Analysis of causes of icing conditions which contributed to the crash of Contentental flight 3407

On February 12, 2009 at 10:20 pm EST, Continental Connection Flight 3407 from Newark to Buffalo crashed 5 miles short of the runway at Buffalo, killing all 49 people on board and one person on the ground. The National Transportation Safety Board (NTSB) is still investigating the crash, but preliminary reports show the airplane was experiencing icing conditions before the crash. Preliminary reports indicate that the pilot did not respond properly to the icing conditions, which contributed to the crash. However the presence of the icing conditions which were significant enough to cause aircraft problems needs further investigation to determine if pilots can be forewarned about these type of aircraft icing problems.

Low clouds covered most of New York, Pennsylvania, and New Jersey on the night of the crash. There were a number of pilot reports of light and moderate icing in the area at altitudes of3,000 – 9,000 feet. There was an AIRMET for Icing issued by the NWS Aviation Weather Center for the New York area at the time of the crash. The Feb. 13 00Z radiosonde sounding from Buffalo showed temperatures below freezing with high relative humidity from the surface to 550 mb (around 15,000 ft.), which is a conducive environment for icing conditions. The surface METAR report at 3Z (10pm EST) for Buffalo showed overcast clouds at 2,700ft with 3 miles visibility, and light snow falling. Radar showed precipitation with 30-40 Dbz reflectivity echoes from Buffalo eastward. The light snow falling indicates a mixed cloud environment with ice crystals present in the clouds in addition to the super cooled cloud droplets. The Bergeron process in a mixed ice-droplet environment will cause the super-cooled water droplets to be depleted, reducing the icing threat. If the icing conditions are to be sustained, then a mechanism is needed to continually regenerate the super-cooled droplets. The focus of this paper will be on what was causing the droplets to be continually regenerated.

The surface analysis shows a cold front in Canada just north of Lake Ontario. The prefrontal winds in Buffalo were from the southwest (240 degrees) from Lake Erie at 15 knots gusting to 22 knots. The 00Z Buffalo sounding showed the temperature profile to be stable, while the NESDIS snow-ice analysis showed Lake Erie to be ice covered. Hence the classic lake effect snow (unstable environment caused by the open lake water) cannot be a cause of the icing. The 00Z Buffalo sounding showed winds of 250 degrees at 20 knots near the surface veering to 300 degrees at 50 knots at 9,000ft. This indicates warm advection in the cloud layer which will generate upward vertical motions resulting in the generation of super-cooled water droplets. In addition the change of friction from the lake to the land should generate a convergence zone near the Buffalo region, which would contribute toward vertical motion. The WRF model will be used to show relative contributions of these vertical motion mechanisms toward the eventual icing conditions that contributed toward the crash.