College of Engineering | Apply to CSU | Disclaimer | Equal Opportunity Statement | Privacy | Search CSU

Ensemble-based Analysis of Front Range Severe Convection on 6-7 June 2012: Forecast Uncertainty and Communication of Weather Information to Front Range Decision-makers

You must be on the CSU network—either physically or using VPN—to watch this or any of the videos on this site.

December 17, 2013
Vanessa Vincente
Hosted by Russ Schumacher (advisor), Dick Johnson, Jorge Ramirez (Civil and Environmental Engineering)


The variation of topography in Colorado not only adds to the beauty of its landscape, but also tests our ability to predict warm season severe convection. Deficient radar coverage and limited observations make quantitative precipitation forecasting quite a challenge. Past studies have suggested that greater forecast skill of mesoscale convection initiation and precipitation characteristics are achievable with an ensemble with explicitly predicted convection compared to one that has parameterized convection. The range of uncertainty and probabilities in these forecasts can help forecasters in their precipitation predictions and communication of weather information to emergency managers (EMs). EMs serve an integral role in informing and protecting communities in anticipation of hazardous weather.

An example of such an event occurred on the evening of 6 June 2012, where areas to the lee of the Rocky Mountain Front Range were impacted by flash-flood producing severe convection that included heavy rain and copious amounts of hail. Despite the discrepancy in the timing, location and evolution of convection, the convection-allowing ensemble forecasts generally outperformed those of the convection-parameterized ensemble in representing the mesoscale processes responsible for the 6-7 June severe convective event. Key features sufficiently reproduced by several of the convection-allowing ensemble members resembled the observations: 1) general location of a convergence boundary east of Denver, 2) convective initiation along the boundary, 3) general location of a weak cold front near the Wyoming/Nebraska border, and 4) cold pools and moist upslope that contributed to the backbuilding of convection. Those members that failed to reproduce these results missed the boundary, produced a cold front that moved southeast too quickly, and used the cold front for convective initiation. The convection-allowing ensemble also showed greater skill in forecasting heavy precipitation amounts in the vicinity of where they were observed during the most active convective period, particularly those near urbanized areas.

A total of 9 Front Range EMs were interviewed to research how they understood hazardous weather information, and how their perception of forecast uncertainty would influence their decision making following a heavy rain event. Many of the EMs use situational awareness and past experiences with major weather events to guide their emergency planning. They also highly valued their relationship with the National Weather Service to improve their understanding of weather forecasts and ask questions about the uncertainties. Most EMs perceived forecast uncertainty in terms of the probability and intensity of the forecasted precipitation, as well as its spatial and temporal variability. The greater the likelihood of occurrence (implied by higher probability of precipitation) showed greater confidence in the forecast that an event was likely to happen. Five probabilistic forecast products were generated from the convection-allowing ensemble output to generate a hypothetical warm season heavy rain event scenario. Responses varied among the EMs in which products they found most practical or least useful. Most EMs believed that there was a high probability for flooding, as illustrated by the degree of forecasted precipitation intensity. Most confirmed perceiving uncertainty in the different forecast representations, sharing the idea that there is an inherent uncertainty that follows modeled forecasts. The long-term goal of this research is to develop and add reliable probabilistic forecast products to the “toolbox” of decision-makers to help them better assess hazardous weather information and improve warning notifications and response.