SWAT+ model generation · United States
Order a SWAT+ model. Get a package you can open and run.
Pick a USGS gage, a HUC12 outlet, or a whole HUC8. SWATGenX clips the national terrain, soils, land cover and climate, delineates the network from NHDPlus HR, builds the HRUs, and returns a complete SWAT+ project — QGIS database, text inputs, generated weather, and a verification run already done.
No desktop GIS, no QGIS install, no code required — and a REST API plus MCP server if you would rather not click at all.
Ordering one, in the actual interface
Not a diagram of the flow — the flow. Both of these are screenshots of the live site, and the dialog tells you what it is about to do before it does it: which model type, which grids, how routing falls back, and roughly how long the build takes.
What is actually in the package
A complete SWAT+ project, not a folder of intermediate rasters and not just output tables. The tree below is a real completed build for USGS 01012515 — 291 files in TxtInOut, the QGIS project database, per-station generated weather series, NetCDF output from the verification run, and a manifest recording which inputs produced it.
SWATGenX_v1/ ├── SWATGenX_v1.sqlite QGIS project database ├── build_manifest.json what was built, from which inputs ├── provenance.json data lineage for every layer ├── Watershed/ │ ├── Rasters/ clipped DEM, land use, soils │ ├── Shapes/ subbasins, HRUs, rivers, aquifers │ └── Text/ ├── Scenarios/Default/ │ ├── TxtInOut/ 291 SWAT+ input files │ │ ├── file.cio codes.bsn time.sim │ │ ├── hru.hru channel.cha aquifer.aqu │ │ ├── soils.sol landuse.lum field.fld │ │ ├── r74_c1345.pcp/.tmp/.hmd/.slr/.wnd │ │ └── channel_sd_day.nc NetCDF output │ └── Results/ shapefiles + output SQLite └── streamflow_data/ observed USGS record
Contents of one real build; total size and file counts vary with watershed size (this one is ~71 MB unpacked, ~20 MB zipped).
Three ways to define the basin
Every order starts by defining a drainage basin, represented as a set of one or more WBD HUC12 polygons. NHDPlus HR catchments inside that basin build the landscape units (HRUs) and connect them to the routed stream network.
| Entry point | What you choose | How SWATGenX turns that into a basin |
|---|---|---|
| USGS station (gage) | A USGS site number | The station is looked up in the platform’s station table and its precomputed list of upstream HUC12s becomes the drainage basin for that gage’s outlet. |
| Catalog outlet (HUC12) | An outlet HUC12 in Watershed Explorer | An upstream walk from the selected outlet gathers every upstream HUC12 draining to it (with safety limits on very large watersheds). |
| Whole basin (HUC8) | An entire HUC8 subbasin | The basin is derived as the HUC12s falling within that HUC8, and the package is assembled for the whole subbasin footprint. |
Prepared and waiting across CONUS: 87,387 HUC12 catalog polygons, 2,246 HUC8 subbasins, 24,750,692 NHDPlus HR flowlines and 24,833,258 catchments across 233 VPUIDs.
Watch one get built
The film opens at the build sequence: a watershed resolved from raw terrain into hydrologic response units, streams and subbasins, then the aquifer and water table for a coupled run.
Nothing is fetched from YouTube until you press play.
Or order it from code
The same job the button triggers is a documented REST call — order, poll, download. There is also a public MCP server, so an AI agent can order and retrieve a model without a browser.
import os, requests
headers = {"Authorization": f"Bearer {os.environ['SWATGENX_API_KEY']}",
"Content-Type": "application/json"}
# a 12-digit WBD outlet HUC12, exactly as picked in Watershed Explorer
r = requests.post(
"https://www.swatgenx.com/api/model-settings/explorer-watershed",
headers=headers,
json={"outlet_huc12": "070500020602",
"ls_resolution": 250,
"dem_resolution": 30},
timeout=120,
)
print(r.json()) # -> task id; poll /api/task_status/<task_id>Or let an AI assistant do it
SWATGenX runs a public MCP server, so an assistant that supports connectors can talk to the platform directly — search models, read calibration results, query the national groundwater and PFAS data, preview a build, and order one. You connect it once by pointing the client at https://www.swatgenx.com/mcp, then choose which tools it may use.
SWATGenX connected as an MCP connector. Each tool carries its own permission, so read-only access and model ordering are separate decisions — you are not handing over blanket access.
What you are getting, stated plainly
Downloads are automated SWAT+ setups for your review and calibration—not “finished” calibrated models unless you complete that workflow separately.
Every build auto-runs a 10-year verification simulation and a SWAT+ Check screening report, so you can see how the model behaves before committing time to it. Calibration is a separate, optional step with its own queue and estimate — how calibration works.