Use of the National Water Model Retrospective Analysis to Provide Context for Real-time Flood Forecasts Issued by the National Weather Service

NOAA’s recently operational National Water Model (NWM) provides streamflow forecasts for approximately 2.7 million points across the United States. The sophistication of the NWM and the volume of data it generates presents a pivotal opportunity to improve losses from flooding. Yet, the magnitude of the advance makes it more challenging to assess the quality of predictions that can be expected from the NWM. National Weather Service forecasters responsible for issuing flood watches and warnings can be justifiably reluctant to use a new tool without an understanding of its performance in their local area. Fortunately, information about the quality of the NWM’s performance can be compiled from the NWM’s 42-year retrospective analysis, which uses observed precipitation as input forcing. The goal of our work is to demonstrate how the NWM’s retrospective analysis can be integrated with the NWM’s real-time forecasts to better provide actionable guidance to decision makers who may be impacted by flooding.

Verification of the NWM retrospective analysis streamflows at USGS gauge sites in the Grand Rapids, Michigan National Weather Service forecast area will be presented. Results depict a best-case scenario relative to real-time forecasts forced by Global Forecast System precipitation, which will introduce errors in the NWM forecasts. In addition, a prototype real-time web application will be shown, which provides the percentile of an NWM predicted streamflow extracted from the 42-year distribution at a given place and time of year. The application can provide context for NWM predictions over the vast majority of reaches that have no gauge observations. The application allows the forecaster to place real-time NWM predicted values in the context of what the NWM has historically predicted, providing forecasters with information that can guide them toward the impacts a predicted streamflow value may have on life and property. A case study demonstrating the application’s capabilities for a heavy precipitation event over southwest Michigan will be presented.