Saturday, September 03, 2005

Unrealistic expectations regarding hurricane forecasts

The post below demonstrates my increasing intolerance for the silly TV reporters who find the need to stand out in the rain and high winds to report from the scene of hurricane landfalls. I'm also progressively more irritated by unrealistic comments and demands regarding hurricane forecasts. Hurricane forecasting is a science, but certainly an inexact one. On 2 September 2005, Glenn Reynolds of Instapundit reported comments found on another site that I decided deserved a response. Here is my take, a portion of which was posted by Prof. Reynolds on this page:

From where I sit, there are too many unrealistic expectations regarding hurricane track forecasting.

The Kaus emailer wanted 72 h warning on Katrina's landfall; "that would have been prescient." It's also too much to hope for, for most storms anyway, for a lot of good reasons. 72 hours out, Katrina had not yet even made landfall in Florida. Hurricanes typically progress through environments with little steering, permitting their movements to be erratic, even capricious. The official forecasts made at that time predicted a second landfall in the Florida panhandle. But Katrina took an unforseen southwestern jog over the Florida mainland, and took a slightly more westerly and southerly course than expected. These were relatively small shifts that can - and did - make a big difference even over a short period of time. It made a big difference for New Orleans, that much is certain.

To see an animation of how the official NHC forecast track evolved, see this page. Click the "5-day" button to see the longest range forecast.

But while it's always a good thing to remind people to check their emergency provisions and plans, predicting danger for a specific place, that far in advance, that far removed from the present storm location, seems like a Cassandra wail. At the time the NHC (forget "the media") was predicting a panhandle landfall, it's safe to say that a large stretch of the Gulf coast, as far away as Texas and eastern Mexico, were as likely ultimate targets as New Orleans and Gulfport. Hurricane track forecast skill has improved markedly in recent years, but it simply isn't that good yet, and the forecasters will be the first to admit it.

But 48 hours out, the NHC forecast was spot on. And it took some amount of faith to put stock in it, for they were calling for a northward turn in the track that had not yet materialized. At that time, Katrina was still moving west (indeed, slightly south of west).

The problem is didn't many -- too many -- people in New Orleans ignored the hurricane warnings when they came because they dismissed them as "hype" and recalled previous forecasts that were didn't pan out. Were 48 hours notice enough to evacuate a large area? No amount of time is enough if warnings are not heeded when they come. But the warnings won't be heeded if they are made too rashly, and if the uncertainties they come packaged with are excised in the media hype.

I've never been to New Orleans. It's a shame that the closest I've been to that city is the Lego version at Legoland, where I took this picture:


Post a Comment

<< Home