On the science side, reasearch obviously continues.

There is a further paper on the topic of calculating Jeffreys\' Priors for climate models recently made available on Arxiv.

Objective Probabilistic Forecasts of Future Climate Based on Jeffreys’ Prior: the Case of Correlated Observables
Stephen Jewson, Dan Rowlands, Myles Allen
http://arxiv.org/PS_cache/arxiv/pdf/1005/1005.2354v1.pdf

The abstract of this and the earlier paper on which this is expanding are quoted below. The papers linked particularly the recent one may well be considered rather technical rather than interesting bedtime reading.


To attempt an introduction to what these papers are about:

A 2005 paper:
Constraining climate forecasts: The role of prior assumptions
http://climateprediction.net/science/pubs/2004GL022241.pdf

discussed the problem of how producing a probability from an ensembles of models is complicated by the way parameter space is sampled. Spacing the samples equally distanced in climate sensitivity will yield different probablities than if the samples are equally distanced apart in another plausible way such as feedback (proportional to 1/sensitivity).

The key feature of a Jeffreys_prior is that it is invariant under such reparameterizations.


Abstracts

To include parameter uncertainty into probabilistic climate forecasts one must first specify a prior.
We advocate the use of objective priors, and, in particular, the Jeffreys’ Prior. In previous work we
have derived expressions for the Jeffreys’ Prior for the case in which the observations are independent
and normally distributed. These expressions make the calculation of the prior much simpler than
evaluation directly from the definition. In this paper, we now relax the independence assumption and
derive expressions for the Jeffreys’ Prior for the case in which the observations are distributed with
a multivariate normal distribution with constant covariances. Again, these expressions simplify the
calculation of the prior: in this case they reduce it to the calculation of the differences between the
ensemble means of climate model ensembles based on different parameter settings. These calculations
are simple enough to be applied to even the most complex climate models.

The previous work was

A new method for making objective probabilistic climate forecasts from numerical climate models based on Jeffreys\' Prior
http://arxiv.org/abs/0908.4207
Stephen Jewson, Dan Rowlands, Myles Allen

We argue that it would be desirable to use Jeffreys’ priors in the construction of numerical model
based probabilistic climate forecasts, in order that those forecasts could be argued to be objective.
Hitherto, this has been considered computationally unfeasible. We propose an approximation that
we believe makes it feasible, and derive closed-form expressions for various simple cases.

If you want to discuss these Jeffreys\' priors papers, can I suggest this thread or create one on this board.

---

Visit BOINC WIKI for help

And join BOINC Synergy for all the news in one place.

FAMOUS v.6.10 MILLENNIUM models

• Available on all platforms.
  
• You may need to select it in the climateprediction.net preferences of your account.
  
• FAMOUS takes about 9 days crunching 24/7 on a fairly fast computer.
  
• 200 model years.
  
• These 200-year segments form part of models lasting 1600 years. After completion the segments are stitched together.
  
• The first 200 years, 599 - 799, are a spinup or settling-down period which allows climatic conditions to stabilise.
  
• The deadline for these models is three months which is much shorter than for other model types. This is because CPDN needs the results back as quickly as possible to stitch the segments together.
  
• Files are uploaded every five model years ie 20 files in total.
  
• Trickles are uploaded every model year ie 200 trickles.
  
• As v.6.10 processes more slowly than the previous CPDN version, credits have been increased to maintain parity with other CPDN models. (If you crunched the earlier version models you will also receive the increased credits for them.)
  
• At the moment CPDN FAMOUS displays the traditional graphics as for other model types. New graphics are being developed in conjunction with Sony Labs in Paris but are still being beta-tested. They will be released on CPDN in a future FAMOUS version.
  
• Each FAMOUS task web page, reached from our computer's task list in our account, displays new interactive graphs. They include clickable red dots for volcanic eruptions. Here is a historical list of Large Volcanic Eruptions.
  The graphs display in most browsers but not in Internet Explorer 7 (Windows). You can upgrade to IE 8 at no cost. If you have XP please upgrade to IE 8 in case you are not able to when Windows support for XP stops soon.
  
• Some FAMOUS models crash, typically with exit code 22 and messages including 'INVALID THETA' or 'NEGATIVE PRESSURE', because of instabilities that develop with some combinations of parameter values. Please do not restore backups of models and rerun them after model instability crashes. On the same platform with the same operating system the models would crash again at the same point. (Reminder: if your models crash with other exit codes eg -186 or -226 you should post on the forum for advice.)
  
• If your multicore computer is also running other model types you may still wish to make backups to help ensure completion of the other models.
  
• FAMOUS does not use enormous amounts of memory. Most multicore computers should be able to load all cores with FAMOUS and run them simultaneously. (Reminder: on a multicore computer with 4 or more cores don't fully load with HadAM3P as they will slow each other down, in a few cases disastrously.)
  
• A future FAMOUS version is expected to be hyperthreaded.
  
• You can read about the Millennium Experiment led by Dr Hiro Yamazaki at Oxford University here. Here is the website of The Millennium Project which is pan-European and multidisciplinary.
  

Thank you to all members who are helping to advance this large research project.

Re-edit: As Geophi has said in his post below, the new CPDN version of FAMOUS can now run on computers without SSE2 or SSE3 processor capability, ie older processors.

---

Cpdn news

Die CPDN-BOINC-Seiten (z.B. auch unsere Teilnehmer-Konten-Details) können auch in anderen Sprachen als nur Englisch angezeigt werden.

Dafür wurde jetzt ein Link zur Sprachauswahl-Seite zum blauen Menue (links) der BOINC-Seiten hinzugefügt.

Dieser Link führt direkt zur Sprachauswahl (nicht alle wählbaren Übersetzungen sind gleichermassen vollständig).


Wichtig : Nachdem bei CPDN das letzte BOINC-Update durchgeführt wurde, gab es ziemlich heftige Probleme mit der deutschen Übersetzung, speziell in den Projekteinstellungen, der zweite Menuepunkt lautete fälschlich "GPU verwenden", das wurde jetzt korrigiert zu "CPU verwenden". Diejenigen von Euch, die dort zwischenzeitlich ein "Nein" eingestellt haben, müssten es wahrscheinlich wieder auf "Ja" ändern, damit sie wieder Arbeit bekommen.


Die Ãœbersetzungsfehler waren Folge eines Problems mit der Sprachdatei beim BOINC-Update.

We can view our CPDN account details in some other languages instead of English.

There is now a link to the language selection page in the blue menu.

This is the selection. Some translations are more complete than others.


When CPDN upgraded its version of Boinc at the beginning of June there was a serious error in the German translation of the second item in the ClimatePrediction preferences of our accounts. It said GPU. Now it correctly says Use CPU. If you edited this second preference and answered No, your computer(s) cannot download new work for CPDN. Some German-speaking members may need to edit this answer which must say Yes.

The error in the translation happened because of an accident with the file.

CPDN with Windows and Boinc service (protected) installation

If your computer has

• Windows and
  
  
• a Boinc version lower than 6.10.49 and
  
  
• a Boinc service (protected) installation
  

it is probably crashing its climate models of every type with the -226 exit code. Members with these computers should upgrade their Boinc version as this problem has been fixed. After an upgrade your models should process successfully. Download your upgrade from Boinc.

Members with Linux and Mac (or Windows with a non-service (screensaver) installation) are not affected.

---

Cpdn news

FAMOUS v6.11 MILLENNIUM models

An updated version of the FAMOUS application was released yesterday. The only change from v6.10 is the addition of the enhanced graphics developed in conjunction with Sony Labs in Paris. The model processing is unchanged and we would encourage users to continue running any v6.10 tasks to completion.

Please note that a problem with the mirroring of the application files was causing a third of download attempts for the new version to fail. This has now been fixed. Apologies to those who were affected.

---

"The ultimate test of a moral society is the kind of world that it leaves to its children." - Dietrich Bonhoeffer

There are two CPDN news announcements, which subscribers to the RSS feed may already have seen:

Myles Allen wins the Appleton Medal

Retirement of HadSM3 and HadAM3P models

CPDN logo mugs

You may have seen from the news on the climateprediction.net index page that climateprediction.net logo + globe mugs are now available from the CafePress shop. I have one and like it. The cost of shipping is the same whether one orders one mug or more. Each purchase will provide US$2 profit towards CPDN's operating costs.

---

Cpdn news

Geoengineering experiment

Many members ran HadCM models for the CPDN Geoengineering experiment.

Nature has published a letter (short article) about the experiment's results. The abstract is here: Regional climate response to solar-radiation management by Katharine Ricke and M. Granger Morgan of Carnegie Mellon University in Pittsburgh and Myles Allen of CPDN at the University of Oxford. The potential political implications of their findings are significant.

There is a forum thread about this publication with further links here.

---

Cpdn news

Members who ran HadSM models while they were available may be interested to read Iain Inglis's final analysis of the geographical distribution of 'iceworlds' here.

Reminder: If you are still running a HadSM or HadSMMH model, please look at its graphics every two or three days if possible to check that its globe shows all the colours. A single colour for the whole globe means that an iceworld has developed and the model should be aborted.

---

Cpdn news

FAMOUS_s3 models

A batch of 599-799 FAMOUS models has been released with s3 in their names. They are for a specific study. The greenhouse gas forcing (effect on the climate) is kept constant for the first 30 years during which the model graphs should only show small fluctuations. In 630AD the greenhouse gases (ghg) are halved in one step; the parameter says ghg_stp_half. Here is one such model.

Or the ghg value is doubled in one step. The parameter says ghg_stp_dble. Look at one model's graph.

These models invent hypothetical changes to study the effects. They do not simulate a real sudden climate change in the year 630.

The FAMOUS graphs were developed at Sony CSL in Paris by Peter Hanappe, Anthony Beurive who is an open source developer and Laurent Saifre, a student who made the initial version of advected texture cloud graphics. As usual, Tolu makes them run on all our home computers. Thank you to all these developers.

Reminder: You can see your own tasks in your account. Click on the Tasks link to see your models.

---

Cpdn news

Tolu is leaving CPDN for another post. Hiro's announcement here:
http://climateprediction.net/board/viewtopic.php?p=89959#p89959

---

"We have met the enemy and he is us." -- Pogo
Greetings from coastal Washington state, the scenic US Pacific Northwest.

HadAM3P regional models

There is now an explanation of the Regional Experiment in the Experiments section of the ClimatePrediction website. You can read about how a small part of the Earth is modelled at higher resolution than CPDN has ever achieved before while incorporating global data. There are three different groups of researchers for the Pacific North West, Europe and Southern Africa models. In the Climateprediction preferences in your account you can if you wish select one or more of these regions. Read more here.

Reminder: HadAM3P models are memory-intensive. If a multicore computer runs a full load of them simultaneously they will probably slow each other down. Try to run some FAMOUS alongside them.

---

Cpdn news

CPDN main project

CPDN Geoengineering Experiment

Many of us ran 80-year 2000 - 2080 HadCM3L models for the Geoengineering Experiment which is a joint project for Oxford University and Carnegie-Mellon University in the US. Kate Ricke and Daniel Rowlands now need more of these models for an extension to this experiment, described in an extra paragraph which has been added at the end of the Geoengineering Experiment page. Milo posted a link to a first publication about this experiment here.

A small number of these HadCM models have already been released on CPDN and if they run well more will follow. These are long models. You can select or deselect them by editing the climateprediction.net preferences in your account. HadSM is fairly long, though not as long as HadCM3L. If you prefer shorter models select HadAM3P EUR, SAF or EUR (though none are available at the moment) and FAMOUS.

Check the Server status page for current model availability.

---

Cpdn news

CPDN main project

Milo has configured the server to prevent creation or modification of profiles unless you have a recent average credit of at least 5. This has been done to stop a bot which has been automatically creating accounts with spam profiles.

---

"The ultimate test of a moral society is the kind of world that it leaves to its children." - Dietrich Bonhoeffer

Apologies for not posting an announcement about this earlier on this forum.

WeatherAtHome

CPDN launched this initiative with the three regional models on 17 November. Read about it on the Guardian website here where CPDN researcher Sue Rosier explains how these models can be used to show extreme weather events in much greater detail than with global models.

You can watch a short video of Sue describing the WeatherAtHome initiative here. There is also a description of WeatherAtHome on the CPDN website where you will see that research groups from three countries will be involved in analysing the regional model results.

WeatherAtHome is not a new Boinc project. It is an initiative within the existing CPDN project. It is aimed only at new members who want to run the regional models.

What does WeatherAtHome consist of for its new crunchers?

1. It has a special registration page. This enables people to choose the regional model type they want before they join the project. Normally when one joins a Boinc project the first task one receives is sent at random from all the types available on the project server.

Existing members do not need to do this. We can select and change the model types we want by editing the climateprediction.net Preferences in our account. A link to our account is in the blue menu on the left side of this page. We can see which model types are currently available for our computer by looking at the links to the Server Status page and the Applications page.

2. New members who join CPDN through the WeatherAtHome registration page will receive a Boinc version pre-configured differently from usual. CPU usage is set at 60%, not the usual 100%. This is to protect computers that have limited memory/RAM because the regional models are memory-intensive. It also protects laptops from overheating.

Existing members do not need this specially-configured version of Boinc. In our CPDN account in the Computing Preferences we can change the CPU usage for all our computers on all projects. If we want to vary the CPU usage for an individual computer we can open its Boinc manager and in the Advanced menu > Preferences > Processor usage we can change the settings.

Edit: Milo, our programmer, has told me that this 60% CPU usage preconfiguration for WeatherAtHome members did not work. It is set as 100% as the default for everyone. If you have a laptop we advise you to reduce this 100% value either in the Boinc manager Advanced menu > Preferences > Processor usage, or in the Computing Preferences in your account. And when using a laptop always raise its little feet to increase air flow and cooling.

3. At the moment WeatherAtHome members can only receive the regional models for Windows. Versions for Linux and Mac may be made available later.

Existing CPDN members do not need this Windows restriction. Although the regional models are only available for CPDN members with Windows at the moment, members with Linux or Mac can run HadCM (which is much longer) for the Geoengineering experiment or FAMOUS for the Millennium experiment.

4. WeatherAtHome members can of course also change the settings in their accounts, for example for CPU usage or to select different model types.

Welcome to all the new WeatherAtHome members.

Existing members should not rejoin CPDN through WeatherAtHome. Please complete all your current models which will all be used.

---

Cpdn news

CPDN WeatherAtHome HadAM3P Regional Models

This is an updated version of the post made when the HadAM3P regional models were released on CPDN 3 months ago to clarify the current lack of applications for Linux and Mac and the name change for the Western United States region (from Pacific North West).

These models all consist of a HadAM3P global model coupled with a higher resolution regional model and run for one model year. The regions covered are Europe, Western US and Southern Africa.

• HadAM3P regional models are currently only available for Windows. We are hoping to make them available for Linux and Mac again at some point in the future.
  
  
• If you joined the project via the WeatherAtHome registration page you will already have set your model type preference. Otherwise you may need to select the model types you wish to run in your project preferences (which now includes a map to allow the selection of a location within each region that will be monitored on task pages).
  
  
• The downloads for these models are large:
  
  
  
  • 103.5MB of region independent data files. These will only be downloaded once.
    
  • 12.4MB of application version stamped files. These will only be downloaded once but will have to be replaced if a new version is released.
    
  • 5.5MB of region dependent data files. These will only be downloaded once for each region.
    
  • 11.7MB of date dependent data files. These will have to be downloaded when you run a model for a date range you haven't already run.
    

  
  

• These models have a much higher memory requirement than FAMOUS (around 250MB compared to 30MB) and the findings in beta testing suggest that they don't scale up very well on multi-core systems (a slow down of between 7 and 10% when 2 tasks are run in parallel on a 2 or 4 core system with 2GB of memory). This slow down might not be as apparent on systems with more memory.
  
  
• The regional models cover different sized areas and have different resolutions. They return a trickle and upload a file at the end of each month and upload a further file on completion.
  
  
• European region
  
  
  
  • Takes about 4 days of 24/7 crunching on a 2.4GHz computer.
    
  • Monthly upload files are about 13MB and the final upload is about 35MB.
    
  • 2,386.39 credits for a completed task.
    

  
  

• Western US region
  
  
  
  • Takes about 5 days of 24/7 crunching on a 2.4GHz computer.
    
  • Monthly upload files are about 6MB and the final upload is about 27MB.
    
  • 3,005.88 credits for a completed task.
    

  
  

• Southern African region
  
  
  
  • Also takes about 4 days of 24/7 crunching on a 2.4GHz computer, but slightly faster than Europe.
    
  • Monthly upload files are about 3.5MB and the final upload is about 32MB.
    
  • 2,244.09 credits for a completed task.
    

  
  

---

"The ultimate test of a moral society is the kind of world that it leaves to its children." - Dietrich Bonhoeffer

Neil Massey, who is one of the CPDN researchers, has said in the Independent Forum News:

Milo's last day

Today is Milo Thurston's last day on climateprediction.net. On behalf of the team I'd like to thank him for all his hard work over the past 4 years and for keeping the project infrastructure running smoothly. We wish him well for his new project, which I'm sure Milo would be happy to let you know about.

Thanks again, Neil and the CPDN team.

---

Cpdn news

Yesterday Sue Rosier, one of the CPDN researchers, posted in the News thread of the independent forum:

Well as you are no doubt aware, today is the 1st of December, which also means it is CPDN's first day without either Tolu or Milo, so I thought I might take a moment to send a bit of an update about recent happenings at CPDN.

First, to re-iterate Neil's message of yesterday, I'd like to add my huge thanks to Milo for doing such a fantastic job over the last four years, looking after the CPDN database and generally ensuring the smooth-running of the project. He will be much missed. We all wish him every success in his new venture.

We are also of course missing the recently departed Tolu and hoping that he is having fun and being successful in his new role. Big thanks also to Tolu for his many years of skilful and dedicated service to CPDN.

It is strange to think of the project without either Milo or Tolu but we hope soon to be welcoming new staff to take their places.

We recently welcomed Cameron Rye as postdoctoral staff on the project. He joins us from Cambridge, where he has just completed his PhD at the Scott Polar Research Institute, modelling glacier mass balance in Svalbard. Cameron will be working on the HYDRA project with William and Myles, investigating the moisture budget in the CPDN ensembles, including the results from the new regional model experiments. He will also probably be performing some extra simulations of his own to accompany these studies. Cameron has done a sterling job getting up to speed very quickly with many aspects of CPDN and we look forward to working further with him.

Jara Imbers Quintana also joined the team relatively recently and she likewise has coped tremendously with the steep learning curve of CPDN, including communicating with the media. A native Spanish speaker, Jara recently spoke live to Colombian breakfast radio about the launch of the “weatherathome” regional modelling experiment and is booked to provide another interview for a Latin American TV channel this week. Jara has a Masters degree and PhD in Theoretical Physics and works some of the time at the Mathematical Institute here in Oxford, but most of the time in AOPP, where she is currently involved in developing a new algorithm for the representation of sea ice in upcoming CPDN experiments.

If you followed the recent launch of the “weatherathome” regional modelling experiment you may now be aware of the climateeducation.net initiative. This is what has kept me very busy over the last many months. In partnership with the Met Office and Oxford University's Department for Continuing Education we are developing two online courses (or “information packs”) which aim to communicate the basics of climate science and modelling, and how to go about interpreting the results of climate modelling experiments. The first course is aimed at the general public, whilst the second more at those who wish, or need, to use a regional climate model in their work (e.g. scientists, climate impacts/adaptation specialists, policy-makers etc). Among other things, the first course presents some of the results from the CPDN transient experiment, for which big thanks are due to you, the participants, for the crunching and Dan Rowlands, CPDN graduate student, for the data processing. The first course is due to launch the week after next and if you are interested, you can register at http://climateeducation.net.

As ever, there is plenty more going on at CPDN and I aim to keep you as up to date as possible via this page.

Best wishes,
Sue

---

Cpdn news

Here's a message from Pardeep:

-----
Dear Seasonal Attribution Project participants,
The results of our work on attribution of the UK autumn 2000 floods have finally been published (http://www.nature.com/nature/journal/v470/n7334/full/nature09762.html), and have generated a fair amount of media interest!
I just wanted to say one final big thank you to all the participants who crunched simulations for us – we couldn’t have done the study without you! And I similarly want to say thanks for the enjoyable discussions I had back on the project’s old message boards.
It's sure taken a while to get this work finished, but now it’s great to see the follow-up weatherathome.net project is well underway, and I hope we’ll see more interesting results come out of that.
Happy crunching,
Pardeep