Re: butterflies over the amazonas jungle

[robert@geoblue.gcn.uoknor.edu (Robert Lindsay)]

In article <9302081124.AA20259@kino.intellektik.informatik.th-darmstadt.de> uli@intellektik.informatik.th-darmstadt.de (Ulrich Grepel) writes:
>hi,
>
>in my latest rambling about astrology I presented an example of butterflies
>in the Amazonas jungle causing a hurricane over Florida. Someone (sorry,
>don't remember who) was asking for an explanation.
>
>This example shows how relevant every single bit of information about
>weather data is when you want to forecast the weather for more than about
>two weeks or so. A small and single event very far away can cause very big
>events somewhere else, If you want more I can provide you with a mathematical
>(admittedly special) example that shows the thing with the need for precision.
>
>The butterfly-example came up when some scientists did some weather simulations
>on a computer. Changing the last digit of just one input parameter caused
>a totally different weather one or two months later. Like snowstorm<->hundred 
>degrees Fahrenheit. That big difference was due to promoting of 'errors' in
>the calculations. It's a real wonder how relatively stable our climate is
>when you think about these facts.

Trust me when I say that the Chaos effect you are decribing may or may not
exist 'in real life' in the weather, but the computer models most meteorologist
use in research bear alarming little relationship to reality. So the effect
above could just as easily be from bad programming, a common practice in
meteorology, where they wouldn't know structured programming if it leapt off
the screen and bit them. IMHO, of course.

Ob Kate: I'm building a radar visualizer called 'cloudbuster'

>Bye,
>
>Uli