Running FDR forecasts is nearly identical to running standard real-time forecasts in CWMS. A few variations and considerations are discussed on this page.
Forecast Time Window
In an FDR forecast, you are solely concerned with what happened in the lookback period of an event that has already occurred, unlike a real-time forecast where the primary concern is the future (forecast period). However, a full forecast time window including start, forecast, and end dates/times is still required for an FDR forecast.
The start time (and extract time) should be set to a time just before the beginning of the event. The forecast time should be set to a time just after the event. Depending on the watershed, this could be soon after the final reservoir has started to recede or as late as when a specific control point downstream is near baseflow conditions after the hydrograph has receded. The forecast time can differ based on the event or the basin, depending on the characteristics of the basin and how well the calibration fits the recession side of the hydrographs in the system. The end time should be set to allow for routing from the most upstream reservoir to the watershed outlet.
Computing
For the first approach, you will compute the FDR forecast as any other forecast and, when complete the FDR results can be viewed from the Reports tab in the CAVI or directly from within HEC-FIA. For the second, more complex, approach, each forecast run is computed separately and the HEC-FIA damage results from each of them are extracted for use in the post-processing to compute the FDR results.
Regardless of the approach, H&H and consequences results need to be scrutinized thoroughly as there are plenty of opportunities for mistakes in model linking and model setup. Additionally, calibration in HEC-HMS is very important and needs to be verified during the compute. More details on setup, computation, and verification of the individual models are found in their respective sections of this document.