A multi-faceted approach to improve short-term, real-time flood forecasting

Heavy rainfall events, which can cause flash flooding, are quite common in Iowa during the warm season. Hydrologic models in combination with short-term precipitation forecasts may help to provide an early indication of the potential for these floods to occur. Further, short-term precipitation forecasts made using multiple modeling components can provide an indication of the uncertainty in both the location and intensity of the predicted precipitation event, both of which are critical to accurate flood forecasting. We will discuss an approach that we have applied when heavy rainfall is anticipated over an 800 x 800 km domain centered on the National Weather Service Johnston, IA office (NWS DMX) office. The Advanced Regional Prediction System (ARPS) three-dimensional variational (3DVAR) radar data assimilation program is used at the initialization of a poor man's 12-h WRF-ARW ensemble. Quantitative precipitation forecasts (QPF) from the ensemble are then used as input to the NWS Hydrology Laboratory-Research Distributed Hydrologic Model (HL-RDHM) to produce forecasts of streamflow in local watersheds. These results are communicated to the hydrologists at NWS DMX to assist in improving real-time, short-term flood forecasting. Results are compared with observations of precipitation and streamflow to assess reliability of this process. This technique is being tested to determine its ability to be a viable option in forecasting flash floods caused by warm-season heavy rain events.