Nerd Vacation to the Earth Simulator

eecue writes "Earlier this year I went on vacation to japan. At the end of my trip I was lucky enough to receive a tour of the Earth Simulator, which is the world's fastest super computer. I took pictures and wrote about it."

Comments

The problem is grid size.

[MyNameIsFred]

I've read in various articles that supercomputer weather prediction systems look at around 200,000 points on the earth surface. With the points being a few km apart. When you think about the size of the earth, this is a very fine grid. However, when you think about a specific person, the spacing is hugh. Hence the problem with weather prediction. A few km can mean the difference between a downpour in a city, or completely missing it.

Holes in the grid

[hughk]

You are quite right and one of the major issues is the holes in the grid, even at 3Km. Satellite based observation dosn't help much as you only 'see' the tops of any clouds and have no way to measure barometric pressure.

Many ships record information for the meteorlogical services, but the trouble is that only works where there are ships. In some of the meteorologically interesting places such as the poles are often shrouded in clouds and have few weather stations.

The truth is that many points must be interpolated. Points closest to civilisation are quite good because there are enough measuring stations. This means that short-term weather forecasts are quite good (except in the UK, where they may be right but delayed or advanced by up to a day) but deteriorates over about three days and over a week or so is extremely difficult.

Forget the calculations, if you don't have data points, you are just speculating.