NPSAT - An Overview

Nonpoint Source Assessment Toolbox (NPSAT)

The Nonpoint Source Assessment Toolbox (NPSAT) is a groundwater modeling framework designed to evaluate the fate and transport of nonpoint source (NPS) contaminants such as nitrate and salts leaching to groundwater from agricultural, urban, and natural land uses. Its primary application is to assess groundwater quality in irrigation, public, and domestic supply wells.

The NPSAT framework – in contrast to other groundwater flow and transport models -  is designed specifically (a) for high-resolution nonpoint source contaminant transport across entire groundwater (sub)basins and (b) to facilitate “on-the-fly” evaluation of dozens, hundreds, or thousands of different user-designed nonpoint source contaminant leaching future scenarios. These scenarios represent user-selected application of alternative source management practices associated with user-selected specific land uses and/or crops.

 

Modeling Framework

NPSAT is a quasi-3D streamline transport model: The complexity of full 3D contaminant transport is represented by millions of independent 1D transport problems along streamlines [1]. To the degree that changes in salinity or nitrate source loading do not affect the flow-paths of groundwater, this streamline approach to groundwater transport modeling is computationally very efficient – allowing for the simulation of a new contaminant loading history scenario in a fraction of the time (seconds instead of hours) of standard 3D contaminant transport simulation tools.

  1. At the heart of NPSAT is a high-resolution groundwater flow model. To enhance spatial resolution where it is most critical (i.e., near wells or regions with steep hydraulic gradients), NPSAT incorporates adaptive mesh refinement (AMR). This approach allows for fine-scale modeling of flow paths in key areas and for appropriately representing hydrogeologic heterogeneity, while maintaining overall computational efficiency across the domain [2].
  2. Contaminant transport is simulated in two steps: First, particles are released from well screens and traced backward through the aquifer until they reach the recharge boundary [3]. This identifies the contributing recharge zones and potential sources of contamination for each well. Second, along each streamline (in1-D), NPSAT solves the transport problem for a rectangular pulse input function (one unit of contaminant loading to groundwater for one year) at the source (start) of the streamline. This produces a so-called Unit Response Function (URF) at the well (end of streamline), which describes how a unit of contaminant travels through the aquifer over time.
  3. The previous steps – (A) flow modeling and (B) particle path simulation and URF calculation – are computationally expensive and are preparatory steps for user scenario modeling. For the simulation of these future contaminant loading scenarios, the URFs are used to very quickly simulate the water quality outcome at tens of thousands of wells over decades to centuries: The breakthrough curve for each streamline is computed by convoluting the URF with the actual, location-specific nonpoint source loading history. Concentrations at the well are obtained by performing a flux-weighted integration of the output concentrations from all streamlines entering the well. Many future scenarios can be evaluated in this way very quickly.

 

CV-NPSAT

CV-NPSAT is a regional implementation of the Nonpoint Source Assessment Toolbox (NPSAT) applied to California’s Central Valley (CV) groundwater basin. This application spans the entire Central Valley floor and utilizes data and flow information from either one of two regional hydrologic models: CVHM2 (Central Valley Hydrologic Model version 2) and C2VSim (California Central Valley Simulation Model).

Current groundwater conditions in the Central Valley are marked by a significant imbalance between groundwater pumping and recharge, resulting in a net groundwater storage loss of approximately 2.3 to 2.9 million acre-feet (MAF) per year.

The goal of CV-NPSAT is to assess the long-term impact of nitrate and salt concentrations under flow conditions that reflect sustainable groundwater use. To support this, two book-end sustainability flow scenarios were developed for each base model:

  • Pumping-adjusted scenario: Groundwater pumping is reduced to match current recharge levels; this is a book-end of future groundwater flow and velocity conditions.
  • Recharge-adjusted scenario: Groundwater recharge is increased to meet current pumping demands; this scenario book-ends recent historic groundwater velocity conditions, but adds the recharge needed to keep water levels from declining.

For each flow scenario, a highly detailed steady-state flow simulation was developed. Key model inputs, such as recharge and pumping, are downscaled from the coarse resolution (~1 square mile) of the base models (CVHM2, C2VSIM) to the finer resolution of NPSAT (150–200 feet): Groundwater recharge is downscaled from model elements to the field scale using high-resolution land use data provided by CV-SWAT, a watershed-scale land surface model for the Central Valley with sub-field scale resolution. Groundwater pumping at the model-element scale from CVHM2 and C2VSim is spatially allocated to individual wells using well data from the California Department of Water Resources (DWR)[4].

CV-NPSAT Verification/Calibration

The NPSAT framework has been evaluated against a full 3D transport model (MODFLOW/MT3D) under various nitrate management scenarios and under both steady-state and transient flow conditions, using an existing groundwater model for the Modesto area [5]. The results indicate that NPSAT’s predicted concentration distributions across an ensemble of wells closely match those with the MT3D, when both are based on the same steady-state MODFLOW groundwater flow model. Both approaches (NPSAT and MT3D) tend to slightly overestimate well concentrations (up to 10%) when compared to MT3D simulations that are coupled with the transient MODFLOW groundwater flow model. However, when assessing long-term changes in nitrate concentrations under different land management scenarios, the NPSAT results are practically identical to results obtained from MT3D simulations with transient MODFLOW.

As described, the hydrogeologic data and groundwater flow budgets in CV-NPSAT are derived from calibrated flow models, CVHM2 and C2VSim. However, the transport parameters, including aquifer porosity and water content of the unsaturated zone, were estimated through calibration using groundwater age tracer data. The calibration process identified optimal pairs of porosity and water content values that yielded simulated groundwater ages consistent with observed measurements.

Finally, CV-NPSAT was validated against measured nitrate concentration data. The results demonstrate that NPSAT is capable of reproducing both the range of observed nitrate concentrations and their temporal evolution, indicating strong agreement between modeled and observed trends.

Limitations of CV-NPSAT and other Models

Applications of NPSAT, like that of other groundwater flow and transport models including MODFLOW with MT3D, is limited by the knowledge about the hydrogeologic architecture of the aquifer system and by knowledge of exact well screen location, of well pumping, of detailed spatial distribution of recharge and associated contaminant (nitrate, salt) concentrations, and of other aquifer boundary conditions influencing contaminant transport and groundwater quality. It is therefore not possible with any model to exactly predict the water quality legacy and future at each of hundreds to tens of thousands of potentially impacted wells.  NPSAT and other nonpoint source model simulations are therefore best interpreted by considering that these models accurately predict the general pattern, trend and variability of water quality across ensembles of wells. The reliability of these predictions increases with the number of wells included in the ensemble, making the model highly effective for regional assessments. However, these models cannot make reliable predictions for specific wells. That would require detailed site investigations.

 

References

[1] Kourakos, G., F.Klein, A.Cortis, and T.Harter (2012), A groundwater nonpoint source pollution modeling framework to evaluate long-term dynamics of pollutant exceedance probabilities in wells and other discharge locations, Water Resour. Res., 48, W00L13, doi:10.1029/2011WR010813

[2] Kourakos G. and Harter T. (2021) Simulation of Unconfined Aquifer Flow Based on Parallel Adaptive Mesh Refinement. Water Resources Research. 57(12), doi: 10.1029/2020WR029354.

[3] Kourakos G., Harter T. and, Dahlke H.E. (2024) Ichnos: A universal parallel particle tracking tool for groundwater flow simulations. SoftwareX 28,101893. 10.1016/j.softx.2024.101893

[4] Kourakos G., Pauloo R.,and Harter T. (2024) An Imputation Method for Simulating 3D Well Screen Locations from Limited Regional Well Log Data. Groundwater. doi: 10.1111/gwat.13424.

[5] Kourakos G., Bastani M and, Harter T. (2025) Simulating Nonpoint Source Pollution Impacts in Groundwater: Three-Dimensional Advection–Dispersion Versus Quasi-3D Streamline Transport Approach. Hydrology, 12(3), 42. https://doi.org/10.3390/hydrology12030042