Hurricane Rita was a monstrous Category 5 storm with sustained winds of 175 miles per hour as it swirled toward the Texas coast on Sept. 22, 2005. By the time Rita came ashore near the Louisiana border two days later, its winds had diminished by 50 miles an hour, making it a still-devastating, but not catastrophic, storm.

A team of scientists spent much of that day flying through the heart of Rita and tracking and modeling features from the ground. They have documented the mechanics of the hurricane's rapid change of intensity and advanced a new model that could help reduce forecast errors of how strong hurricanes will be when they hit land.

They found that the tightly wound storm's eye fell apart and was replaced by a band of surrounding thunderstorms that went on to form a new, but less powerful, eyewall.

"Many times, the replacement of the eyewall makes a strong storm even stronger, but fortunately Rita didn't have time to do that before landfall," said Robert Houze, a professor of atmospheric science at the University of Washington and lead author of a report on the research published Friday in the journal Science.

"Three weeks before, we had watched Hurricane Katrina take almost the same track through the gulf (landfall was about 150 miles apart), but Katrina's eyewall stayed intact until just before it came ashore. We're still working to analyze all the data on the two storms to compare what happened inside them," Houze said.

Forecasters have known for decades that many hurricanes, particularly those with steady winds over 110 mph, undergo one or more eyewall replacements and that those changes can both weaken and strengthen storms, but the new research is the first to get such an inside look at how this happens.

The project, called the Hurricane Rainband and Intensity Change Experiment, was important because while forecasts of hurricane tracks have improved by about 43 percent since the 1990s, forecasts of how strong storms will be have improved only by 17 percent.

Emergency managers along the coast know that many residents will try to ride out a Category 1 or 2 storm but are more likely to evacuate if told a Category 4 or 5 is looming. Warning credibility is hurt if a big storm fizzles, but things are worse if a storm intensifies just before landfall, as did Andrew in 1992 and Opal in 1995.