport, denitrification, and discharge to surface waters in a hillslope adjacent to a wetland and stream discharging into Lake Taupo, Australasia’s largest lake. AquiferSim 2DV is a steady state model using the finite-difference stream function method for flow modelling and finite-volume mixing cell method for contaminant transport modelling. The ratio of horizontal to vertical hydraulic conductivity must be specified within the aquifer domain, as must effective porosity and denitrification rates. Boundary conditions consist of recharge fluxes and contaminant concentrations, as well as the assumed zone of discharge. Hydrodynamic dispersion is simulated through numerical dispersion, which depends on grid resolution. Denitrification reactions within each computational cell may include both zero-order and first-order rates. All parameters may be spatially heterogeneous. Previous applications of this model have been to essentially horizontal aquifer systems. By contrast, this hillslope system has sloping material layers and a dynamic and sloping water table. Extensions were made to AquiferSim 2DV, including representation of converging/diverging flow, which allowed a 2D steady-state model of this system to be developed. Comparison of model predictions with detailed water level and hydrochemical data from the site, however, showed that the model’s attractive simplicity in this case precluded adequate characterisation of what is essentially a 3D, transient system. While the model produced reasonable agreement with the concentration patterns under an average water table profile, predictions of oxygen and nitrate concentrations under low summer and high spring water table conditions were poor. The seasonal changes reflected an annual recharge pulse of fresh, oxidised water followed by gradual oxygen depletion till the next recharge pulse occurs in the following year, an essentially transient phenomenon which could not be represented using a steady state model. This in itself has provoked fresh thinking about the dynamic nature of flow and chemistry at the site.