« Dignity, Humiliation and Changing the World | Main | LinuxChix Africa »

Climate Prediction, by way of the BBC

bbcclimateprediction.jpgClimateprediction.net, one of the largest distributed computing projects going, has started a new forecast in concert with the BBC. As before, users download code that runs in the background, allowing thousands of computers around the Internet to process parts of the whole project. In this new effort, climateprediction.net is looking at the changes to come over the next 75 years, but this time its model -- based on the climate and weather forecasting code used by the BBC -- includes a "fully dynamic ocean," allowing for a more complex interaction between the atmosphere and the seas.

Climateprediction.net garnered a bit of attention last year when researchers presented its first set of findings -- which included results showing that the range of temperatures towards the end of this century could be substantially higher than the IPCC baseline case.

Interestingly, rather than each participating computer working on a small chunk of the overall process, every computer in this BBC/climateprediction.net project will have its own version of Earth to examine. The simulation starts at 1920, and pauses at 2006; if the result is inaccurate (such as an iceball or runaway greenhouse Earth), it then ends. If the 2006 result is more-or-less comparable to the real world, the project continues, mapping out changes through 2080. Climateprediction.net estimates that the whole process will take 3-4 months. How many Earths you chew through in that time depends upon the speed of your computer and how idle you leave it.

What makes this version of the software particularly fun is that the "screensaver" mode shows your current Earth model in action, and has shortcuts for looking at current temperatures, cloud cover, and rainfall, among other bits of info. Since the software runs in the background, that's the extent of the interactivity -- but still, you do get to see "your" planet evolve.

(A minor annoyance: the BBC/climateprediction.net software is only available for Windows and Linux. Given that the core code remains the open source BOINC, and both BOINC and the previous climateprediction.net project supported the Mac, I can only presume that the current processor architecture transition at Apple has delayed the new code's introduction.)


Listed below are links to weblogs that reference Climate Prediction, by way of the BBC:

» BBC Climate Change Experiment from Capital Region People
If you missed getting on that SETI thing with Art Bell and company, you have another chance to add your computer to a network of computers set up to perform research! Climateprediction.net, one of the largest distributed computing projects going, has s... [Read More]

» BrÚves from Le Chialeux de Salon
Il vous reste à peu prÚs une semaine pour donner votre avis au gouvernement québécois sur le remboursement de la dette. Ça se fait trÚs rapidement, trois questions, deux commentaires. Je sais que plusieurs considÚrent ces manœuvres comme de la... [Read More]

Comments (10)


wow. a SETI for the climate set!! and you say it has oceans too? just like teh NUGEM? Wow ...GCMs change so much when you integrate oceans into them. So I am curious, do you think one can run both this, and EdGCM at the same time? I mean I heard that EdGCM alone can "heat up" your PC if left unattended (no seriously, not kidding)


Maybe it's just me...but this sounds a lot like Seti which I had downloaded all through college to try to help look for radio signals of other planets (a la the movie Contact).

maybe it's totally different, but at least it sounds similar.

--RC of strangeculture.blogspot.com

Michael Slavitch:

BOINC has had SETI, Climateprediction.net, and various Einstein projects for some time now as very wonkish experiments with little glitz.

This is a far more media savvy BEEB-influenced version, one that may encourage not just more participation but more awareness.

It's not very useful if it's only the /. crowd that gets to play.

Either you believe chaos mathematics is a crock and that the weather is not a complex system (at least three variables) or you have to buy in to the thought that this exercise is a waste of processing cycles because getting closer to reality does not increase accuracy at all on the multi-decade scale that is needed to get anything interesting regarding global warming.

Initial condition dependent systems are not going to be able to be modeled with any accuracy past a few months. A decade is just impossible, a century laughable. Not nailing down, exactly, any of the relevant variables just kills accuracy. Today, we're still identifying relevant variables and plugging them into the models. This is absurd given the mathematical reality.

The BBC effort will provide some nice pictures and an interesting fantasy but unfortunately some will want to legislate on the basis of that fantasy and the types of adjustments that get made on global warming are of such a nature that people are severely discomferted throughout the globe and, at the margin, die.

First of all, it's climate that's being modeled, not weather, but let's set that aside.

If this was running just one simulation dataset, you'd have a point. But it's running tens of thousands (and potentially more) datasets, and only using those that manage to maintain multi-decadal accuracy over the 1920-2000 time frame. No one simulation is going to be 100% correct, but the combined mass of results is likely to provide a pretty good estimate -- to the extent that the simulation itself is a good model.

The implication of your last paragraph is that we should hold off on any social/systemic responses (e.g., legislation) until we know exactly what the effects of global warming will be. This only would make sense if there was a plausible chance that refined models would show that the effects will be minimal or non-existent. However, each model refinement -- each time we get closer to a high-quality simulation of the effects of general atmospheric warming on the global climate system -- the estimated effects get worse, not better.

Recovering from an economic slowdown -- which, by the way, is also a model-based estimation, and one that has not been borne out by real-world data -- would be a helluva lot easier than recovering from a global climate catastrophe.

Pace Arko:

Whenever I read ill-informed comments from people who neither understand nonlinear dynamics nor climate modeling, my response simply bypasses that. I go straight to the argument from efficiency:

"Never mind global warming, Mr. John Public. It makes the hardest kind of economic sense to have everything be as energy efficient as possible. That's savings you'll see directly in your heating bills and at the pump."

That usually shuts them up and the rest of us can go back to taking care of the undeniably real problem of global warming.


""model refinement -- each time we get closer to a high-quality simulation of the effects of general atmospheric warming on the global climate system -- the estimated effects get worse, not better.""

...and then I woke up, saw partick duffy in the sohwer and realized that this whole climate scenario was just a bad dream (sorry)...
but seriously, a) your bang on that with each refinement (mixing atmos + ocean models, subgridscale parameterizaion...) the uncertainties shrink and the scenarios get more 'chilling' but also b) if we never spent any funding or acted until there was total certainty, um would we be in Iraq (or less contentiosly would there be any defense budgets at all)??

Lee, we might not -- but then again, we might anyway, given that the administration did not admit to any uncertainty about the stated rationale for invasion (WMDs).

It is an example of seemingly "certain" predictions (WMDs, welcomed with open arms, oil will pay for reconstruction, etc. etc.) turning out to be 100% wrong, but it is also a case where the predictions concerned human behavior, not geophysical systems. Humans are less prone to follow well-understood mathematical principles of behavior. Hari Seldon, where are you when we need you?

A reasonable guideline for situations like this is the "what if we're wrong?" experiment. If we do this action, and we're wrong in our analysis, how bad could it plausibly be? If we aggressively bring down carbon emissions but we're wrong about global warming, the most plausible worst case scenario is a global economic recession for a period of time. If we decide that global warming isn't real/isn't a threat/should be "adapted to" instead of ameliorated, and we're wrong, conversely, the plausible worst case scenario is a global climate catastrophe that potentially kills billions through flooding, drought, disease, starvation, and war.


i hear ya! Stuart Brand talks well about following various scenarios for the same event, and then plotting out... quite prudent!

and i agree that with iraq it would have probably been like you describe... but that "uncertainty" stalls action seems to play out in this arena much more than in other areas of socio political economic thinking/planning....

and i like the bestcase/wortscase question, but think that you too, are proably in agrement that the whole recession concept is itself uncertain, as there are great arguments that suggest that a rational restructuring of the economy would have different losers and winner but that in agregate things may actually be much better, negawatts and all..... also i wish that adaptation was included in the discourse already(!) what happens when things ramp up and there are a couple of cities that have to get relocated in the same year....

you know william calvin, right? UofW Evolution/Climate Change expert... he shows how our very own evolution has been linked pretty much directly to climate change and the past 2 million year of ice age/interglacial changes... so if we are to evolve once more we have to start adapting. and the scary thing is he once told me that we need at least 50 years to actually adapt. I may be bit more hopeful than that but without going past emission reductions and 1990 levels and fingerpointing i fear sometimes that we will ever get close to it. And the thing that realy sucks about this is that we do have technolgies in place (from modular stackable organic hydroponic agribot systems and analog forestry, to remineralization et cetera...) that could be implemented NOW and ensure a long healthy future.... How do we leap beyond this present stalemate and get to the stuff that really matters? Will it take a couple more new orleanses??

I think it sounds like a really cool project. It has been my experience that mathematical modelling can be a bit subjective (look at what a load of crap the field of economics, and in particular econometrics, is).
I really agree with the rebuttal posted by Jamais Cascio though. A model that is designed with the proper axioms and carried out with a lot of good data will add something worthwhile to the discourse on climate change (which is PR for global warming).
Here's hoping.


This page contains a single entry from the blog posted on February 14, 2006 3:16 PM.

The previous post in this blog was Dignity, Humiliation and Changing the World.

The next post in this blog is LinuxChix Africa.

Many more can be found on the main index page or by looking through the archives.

Powered by
Movable Type 3.34