Re: U.S. to build world's fastest computer

From: Lassi Hippeläinen (lassi.hippelainen_at_welho.organized.invalid)
Date: 05/20/04 

Date: Thu, 20 May 2004 11:25:16 +0300

Harris Georgiou wrote:
>
> Yes you are right. In fact, the reason why weather prediction (in normal
> conditions) has gone so well over the last 10-15 years is the fact that
> there is now much more computing power available for simulations rather than
> pattern-based prediction. The real problem arises for very large scale
> models, like hurricane effects and oceanographic models (roughly a factor
> x10 more complex than normal weather simulation).

Yup. But the problem with oceans is lack of accurate real time data. No
weeather stations, and no sensors about heat transfer in sea currents. I
think weather prediction is currently limited by availability of data,
rather than availability of computing power.

-- Lassi