Feature: LEADing Weather Research
Severe storms in the United States take the lives of hundreds of people and cause more than 13 billion dollars in damage every year.
Researchers who seek to improve storm forecasting, and to lessen some of the social cost inflicted annually, are hampered by the complexity of their task. The atmospheric models and data collection tools used today run essentially independent of weather conditions, they do not respond to rapid changes as they occur.
"Could we do a better job forecasting and understanding the weather if we adapted to the weather as it evolves?" asks Kelvin Droegemeier, the project director of Linked Environments for Atmospheric Discovery, which explores that very question.
Currently, weather models run in a static mode-independent of changes in temperature, wind direction and velocity that occur while the model is running. The team at LEAD believes predictions will be more accurate if the models can include and adapt to new data as it is collected.
LEAD's goals are twofold: to improve the way weather is predicted and to make forecasting and simulations more accessible and interactive for students of meteorology. The LEAD Portal seeks to gather complicated meteorological tools, forecast models and data repositories tools in a single, user-friendly environment-one similar to Web services such as Amazon.com.
Students seeking to use predictive models and other complex tools must learn them intimately-a process that takes months-before they are able to launch a forecast, taking away from their time to do actual research.
"We're trying to manage the tremendous complexity of these things," says Droegemeier, "to allow researchers and students to focus on the science they want to do rather than learning a lot of complex infrastructure. This allows them to focus on the science and understand the meteorological aspects of the tools they're using."
LEAD researchers use grid technology to develop, integrate and test the components of their software. A main function of their grid, with sites located at six of the nine participating institutions, is to host the large data sets needed for their work.
During summer 2006 LEAD released the first version of its software, demonstrating it for researchers, professors and students.
"We got great feedback," says Droegemeier. "One researcher said, 'I just did in 20 minutes what it took me three months last summer!'"
LEAD's plans for this spring include a pilot experiment during the southern Great Plains' tornado season. Once a tornado watch is issued by the National Weather Service, LEAD will automatically launch a fine-scale, adaptive forecast of the area. Droegemeier hopes this will give meteorologists a better understanding of how tornadoes develop and move-ultimately lessening the impact these storms have on the lives of people.
To learn more visit the LEAD Portal Web site. LEAD is one of the TeraGrid Science Gateways.
- Danielle Venton, iSGTW Editor