Colorado High Elevation Flash Flood Producing Thunderstorms: Radar, Rainfall and Atmospheric Characteristics

Colorado has been the site of many recent deadly flash floods with the Big Thompson Flash Flood of 1976 which killed at least 140 people, the Buffalo Creek Flash Flood which killed 5 people and the recent 1997 Fort Collins Flash Flood which killed 5 people heading the list. Considerable research has been directed at understanding the causes and describing these storm events which occurred in the foothills of the Colorado Front Range generally at elevations between 6,000 feet and 8,000 feet.

However, little research has been directed at understanding and describing in a quantitative manner similar characteristics of thunderstorm produced flash floods which have occurred west of the Continental Divide at elevations of 7,500 feet or higher elevation. The recent availability of cloud-to-ground lightning and WSR-88D Doppler radar observations west of the Continental Divide has facilitated this study and made evaluations possible.

The flash floods investigated included Rifle Creek Flash Flood (5-15-93), the Robideau Creek Flash Flood (8-10-93), the Saguache Creek Flash Flood (July 31, 1999), the Dallas Creek Flash Flood (July 25, 1999) and the I-70 Mudslide and Flash Flood of July 27, 1999. A complete set of conventional surface , upper air, cloud-to-ground lightning and WSR-88D Doopler base reflectivity observations were obtained for each event. These observations were supplemented by automated SNOTEL weather/rain observations and cooperative NWS observers. The first two events did not benefit from the WSR-88D obsrvations but creative use of cloud-to-ground lightning observations provided the needed quantitative detail. Analyses of these observations revealed substantial differences in atmospheric structure, storm development patterns and rainfall production from storms which formed east of the Continental Divide.

Key differences in atmospheric conditions of the storms indicated that the estimated depth of the warm layer(>0C) was about 2km and the Surface to 500MB precipitable water index was 0.98 inches which were both about 25 % less than eastern storms. Atmospheric temperatures tended to be 1-2C at 700MB, 500MB and 300MB than east of the Divide. Low level forcing was controlled by local topographic influences on both sub-cloud and cloud-layer winds. The sub-cloud layer forcing appeared to be influenced by sun exposure to surface heating and upslope wind factors. Low level jet, baroclinic boundaries and moisture tongue influences were not detected. Instead outflow boundaries moving downslope influenced storm development sequnces equally with topographically forcing echo training within a very moist atmsophere. Unequal heating of mountain slopes played an apparent significant role in storm development and evolution.

Weak mid-level sub-tropical disturbances associated with the Arizona monsoon and upper level short waves associated with jet streaks were also noted. Cloud layer wind shears were 15% more with the western storms.

Storm peak rainfall production was generally in the 2.00 to 3.50 inch per hour range for durations of three hours or less. Storm total rainfalls were in the 4.00 to 5.50 inch range or roughly half of the eastern storms. Exceptions were related to repeated topographic echo training and enhanced vertical motion fields associated with passing upper level features.

Thus it appears the the prediction of high elevation flash flooding events will require emphasis on different analyses which are topographically sensitive. Application of some forecast tools used on storms east of the Divide will prove useful but will have to be adjusted to observed differences.