Earth Simulator Now Predicting Hurricanes?

GeoGreg writes "The BBC is reporting that the Japanese Earth Simulator supercomputer is producing results showing that it is possible to model climate down to the level of severe weather events such as hurricanes. This computer has been discussed on Slashdot previously, and it sounds like at least some of the hype around this beastie was justified."

Comments

Model

[CGP314]

the Japanese Earth Simulator supercomputer is producing results showing that it is possible to model climate down to the level of severe weather events

Sure, you can model it, but how accurate is the model? I can model a cow as a sphere, but I haven't told you if that is appropriate for what I need.

Re:Model

[Davak]

In my belief, this is an excellent point. Being in a scientific field, I have a tendency...err...not to believe people doing research!

This thing is easy enough to test. Plug in a the variables today... and see if it predicts the weather currently tomorrow, or the next big hurricane, or whatever. They haven't published this type of research yet... why not?

Pretty graphics and powerful computers do not insure success.

Show me the data.

Davak

Re:Model

[snarkh]

Being in a scientific field, you might have taken a minute to read the article, where it says that the computer is designed for climate not weather forecast. I.e., you might get an accurate estimate for the probability of a hurricane within a given month, but don't expect to find out the weather for tomorrow.

Re:Model

[2marcus]

Actually, what you want to do is take the best data you can find for "exogenous" variables over the last 100 years (solar cycle, volcanoes, anthropogenic emissions), and plug them into the model and compare the overall trends to reality (rather than trying to predict a "specific" hurricane, which is not what the model is designed for): One would expect that you can match global average temperature and sea level rise pretty well since current (100 km resolution) models already do so (see IPCC report). The question is whether you can match extreme weather events. Unfortunately, I don't know if we have good enough data about hurricanes going back long enough to really see trends... And this is a controversial area, because many climate change scientists believe that an increase in radiative forcing due to greenhouse gas emissions will lead to increased latent heat (ie, more water in gaseous state) in equatorial regions, which should increase the frequency of events like hurricanes... Which would be yet another reason to reduce GHG emissions. But others disagree. And when it comes down to it, you are trying to validate your model either because you believe you have done a good job on the fundamental physical processes, or by matching external data, and I think we may not be confident in either of those areas yet. But it is still a useful exercise to see what our best predictions show, even if there is significant uncertainty in those predictions.

10km resolution

[CriX]

Wow, I'm surprised that a resolution of 10 cubic kilometers is enough to actually make any predictions besides the most general of weather trends.

Think of the variation between the state of air at sea level and then at the ceiling of a 10km cell... that's some severe approximation.

Re:10km resolution

[rfovell]

Wow, I'm surprised that a resolution of 10 cubic kilometers is enough to actually make any predictions besides the most general of weather trends.

Think of the variation between the state of air at sea level and then at the ceiling of a 10km cell... that's some severe approximation.

It would be a horrifically bad approximation, yes, but you cannot compare horizontal resolutions and scales with vertical ones. The temperature variation over the lowest 10 km is about 70C (130F). At that height, pressure and density are both about 20% of their sea level values. You'll never find that kind of variation in the horizontal over any distance, never mind adjacent 10 km grid squares.

There is much that cannot be resolved at 10 km, but at this point in time 10 km horizontal resolution on a global scale is fantastic.

Output

[mmmjstone]

I wonder if the system releases only one pattern that the weather will follow or if it returns many different ways that the weather could go. From what I've heard previously about hurricanes, they have the tendancy to change paths when they feel like it - would the machine give more than just one pattern that the hurricane could take or do you think that it gives what it has discovered is the best answer?

Re:Act of God?

[bhima]

Given that they accept these things on faith. I expect that another solution for their need to rationalize the beliefs will show itself.

Don't frustrate yourself! Don't worry about these folks!

Probably off topic but my daughter recently learned about how hurricanes form, and what powers them in school and was quite fascinated. My point is that the truth about our natural surroundings can be as interesting to those recently exposed to it (But I suppose it requires an open mind) as the fantasies concocted by the zealots!

Re:Predicting Hurricaines..

[girouette]

You are right, the forecasting of individual hurricanes or storms is completely besides the point of a climate model.

The application here is in the area of climate forecasting, attempting to forecast trends in upcoming decades. It's not even important whether the model gets individual storms right, as long as the averages are realistic.

The advance is in becoming able to incorporate hurricanes in the simulation. This should help improve the realism of those trends and averages.

Nonsense in article

[Brian Blessed]

The article contains a photo of a train with the caption: Fewer tracks may buckle in heatwaves

The recent track buckling problem in the UK was caused by the use of cheap lightweight tracks (which is why our European neighbours were unaffected). I have to wonder though how the author of this article reaches the conclusion that simulating climate models will actually lead to less track buckling. It was already known that the tracks would buckle occaisionally, but those in charge of the railways planned for drivers to slow down and try to see buckled lines ahead (as if derailing at 60mph is acceptable!).

Re:Act of God?

[Rostin]

This comment is akinned to those that people make who are convinced that Science (tm) has somehow disproven the existence of God, simply because we now have a better understanding of the physical mechanisms for certain phenomena that used to be explained by divine activity (a popular example is Zeus casting down lightning bolts). Whether or not individual lightning "bolts" (or hurricanes) have some divine purpose is not a question that we can answer by understanding atmospheric science. In a rough parallel, it might be asked, "Why did Phigrin type what he typed?" And in response someone will say, "Because the electrical/chemical impulses transmitted via his nervous system caused the muscles in his fingers to move in such and such a fashion." It's a valid answer, but says nothing about Phigrin's intents or motives.

Better models please

[Pedrito]

I live in Southern Mexico on the coast. About 5 years ago Hurrican Mitch, a category 5 hurricane, was sitting out at sea not too far away. For nearly a week it hardly moved, just hanging out in the middle of the ocean building up strength. The whole time, the NOAA/NHC was predicting it would hit us dead on in 3 days. Yet the hurricane just stayed there.

Suddenly, the hurricane turned south and hit Honduras. Where it again stalled and hung out for 3 days. In the end, about 11,000 Hondurans died, primarily from massive mudslides that consumed enitre villages.

I really hope they improve the models significantly so that things like this don't have to happen. If hurricanes could be predicted with more accuracy, to the point that people and countries could trust the predictions, these areas could be evacuated.

Unfortunately, with the level of accuracy, there's such a wide area in the predicted path that it's impossible to evacuate everyone that could potentially be in the path in time to save them.

When I first moved down here, I though, "Gee, I'd like to see what a hurricane is like." Then Mitch showed up. When you have a category 5 hurricane on your doorstep, you start to re-evaluate your life a bit. The town I live in would have been leveled. I would have been one of the lucky ones. I had a car and would have simply driven inland to avoid it. A lot of people couldn't have afforded to do that.

With more accurate predictions, the government could sponsor the evacuations and save a lot of lives.

Re:Who needs a supercomputer?

[nat5an]

Well, it's not really so much of an urban legend, as a way of explaining chaos theory (thanks Jurassic Park). The idea is that, no matter how accurate your initial data are, there will still be some round-off errors (basically) in your numbers. When you do a lot of calculations on these data, small differences in the initial data manifest themselves as large-scale phenomena down the road. Hence, a butterfly's position now determines whether or not a hurricane occurs three weeks later.

The New York Times has a zoo animal mascot?!?

[ianscot]

If you left out the BBC-sexy-up-styling to comment, I might have taken you seriously. So go back and keep getting your news and political insight from a webpage that has a zoo animal for a logo.

The article quoted from was in the New York Times originally, wasn't it?

I don't gather how you can say that America know the LEAST about climate science because the EU/Japn have fast computers.

Hey, here's an idea for anonymous cowards everywhere -- If you actually read the article, all the way through, you might not have to post anonymously. Then you might see exactly what the article says about the US's ability to predict long-term climate, see, instead of just responding to one paragraph out of context. Zoinks, wouldn't that be informative?