Prediction and Identification of Flash Flood Storms in Colorado, Part I - Nowcast skill Using NWP, Blending and Heuristic Techniques

Heavy rainfall and hail frequently occur in association with intense, summertime convective storms that form along the foothills and eastern plains of the Colorado Rocky Mountains. Heavy rainfall amounts over localized regions can result in flash flooding in mountain communities and in the dense urban areas along the Front Range, disrupting traffic, causing damage to property and in extreme events, resulting in loss of life. Various approaches have been taken over the years to provide the best possible estimations of quantitative precipitation (QPE) and nowcasts and short-term forecasts of heavy precipitation (QPN and QPF, respectively) in order to assess the potential for flash floods over the 0-6 hr time period and to accurately model and predict stream flow increases and runoff. Ten Colorado flash flood and hailstorm events that occurred during the period from 2008-2012 are examined in Parts I and II of this study to benchmark our current understanding of the attributes and evolution of flash flood events and determine how to improve our prediction and identification of those storms that are likely to produce heavy rainfall of short duration over very specific regions and basins sensitive to flooding.

In Part I of this presentation, we document the current capabilities (strengths and weaknesses) of various NWP, blending and heuristic techniques in predicting these Colorado heavy rainfall events. These techniques include five WRF data assimilation techniques (cold start, RTFDDA, RUC, HRRR and 3DVAR), a blending system that combines radar echo extrapolation and HRRR (CoSPA). One version of the 3DVAR technique will include assimilation of radar reflectivity and Doppler velocity data. Without radar data assimilation the NWP techniques lack skill in predicting the precise timing and location of heavy rain on the scale of a city or small watershed, but provide some useful skill in predicting storm structure and intensity. The results with radar data assimilation are not yet completed but will be presented at the conference. In addition planned efforts to test and enhance these techniques in a Colorado Front Range testbed will be discussed.