Summary

The goal of this study was to develop an HEC-HMS model of the Kaskaskia basin and create forecast alternatives to view and analyze the results of two precipitation forecasts. Model development included developing a basin model of the Kaskaskia River watershed. Multiple parameter sets were developed to represent both a wet/typical and a dry condition. Both watershed conditions were calibrated to historic events to develop final parameters.

Forecast alternatives allowed the forecaster to analyze two precipitation forecasts and quickly and easily compare the results. The forecaster developed two zone configurations; the gage configuration was assigned to transform and baseflow parameters, and the soils configuration was assigned to loss parameters. The forecaster initialized baseflow and reservoir elevations using observed stream gage and stage gage data. Loss and transform parameters were efficiently calibrated using the Forecast Parameter Adjustment editor and the Slider Bar editor. Computed flows were compared to observed flows by systematically calibrating the model from upstream to downstream utilizing the compute to point compute option. The built-in model computational efficiency measures in HEC-HMS limited the simulation run time by only computing results for those elements that required results to be updated. Forecast blending was used to replace computed flows in the look back period with observed flows. The results of each forecast alternative were quickly compared and were ready for use as inflows for reservoir operations and inundation mapping.

Key Takeaways

Key Steps in HEC-HMS

  1. Configure a Forecast Alternative with the basin and meteorologic models, computation interval, start time, forecast time, and end time.
  2. Use Zones to organize parameter adjustments, and initialize subbasin baseflow, reach outflow, and reservoir pool elevation from reviewed observations where appropriate.
  3. Check unblended results during the lookback period. Use the Combined Parameter Editor to adjust zones and element overrides, and consider forecast optimization with a linked Optimization Trial to estimate zonal parameter scale factors within realistic bounds. Review the resulting parameters and hydrographs from upstream to downstream, using computation points to limit recomputation.
  4. Configure Blending only after checking the observed data and the underlying model fit; review the transition after the forecast time.
  5. Compute the alternative meteorologic scenarios and compare peak flow, timing, runoff volume, and reservoir conditions. Record inputs and adjustments when providing hydrographs for reservoir operations or hydraulic modeling.

Common Pitfalls

  • Unchecked observations. Review time references, units, vertical datums, missing data, and implausible changes before initialization or blending. Inspect estimated and final initialization values and document any defaults or substitutions (EM 1110-2-3600 (USACE, 2017), paragraphs 5.3.2.6 and 5.3.2.7).
  • Blending mistaken for calibration. Blending replaces computed flow with observed flow during the lookback period and can conceal model errors. Check unblended hydrographs and the selected Step, Taper, or None transition; a 72-hour taper is a case-study choice, not a general requirement.
  • Scenario spread treated as probability. Zero future rainfall and a selected rainfall depth are conditional scenarios. Their spread does not define forecast probabilities; EM 1110-2-3600 (USACE, 2017), paragraph 6.3.1.3, discusses ensemble approaches to meteorologic uncertainty.
  • Unreviewed reservoir assumptions. Check initial pool elevation, any reset at forecast time, and the Forecast Reservoirs treatment of missing future discharge data. Document these assumptions before using the hydrographs to evaluate operations; releases remain subject to the approved water control plan (ER 1110-2-240 (USACE, 2016), paragraph 3-3a).
  • Outdated state or forecast window. Match the starting parameter set or saved state to current watershed conditions and inspect observation coverage relative to the forecast time. Choose a lookback period and forecast horizon suited to basin response and available meteorologic data; the case-study time window is not a universal setting.