cc: "Cox, Peter" , Karl Taylor , bryant.mcavaney@lmd.jussieu.fr, Curtis Covey , "Mitchell, John FB (Chief Scientist)" , mlatif@ifm-geomar.de, Tom.Delworth@noaa.gov, Andreas Hense , Asgeir Sorteberg , Erich Roeckner , Evgeny Volodin , "Gary L. Russell" , Gavin Schmidt , GFDL.Climate.Model.Info@noaa.gov, Greg Flato , Helge Drange , Jason Lowe , Jean-Francois Royer , Jean-Louis Dufresne , Jozef Syktus , Julia Slingo , Kimoto Masahide , Peter Gent , Qingquan Li , Seita Emori , Seung-Ki Min , Shan Sun , Shoji Kusunoki , Shuting Yang , Silvio Gualdi , Stephanie Legutke , Tongwen Wu , Tony Hirst , Toru Nozawa , Wilhelm May , Won-Tae Kwon , Ying Xu , Yong Luo , Yongqiang Yu , Kamal Puri , Tim Stockdale , Gabi Hegerl , James Murphy , George Boer , Myles Allen , claudia tebaldi , Ben Santer , Tim Barnett , Nathan Gillett , Phil Jones , David Karoly , Dáithí Stone , "Stott, Peter" , Francis Zwiers , Ken Sperber , Dave Bader , Mike MacCracken , boyle5@llnl.gov, Stephen Klein , "A. Pier Siebesma" , William Rossow , Chris Bretherton , George Tselioudis , Mark Webb , Sandrine Bony , James Hack , Martin Miller , Ken Kunkel , Christian Jakob , Kathy Hibbard , "Eyring, Veronika" , pasb@dsm-mail.saclay.cea.fr, giorgi@ictp.trieste.it, c.lequere@uea.ac.uk, naki@eeg.tuwien.ac.at, stephen.griffies@noaa.gov, Pierre Friedlingstein , Olivier Boucher , Bala Govindasamy , Jonathan Gregory , Chris Jones , "Jones, Gareth S" , David Lobell , peter gleckler , Cath Senior , Keith Williams , "stephen e. schwartz" , David Easterling , Inez Fung , Duane Waliser , William Collins , Ken Caldeira , Dave Randall , Joyce Penner , Anna Pirani , Bjorn Stevens , Ronald Stouffer date: Tue, 29 Jul 2008 23:42:45 +0200 from: Marco Giorgetta subject: Re: Proposed experiment design for CMIP5 to: Jerry Meehl -----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 Dear Jerry, I won't be at the Snowmass meeting. Here are a few thoughts about 1% vs. RCP runs. The experience of the Japanese colleagues indeed indicates that the concept of the concentration driven C cycle experiments is useful. Within the ENSEMBLES project we are going through a similar exercise, essentially running our ECHAM5/MPIOM based C cycle model in two modes: (1) forced by "RCPs" (ENSEMBLES uses SRES A1B and a new 450 ppm stabilization scenario produced by the IMAGE group) and (2) forced by emissions. In both cases land use change is prescribed and accounted in the C budget. Our experience is that the 20C emissions derived from the "RCP" runs approximate well the emission data used for the "emission" runs. We learned from this (1) that this concept is interesting and experiments are now on the way for the "future", and (2) that the model as a whole can be tuned in an acceptable manner such that a reasonable transient 20th century climate and C dispersion to the different pools can be simulated. As you see we are more or less on the way to explore the discussed experiments (emission driven, RCP driven). No-climate feedback experiments could follow along this track, and they would be politically more relevant than experiments based on 1% CO2 increase per year. This would support the idea to change from the 1% runs to RCP rund. The only disadvantage is the complexity of the simulations, in terms of the many different forcings. The setup of the 1% runs is simple and it is likely that the participating groups in the end really have a comparable forcing. This simplicity also facilitates the interpretation and the model intercomparison, which I think was an important reason for such experiments in C4MIP. The same advantages would hold for a slower fixed increase experiment, for example 0.7% instead of 1% annual increase. If the 1 % increase is unrealistic. The complex setup of 20C or scenario runs may result in a number of differences between different models (as it happened for example with O3 in AR4) and therefore make the "understanding" of the results and of model differences harder. A good point of Peter Cox is of course that the dynamics of the C cycle depends on the degree of disequilibrium, which is different in a 1% run, an RCP run with variable increase of CO2, or for example a 0.7% run. If you think of "no climate feedback" RCP runs, would you propose these in addition to expts. 4.2 and 4.3c of Karl's list of experiments, or as replacement? Would you propose these experiments for the the future only or also for the 20th century? Would you chose the RCP8.5 scenario, because it is also used in expt. 3.3? Best regards, Marco Jerry Meehl wrote: > Hi Peter, > > How long will you be in Snowmass? I get there tomorrow late afternoon > and will be there for the sessions Thursday and Friday. Ron and I were > planning on re-visiting the experimental design more then, and if you > could join in that would be great. > > Regarding your point in favor of using the RCPs for carbon cycle > feedback, I think Ron and I arrived at this conclusion independently > while we both attended a US-Japan workshop in Colorado a few weeks ago. > The Japanese have performed a proof-of-concept experiment using two > idealized mitigation scenarios and basically computed numbers for the > Aspen experiments you originally proposed in 2006. There were two key > additional points that we noted--one was that they started from a > pre-industrial control run so they had 20th and 21st century in the > "climate-carbon feedback" contrasted to "no-climate carbon feedback" > allowable emissions plots. Second, they had some kind of 20th century > "observations" of carbon emissions they plotted on their allowable > emissions graphs to show that their model with carbon-climate feedback > actually tracked those observations for 20th century. Since there are > so few observations to compare carbon cycle feedback to, this seemed > like a fairly compelling reason to use RCPs, which is what you also note > below. > > I think Karl and Ron had lumped the carbon cycle feedback experiments in > the 1% runs both because this had come up as a possibility in the > post-Aspen WGCM meeting in Victoria in 2006, and because it could > possibly present a more pleasing context to evaluate all feedbacks, > carbon cycle and all others. However, on further review, in addition to > the points you raised, deriving allowable emissions from RCPs allows a > check to what the IAMs used for emissions in the first place (and used > to derive concentrations used in the ESMs). Also, it seems to me that > carbon cycle feedback falls into a new category of feedback that we in > the AOGCM world are not used to evaluating. We must depend on the > advice from you and others in that community. Though it's tempting to > think that everything can be boiled out of 1% runs, I think those are > most useful for feedbacks basically "managed" by the atmosphere (like > clouds, water vapor, etc.). The original Aspen concept for carbon cycle > feedback always depended on using actual mitigation scenarios, and I > think we're coming around again to agreeing on that. > > Another point is that the cloud feedback community will make a proposal > to WGCM to enlarge the idealized 1% feedback experiment list, so that > makes separating out the carbon cycle feedback experiments in a separate > category using RCPs more compelling. > > Hopefully we can discuss this more Thursday. > > Jerry > > Cox, Peter wrote: >> Dear Karl and Ron >> >> Thanks for this very thorough document. >> >> Generally speaking I think we should be focusing much more on >> realistic policy relevant scenarios rather than 1% per year type >> experiments. There are two reasons for this: >> 1) Most now consider a ("business as usual") 1% per year scenario not >> to represent a viable future. So detailed information on these >> scenarios is less and less relevant to people outside of the GCM >> modeling community. >> 2) More realistic scenarios allow us to utilize observations to >> validate models/reduce uncertainties in a way that idealized scenarios >> do not. >> So I am in favour of diagnosing feedbacks in the more policy-relevant >> RCP scenarios wherever possible. I say this even though Ron, who is >> sitting beside me here now in Snowmass, has told me that this makes >> identifying model differences more difficult. Ron also tells me that >> this is a fight not worth fighting, but I can't resist commenting >> anyway..:-) >> >> More usefully I would like to respond to your PS. regarding the >> diagnosis of carbon cycle feedbacks. I strongly believe these should >> be diagnosed relative to the RCP scenarios. Carbon cycle feedbacks >> cannot easily be reduced to an equilibrium response plus a timescale. >> Carbon uptake essentially relies on disequilibrium and is therefore >> dependent on scenario, so I don't think it is very helpful to define c >> cycle feedback relative to idealised 1% per year runs. There are also >> the potential for significant "cold-start" problems with the carbon >> cycle (as land and ocean uptake are both highly dependent on history). >> So I vote for diagnosing carbon cycle feedbacks (at least) relative to >> the RCP scenarios. >> >> All the best >> >> Peter >> >> PLEASE NOTE NEW MOBILE NUMBER >> Prof Peter Cox, >> Met Office Chair in Climate System Dynamics, >> Room 336, Harrison Building, >> School of Engineering, Computing and Mathematics, >> University of Exeter, >> Exeter, >> EX4 4QF, >> >> Email: P.M.Cox@exeter.ac.uk, >> Tel (univ): 01392 269220, >> Tel (mob) : 07827 412572 >> >> >> >> -----Original Message----- >> From: Karl Taylor [mailto:taylor13@llnl.gov] >> Sent: Tue 22-Jul-08 09:25 AM >> To: bryant.mcavaney@lmd.jussieu.fr; Curtis Covey; Jerry Meehl; >> Mitchell, John FB (Chief Scientist); mlatif@ifm-geomar.de; >> Tom.Delworth@noaa.gov; Andreas Hense; Asgeir Sorteberg; Erich >> Roeckner; Evgeny Volodin; Gary L. Russell; Gavin Schmidt; >> GFDL.Climate.Model.Info@noaa.gov; Greg Flato; Helge Drange; Jason >> Lowe; Jean-Francois Royer; Jean-Louis Dufresne; Jozef Syktus; Julia >> Slingo; Kimoto Masahide; Peter Gent; Qingquan Li; Seita Emori; >> Seung-Ki Min; Shan Sun; Shoji Kusunoki; Shuting Yang; Silvio Gualdi; >> Stephanie Legutke; Tongwen Wu; Tony Hirst; Toru Nozawa; Wilhelm May; >> Won-Tae Kwon; Ying Xu; Yong Luo; Yongqiang Yu; Kamal Puri; Tim >> Stockdale; Gabi Hegerl; James Murphy; Marco Giorgetta; George Boer; >> Myles Allen; claudia tebaldi; Ben Santer; Tim Barnett; Nathan Gillett; >> Phil Jones; David Karoly; Dáithí Stone; Stott, Peter; Francis Zwiers; >> Toru Nozawa; Ken Sperber; Dave Bader; Mike MacCracken; >> boyle5@llnl.gov; Stephen Klein; A. Pier Siebesma; William Rossow; >> Chris Bretherton; > George Tselioudis; Mark Webb; Sandrine Bony; James Hack; Martin Miller; > Ken Kunkel; Christian Jakob; Kathy Hibbard; Eyring, Veronika; > pasb@lsce.saclay.cea.fr; giorgi@ictp.trieste.it; c.lequere@uea.ac.uk; > naki@eeg.tuwien.ac.at; stephen.griffies@noaa.gov; Cox, Peter; Pierre > Friedlingstein; Olivier Boucher; Bala Govindasamy; Jonathan Gregory; > Chris Jones; Jones, Gareth S; David Lobell; peter gleckler; Cath Senior; > Keith Williams; stephen e. schwartz; David Easterling; Inez Fung; Duane > Waliser; William Collins; Ken Caldeira; Dave Randall; Joyce Penner; Anna > Pirani; Bjorn Stevens >> Cc: Ronald Stouffer >> Subject: Proposed experiment design for CMIP5 >> >> Dear all, >> >> As most of you know, plans are well underway for a coordinated set of >> climate model experiments, which will constitute the Fifth phase of >> CMIP. Attached is a description of the proposed experiments. As >> members of the CMIP panel, which was established by the WCRP's Working >> Group on Coupled Modelling (WGCM) to help coordinate this activity, we >> are seeking your comments. Considerable thought and input from a wide >> community of scientists have already contributed to the CMIP5 design, >> and therefore major changes are not envisioned. Competing interests >> and various tradeoffs have been carefully considered before coming up >> with the proposed suite of experiments. Please keep in mind that >> modeling groups have limited resources and the experiment must >> represent a compromise among various priorities. We will not be able >> to please everyone. >> >> The CMIP panel must present a final design plan for CMIP5 to the WGCM >> at its annual meeting in September, just two months from now. Given >> this tight deadline (which cannot slip if the CMIP5 results are to be >> available in time for the IPCC's Fifth Assessment Report). For this >> reason, we ask that you send us (taylor13@llnl.gov and >> Ronald.Stouffer@noaa.gov) any comments and suggestions you have by >> September 1, 2008. >> Feel free to pass this document on to anyone you think will have an >> interest in it. We invite comments from scientists associated with >> all aspects of the climate change issue, spanning the three IPCC >> working groups. >> >> With best regards, >> Karl Taylor (PCMDI) and Ron Stouffer (Chair, CMIP panel). >> >> P.S. Please note that there are remaining details yet to be worked >> out. In particular it has been suggested that experiments 4.2 a&b >> described in the document should be performed in conjunction with the >> so-called RCP-driven experiments given in Table 2 rather than with the >> idealized (1% CO2 increase per year) experiments of Table 4. >> Experiments 4.2 allow us to separate out the climate-carbon cycle >> feedback. The original proposal was in fact to do this separation for >> the RCP runs, but several scientists offered compelling arguments for >> switching this diagnostic analysis to the 1% runs. Some of the >> reasons for making this change from the original proposal can be found >> in section 9. Still, there are some scientists who continue to >> express a preference for the original design. Please let us know what >> you think about this. >> >> >> >> >> >> >> >> >> > - -- Marco A. Giorgetta Max Planck Institute for Meteorology Bundesstr.53, 20146 Hamburg, Germany Tel./Fax : +49 (0)40 41173-358/-298 Email : marco.giorgetta@zmaw.de -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.4.8 (MingW32) Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org iEYEARECAAYFAkiPjtUACgkQXnnmHpkSX2RISACfbchEzrKdzJMIrkNHK8frQE92 JZwAoI1N9pna6QUih+ApNxZ+0hjOx/YP =vy8G -----END PGP SIGNATURE-----