The Jury Talks Back



Filed under: Uncategorized — Patterico @ 5:48 pm

This thing looks ugly.

Not sure I’d stay in southern Florida if I lived there right now.


  1. Indeed, Irma is looking very bad for south Florida.
    If, that is, the National Hurricane Center is on the level.

    I know that sounds like a conspiracy theory – and it is. However, they’ve been caught lying before. A major hurricane that tracked close in on the east coast (but the eye did not make landfall) a couple of years ago was forecast to make landfall, even though the averaging of their own models showed it not making landfall. They weighted their forecast in order to encourage people to evacuate. This was admitted. A further example was a “hurricane” force storm they (National wather service in this case)forcasted to hit the Oregon coast last fall. They predicted the worst storm in almost a century, and hyped it as such. They continued to do so when their own models and data showed a weakening as it approached the coast. I was on vacation there, on the shore, and saw it all firsthand; the hype, the forecast not matching the actual data, and the storm itself (ONE stormy day, and not that bad a one). This, too, was admitted by a few involved; they did it to encourage people to evacuate.

    They have clearly never heard the story of the boy who cried wolf. (the moral of which is, when the wolf was real, he was ignored.)

    So, now I’m looking at the “spaghetti”, a charting of multiple model runs using various models. Usually, an averaging of those is the forecast track – or close to it. In this case, the models diverge quite a bit, but the current forecast track (which indicates a direct hit on Key West, with far south Florida taking the deadly righthand side of the storm (which contains the surge). If the model average is correct, I’d be more worried about the the Carolinas than Key West.

    If I were to make a prediction based on the multimodel averaging, I’d say IRMA will make a close northbound pass up Florida’s east coast. (However, no one at risk should put any merit in my words; I’m not a meteorologist!)

    I earnestly hope that what I wrote above proves false, because we need a reliable weather service, not one that manipulates forecasts.

    Comment by Arizona CJ — 9/6/2017 @ 2:25 am

  2. Since I commented above, the NHC forecast has moved the track eastward, though it still shows a direct hit on south Florida. If I’m right, they’ll move the track eastward again within 30 hours, to the track I outlined (a close pass up Florida’s east coast).

    If this plays out as I predicted, I think it’s evidence that they are weighting their forecasts again, based on evacuation considerations. If so, that’s scandalous, because regardless of intentions, it’s the wrong thing to do because it will result in people, for good reason, putting less and less credence in the forecasts (and thus not evacuating when there’s a real danger, because they’ll have no way to know it’s real).

    Comment by Arizona CJ — 9/6/2017 @ 10:56 am

  3. They should be honest but I wonder if they also worry about the consequences if they are wrong — and frankly there is a good chance they will be wrong. Weather forecasting still seems like an art with some science here and there.

    Comment by DRJ — 9/6/2017 @ 11:31 am

  4. @ DRJ,

    The ugly fact is that they cannot predict a track beyond 72 hours. The average error for their attempts to do so is around 200 miles. They are equally unsure with intensity.

    What they need to do IMHO is, first and foremost, stop weighting forecasts (making them sound worse than they actually are regarding danger in an area) and give honest forecasts. If they do not, people will soon start ignoring them (the Boy who cried Wolf effect).

    What they also need to do IMHO is increase the breadth angle of their forecast cone to better reflect the actual uncertainty. A good example of what those would look like are the ones they used until a few years ago.

    I’d also like it if they made the actual model runs they use easier to find. There’s no current way to find the ensemble (Spaghetti) from their site. Here’s a link that updates with the most recent runs for this storm;

    Comment by Arizona CJ — 9/6/2017 @ 6:35 pm

  5. This is interesting.

    Comment by DRJ — 9/6/2017 @ 6:39 pm

  6. Well, rather clearly, I was flat out wrong regarding the model ensemble.

    I have to commend the forecasters and modelers; they came closer to nailing the long range forecast than I’ve ever seen for a hurricane. The only thing they missed (and there’s no way they could have accurately forecast such a fine detail) was a tiny course shift to a slightly more southerly track, resulting in storm core interaction with the mountains of Cuba – and thus lower intensity at Florida landfall.

    Comment by Arizona CJ — 9/11/2017 @ 11:54 am

RSS feed for comments on this post. TrackBack URI

Leave a comment

Comment moderation is enabled. Your comment may take some time to appear.

Live Preview

Powered by WordPress.