1996-11 -- FOR IMMEDIATE RELEASE: May 20, 1996
The crash raised questions about how seemingly nonthreatening weather conditions could have caused hazardous icing. Because strong winds and heavy rain at Chicago O'Hare International Airport were deemed too risky for landing, the American Eagle circled for a half hour at 10,000 feet. Suddenly the plane flipped sideways, eventually crashing head-first into the ground. All 68 passengers and crew members died. The ATR-72 aircraft was certified to fly in icing conditions, which were forecast for northern Indiana at that time.
Marcia Politovich and Ben Bernstein of NCAR, John Marwitz of the University of Wyoming, Martin Ralph and Paul Neiman of the National Oceanic and Atmospheric Administration (NOAA), and James Bresch of the University of Washington worked together to analyze the weather associated with the accident. Their mission was to identify the atmospheric parameters at the time and place of the crash, including temperature, humidity, winds, liquid water content, and possible droplet size. They used a special computer model, called the NCAR/Penn State mesoscale model, to gain further insight that the available observations could not provide.
"Fortunately," says Bernstein, "forecasters on duty made hard copies of the weather radar images immediately following the crash. Otherwise, we wouldn't have had that information to work with, because normally it's not recorded."
The research team concluded that the relatively warm temperatures at cloud top (-5 degrees Celsius) combined with vertical wind shear could have produced large drizzle drops of supercooled water--water that is still liquid even though temperatures are just under freezing, or 0 degrees C. These supercooled drops would have frozen when they hit the plane. Although the resulting ice on the front of the wing may have been removed by deicing boots, other large drops could have flowed back over the wing before freezing, creating crusty ridges of ice that can disrupt airflow and ultimately destabilize a plane.
"We don't know for sure that these large drizzle drops existed at the time the plane went down," says Politovich. "Direct proof is lacking, but the weather evidence and the behavior of the airplane suggest their presence."
With these findings in hand, last December the FAA asked the AWC, the part of the National Weather Service (NWS) that creates the official icing forecasts for aircraft, to produce advisories that warn pilots specifically of freezing drizzle. General advisories, called AIRMETs (airmen's meteorological information), already existed as a routing product of the AWC to inform pilots of moderate icing and other significant aviation weather conditions. Historically these have been too general to be used for avoiding in-flight icing caused by localized freezing drizzle.
Meanwhile NCAR's Bernstein had been working on creating an algorithm, or mathematical problem-solving procedure, to automate freezing-drizzle/icing advisories for areas smaller than those covered by the AIRMETs. Using airport surface readings of precipitation type and computer model forecasts of temperature and humidity, Bernstein could identify the location of much of the freezing drizzle in the U.S. airspace. He called his new algorithm the "stovepipe" because it uses data for columns of air above the observations taken on the ground.
With the new algorithm, freezing drizzle aloft can be automatically diagnosed across smaller areas as well as for the whole country. To check the new system's accuracy, Bernstein is comparing his predictions with reports from pilots flying through these areas. Pilots can see signs of ice on the plane's nose and wings through the cockpit window, and they can feel a difference in how the airplane flies as the ice builds up. Each time a pilot reports icing within a stovepipe volume of air, it counts as a hit. The smaller the volume of air, the more precise and successful the prediction.
So far, says Bernstein, "a high percentage of pilot icing reports are falling within these targeted prediction areas. That's exactly what we wanted--to zoom in on the freezing drizzle in time to warn pilots about possible icing."
According to Ron Olson, the AWC's science and operations officer, Bernstein's stovepipe algorithm is one of five new tools and procedures employed by AWC to meet the FAA's request for more specific icing advisories. The other four are an earlier NCAR algorithm; a manual prepared by Bernstein and Olson on how to detect and forecast freezing drizzle; daily briefings in which the latest scientific research and weather data are presented to forecasters, researchers, and managers; and Neural Net. The last is an artificial intelligence tool that produces maps of icing intensity using temperature, relative humidity, and convection data from two new, high-resolution weather models running at the NWS's National Centers for Environmental Prediction in Camp Springs, Maryland.
"My job is to keep the scientists and forecasters interacting with each other and to make sure the research is operationally directed," says Olson. "That way we're poised to meet new FAA requirements as quickly as possible."
The new AIRMETs are especially important for smaller aircraft, says Politovich. Big planes can climb through the drizzle layer quickly and then fly above it. Smaller planes, which remain at lower altitudes, may never get above it.
This research is sponsored by the National Science Foundation through an interagency agreement in response to requirements and funding by the Federal Aviation Administration's Aviation Weather Development Program. NCAR is managed by the University Corporation for Atmospheric Research under sponsorship by the National Science Foundation.