The Warning Before the Wind

On October 6, 2016, that translation was urgent. At 5 p.m. EDT, the National Hurricane Center placed Hurricane Matthew about 100 miles east-southeast of West Palm Beach, Florida, with maximum sustained winds near 140 mph. It was a Category 4 hurricane approaching Freeport in the Bahamas. Its eye was forecast to move close to or over Florida’s east coast through Friday night.

The accompanying forecast discussion explained an uncomfortable geometry: a storm moving roughly parallel to a coastline can produce sharply different local impacts after a small change in track. A slight shift left could bring the major hurricane’s core ashore in Florida and Georgia. A modest shift right could keep much of its hurricane-force wind offshore.

Uncertainty was not the absence of useful knowledge. It was part of the information people needed to act on.

The clock starts before the worst weather

The public advisory defined a hurricane warning as hurricane conditions expected somewhere within the warning area. Such warnings were typically issued 36 hours before the anticipated arrival of tropical-storm-force winds—not before the strongest winds reached shore.

That distinction supplies the mechanism. Preparations become difficult or dangerous before the hurricane’s core arrives. A useful warning must anticipate the closing of the preparation window, not merely announce the arrival of the hazard. The atmosphere does not grant extensions.

Nor did the center’s forecast line describe the whole danger. The advisory warned that large waves could cause water rises well ahead of, and far from, the storm’s center. It described possible water heights of 7 to 11 feet above ground from Sebastian Inlet, Florida, to Edisto Beach, South Carolina, including portions of the St. Johns River, if peak surge coincided with high tide. These were conditional projections, not measurements of flooding already observed.

The forecast discussion also explained what its Potential Storm Surge Flooding Map meant: a reasonable worst-case scenario, with the depicted inundation amount having a 10 percent chance of being exceeded. It was explicitly not a map of expected inundation. A precautionary picture and a prediction of the most likely outcome are different instruments.

The distance between knowing and leaving

Contemporaneous reporting shows why the next step could not be reduced to a weather graphic. The Guardian’s October 6 report described bottled-water shortages and several filling stations in Coral Springs, Margate and Fort Lauderdale that had closed after running out of fuel on Wednesday morning.

Those observations establish a constraint on preparation. They do not establish how many people were unable to evacuate, whether particular households received warnings, or how access differed across communities. That would require evidence from the people and local services involved—not assumptions pasted onto a storm track.

The NHC itself urged residents of mobile homes under evacuation orders to heed them, emphasizing that evacuation was not exclusively a coastal concern. But an instruction to leave and the practical ability to leave remain separate things. Transport, a reachable shelter, accessible communication and assistance are questions a protective system must answer, not accomplishments a forecast can certify.

This is the coherenceist distinction that matters here: a technically sound component does not prove the whole arrangement works. The forecast must connect to decisions, and those decisions to actions people can actually take. Looking for the break in that chain is not criticism of prediction. It is how prediction becomes useful.

Matthew’s October 6 advisories offer no warrant for claiming everyone was reached or protected. They do show forecasters making uncertainty legible before conditions worsened. The humane completion of that work lies beyond the map: someone has understood the warning, has somewhere safer to go, and has help getting there.