An Assessment of National Water Model Streamflow Forecasts during Tropical Storm Cindy

Floods are among the most destructive natural hazards and are responsible for significant damage to life and property worldwide. Accurate forecasts of flood occurrence and evolution are critical for issuing flood warnings and watches, and reducing human and economic losses. The National Weather Service's National Water Model (NWM) Version 1.0 was operationally implemented in August 2016 and was updated to Version 1.1 (V1.1) in May 2017 to provide nationwide high-resolution water resource forecasts over multiple forecast horizons. The objective of the present study is to assess the capability of the NWM V1.1 in forecasting streamflow during the high-impact flooding event caused by Tropical Storm Cindy, and to gain perspective on the model’s information for potentially supporting flood control and emergency management.

Widespread flooding occurred during 19-25 June 2017 over the Gulf Coast and parts of the Midwest and northeast due to heavy rainfall associated with Tropical Strom Cindy. The preliminary evaluation of the streamflow output from the NWM V1.1 at select United States Geological Survey (USGS) gaging stations suggested that the short-range forecast performs better than the medium-range forecast with the ratio of correct forecasting of flooding location exceeding 60%. Furthermore, the calculated statistical measures from short- and medium-range streamflow forecasts indicate that the model has high correlation with the USGS observation especially during the early forecast period. For both forecast time horizons, positive bias with large magnitude is mainly found over the heavily impacted Cindy areas including the South and the Midwest where peak streamflow was overestimated and the timing of peak flow is overly early in the short-range forecast but late in the medium-range forecast. Future analyses will examine the accuracy of the meteorological forcings used as input to the NWM V1.1, such as quantitative precipitation forecasts and identify the sources of errors in resulting forecasted streamflow. Moreover, additional cases of flood events will be investigated to provide further insight into the level of skill of the NWM streamflow forecast.