HRRR model performance for the September 2013 northeastern Colorado floods

In mid September 2013 much of northeastern Colorado experienced an historic flood event. Although rains occurred over a prolonged period from 9 through 15 September, a key part of the event happened during the evening of 11 September, when very heavy rains fell for several hours in and near the foothills of Boulder and Larimer Counties. Examination of this period revealed that a small scale circulation stalled over far southeastern Boulder County, sending convective cells west-northwestward into the foothills. About half the total precipitation in the nearby foothills and Front Range for the entire event fell during this six to ten hour period on the evening into the night of the 11th, and while significant rains also fell on the following day, it is likely that the magnitude of damage would have been considerably lower without the rains on 11 September. This study focuses on this key date of 11 September and the predictability of the circulation that led to the heavy rainfall in terms of a high-resolution (3 km horizontal grid resolution) convection-resolving model, the HRRR (for High Resolution Rapid Refresh). The HRRR has been run in real-time at NOAA/GSD for several years, and as of September 2014 will become part of the operational model suite run at NCEP.

First the observations for this period will be reviewed to document the evolution of the small-scale circulation feature, which developed during the afternoon over the western Denver suburbs and then drifted northwards and stalled. It appears to be much like the well-documented “Denver Cyclone”, except in this case played an important role in precipitation distribution and not severe weather. Also available during this period is total lightning data from the Colorado Lightning Mapping Array (CO LMA), which helped to reveal where the most intense updrafts were developing in the foothills. Numerical model forecasts for this event, as documented in the recently released National Weather Service (NWS) Assessment Report, were not particularly accurate in their details (although certainly gave a longer-range indication of a wet period), and tended to focus precipitation away from the foothills as one got closer to the event. The type of circulation that was noted on 11 September is on a scale that might not be readily resolved by the larger scale models like the GFS or ECMWF. In the case of the HRRR, the model did forecast the circulation, but it was displaced to the east. Because of this displacement heavy rains were not predicted for the foothills and nearby Front Range but instead over the adjacent plains, and below levels that were felt by forecasters to be of great concern. Eventually the HRRR solutions (the HRRR is run hourly out to 15 h) did more accurately position the circulation and as a result shifted the heavy rains into the foothills, but generally these forecasts came too late to be of utility to forecasters at the Boulder NWS Weather Forecast Office (WFO). The evolution of the HRRR forecasts for 11 September will be detailed and compared to the observed flow and reflectivity. But an additional aspect of this study will be to rerun the HRRR using data assimilation that was not available for the real-time HRRR runs in September 2013 and document whether this made a difference in the position of the circulation and distribution of the heavy rainfall.