The Herald researched every hurricane that struck land since Hurricane Andrew leveled South Miami-Dade 13 years ago.
Using the National Hurricane Center's own forecast verification reports, the newspaper studied how often the track, strength or timing predictions were off during the more than 40 hurricanes that made landfall.
Though it's nearly impossible to identify exactly why any one forecast goes wrong, one trend quickly emerged: Essential forecasting equipment was often broken or unavailable as the Hurricane Center struggled to craft predictions.
Over eight months, The Herald conducted more than 200 interviews with meteorologists and researchers, social scientists, computer modeling experts, staff at the Hurricane Center, emergency managers, former directors of the Hurricane Center, officials with the National Weather Service and families from Belize to Punta Gorda to Pensacola to Richmond, Va.
Sign Up and Save
Get six months of free digital access to the Miami Herald
The newspaper also reviewed hundreds of government audits and studies, flight logs, congressional testimony and Weather Service memos and e-mails. The newspaper obtained maintenance logs and equipment status reports for radars and buoys, a searchable database for upper-air observations, and a National Climatic Data Center database of coastal weather stations from Florida International University Assistant Professor Forrest Masters.
The Herald presented its findings, along with all of the Hurricane Center's forecast verification reports, to two nationally known meteorologists and longtime advisors for the National Research Council of the National Academy of Sciences, which advises Congress on the Weather Service and other issues.
''Your assessments are on track,'' said retired Air Force Col. Floyd Hauth, who in the 1980s served as senior science officer for the Air Weather Service's 300 units around the world.
Hauth said chronic gaps in weather-observing equipment have repeatedly stifled the nation's hurricane forecasters. At the National Research Council, Hauth supervised eight major studies about the performance of forecasting equipment at the Weather Service, including radars.
Charles Hosler, a retired meteorology professor and administrator at Penn State University, agreed.
''At critical points, the Hurricane Center is wrong at landfall,'' Hosler said. ``But the accuracy and the frequency of their data puts a limit on how much they can do.''