🌊 SwaNET · Watershed Modeling Tutorial

Building a SWAT Watershed Model in One Click

Generate a watershed model from DEM, land use, soil, climate, and monitoring data; then use the saved time for refinement, calibration, visualization, and management insight.

SwaNET Tutorial S8Prereq: SwaNET account and login8 major sections · nested workflow

1Watershed modeling in one click

Capability: SwaNET can generate a ready-to-run SWAT watershed model from a user-selected watershed boundary or map extent. The workflow automatically assembles terrain, stream network, subbasins, HRUs, weather-generator inputs, simulation files, and visualization outputs so the user can focus on refinement, calibration, scenario testing, and management interpretation.

This tutorial demonstrates the single-click SWAT model builder using the Kalamazoo River Watershed in southwestern Michigan. The same workflow can be applied to other watersheds where the required global or regional datasets are available.

The original tutorial states that the one-click process creates a SWAT model, runs the simulation, and generates streamflow time series, subbasin water-yield maps, 3D visualization, and a Sankey diagram showing overall watershed water dynamics. Optional USGS/Canada streamflow data can be added for comparison.

Hydrologic foundationDEM-based watershed delineation, stream network creation, and subbasin definition.
Process representationLand use, soil, slope, HRUs, weather data, runoff, aquifer exchange, and streamflow.
Decision intelligenceWater balance, water yield, observed-vs-simulated comparison, and Sankey process connectivity.
Modeling mindset: the one-click model is not the end of modeling. It is the starting point. The value is that users spend less time wrangling input files and more time refining assumptions, adding local data, calibrating, testing scenarios, and explaining watershed behavior.
Tutorial context and Kalamazoo River Watershed modeling area.
Tutorial context and Kalamazoo River Watershed modeling area.

2Optional: download observed streamflow data

Observed USGS or Canadian streamflow data are optional, but they are highly useful because they let the model output be compared against real gauging-station records. In this example, the tutorial uses a USGS station for comparison after the model is built.

  1. Open Utilities -> USGS/Canada monitoring data -> Extract data.
  2. Draw a rectangle around the Kalamazoo area and click Save shape.
  3. Choose the temporal range and station-data options, then click Extract Monitoring Data.
  4. When prompted, save the downloaded data locally. Later, reload it using Utilities -> USGS/Canada monitoring data -> Load local.
Why this matters: observed streamflow turns a generated model into a model that can be evaluated, calibrated, and defended.
USGS/Canada monitoring-data extraction window for station and climate data.
USGS/Canada monitoring-data extraction window for station and climate data.

3Load the model boundary

The model boundary defines the spatial extent of the SWAT model. In this tutorial, the Kalamazoo River Watershed is loaded directly from the SwaNET server rather than manually uploading a shapefile.

  1. Click Load Model, then Create a New Model.
  2. Confirm deletion of the existing delineation file. If asked to load the latest project, click Cancel.
  3. Select Extract watershed map from server.
  4. For this US example, choose DataCenter 1 (US only) and Level04 (HUC 8-digit).
  5. Click near the Kalamazoo area so latitude and longitude populate automatically.
  6. Click Load Shapefile.

The Kalamazoo River Watershed boundary should appear on the map. If the displayed area is incorrect, zoom to the Kalamazoo region and repeat the extraction.

Boundary selection interface using server-extracted watershed data.
Boundary selection interface using server-extracted watershed data.
Kalamazoo River Watershed boundary loaded on the map.
Kalamazoo River Watershed boundary loaded on the map.

4Select one-click model options

The SwaNET One-click Model Options window controls how terrain becomes streams, how subbasins become HRUs, what weather generator is used, and whether observed stations are added for comparison.

One-click model options for delineation, HRUs, SWAT inputs, and station comparison.
One-click model options for delineation, HRUs, SWAT inputs, and station comparison.

DEM and stream threshold

Select DEM resolution based on watershed size. Larger basins usually work better with coarser DEMs such as 300 m or 1000 m; smaller areas may justify 90 m or 30 m DEMs where available. This example uses the default 300 m DEM and 1.5% DEM cell threshold.

Interpretation: a lower DEM threshold creates more streams and subbasins; a higher threshold creates a simpler network.

HRU creation and selection

HRUs combine land use, soil, and slope. This example uses 400 m land-use and soil maps, one slope band, and a target HRU count equal to four times the number of subbasins.

Weather generator and SWAT inputs

For US watersheds, a US weather generator may be used; outside the US, select the global CFSR option. This tutorial uses WGEN_US_FirstOrder, no dummy point-source files, and heat units calculated from the weather database.

Observed-station comparison

If station data were downloaded, load stations, select stations to create outlets, and use a snap threshold. The example uses USGS-04108670 and a 300 m snap threshold. Then click Build Model.

5Understand the automated build process

Although the workflow is launched with one click, SwaNET performs a complete watershed-model construction sequence.

Watershed delineation

  1. DEM extraction: selected DEM is extracted for the model boundary.
  2. Stream network creation: TauDEM uses the DEM and threshold to generate streams.
  3. Outlet snapping: valid stations are snapped to the stream network.
  4. Final watershed: TauDEM finalizes the watershed and subbasins.
Automated DEM extraction and stream-network generation.
Automated DEM extraction and stream-network generation.
Snapped outlets and finalized watershed delineation.
Snapped outlets and finalized watershed delineation.

Land use, soil, and HRU creation

SwaNET extracts land-use and soil maps, combines them with elevation and slope, creates potential HRUs, and selects the final HRUs based on the selected option.

Extracted land-use and soil maps.
Extracted land-use and soil maps.
Potential HRU map and summary of final HRU, SWAT input, simulation, and output creation.
Potential HRU map and summary of final HRU, SWAT input, simulation, and output creation.

Write SWAT inputs and run simulation

SwaNET writes SWAT input files and runs the simulation. If observed station data are available, the simulation period is matched to the available data; otherwise the tutorial notes a 1990-1995 run.

6Review generated outputs

When the process completes, SwaNET loads the model and displays 2D map results, streamflow, water balance, and 3D visualization outputs.

One-click model-building outputs including map, water balance, and streamflow plot.
One-click model-building outputs including map, water balance, and streamflow plot.

Water-yield maps

Subbasin water-yield maps help identify which areas generate more runoff or water yield, providing a first screen for management priorities.

Observed-vs-simulated streamflow comparison and subbasin water-yield map.
Observed-vs-simulated streamflow comparison and subbasin water-yield map.

Observed vs. simulated streamflow

If a gauging station has observed data, SwaNET plots observed and simulated streamflow together. This first comparison helps identify timing errors, peak-flow bias, baseflow mismatch, and volume imbalance.

7Interpret water-balance and Sankey diagnostics

Water-balance outputs summarize how precipitation and stored water move through overland flow, shallow aquifer, deep aquifer, and in-stream compartments.

Overland balanceRunoff, infiltration, evaporation, and routing to streams.
Shallow aquifer balanceRecharge, lateral flow, return flow, and groundwater storage.
Deep aquifer balanceDeep recharge, storage, pumping, and baseflow contribution.
In-stream balanceUpstream inflow, lateral inflow, losses, evaporation, and downstream flow.
Overland and shallow-aquifer water-balance graphics.
Overland and shallow-aquifer water-balance graphics.
Deep-aquifer water-balance graphic.
Deep-aquifer water-balance graphic.
In-stream water balance and Sankey diagram of watershed water balance.
In-stream water balance and Sankey diagram of watershed water balance.

Sankey chart as watershed intelligence

The Sankey diagram turns water-balance accounting into a process-connectivity map. Flow-path width communicates relative magnitude, making it easier to see dominant processes and how water transfers among compartments.

Why this is powerful: as users refine data, calibrate parameters, or test BMP and climate scenarios, the Sankey diagram evolves into a diagnostic lens, design tool, and stakeholder communication device.

8Refine, calibrate, and manage the watershed model

The generated model is a strong starting point. The next stage is to improve it using local data, observed records, management knowledge, and scenario assumptions.

Add observed or projected weather

Open Refine Model -> Edit Weather Data. Users can add precipitation, temperature, relative humidity, solar radiation, and wind speed.

Weather-data refinement interface.
Weather-data refinement interface.

Manual or automatic calibration

Open Refine Model -> Calibrate SWAT Model. Users can use one-at-a-time calibration or Auto calibration using a genetic algorithm.

SWAT model calibration interface.
SWAT model calibration interface.

Management scenarios

After calibration, the model can support scenario testing for land-use change, climate variability, agricultural BMPs, erosion control, water quality management, and watershed planning.

The MAGNET difference: rapid prototyping meets deep customization. Users move from global data to local insight, from automatic model generation to calibrated decision support, and from static outputs to process-based watershed storytelling.