84th AMS Annual Meeting

Tuesday, 13 January 2004: 2:00 PM
Performance of various operational and experimental numerical forecasts for the March 2003 Colorado snowstorm (Formerly Paper Number 5.3)
Room 607
Edward J. Szoke, NOAA/ERL/FSL, Boulder, CO; and B. L. Shaw, P. Schultz, and D. Barjenbruch
Poster PDF (2.4 MB)
A vigorous snowstorm struck the Colorado Front Range Urban Corridor and adjacent mountains beginning on the evening of 17 March 2003. In terms of snowfall amounts and socio-economic impacts, this was the worst snowstorm to hit the area in nearly 100 years. At the official observation site for Denver, 31.8 in of snow fell in the 17–20 March period. Similar totals were reported all along the Front Range from Denver north to Fort Collins, and amounts in excess of 80 in were reported in the northern mountains just west of the Denver-Boulder area.

As is often the case with spring storms in the central United States, the challenge was in determining whether or not surface temperatures would fall enough at lower elevations (below 6000 ft) for the precipitation to change from rain to snow. To do so would necessitate other processes besides cold-air advection and upslope lift to occur. Although forecasters generally anticipated such processes, the various numerical weather prediction models available did not fare as well in predicting the relatively localized band of colder temperatures that developed along the Front Range Urban Corridor.

An overview of the storm evolution and the processes that likely combined to create heavy snowfall are discussed, with an emphasis on how various operational and research model runs performed. Deficiencies of the various numerical forecasts of this situation are discussed with regard to the initial conditions and representation of the microphysical processes that may have contributed to the cooling. Results of research model simulations using additional surface observations during the data assimilation as well as microphysics “tuned” for the situation are presented in order to draw some conclusions on how models could be improved to handle similar situations.

Supplementary URL: