SWATGenXSWATGenX
ResearchDocs
Open Explorer Work with us

Watershed-to-site PFAS fate & transport modeling

Five decades of an AFFF legacy at Wurtsmith AFB, simulated from a prescribed source rather than a flux fitted to the plume: a SWAT+ watershed model whose daily percolation drives a 250 m MODFLOW 6 model of the unsaturated zone and the aquifer, which carries the reported transport, and a nested 30 m site model around the monitoring wells.

1
Pick a point
anywhere in CONUS
→
2
Watershed model
HUC12 · SWAT+/MODFLOW 6 · 250 m
→
3
Site model
Catchment set · nested · 30 m
→
4
Compare with wells
observed PFOS · 250 m transport

Two scales, nested

At Wurtsmith the study works at two scales, nested: the SWAT+ watershed model, whose percolation drives the 250 m MODFLOW 6 model, and a 30 m site model of the catchments around the wells, whose single inlet the watershed model supplies. The 30 m flow solves, but its transport run does not converge, so every simulated Wurtsmith PFOS concentration on this page comes from the 250 m model.

This is one nest, chosen for one site. Where to place the cut in general — at a gauge, a HUC12 outlet, or a single catchment — what can cross it, and the cases where nesting is physically indefensible are set out on nested watershed modeling; the site-scale build itself is described on site-scale modeling.

Each stage has PFAS transport inside its own model: land and soil in our SWAT+ engine, the vadose zone in MODFLOW 6 UZF/UZT (which we extended for PFAS sorption), the aquifer in MODFLOW 6 GWT, and streams in the SWAT+ engine. In this study the vadose zone and aquifer run as one MODFLOW 6 simulation, offline from the engine, from a prescribed source; this study does not use the engine's daily two-way coupling of SWAT+ and MODFLOW 6. No WASP, RT3D, or other add-on water-quality model is used anywhere in this workflow.

Watershed scale
HUC12 · 250 m
The SWAT+ watershed model, whose daily percolation drives a 250 m MODFLOW 6 model with three layers kriged from state well records. Every simulated Wurtsmith PFOS concentration on this page is from that 250 m model.
Site scale
Catchment set · 30 m
The 118 NHDPlus HR catchments holding all 35 plume wells, most of the surface-water stations and the three source areas, rebuilt at 30 m, with the watershed model supplying its single inlet. Its flow solves; its transport run does not converge, so it carries no concentration on this page.
Vadose → aquifer
explicit UZF/UZT
From a prescribed source, PFAS travels through the unsaturated zone to the water table, into the plume and on to the stream (a leg the study does not validate), in one MODFLOW 6 simulation. Under Wurtsmith's 3.5 m of sand, sorption in the unsaturated zone, not water movement, sets the delay.

Three sites, three depths to water

The three sites differ in depth to water, infiltration and water content: two firefighting-foam sites and a tannery legacy source. Wurtsmith sits at the shallow end, under 3.5 m of sand, and even there most of the PFOS released while the source was active was still above the water table in 2024: the delay is sorption, not water movement. The deeper sites below run on the same method.

Four-panel study-area figure: a locator map of the three modeled sites, then each site's model domain — Wurtsmith AFB in Michigan, Gabreski ANGB in New York, and the Stegman Creek site in Michigan — with PFAS source zones marked.
The three modeled sites, two firefighting-foam sites and a tannery legacy source, spanning depths to water: (a) where they are, then each site's model domain and source zones — Wurtsmith AFB, MI (3.5 m to water), Gabreski ANGB, NY (10–12 m), and Stegman Creek (Rogue), MI (water table a median 8 m below land surface in its basin model, 0.8–26 m).

What stays fixed across the three sites. The claim above is that the method does not change between them — so here is the method that does not change.

Source term
Prescribed, not fitted
At each site the source is prescribed, not fitted to the measured PFOS. At Wurtsmith it is a single cell, located from the two monitoring wells with the highest PFOS, whose concentration is inverted from the PFOS measured in the source-area soil through an assumed partition coefficient; at Gabreski the mapped fire-training area is the source, and its release dates and footprint were checked against the observed plume.
Sorption
Fixed before the models were built
A linear partition coefficient for low-organic-carbon sand, from the literature, fixed before the models were built and not adjusted to the observations. In the unsaturated zone at the two firefighting-foam sites, PFOS is also held at the air–water interface, with declared values, not measurements at these sites; we modified MODFLOW 6's unsaturated-zone transport to carry both sorption terms, each dependent on water content. Below the water table there is no air–water interface, so only the solid phase holds it.
Advection & dispersion
MODFLOW 6 transport on a flow field calibrated to heads
MODFLOW 6 groundwater transport carries PFOS on a flow field whose hydraulic conductivity was calibrated to measured water levels, data that carry no PFAS information. The dispersivities were fixed before the models were built; no site measures them.

In 3D: the second site, Gabreski ANGB, New York

This 3D view is Gabreski, the second site in the figure above, not Wurtsmith, the site this page describes. It is a 250 m MODFLOW 6 model with unsaturated-zone transport, and it steps through 54 years of simulated PFOS in groundwater.

PFOS from a fire training area, Gabreski Air National Guard Base, New York: a still of the 3D scene (2024)

Checking your access…

PFAS fate and transport

PFOS from a fire training area, Gabreski Air National Guard Base, New York

Model
MODFLOW 6 with unsaturated-zone transport, 250 m cells, 3 layers
Quantity shown
Simulated PFOS in groundwater, ng/L, as shells at 4, 40 and 400 ng/L
Period
1971 to 2024, one frame per year (54 years)
Source
An assumed infiltration of 50,000 ng/L over 3 x 3 cells at the base's fire training area (position approximate), 500 ng/L from 2016
Flow
Recharge and boundaries are constant; the simulated water table settles in the first three years (by up to 0.9 m) and is steady after that, so the scene draws one flow field
Wells
8 of the report's 25 sampled wells (those with a known position), PFOS observed in 2020 (6 with a value; 2 below the report's screening level); positions approximate

Scope of this model

  • The model covers one HUC12 watershed, 020302020603, on the south shore of Long Island, New York.
  • One source area: the base's fire training area (position approximate).
  • It shows PFOS in groundwater only. It gives no prediction for the bays or for surface water.
  • The source concentration is an assumption of the research configuration, not a measured release.
  • A research configuration, not a calibrated site prediction.

The result — plume and vertical structure

Simulated 2017 PFOS plume at Wurtsmith AFB, with the prescribed AFFF source cell and the monitoring wells
Simulated 2017 PFOS plume: the prescribed AFFF source cell (star), the monitoring-well network sized by observed PFOS, and the model boundary (dashed).
North–south cross-section through the simulated plume core
A vertical slice through the simulated plume core, from the 250 m model; the blue line is the water table.

What this model does not do. The limits the manuscript states, kept here beside the plume rather than in a footnote.

Not calibrated to PFOS
No parameter was adjusted to improve the agreement between simulated and measured PFOS, so simulated concentrations are not calibrated magnitudes. The comparisons with wells test the plume's extent and depth, not its fit, and they are weak tests of the aquifer leg.
The stream leg is not validated
MODFLOW 6 carries the groundwater discharge into the streams. That leg conserves mass, and it is not validated.
A coarse unsaturated zone
The unsaturated zone is one or two mixed cells per column, so the PFOS it holds and the delay it causes are lower bounds. The first arrival comes from a separate steady-flux calculation that depends on a dispersivity no site measures.
Unknown recharge after closure
After the Wurtsmith source stopped, the water that reached it is not recorded, so the study brackets it with two rates.
Not every run at every site
Gabreski has no run that delivers the same PFOS straight to the water table for comparison. The tannery site's runs carry no air–water interface term, and they end in 1989, so that site's comparison covers the source period only.
The plume above is from the 250 m model
A 30 m Wurtsmith transport run does not converge, so the plume and section above come from the 250 m model.
PFOS only
The simulations are for PFOS. Other PFAS enter the same screening calculation through their own solid-phase and interfacial coefficients.
Depth on three layers
The aquifer has three layers, so depth is read at three levels.

By the numbers

77 km²
site domain
the 30 m site model: the 118 catchments holding all 35 plume wells; the 250 m MODFLOW 6 model reaches 16 of them
1971–2024
five decades simulated
from the first year of release to 2024, from a prescribed source, with no flux fitted to the plume
250 m
transport grid
every simulated Wurtsmith PFOS concentration on this page is from the 250 m model; the 30 m transport run does not converge

Not calibrated to PFOS, and presented as a research demonstration rather than a regulatory product. The study is published in the Journal of Contaminant Hydrology 284, 105141. Its source is prescribed: at Wurtsmith, one cell located from the two highest-PFOS wells, with a concentration inverted from measured source soil through an assumed partition coefficient, and no parameter was adjusted to improve the match to measured PFOS. Groundwater observations were extracted from public agency reports.

Model your own watershed

The SWAT+ watershed model here was built by the same pipeline that builds every model on SWATGenX. With model access, pick a point to build a SWAT+ model; a groundwater study (offered in Michigan Lower Peninsula, Indiana, Minnesota, Ohio, Wisconsin) adds a steady-state MODFLOW 6 model. PFAS transport studies like this one are arranged directly.