What strikes me about these predictions is, again, they're very specific with the main question not being if, but when.

One question I have is, as these predictions get more and more specific and accurate, will they have any impact?

I remember we discussed something like this a long time ago - to do with earthquakes - got to the question of: Let's say you could predict earthquakes of >M5 with 60-70% accuracy to within a 3-4 day time-frame. As a government, what are you gonna do with that knowledge? Do you evacuate Hualien for 3-4 days? Evacuate them to where?

Or, to avoid public panic, do you sit on that information, let it happen?

On the other hand, we do have pretty accurate typhoon forecasts. You can usually see them coming a week away vaguely towards your vicinity. As Dan says, when it's several days away, the landfall prediction is almost certainly where it won't be - too many variables - but that tightens up to the point where the last 24-hrs are usually accurate to within 50km.

As to whether accurate predictions have impact? Well, it didn't in Australia during the floods where the storm carved out a slow, predictable march south, but later the authorities who failed to act, said: Who knew? It was unprecedented!

You can't keep saying every disaster that happens is unprecedented, then use that as your cop-out. We can safely say there's a precedent that's formed for extreme weather.

Guess it depends on the government you've got as to how accuratization of disaster modelling is used, informs building standards, mitigation measures, etc. Taiwan seems better at than than Aus.

As to what extent that information is classified ? That's an interesting question.

Also has warfare implications...