Employing Quantitative Precipitation Forecasts into the Operational Colorado Flood Threat Bulletin

- Indicates paper has been withdrawn from meeting
- Indicates an Award Winner
Tuesday, 6 January 2015: 4:30 PM
127ABC (Phoenix Convention Center - West and North Buildings)
Dimitry Smirnov, Dewberry, Denver, CO; and S. Geiger, B. Workman, and J. F. Henz

A quantitative precipitation forecast (QPF) is a key ingredient in any flood forecasting tool. Currently, operational QPFs are generally produced by some combination of objective numerical weather prediction guidance blended with subjective forecaster experience and intuition. However, the relatively extreme and often isolated nature of heavy precipitation (and flooding) events makes this relatively subjective approach susceptible to both false alarms and un-forecasted events. Furthermore, advances in numerical weather prediction output have made great strides in at least two areas relevant to flood forecasting: increased horizontal resolution and ensemble forecasts capable of probabilistic guidance. However, it is difficult to gage how well these advances percolate down to the level of day-to-day forecasting.

Here, we compare the performance of our subjective daily Flood Threat Bulletin prepared for the Colorado Water Conservation Board, with a purely objective analogue developed from a 23-member ensemble of atmospheric models with horizontal resolution ranging from 4-km to 16-km. Validation spans the 2014 warm season, 1 May 31 Sep, and is based upon both individual National Weather Service flood reports as well as gridded precipitation estimates, to ensure coverage in poorly observed areas. We show a range of validation metrics assessing both probability and spatial coverage of QPF. Broadly, results suggest that a purely objective approach is not yet operationally feasible due to an underestimate of short-term heavy rainfall, likely due to insufficient model resolution.