Bet The World on Secret Black Boxes

We frequently see various reports of this or that General Circulation Model (GCM) or the “ensemble mean” of several of them (the average of a bunch instead of just one) showing the earth is going to heat up and it’s the end of life as we know it.

They are the backbone of the IPCC / ARx reports fear mongering.

Recently, several folks have pointed out that actual temperatures are now well outside the 95% confidence interval of many of the models, and the “ensemble mean”, and that statistically there has been no warming for, variously, 12 years to 16 years (depending on the model and the “statistical significance” test used).

In short, the major driver of climate panic, The Models, are no longer giving results that match reality.

Yet they are being used to herd public opinion and lawmakers into the pen of Carbon Credits, high Energy Taxes, reduced quality of life (via higher prices and taxes for fewer goods), and generally force feed the world a Watermelon Green Agenda. Despite having no skill at predicting our present decade and a half of no warming. (And, for those of you under snow on this ‘start of summer’, it’s a bit less than “no warming”…)

So while the discussion mutates from Global Warming through various other “rebranding” events to now “Extreme Climate” and “Climate Chaos” (still trying to find a name to scare folks enough, while not being obviously out of touch with reality as “warming” has become); we are treated ever more to “model results”.

OK, so what’s inside these things? I ponder.

I set off to download the models and look under the hood.

No Can Do.

The One

I did find one model that you can download. So I have. GISS “ModelE” (that presently comes in 3 flavors and a dozen snapshots… it would seem it is a ‘work in progress’ at the moment… Including a ‘special’ one for AR5…)

http://simplex.giss.nasa.gov/snapshots/

It has an online manual and an archive of the version used for AR4:

http://www.giss.nasa.gov/tools/modelE/modelE.html

http://www.giss.nasa.gov/tools/modelE/modelEsrc_ar4/

So I find myself in the odd position of saying: Kudoes to GISS for making their source code available. Good on you.

As it seems I have a “choice of one”, it is their Model E that I’ll be taking a look at. It is in FORTRAN (which I have installed here already) and I’m comfortable with FORTRAN.

More information about it here: http://www.giss.nasa.gov/tools/modelE/

I don’t know that I’ll do more than just unpack it, see if it compiles, and maybe spot check some parts of the code. I expect that the major issues will not be implementation, but assumptions. Building into the model an assumption that CO2 is the big driver, and leaving out things like clouds, lunar / tidal ocean cycles, etc. The disconnect from reality more than the manner in which it is coded. So don’t expect a whole lot from me on it.

The description particularly points out that the parallelized parts expect shared memory, not a Beowulf style MIMD machine (Multiple Instructions Multiple Data). As I don’t have a MISD (Multiple Instruction Single Data) supercomputer available, I’m unlikely to try running the model on a single CPU dinky box. Also, as MIMD is what I like building at home, I’m unlikely to be all that enthused about exploring DIY MISD (and don’t have the $Millions for a SISD or SIMD box; i.e. a giant single CPU).

So mostly I expect to just do a vague overview, attempt to compile, and maybe try running one or two parts of it. Likely on the Raspberry Pi, just to see how much it can handle.

The Other

The page also says:

IPSL-CM4 (Institut Pierre Simon Laplace, France). Here we hit the jackpot: open web access to the subversion repository and Trac database.

I looked around a bit, but the source code didn’t stand out. I presume it just takes more than 2 minutes to find it; and I’ll go back to look again after dinner.

So likely some Kudoes to the French as well. Just not a ‘done deal’ yet.

The Black Box Mob

I ran into an interesting page that did my work for me. It looks at what models exist, and what is the status of the download / source available.

http://www.easterbrook.ca/steve/2009/06/getting-the-source-code-for-climate-models/

As they have already done the work, I see no reason to do more than quote them and point to them. The Author seems a nice fellow, from other articles on the site. There are other interesting articles, like a pointer to his presentation on moderating climate blogs; and a history of climate models video (that I’ve not watched yet – but the reviews were nice).

This is from 2009, and at the end points to a RealClimate link that claims to list more. Having been abused at RealClimate for asking simple questions (when first just finding out about “Global Warming” and having simple real questions, I was insulted and shunned and had innocent questions deleted. But they DID accuse me of being influenced by Wattsupwiththat, and that was the first I ever heard of them… so RealClimate must be credited with sending me there to get polite answers… that had a lot fewer “loose ends”…) I don’t know that I can stomach another visit to RealClimate, so that will have to wait… I doubt things have changed all that much in the last couple of years.

NCAR shows as available “if you register”. I’ve left it in with the Mob as I’m not so interested in registering. Registration usually comes with strings / tracking things.

24. June 2009 · 9 comments · Categories: climate modeling, climate science

First, we have to be clear what we mean by a climate model. Wikipedia offers a quick intro to types of climate model. For example:

zero dimension models, essentially just a set of equations for the earth’s radiation balance

1-dimensional models – for example where you take latitude into account, as the angle of the sun’s rays matter)

EMICS – earth-system models of intermediate complexity

GCMs – General Circulation Models (a.k.a Global Climate Models), which model the atmosphere in four dimensions (3D+time), by dividing it into a grid of cubes, and solving the equations of fluid motion for each cube at each time step. While the core of a GCM is usually the atmosphere model, GCMs can be coupled to three dimensional ocean models, or run uncoupled, so that you can have A-GCMs (atmosphere only), and AO-GCMs (atmosphere and ocean). Ocean models are just called ocean models :-)

Earth System Models – Take a GCM, and couple it to models of other earth system processes: sea ice, land ice, atmospheric chemistry, the carbon cycle, human activities such as energy consumption and economics, and so on.

Current research tends to focus on Earth System Models, but for the last round of the IPCC assessment, AO-GCMs were used to generate most of the forecast runs. Here are the 23 AO-GCMs used in the IPCC AR4 assessment, with whatever info I could find about availability of each model :

BCC-CM1 (Beijing Climate Center, China). The only mention of source code I can find is a link to an email address for the atmosphere model. I’ll fire off a message (Response: “it’s not ready for release yet”).

BCCR-BCM 2.0 (Bjerknes Centre, Norway). I can’t find any info about the model in the BCCR website, but I did find a 2003 paper describing the development of the model.

CCSM3 (NCAR, USA). Model source code is available for download, if you register.

CGCM3 (Canadian Centre for Climate Modeling and Analysis, Canada). Source code not publicly available.

CNRM-CM3 (Meteo-France). Not much information on the GCM, but the source code for the NEMO ocean model is available if you register.

CSIRO-MK3.0 (Commonwealth Scientific and Industrial Research Organisation, Australia). Can’t find any info about the model at CSIRO’s website, apart from this report on the model development. [update: The MK3.0 isn’t publicly available, but a lower resolution version, the Mk3L is – see comments below]

ECHAM5/MPI-OM (Max Planck Institute, Germany). The source code for the models is available if you sign the licence agreement.

ECHO-G (University of Bonn, Germany and Korea Meteorological Administration, Korea). Here’s a technical report describing ECHO-G, but I can’t find much else.

FGOALS-g1.0 (LASG/Institute of Atmospheric Physics, China). Not sure about availability, as most of the documentation is in Chinese.

GFDL-CM2.x (GFDL, USA). Source code for AM2.1 (atmosphere only) is available if you register, as is the MOM ocean model.

GISS-AOM, EH, and ER (NASA Goddard Institute for Space Studies, USA). Source code for various configurations of GISS-AOM is available. You can also browse the source code for the latest model, ModelE, either using the f90toHTML tool, or directly in its repository.

INM-CM3.0 (Institute for Numerical Mathematics, Russia). Can’t find much about this model at all.
[…]

MIROC3.2 (U Tokyo and JAMSTEC, Japan). I found documentation for MIROC, but not much else.

MRI-CGCM2 (Meteorological Research Institute, Japan). No info about model availability.

PCM (NCAR, USA). Unlike the CCSM above, the code for PCM doesn’t appear to be available.

UKMO-HadCM3 and HadGEM1 (Met Office Hadley Centre, UK). Last, but definitely not least. The Met Office models are built from a shared code base, known as the Unified Model. Documentation is available, but the source code is only distributed to collaborators under a restricted licence.

Now, if you were paying attention, you’ll have noticed that that wasn’t 23 bullet points. Some labs contributed runs from more than one version of their model(s), so it does add up somehow.

Short summary: easiest source code to access: (1) IPSL (includes Trac access!), (2) CCSM and (3) ModelE.

Future work: take a look at the additional models that took part in the Coupled Model Inter-comparison Project (CMIP-3), and see if any of them are also available.

The original text is full of live links to the various “providers”. But since they mostly (except for The One of GISS Model E, and the other from France) are made of “unobtanium”, and not available to the un-indoctrinated and un-washed; I’m not seeing much use in the links. If you want to go exploring, hit the article and pick up the links there.

In Conclusion

It sure looks to me like we have a collection of Black Boxes, guarded by a Priesthood, and being used to drive policy and the people. Last time I looked, our form of government did not include the Branch Of Government of “Black Boxes”…

I can make computer code print out anything you like. It’s not hard. Without any independent and PUBLIC code review; we have no idea what the models do. I suppose I could head down the whole FOIA route, but I’m not a lawyer, and can’t afford one. I have to make a living, and this is just an unpaid public service on my part, at this point.

Will inspection of ONE of the models be ‘enough’? Can we play “Bet The World” on a sample of one, out of dozens? Can we trust that these models have any use when they have now got a dozen years history of being wrong and present temperatures are outside of the statistical confidence level? Can we afford to play “Bet The World” on a collection of secret black boxes?

My opinion is “No.”. That’s why I decided to take a look “under the hood”. The “Harry Read Me File” gave us all a pretty good clue that generally the code quality was not very good at UEA, and not very robust, and certainly not well maintained nor archived. Does that extend to the models? Most likely. Organizations have a ‘style’ that tends to be pervasive. Furthermore, the FOIA-2011 / Climategate emails showed clearly that the participants were quite happy to use sub-ethical means to advance a political agenda. (Blackmail of journal editors and suppression of dissenting articles, for example.) So absent the code to look at, all I can look at is the surrounding context. Thanks to HarryReadMe and FOIA-2011, we know that context; and it does not inspire confidence nor trust.

So IMHO, the only prudent thing is to ignore the models. In court, the question “What would a prudent man do?” forms the Prudent Man Standard for reasonable decision making. Given the context of known poor code quality in other work products from the climate schools, the deception on FOIA requests, the thwarting / manipulation of peer review, etc… A Prudent Man can only conclude that the source code is hidden as they have something to hide. So the entire package must be rejected in any claim of “proof” or policy making.

Take the AR4 or AR5, remove the models, and what is left?

Not a whole lot…

Subscribe to feed