What teams build with SWATGenX

The core product: an Editor-ready SWAT+ package for any USGS streamgage, HUC12 outlet watershed, or full HUC8 basin in the contiguous U.S. The pipeline preprocesses NHDPlus-HR hydrography into a valid routing network, builds HRUs from land use, soils, and slope, attaches PRISM/NSRDB climate and observed USGS streamflow, and runs automated quality checks before delivery — with every dataset's lineage documented in the methodology. Inspect real packages (HRU counts, delineation source, build timing) in the example-model catalog before spending quota.

End-user walkthrough: how it works. Packages are standard SWAT+ projects — no platform lock-in.

For facility- and contamination-scale questions, drill inside a HUC12 in the Explorer and hand-pick individual NHDPlus-HR catchments (with recursive upstream selection). The model domain is delineated at 30 m resolution, so a single site — an airfield, a plume, a monitored reach — gets its own nested model while the surrounding regional model supplies context. This is how the coupled site models at AFFF-impacted facilities are built.

Full guide: site-scale watershed modeling.

Where a study needs a real aquifer, SWATGenX constructs the MODFLOW 6 model automatically — layer geometry, hydraulic conductivity, and starting water table kriged from state water-well records for the selected watershed — and couples it to the SWAT+ water balance, with simulated heads evaluated against observed wells in the pipeline. The open SWATGenX engine also supports live daily two-way coupling through the MODFLOW 6 BMI/XMI interface.

How the coupling works, with validation results: SWAT+ ↔ MODFLOW 6.

Hosted calibration runs MMPSO — a multi-memory, role-aware particle swarm optimizer with a separate memory per gauge — against USGS streamflow, summing a floored daily-plus-monthly NSE objective across every assigned gauge, with optional regionalized parameters over tributary sub-watersheds. You see a cost estimate before launch, the job dispatches to a right-sized AWS EC2 instance (your model never queues behind someone else's), and convergence, ensemble envelopes, and holdout validation land on your dashboard. Compute is billed at cost.

Method detail: calibration methods; infrastructure: AWS EC2 calibration; measured results: calibration examples.

Calibrated SWAT+ fleets are the starting point for nutrient and sediment work: observed water-quality records are extracted per channel alongside streamflow, so loading studies inherit a hydrology that has already been defended. The flagship is the west-central Florida program — Peace, Myakka, and Tampa Bay basins, platform-generated and expert-finished, with streamflow calibration complete and nutrient baselines the next rung on the page's live status ladder.

Follow the program: Florida nonpoint-source pollution; tool landscape: water-quality modeling software.

The coupled engine simulates watershed-scale PFAS: Freundlich sorption and air–water interfacial partitioning in soils, advective–dispersive transport in the aquifer, and in-stream routing of groundwater discharge — demonstrated at AFFF-impacted sites and described in published and under-review research. PFAS simulation is a study-specific workflow on the open engine (with site data and expert review), distinct from the Explorer's screening layer of measured PFAS concentrations.

Start with the PFAS fate & transport guide, the watershed-to-site demonstration, and the publications behind it.

Before (or instead of) any model, the Watershed Explorer answers screening questions in the browser: a CONUS map with USGS gauges, bulletin-style flood-frequency classification, recent rainfall against PRISM percentiles, population and land-cover overlays, and consistent PDF exports for the gauge, HUC12, or HUC8 you are reviewing. Teams use it to short-list basins with the same filters and report layout across an entire portfolio — and the identifiers carry straight into model orders.

Jump in: Watershed Explorer; gauge-level flood screening: flood analysis.

The Developer API mirrors the workspace: everything you can do in the browser — order models, poll build status, fetch packages — can be driven from scripts, notebooks, or internal tools, against the same queue and the same model catalog your analysts see. That makes SWATGenX a build stage inside larger pipelines, not just a website.

Endpoints and examples: Developer API.

The Education hub turns the platform's process representations into interactive lessons: The Water Journey follows a drop of rain through canopy, soil, aquifer, and channel, and the PFAS journey does the same for a contaminant. Built for courses, outreach, and anyone who wants the physics before the model — no account required.

Start at Education.

The research programs behind the platform publish their data: a national groundwater inventory of harvested well-log lithology and water-level records, and a national PFAS soil inventory compiled from agency reports with documented QA. Both are browsable on the site and citable in your own work — as-harvested source data, not modeled fills.

Explore the groundwater inventory and the PFAS soil inventory.