National PFAS Soil Inventory | SWATGenX
Where does PFAS actually sit in the soil — and how much? We harmonized measured soil concentrations from public agency reports across 41 states into one depth-resolved, source-anchored inventory.
The inventory at a glance
Why it exists
No national layer of measured soil PFAS exists — the data is scattered across thousands of pages of state and federal investigation reports, in inconsistent formats, mostly as PDFs. A watershed or groundwater PFAS model that wants a real source term instead of a guess has nowhere to look.
This inventory is that missing layer. Every value is transcribed as-reported from a public document — no interpolation, no modeled substitutes — and kept depth-resolved, so a model can read the actual concentration profile from the surface down toward the water table. The dominant setting is AFFF fire-training areas, the same legacy sources that drive most groundwater PFAS plumes.
The extraction combines deterministic table parsing with a vision-language pipeline for scanned reports, behind ten quality-assurance gates. The method and dataset are described in our national PFAS soil inventory manuscript.
What it feeds
The inventory is the source-term half of coupled PFAS modeling on SWATGenX. Paired with the national groundwater lithology inventory, it lets a SWAT+MODFLOW-6 model start from measured soil concentrations and carry them through the unsaturated zone, the aquifer, and back to the stream — instead of prescribing a plume out of thin air. At sites like Gabreski (Long Island) and Wurtsmith (Michigan), the measured soil profile is the model's input.