Putting the NPSAT modeling framework to test: comparison to a fully three-dimensional, transient MODFLOW-MT3D model of nitrate transport in the northeastern San Joaquin Valley

Degradation of groundwater quality due to nonpoint source (NPS) contaminants from the land surface, especially nitrate in agricultural aquifers, is a major environmental concern. Developing numerical models to support policy decisions on the management of diffuse pollution sources across large landscapes is an important step to improve water quality in public supply and irrigation wells. This paper investigates the suitability of our Nonpoint Source Assessment Tool (NPSAT), a promising assessment tool for NPS pollution transport in large scale agricultural aquifers.

NPSAT simplifies the complex three-dimensional transport problem into a series of 1D transport simulations solved along 3D streamlines in the steady state (average) groundwater flow field (Kourakos et al., 2012). Each streamline connects the source area to the discharge point of interest and the contaminant level is evolved through time along the streamline by using the concept of unit response function (URF). Using the steady state flow condition and ignoring transverse dispersion are two key assumptions of this NPS transport approach.  How good, then, can NPSAT be in predicting nitrate across a group of wells in a region of interest within, e.g., the Central Valley of California?

For the purpose of demonstrating the suitability of NPSAT, we considered a real regional nitrate transport case study in Eastern San Joaquin Valley, CA, with spatial heterogeneity in aquifer properties and in nonpoint source nitrate loading.  The region was modeled using what is among the most common tools to assess detailed grounwdater pollution: the USGS' software suite MODFLOW and MT3D.  Using this software, two cases were simulated:

  • nitrate transport under steady state flow conditions and
  • nitrate transport under more realistic transient flow boundary conditions.

The NPSAT simulation results are then compared with both MT3D model outcomes, for different nitrate management practice scenarios.

The results show that the steady-state streamline approach has very similar outcomes to the ADE approach with steady state flow simulation, while both methods tend to overestimate the concentrations at the wells by up to about 10% compared to the ADE with transient flow field.

Further, the long-term attenuation of nitrate under alternative land management scenarios is the same between the streamline and the transient ADE results. Therefore, both methods are useful to reveal long-term effects of different ameliorating management practices on the contamination system.

Importantly, NPSAT's streamline approach offers several orders of magnitude efficiency gains in computational runtime, especially at large scale basins. In general, this work justifies the streamline approach as a reliable and effective alternative tool for groundwater quality assessment of nonpoint source contaminants at regional scales. The efficiency of NPSAT, relative to that of MODFLOW-MT3D, allows for the evaluation of hundreds to tens-of-thousands future nutrient management scenarios, a key to identifying viable future agricultural practices that meet sustainable groundwater quality goals.