Numerical simulation of a regional icing event by a mesoscale model

Abstract

A control run and a series of sensitivity studies were performed for this study on the developing extra-tropical cyclone in the central plans of the United States on 31 October 1994. During this storm, a commuter plane crashed near Roselawn , Indiana, killing all 68 passengers and aircrew on board. A suspected or contributing factor in the crash was the presence of a high amount of supercooled liquid water which could have resulted in creating an extremely hazardous icing event. The RAMS model was initialized using standard synoptic scale atmospheric data at 1200 UTC 31 October 1994, with a total of three stationary grids centered on Illinois and Indiana, which was the primary area of interest . The resulting extra-tropical cyclone very closely resembled observations and a supercooled cloud water field was produced in all the simulations performed at the flight level of the aircraft. The sensitivity studies entailed varying the number concentration of activated cloud condensation nuclei (CCN) from as low as 150/cm3 to as high as 1000/cm3. Additional sensitivity studies were performed in which the shape parameter of the gamma distribution function was switched to z,·=3 (versus the default value of 1), as well as a no hail case. The resulting droplet size distributions were then examined to determine icing potential based on the mass of cloud water in ~ach bin of the droplet size distributions. Droplet size distribution is a critical element in determining icing potential. Large (D > 30 μm) droplets , having a higher collection efficiency than small droplets can, in certain types of distributions (low number concentration Nt) comprise most (over 80'l.) of the total mass of cloud droplets, resulting in potentially severe icing. Larger values of Nt require the droplet sizes to be smaller and thus appear to reduce the icing potential. Various methods of calculating the expected potential accumulation were used and resulted in possible accumulations of around 1 cm thickness of ice to over 20 cm, depending upon the simulation performed. This illustrated that high-resolution (-6.X = .6.. Y 5 km) cloud models can provide useful guidance in forecasting aircraft icing conditions.