Storm Tracking with OMEGA

Because their winds can pack a 200-mile-an-hour punch and they are very hard to predict, hurricanes cause billions of dollars in property damage every year. The costs in human misery are incalculable.

A critical part of protecting life and property as hurricanes form is better storm track forecasting. SAIC has responded to this need by developing OMEGA, a high-resolution atmospheric modeling system.

The OMEGA system represents a significant departure from traditional weather prediction methods because it melds new computational "gridding" technologies from the fluid dynamics realm with numerical weather prediction techniques. OMEGA dynamically adapts its grid to provide high resolution in regions where storms can potentially develop and it resolves important features of atmospheric circulation and cloud dynamics down to one kilometer. (This flexibility allows more accurate forecasts by resolving at higher resolution both the physical processes in the region and the surface properties that impact the weather.)

In his ESTC Award-winning paper, SAIC's Sundararaman Gopalakrishnan (and SAIC co-authors) details simulations of several hurricanes and tropical storms and compares OMEGA's predictions with those of the National Hurricane Center. (This includes simulating Hurricane Floyd, which hit the U.S. Atlantic Coast hard in September 1999. With 57 people dead and $3 billion in property damage, Hurricane Floyd ranks as one of the worst natural disasters to hit that area.)

Under contract with the Defense Threat Reduction Agency, the SAIC researchers used OMEGA to predict Hurricane Floyd's landfall near Cape Fear, North Carolina, very accurately 96 hours in advance. OMEGA also performed after-the-fact simulations that modeled the structure of the hurricane's inner core and simulated the larger-scale environment in which Hurricane Floyd was embedded.

In addition, the storm tracks from Hurricane Floyd (and seven other hurricanes and tropical storms) that OMEGA produced were 20 percent more accurate than the 1998 National Hurricane Center average. This indicates significant skill of the model to predict a storm's path 72 hours in advance.

The article that Gopalakrishnan co-authored with SAIC's David Bacon, Nash'at Ahmad, Zafer Boybeyi, Thomas Dunn, Yi Jin, Mary Hall, Pius Lee, Douglas Mays, Rangarao Madala, R. Ananthakrishna Sarma, Mark Turner, and Tim Wait, "An Operational Multiscale Hurricane Forecasting System," appeared in Monthly Weather Review.