"The problem of predicting when convective storms will form is huge," said
Dr. John Mecikalski, the system's creator and a UAHuntsville Associate Professor of Atmospheric Science. "Thousands of cumulus clouds form every day and only about 1 percent of those develop into storms. The challenge was to find a way to predict which clouds are about to turn into storms."
After a seven-year development program, SATCAST went live as part of the Federal Aviation Administration¹s Corridor Integrated Weather System (CIWS) this summer. CIWS is an automated forecasting system used by air traffic flow managers, airlines and other people responsible for flight planning.
CIWS combines data from U.S. and Canadian weather radars with satellite data, surface observations and numerical weather models. The system provides several weather analysis and forecast products to the aviation industry.
"CIWS does fine when it knows where the storms are," Mecikalski said, "but Doppler radar by itself only mostly tells you us what is happening now, not what's going to happen an hour into the future."
For more than three years UAHuntsville scientists operated SATCAST for the National Weather Service's forecast office in Huntsville, and more recently for NWS offices in Florida. During that time it was accurate in its storm forecasts between 65 and 75 percent of the time. It was also tested at the FAA's air traffic control center for New York City.
Although the SATCAST algorithm was successful at the local level, it had to be massaged to make it ready for use by the FAA. When MIT Lincoln Laboratory evaluated SATCAST using an expanded coverage area, the system took as much as 20 minutes to run through one cycle. MIT LL software engineers and scientists cut the run time to two minutes or less.
Early in the development process, Mecikalski's team determined that an important factors in predicting thunderstorm formation are cloud top temperature change, in addition to and determining when the tops of cumulus clouds turn to mostly ice, which these factors can be monitored using multiple satellite sensor channels. The temperature at the top of a cloud is related to its altitude, with temperatures dropping as you go higher. If the top of a cloud cools by 4 C (about 7.2° Fahrenheit) or more in 15 minutes, that means the cloud is growing quickly enough, and there is a rising to raise the probability of rain beginning within 30 minutes to an hour.
SATCAST uses data from the GOES East visible and infrared sensors to track changes in cloud temperature every 15 minutes. Other infrared sensor channels help identify the size, depth and longevity of a new storm's main updraft, which is tied to its overall intensity.
In the future, as new weather satellites come on line, the SATCAST coverage area can be expanded to give warnings of growing thunderstorms over the western U.S. and the North Pacific. At present, the GOES East satellite is the only satellite over the U.S. that has the instrumentation that SATCAST needs using data from additional, more sophisticated satellites, giving pilots nationwide (including extended oceanic locations) additional warnings of growing thunderstorms before they occur.
While SATCAST is part of a sophisticated robust and extensive network of weather monitoring systems in the U.S., it is expected to have its greatest impact value in regions where storm forecasting and monitoring systems have been limited or non-existent. The SATCAST system is relatively inexpensive to install and operated, since it uses freely available weather data from existing satellite sensors.
In areas where Doppler radar networks do not exist, SATCAST might be used to track storm systems and provide severe weather warnings that are not now available, Mecikalski said. "This makes SATCAST and satellite-based rainfall predictions very relevant in developing countries where ground-based radar is absent but high-quality satellite data are in place."
Organizations evaluating SATCAST's potential include the European
Organization for the Exploitation of Meteorological Satellites, meteorological satellite agency and the South African Weather Service, whereas the NASA's Short-term Prediction Research and Transition Center (in Huntsville) is helping to train National Weather Service forecasters on its forecasting capabilities.
The UAHuntsville team is also working on a next generation SATCAST, which will take advantage of improved sensing systems on NOAA's forthcoming GOES-R satellites starting in 2016. Sensors on those satellites will collect data in more channels, more often and at higher resolution.
SATCAST development was supported by grants from NASA's Applied Sciences Program, and from the National Oceanic and Atmospheric Administration and the National Science Foundation.
For more information:
Dr. John Mecikalski,