Calibration and scenarios
Compare the model with flow monitor, level logger and rain gauge records (NSE, RMSE, peak, volume, timing), run a sensitivity sweep and an automatic calibration, and save and compare scenarios.
Load field records
Tools > Calibration... opens the Calibration window. Under Field records, pick a CSV or Excel file with one reading per row: a time (date and time, hh:mm, or decimal hours from the first report day) and a value. Name the conduit for a flow monitor, the node for a level logger, and the subcatchment for a rain gauge.
- Pick the file. For a workbook, pick the sheet.
- Check the columns DrainPilot found: Time and Value are required; Location ID, Date (if separate), Type and Unit are optional.
- For a sheet from a single monitor with no location or type column, type the Location (for example C12) and choose the Type and Unit.
- Press Read records. The file is listed with its number of readings and any rows that could not be read. Load more files the same way.
- If a logger ran on a different clock, set the Clock offset in minutes.
- Run the model. The window places each record at the report times and says which locations are not in the run and how many readings fall outside the simulation period.
| Record type | Units accepted |
|---|---|
| Flow | L/s, m3/s, m3/h, MLD, CFS, GPM, MGD |
| Depth | m, cm, mm, ft, in |
| Water level (elevation) | m, ft |
| Rainfall | mm/hr, in/hr |
Read the fit
The Statistics tab gives one row per record: number of points, NSE, RMSE, R2, observed and simulated peak, peak error %, volume error % (flow only), time to peak error in minutes (positive when the model peaks late), and whether flow and depth records are within the storm bands. Click a row to plot it, or map to find its location.
- NSE (Nash and Sutcliffe 1970) above 0.75 is rated very good, above 0.65 good and above 0.5 satisfactory (Moriasi et al. 2007). Those ratings were set for monthly data, so for storm hydrographs treat them as a guide.
- Storm bands: peak flow within +25 % to -15 %, volume within +20 % to -10 %, peak depth within 0.1 m, as commonly quoted from the WaPUG 2002 Code of Practice. DrainPilot could not confirm them against the printed code.
- The Observed vs simulated tab plots the chosen record over time beside a scatter of observed against simulated values.
Choose the parameters
Under Parameters to calibrate, Add a parameter... lists the parameters the model has: Subcatchment width, Imperviousness %, Subcatchment slope %, Manning n, impervious, Manning n, pervious, Depression storage, impervious, Depression storage, pervious, Manning n, conduits, Horton max rate, Horton min rate, Horton decay constant, Green-Ampt suction head, Green-Ampt conductivity, Green-Ampt initial deficit, Curve number, RDII R (fraction of rain), RDII T (time to peak), RDII K (recession ratio). Infiltration parameters appear for the infiltration method in use. Each varies by a multiplier around the project's own values, within a +- range (30 % to start). Leave IDs empty to change every object, or list the IDs to change, separated by commas.
Sensitivity sweep
On the Sensitivity tab, set how far to move each parameter (20 % to start; the steps are shown, for example -20, -10, 0, 10, 20 %) and press Run sweep. Each cell is a full SWMM run and shows the objective at that move (lower is better; hover for NSE, peak and volume error). The last two columns give the relative sensitivity of the simulated peak and volume: 1 means a 10 % change in the parameter moves the result by 10 %. Cancel stops it.
Automatic calibration
- On the Auto-calibrate tab, set the weights of the objective: on 1 - NSE (1), on volume error (0.5) and on peak error (0).
- Set the genetic search: Population (16), Generations (12), Seed (1) and Polish (on). The window says how many engine runs that can take.
- Press Calibrate. Progress shows the generation, the best objective so far and the runs made; Cancel stops it.
- Read the objective before and after, the convergence curve, and the multiplier found for each parameter with the number of objects it changes.
- Press Apply N changes to the project to write the new values into the model as one edit, then run the model again to see the new fit.
The search is seeded, so the same records, parameters and seed give the same answer. The objective is the weighted mean over the flow and depth records of (1 - NSE), the absolute volume error and the absolute peak error; rainfall records are not fitted. It needs at least one flow or depth record at two or more report times.
Scenarios
Tools > Scenarios... saves complete copies of the open project as named scenarios, opens them, and compares two runs. The type list offers starting names (Base 2026, Design 2056, 2-year storm, 5-year storm, 10-year storm, With LID, Without LID, with the current year in place of 2026); you can type any name.
- Save open project as scenario stores the open project under the name in the box.
- Each saved scenario shows run stored or not run, and Open, Update and Delete buttons. Update replaces the scenario with the open project and clears its stored run. Delete asks you to confirm.
- Open replaces the open project with the scenario. The open file is then named after the scenario (for example 5-year storm.inp).
- To compare, pick two of Open project and the saved scenarios and press Compare. A scenario's stored run is reused unless you tick Rerun stored; a fresh run is stored for next time. The two runs go one after the other.
The comparison gives a summary line (flooded nodes and flood volume, conduits over capacity, how many nodes got deeper or shallower) and four tables, largest changes first: node peak depth, node flood volume, link peak flow, and peak over full flow. Increases are red, decreases green.