Slashdot Mirror
Japan Plans 30-Year Supercomputer Forecasts
BaltikaTroika writes "According to a ministry representative, 'Japan is planning ultra long-range 30-year weather forecasts that will predict typhoons, storms, blizzards, droughts and other inclement weather.' Maybe they should tell their secret to my local weatherman, who usually can't even get tomorrow's weather right. Whatever happened to chaos?"
Whatever happened to chaos?
Pfft. Chaos is so predictable.
Don't just game, Dungeoneer
Everyone is going to talk about how the buttefly effect makes this useless, and that is true for any sort of instantaneous weather. However, there are many things that affect weather cycles that are much more predictable. First is El Nino/La Nina which oscillates every few years. Then there are other oceanic oscillators that operate on a decade or longer cycle. Also there is solar output and human output. Add all of these up and you may be able to predict the frequency and severity of storms, the probablility of different weather patterns, etc. You will be able to plan for these events which will be 30 years down the road, and be able to do something about them - like build buildings capable of withstanding stronger typhoons, or rising sea levels, or what have you.
But never, in no way, will someone be able to tell you if it will rain in 3 weeks, let alone 30 years. I've studied the accuracy of forecasts quite a bit (as an energy analyst), and you can't get much better than climatology once you go 2 weeks out.
I submitted this story last night, and it didn't get posted.
All it takes is one large volcano to erupt and it'll throw off all your predictions. There are plenty of factors involved with the weather outside of normal weather-type things.
-mrxak
Onions Will Kill You
It might depend where you are.
In Michigan, sure, sometimes they get a week right.
On the other hand, sometimes they're so far off you can barely recognize the week. What seems to happen is a lot of storms stall that they don't expect, or they expect something to stall and it doesn't.
Probably the funniest was the recent "hurricane" over Michigan (about a month ago), which even made Fark. This storm complex stalled for a week and change, and basically every day of the week, the prediction was that it would move away by tomorrow.
Michigan seems to be at a meeting point for storm systems coming from the West, cold air coming from Canada, and wet, moist air coming from the Gulf. Predicting which will "win" for any given day seems to give the models fits. For example, the worst winter storms for us are when the cold Canadian air meets the warm, moist Gulf air, but predicting exactly where they will meet and drop all the snow seems to have an error bar of several hundred miles (i.e., for a prediction of hitting Lansing, smack dab in the middle of the lower peninsula, you're looking at it actually hitting anywhere from mid-Ohio to the top of the UP.) I've noticed that for predicting precipitation, you're almost better off just watching a couple of hours of the radar loop and making your own prediction.
1) The computer will be doing CLIMATE modeling, not weather prediction. That's a different bird. It's like the difference between the average score on a test and your score on the test. Or like describing the flow of heat, but not knowing the underlying collisions that result in the transfer of energy.
2) Higher precision does help you model chaotic systems longer, but... If you run your model until the difference between your prediction and the actual system is larger than a tolerance, the time when this happens is called the horizon time. If you improve your accuracy (let's say your computer system is perfect and errors only occur in getting the initial state right), you only improve the horizon time as the LOG of your improvement. In an age where quadratic methods are just adequate in scientific computing, this is unbearable.
3) Another weather (not climate) prediction option is to use a statistical cohort model. Such a model just takes in data and tries to predict what will happen next based on past trends. It doesn't know any physics, and can take a while to train. This means that the cohort you train in London is useless in Paris. Such "models" often beat physical models in predictive ability, but don't give any insight into why. If you want to fly a plane, they're fine. If you want to do science, see (1) or (2).
Also, this computer is way, way cooler than the one predicting nuclear bomb blasts. But that's, just like, my opinion, man.
Use the Firehose to mod down Second Life stories!
You won't be able to "predict" anything; weather is driven by a complex set of forces, of which we have a very incomplete understanding. It isn't just a matter of temperature, pressure, moisture content, UV radiation, and infrared radiation, which are the main variables your local forecaster uses to try and predict weather trends. Solar wind, ground cover, cloud formation, cosmic rays, vulcanism, atmospheric electrodynamics: these are extra variables that influence the weather in ways we can't understand. And just to screw up the mixture a bit more, add global warming.
You can build more and more sophisticated models and run them on faster and faster hardware, but in the end, you can't really account for all the possible variables to any degree of accuracy. The more variables you add, each with its own degree of accuracy, the more soupy the predictions become. We know in general terms how systems work, but we have no idea how all these forces interact to create weather. I think the Japanese should stick to trying to determine what actually drives the weather and stay out of the prediction business.
GetOuttaMySpace - The Anti-Social Network
That's true. Do you think they're going to install GPS trackers on all the butterflies in the world?
"I like to lick butts!" by MobileTatsu-NJG (#32700246) (Score:5, Informative)