Improving storm surge forecasting means harnessing large amounts of computing power that can quickly analyze meteorological and oceanographic data needed to develop forecast models.

"The real challenge here is to be able to create a product far enough in advance of a storm hitting the coast to actually take action," said Gary Crane, director of IT initiatives for the Southeastern Universities Research Association (SURA). He said surge forecasts are now accurate about 24 hours ahead of time; one goal is to extend that forecasting window to 72 hours.

In New Orleans, for instance, officials are trying to predict the best time to lower the flood gates on the Lake Pontchartrain canal system -- and data generated by the grid could be used to bolster those kinds of predicative capabilities, said Crane.

There are about 900 CPUs in a heterogeneous environment on the grid, but the schools recently purchased IBM Power servers that will roughly double the number of CPUs and boost the computing power from about 3 teraFLOPS, or 2 trillion calculations per second, to about 10 TFLOPS. IBM announced the hardware deal today.

Crane said the grid has the potential of growing many times in size -- depending on the number of other universities that contribute resources to the computing pool behind it.