Slashdot Mirror

I think our miscommunication is not on ANNs, but on climate models. In fact, like SETI@Home, you can donate cycles to trying to machine develop models.

I understand the methods are different, but I'm not sure the results are distinguishable.

http://www.epic.noaa.gov/epic/...
http://www.giss.nasa.gov/tools...
http://edgcm.columbia.edu/
http://nomads.gfdl.noaa.gov/CM...

Some data: http://www2.cesm.ucar.edu/

Some background info:
http://www.ec.gc.ca/ccmac-cccm...
http://www.climateprediction.n...
http://www.climate.uvic.ca/
https://www.ipcc.ch/pdf/techni...

This one has videos: http://vimeo.com/user12523377/...

In this age of information, ignorance is a choice.

Since we're on the topic, I'll tell you what the biggest weakness is of the IPCC report WGI (which is more reliable): it doesn't establish anywhere that computer models are accurate. This is understandable, because really they aren't. Unfortunately so much of the case for global warming comes from computer models. If you take away their predictions, then most of the serious problems of global warming go away.

Sorry, that myth has been comprehensively debunked. Here is one of many debunkings written by climate scientists:

climate-myths-we-cant-trust-computer-models

The climate models I am running on climateprediction.net begin in 1820. They do that to correlate the various models with the climate record since 1820. Only models that show a good correlation are used to predict the future. There are plenty of links on the site showing this correlation, take a peek.

If you have a problem with any of the individual statements they make, they look ready to respond to any questions. Given that this individual post was endorsed by Nature, and that one of the top contributors is a NASA climatologist, I'd like to ask if "left-wing" can be supported by anything other than the topic of defending the concept of global climate change itself.

And yes, scientists have acknowledged as a non-anthropogenic warming phenomenon as a drop in the intensity of sandstorms in the Middle East. The sand in the air lowers the albedo of the Earth's surface, causing it to reflect less light and warm the surrounding area. The opposite effect occurs on Mars, when dust storms cause huge temperature increases, which I would then assume is due to the Martian dirt being darker. However, the most credible study of solar cycles contributing to global warming of which I have heard estimated the effect to be as high as 25%, but this was refuted by other scientists and the study was withdrawn by its authors.

With regard to skepticism concerning computer models, would you accept that if a computer model can accurately model climate changes from 1900 to 2000, and is fed accurate data, then if it is powerful enough, it can predict future climate changes to a limited extent? There is such a project that exists using distributed computing, where you download a "module" that starts in 1900 and finishes in 2000 before going into future decades. The results are sent back to the scientists, and each one helps improve future models.

If one day a model from 1900 to 2000 is perfectly correct, might you be inclined to give it some credence? At first the project did things like use a solid slab ocean, without currents, as a developmental stage, but it has progressed far since then. It's hosted by Berkley and you might care to look into it, since their methods are so transparent.

If you're worried about that, try a real mindfuck with http://www.climateprediction.net/.

Yep! Run Climate Prediction client. http://www.climateprediction.net/

You could always run Climateprediction.net.

I'd personally still prefer Folding@Home - climatology is way too complex, with lots of unexplained and speculative stuff. I'm not a scientist, but I'd guess this needs more basic research of underlying principles before brute force number crunching starts yielding useful results (any climatologist here?), not mentioning this project screams "junk science" out loud. And if they want internet community to get interested maybe someone should enlighten them about possibilities of different picture formats than 22 MB .bmp for high resolution histogram of global temperature change.

Folding@Home is useful and brings actual results - you'll get a chance to throw your own pack of frozen pea against Africa's hunger, instead throwing it into wastebasket of "well, it seemed as a way to go then".

As for SETI, well, yes there's a lot of space research fans here and way more Star Trek and Star Wars fans, who just secretly wish aliens to exist because it would be so cool if they existed even if without a chance to get into a hot threesome with Spock and E.T, but let's face it - aliens don't exist. And if they do, hoping to get some proof from SETI is like going to the sea coast once in your life, step on the shore with closed eyes and reach into the water in hope you'll get a grasp of bottle with a message from boat wreck survivor.

If you gonna donate spare cycles, donate them on something useful instead of something cool or guilt relieving.

Not to mention climate change prediction at home via climateprediction.net.

You mean something like this?

You joke, but two years ago I kept my college room warm by running distributed climate prediction on all my boxes.

Some poster mentioned it earlier: If you priorities is to spend youd budget on the best way to save lives then research into Cancer or AIDS isn't the best place to put it, even within the medical research field. There are other diseases that kill far more people but get far less research dollars than Cancer/AIDS already! The money goes into areas where the research companies think there will be the best return on the investment!

That said, it is a fallacy to suggest that SETI might also result in a cure for all known ills by finding the aliens who already have the cures! Again, from another poster, the best thing SETI could do is offer a wake-up call to the religiously infatuated, perhaps providing some coffee flavoured smelling salts at the same time.

FWIW, I used to run SETI, before and after BOINC. I also ran a number of other BOINC clients, including:-
SETI,
Folding,
Climate Prediction,
Einstein searching for gravitational waves,
LHC helping with the Large Hadron Collider,
Predictor trying to predict protein structure from protein sequences,
QMC,
Rosetta,
Stardust,
yada yada yada
but removed it a year or so back as it did seem to get in the way rather too often.

BOINC was just too clunky. Why did you have to register individually with each BOINC project, be given yet another HUGE number, have to search for the interesting projects yourself. BOINC should have taken care of the registration once, then offered a drop-down of active projects. Selecting something interesting would do all the install stuff for you and allow you to control the shares from the Client - currently (or at least when I left it) if you wanted to alter the share of one particular project got you had to go to each Project's website rather than just set it within the client. Just clunky!

Anyway, I moved on, but I'd have to say I'm sort of interested again and may fire up SETI again for a while to see how things have progressed since I last offered some cycles!

He launched a blog to attract attention to his research and created a website where he posted his manuscripts that had been rejected by Nature. But in early January of this year, he finally had a paper accepted into a real science journal--Geophysical Research Letters (GRL).

Decades of research have created a massive body of scientific literature on climate change, and thousands of new studies on the subject appear every year in different science journals. Yet, within weeks of publishing his first peer-reviewed study, McIntyre was profiled on the front page of the Wall Street Journal. The article ran 2209 words and was written by reporter Antonio Regalado.

climateprediction.net are doing a really good job of modelling climate change, and are getting lots more data all the time (due to their distributed computing project). Check out the Results and Climate Science sections (although I couldn't find anything about adherence to historical temperature changes in the 5 minutes I spent flicking through).

The BBC ran a programme in January using some of their data.

I've recently created a fairly low power, high spec PC. Unfortunately, I needed 64bit, and HW virtualisation assitance, so I couldn't get the power that low.

2x 60G laptop hard disks (1 watt each) as a mirror
3x 500M Maxtor (7 watts each spinning, 1 watt on standby), RAID5 (less frequently accessed data - spun down >90% of the time)
1x Asus N4L-VM (~ 3 watts, 945GM chipset + graphics) "Socket 479" motherboard.
1x Core 2 Duo Mobile T7400 2.16GHz (30 watts full speed / both cores flat-out, but about half this with speed step, and less if idle)
- you could use one of the much lower power ~6 watt ULV (ultra low voltage) Celerons, or Cores.

And when it's idle, it runs http://climateprediction.net/.

If you don't need that much horsepower, go for a VIA MiniITX, or an ARM for your always-on box, as other people have said. All-in ~65watts, not as good as I'd hoped, and I may be able to get it down a bit further using a more efficient PSU, but at least it's doing good science whilst it's on!

Very interesting post...got me thinking.

My ClimatePrediction.net client is currently chewing away on a pair of scenarios which, the app tells me, have approximately 3000 hours (wall-clock) to complete. By your figures, I'm adding roughly a ton of CO2 to the atmosphere for every four iterations I process (assuming I only leave the computer on to run CPDN, which is questionable).

If this is correct, it would perhaps be well for the Boinc people to call this out on their FAQ...

Have you ever done any serious mathematical modelling? Do you imagine you just put in one set of values and say "that's the answer"? Take a look at this to see how it's really done.

Join climateprediction.net. Americans obviously need not apply.

I don't care if you're folding proteins to cure cancer, searching for aliens, sticking a finger up at the RIAA or just keeping it running to tell all your friends about your big swinging dick uptime - it's all a waste.

Maybe it's a waste to you, but different people have different priorities. Science always takes energy and time from other activities, but it often results in net energy savings in the long run. For a rather ironic example, take a look at http://climateprediction.net/. The project expects you to keep computers running in order to deal with global warming; they estimate that the energy thus wasted is not significant to global warming itself. I can't remember the exact reference, but the mere existence of that project should be indicative of the general thinking involved.

As for me, one tiny reason for keeping a computer turned on is to host my personal website, since my university-provided web space is too limited. This is also a matter of control; you could always pay for someone for hosting, and pray that they do things the way you like. I guess this is the kind of thing that is economically inefficient, as opposed to specialization, but many people prefer such irrational DIY ways of life.

Wait no longer: climateprediction.net

One group is actually modeling climate change using the BOINC distributed client. I don't participate myself, as most of my home computers are old and too slow to run the client-- and I'd rather just switch off the box.

http://climateprediction.net/

What is climateprediction.net?

Climateprediction.net is the largest experiment to try and produce a forecast of the climate in the 21st century. To do this, we need people around the world to give us time on their computers - time when they have their computers switched on, but are not using them to their full capacity.

"Instead, he donates his spare computer power to a global warming project"
Does this attempt to determine how much global warming is being caused by donating CPU cycles.

I think that issue is answered pretty well in the FAQ. When it comes to the real experiments being run by that particular project and their results, you can start here.

"Instead, he donates his spare computer power to a global warming project"
Does this attempt to determine how much global warming is being caused by donating CPU cycles.

I think that issue is answered pretty well in the FAQ. When it comes to the real experiments being run by that particular project and their results, you can start here.

Here's my problem with your statement, you ask me to provide cold hard facts that global warming isn't all it's cracked up to be. Well, I can do that with a dozen studies and web sites http://www.junkscience.com/, http://www.john-daly.com/, http://www.climateaudit.com/ are all quick and easy to pull off the top of my head.

In addition, it's recently been pointed out that there's no Nobel Prize waiting for the person who proves anthropogenic global warming (AGW) is a crock, in fact it's like a death knell to your carreer to pursue global warming skepticism, even if you are totally right.

McIntyre and McKittrick, the two people who have (respectively) a PhD in Statistics and PhDs in Math and Geology were told that they had no qualifications to argue the quality of Mann's "Hockey Stick". This work was done by climate change scientists who had degrees in, hmmm, one has a PhD in Math and Geology (Mann) and the other has a degree in Statistics (the et. al. of the report.) McIntyre and McKittrick have received dozens of death threats from the AGW crowd, especially after they proved that Mann's equation would produce a hockey stick, even with totally random data.

The reason gravity, and relativity, and evolution haven't been "shot down" is very simple. They are falsifiable theories. It takes a single fact that lies outside their purview to devastate the theory. Gravity - at least Newton's version - was ruined by the fact that the planet Mercury was in the wrong place. Look up the Planet Vulcan sometime and see why Relativity knocked Newton out of the ballpark.

The current AGW debate is based on two facts. CO2 concentration in the atmosphere has increased by approximately 80PPM over the last 160 years, and during the last 140 years, there's been an increase in temperature of about 0.6 degrees C. However, there's a big caveat in these two pieces of data. It's called "Correlation does not imply causality." It's one of the first things any good statistician should be taught. However, it's plain that the climate scientists decided to jump on the bandwagon and scream "CO2 is wrecking the Earth!".

To "prove" this, they've turned to computer models of the atmosphere. These, they say, prove that global warming is real, yet even they admit that most of their models "go runaway" and have to be thrown out. I'm sorry, but if your model is so fragile that given the same inputs it can "go runaway", then the model isn't accurate. It's equivalent to tossing a coin. It's meaningless. Who decides what is a "runaway result". Climate Prediction even admits that they threw out any run of their model that showed cooling with an increase in CO2.

Even the most powerful simulator in the world, the NEC Earth Simulator, only works on 50 kilometer wide grids. They had never even seen a hurricane on their "simulated Earth" until two years ago, and even then, they didn't call it a hurricane, they called it a "hurricane precursor" known as a "curl" because the simulation wouldn't support the actual hurricane formation and flow. Now, hurricanes are responsible, annually, for 30-40% of the rainfall in portions of the Southern United States. Their model admittedly doesn't handle this, those areas never receive that rainfall, and precipitation is responsible for a large amount of ground-cooling in models, as well as hundreds of other effects that simply aren't modeled. And that's just one of a dozen things I could list that are wrong with computer models. I've had this argument before. (And it's dropped off my lis

I'm pretty sure these guys - http://boinc.berkeley.edu/ - are using the CPU for their modeling, but why not the GPU? I think the way the Climate Prediction project (http://climateprediction.net/) works is the software crunches numbers for a while (e.g. how did the weather progress on the morning of June 3, 1811) and then takes a snapshot of the result at periodic intervals. Seems just right for a graphics card, given its modeling strengths and download/upload constraints.

If you don't trust the BBC, then sign up via the original web site at Oxford University. This distributed computing experiment as been running for a couple years (I was a beta tester). They have already published some initial results.

If you don't trust the BBC, then sign up via the original web site at Oxford University. This distributed computing experiment as been running for a couple years (I was a beta tester). They have already published some initial results.

So I'm not being sarcastic here, my genuine questions is : why should I spend my free computing power on calculating prime numbers instead of research to cure cancer?

Yeah, I was surprised to find that SETI@Home on BOINC was news— I made the transition almost a year ago and added Climate Prediction. It seems like setting up BOINC was more trouble than the old SETI@Home client, but I got over it. Now adding Folding@Home...

I think this suggestion has been made before, and it may a good idea. However the initial response mentioned that the code wasn't owned by the project or the university, but is on-loan from the UK's Met Office. The CPDN-distributed models are even named "Hadley", the Met's climate research center that has developed these models over the course of decades.

You seem prepared to make a well-reasoned argument for open sourcing the models, so you may want to start at the discussion board. Perhaps project members can offer insight on dealing with the Met Office.

Jeez man, put some of that idle CPU power to good use!

When did I suggest that we should go off try any old half-baked idea that some idiot proposes? Why would you assume that I would support something as silly as that?

I'd prefer a bit more understanding of the complexities of the climate... The I hope that your are participating in the ClimatePrediction distributed computing experiment, that is trying to develop a scientific model for climate change.

That is what the Climateprediction.net project does:

It uses distributed computing (ala SETI@Home) to test climate models against the past and present in order to hone its climate forcast for the future (post-2050).

http://www.climateprediction.net/

How ironic :) but true.

That may be common on some BOINC-type projects, but I've found that people crunching for ClimatePrediction.net to be more of the concerned variety.

There is still some emphasis on stats, but overall the activity surrounding the related Open University course and discussion of climate change and ecology tend to eclipse competition for its own sake.

CPDN is the most demanding distributed computing research project I've seen and narcissists fall by the wayside pretty quickly. What we COULD use are more geeks. ;-)

What would be amusing was if global warming research was being done with the 'spare' cycles

Putting your joke aside, look here
http://climateprediction.net/

at first, weather in two weeks != climate in 100 years, which is FAR better predicable ( see http://www.climateprediction.net/ ) its not about technology, its about fact that weather forecat is expected to be really precise in time, location and efect, but in global climate, one can ommit time and location preciseness

seccond, everytime someone suggest, that we should not care about our envroment because some economical issue, i have to punch him into face.

i mean, it is short sighted, there are issues that has to be resolved right now.

let me put it this way: every day, we human race harvest natural resources worh of triple budget of usa, if we had to produce them by industry. if we are not carefull and reguating, we can soon find ourcelwes in reall ecomonical trouble, because there will be no funcking way to produce energy, food and enviromental necesities like clean air, since global economy canot handle simulationg even percent of nature.

(note: im from post comunistic coutry, i know how much unrepairable damage can do "glorious industry and economy", damage that peoplbe will have to suffer a long time. damage that will cost tenfold to undo than was worth of products produced)

Sigh, from the "you guys" comment, it's obvious that you didn't take the time to actually read my post. I could tell you that I've run McKitrick and McIntyre's algortithm, I could tell you that I have a hundred results with everything from red noise to sine waves as the input and that Mann's equation continually produces "hockey sticks" on all of them. I could publish code and graphs and raw data, but you would choose not to read them or pay attention, just as you failed to read my previous message.

I could tell you how I have reverse engineered the http://climateprediction.net/ source code to see how their model worked and found it woefully inadequate for even the simplest of models.

I could tell you all these things, but you wouldn't read them, just as you failed to read my previous post. For you, the conclusion is already there and messy facts must be ignored to support the conclusion. Like most Anthropogenic Global Warming advocates, you have made your cause a religion, rather than a Science. Mann is your gospel and no amount of proofs will change your mind.

Feel free to save the world from Dihydrogen Monoxide while you're at it.

But my CPU cycles go to climateprediction.net now.

Please come join Folding@home, we're actually doing something worth all that waste heat. :)

Oh yeah? How about using that waste heat to help fight more waste heat?

From the Climateprediction.net FAQ:

a rough calculation suggests that 100,000 computers running 24hrs/day for one year at a power consumption of 50W will contribute approximately 0.0001% of the total amount of CO2 generated in one year. This is not an insignificant amount, but seems (to us) a worthwhile investment to better understand the climate system.

My humble opinion is that Folding@Home is not the only worthwhile distributed project out there. As long as your computer is doing something useful with its spare cycles, I'm happy. Besides, what good is protein folding when our brains are coagulated due to global warming?-)

Yes, you are wrong. Weather models attempt to simulate physical chunks of air. Climate models also attempt to simulate physical chunks of air. Climate models are not just interpolations of past data; they represent an attempt to model physical reality.

Most climate & weather models are GCM-based (general circulation model) for the atmospheric portion of the model. As an example, here is a description of how a specific climate model works.

The biggest problem with reports like these is they give no indication of the scientific error in the results. They just report a the high, low and maybe the average. No mention at all that it isn't a probability distribution of any sort.

So why not go read the actual paper that Nature published instead of spouting off uninformed crap on subjects you obviously know nothing about?

The thing you said about the 11 degree increase reminded me of a radio article I heard this morning on the way to work, web version is still at the BBC news site. The BBC say that 11 degrees is a worse case and the minimum is 2 degrees. They link to climateprediction.net which are the people performing the experiments. I'm not sure if this is the same as your source.

Do you know anything on how long an event like this would take to reverse? From the stuff I've previouly read shutting down the gulf stream is a lot easier than starting it back up again but I don't remember anyone saying how long it would take to restart.

Quoting the site http://climateprediction.net/info/part_faq.php/ ,

"The programming and resources involved in running a full-scale climate model is tremendous, as well as the peer-review required over years of academics using something like the UK Met Office model. The UK Met Office model is not "open source" -- it is an extremely large, complicated system (something like 500,000 lines of Fortran; two miles of continuous paper if printed out and laid end-to-end). The UK Met Office has been superb in allowing us to bring it over to run on a Windows platform and distribute to the world. They are certainly not getting any money out of it."

I think they ought to disclose what Fortran compiler and what options were used to create the Windows executables.

According to http://www.meto.gov.uk/research/nwp/numerical/fort ran90/ ,

"The Unified Model was originally written in the Fortran 77 programming language with some low level routines written in C to aid portability. Now however, Fortran 90 is increasingly used to take advantage of its new features and to facilitate exchange of code between different international meteorological organizations. Some components of the Unified Model such as the observation processing system and the variational data assimilation system have been written entirely in Fortran 90."

These results were collated from approx. 60,000 separate climate model runs. Here's a link to the actual paper published in Nature (PDF). ClimatePrediction.net passed the 50,000 run mark only a month ago, so it looks like participation is on the up. Kudos to everyone running it! Personally I've switched from SETI@Home to this project. (Of course, you may feel that cancer research into protein folding is more important. One of the nice things about the BOINC framework is that you can contribute to multiple projects at the same time.)

The 'eleven degrees rise over the next century' is of course the worst-case scenario. Of course, climate disruptions of that magnitude really would be catastrophic to human civilisation - for one thing, massive loss of agricultural production, the loss of large areas of expensive real-estate (many of the world's great cities would certainly be under water. I don't know precisely what magnitude of sea level rise 11 degrees would produce but consider that the Greenland ice sheet, which is already showing signs of increased melting, would produce approx. 7m rise - that's goodbye to London and New York and Amsterdam for starters.) Here's a chart from the IPCC's 2001 report showing the various scenarios they based their predictions on. As you can see, the worst-case they foresaw was about 5 or 6 degrees C. The significant thing about these results is that the upper bound of the range of possible temperature rises is shown to be about twice as severe as previously thought. Not only is more and more solid evidence being produced to back the fundamental prediction that human CO2 emissions are causing significant changes in our climate, but the magnitude of those predicted changes is getting greater and greater as time goes on. Note as well that the charts don't suddenly flatline at the year 2100...

Finally I'm looking forward to a discussion on RealClimate.org on this. I've found it to be utterly addictive to see discussions amongst actual researchers in the field, not only showing the areas of legitimate disagreement, debate and uncertainty, but also the solidity of the scientific consensus, as well as busting various common myths - the Crichton garbage, the hockey-stick stuff etc etc. Strongly recommended.

These results were collated from approx. 60,000 separate climate model runs. Here's a link to the actual paper published in Nature (PDF). ClimatePrediction.net passed the 50,000 run mark only a month ago, so it looks like participation is on the up. Kudos to everyone running it! Personally I've switched from SETI@Home to this project. (Of course, you may feel that cancer research into protein folding is more important. One of the nice things about the BOINC framework is that you can contribute to multiple projects at the same time.)

The 'eleven degrees rise over the next century' is of course the worst-case scenario. Of course, climate disruptions of that magnitude really would be catastrophic to human civilisation - for one thing, massive loss of agricultural production, the loss of large areas of expensive real-estate (many of the world's great cities would certainly be under water. I don't know precisely what magnitude of sea level rise 11 degrees would produce but consider that the Greenland ice sheet, which is already showing signs of increased melting, would produce approx. 7m rise - that's goodbye to London and New York and Amsterdam for starters.) Here's a chart from the IPCC's 2001 report showing the various scenarios they based their predictions on. As you can see, the worst-case they foresaw was about 5 or 6 degrees C. The significant thing about these results is that the upper bound of the range of possible temperature rises is shown to be about twice as severe as previously thought. Not only is more and more solid evidence being produced to back the fundamental prediction that human CO2 emissions are causing significant changes in our climate, but the magnitude of those predicted changes is getting greater and greater as time goes on. Note as well that the charts don't suddenly flatline at the year 2100...

Finally I'm looking forward to a discussion on RealClimate.org on this. I've found it to be utterly addictive to see discussions amongst actual researchers in the field, not only showing the areas of legitimate disagreement, debate and uncertainty, but also the solidity of the scientific consensus, as well as busting various common myths - the Crichton garbage, the hockey-stick stuff etc etc. Strongly recommended.

The climates models are computed using the BOINC platform (distributed computing in your PC, similar to SETI, etc.).

Please, help the project donating your idle CPU cycles, go to: the homesite of the project and download the client.

The client (BOINC) supports Linux, Windows, MAC OS, etc.

You might also want to check out the following (Distributed Computing) project:

ClimatePrediction.net

If you want to support real climate change research instead of listening to this fear mongering, head over to climateprediction.net and donate some of your idle CPU time to their distibuted computing project.