cc: m.hulme@uea.ac.uk, tim.carter@vyh.fi, t-morita@nies.go.jp, s.raper@uea.ac.uk, parryml@aol.com, ssmith@pnl.gov, naki@iiasa.ac.at, John@jvmitchell.demon.co.uk date: Mon, 17 Sep 2001 12:48:34 -0600 from: Tom Wigley subject: Re: TGCIA to: Rob Swart Dear all, With four ways to handle non-CO2/SO2 emissions as suggested by Rob, this gives 105 times 4 equals 410 runs. This is nothing really, since I had to do about 200,000 runs for the Science paper. The real issues are: examining and comparing all the results; and communicating the main conclusions to policy makers/advisors. Naki has suggested a way of cutting down the number of MAGICC runs. For non-CO2 gases, I have already done some sensitivity runs similar to what Rob suggests using base case and stabilization data for A2 and A1B developed by Steve. Studies like this are probably all that is required. For the runs with MAGICC, the critical issue seems to be getting the aggregated SO2 emissions -- which, as I said before, is best (and perhaps only) done by Steve. Just what is the deadline that you are working to? For the GCM runs, which modeling groups are going to be (or have agreed to be) involved? You are right that even a single group of ensemble runs is about all one can hope for -- but, if two or more GCM groups are involved, then there is some value in this for the GCM community. An example is the different conclusions that one can draw from the Hadley and PCM stabilization experiments already performed. There are three important issues here. First is just what scenarios should be used. Results from MAGICC will give a guide to this. An alternative to using any of Morita-san's results is to use something more idealized; as we did with PCM. I am not advocating this directly, but there is some middle ground. Suppose, for the sake of argument, the A1B case is chosen as the 'ball' (base case) to run with. Then, a good experiment would be to change one thing at a time. So, the first set of runs would be the base case ensemble -- perhaps already done by some GCM groups. The second would be to change only CO2. The third would be to change both CO2 and SO2. (It would be nice to do a 'change SO2 only' case, but this would probably not be possible.) The fourth would be to change CO2, SO2 and the other gases. For the fourth case, I or Sarah would have to provide the projections for the other gases -- and we would have to select one of the four possibilities that Rob has outlines (or some alternative, which could be decided on the basis of MAGICC results). This makes a minimum of 16 simulations!! The second issue is whether some runs should be extended to 2200, as we did with PCM -- but this could be done perhaps in a less ad hoc way. One of the beauties of the GCM runs is that end of this century data could be save as start up files for future extensions out to 2200, if this were deemed valuable. The third issue is providing the projections for non-CO2/SO2 gases. MAGICC can do this at the global-mean level for all but stratospheric ozone (where MAGICC only calculates the forcing directly). For halocarbons, however, it is a bit tricky, since the full suite of halocarbon (and related gases) results was calculated off line with other software of mine (which no-one else has). If you read the Dai et al. papers, you will see that we only consider CFC11 and CFC12, scaling one of these up to account for other related gases. Not being part of the IPCC GCM exercise, I do not know what was done with all these other gases in the TAR runs. Different models have different capabilities, so how was it assured that the results from different models were comparable? Perhaps John Mitchell can fill me in on this? Whatever, for at least some of these gases (viz. the halocarbons and stratospheric ozone), there will be no difference between the base case and stabilization runs; and I presume (hope) that consistent base case data sets already exist. Tropospheric ozone, by the way, is tricky, because (at least in our runs) the changes in this gas are spatially variable. Again, I am not sure what was done for the IPCC TAR -- John?? A final issue is just how much of this Sarah can do, and how much might I have to do. I will be away in Australia from Sept. 23 through Oct. 18. Sarah needs to get back to me on this some time soon. I hope these comments are useful. With care, this can be a very valuable set of GCM experiments. The hardest part might be in making the inputs in different models fully consistent and comparable. I don't know how this was done in the TAR, but I suspect it was done in a somewhat cavalier fashion through lack of time (apologies if I am wrong here). One final point, it would be very valuable to have, as a product from all models and all simulations, radiative forcing time series at the hemispheric and land/ocean level. This is a non-trivial task, which few other than Hadley seem to have done (and apologies here, too, if I am wrong on this). Cheers, Tom. ********************************** Rob Swart wrote: > > Dear Sarah, Steve, Tsuneyuki, and others, > > Martin asked me to summarise the email discussion of the last few weeks on > possible TGCIA recommendations for GCM stabilisation runs in order to be > able to have concrete proposals at the next meeting in Barbados. I > understand that some DEFRA (?) money is now available to support Sarah > doing some work for this purpose. I also spoke to Mike Hulme about the > issues briefly. Let me summarise at the risk of stating the obvious, or, > maybe, nonsense as a non-climate modeller: > > After all the debate, I still believe that it would be useful for the > TGCIA to attempt to recommend a few stabilisation scenarios to GCM teams > to encourage work that could be useful for the Fourth Assessment Report. > It was frustrating that the policy questions for the TAR Synthesis > Report on the advantages of stabilising GHG concentations could not be > addressed adequately. > It may be more useful if a few GCM teams would run ensemble runs for one > particular stabilisation scenario to enhance the statitical significance > of the results, than many teams doing single runs for several > stabilisation levels. The results could be used for additional analysis > using pattern scaling techniques. Even if the results would not show any > significant difference between the reference and stabilisation scenario > until well into this or maybe even the next century, this would still be > useful information. I must admit that I do not fully understand Steve > Smith's suggestion to "compare a base case with a 'synthetic" model run, > derived from a combination of the GCM run and simple model results > (perhaps with artificial noise added to derive statistical results)". > In order to facilitate the selection of stabilisation scenarios, Sarah > would run the 15 post-SRES cases, taking Morita's post-SRES GHG > emissions and calculate radiative forcing with MAGICC. I attach the > latest file from Morita-san again. To allow this to happen, we still > need two things: > First, I believe only global CO2 and SO2 emissions were made available. > In addition, the regional SO2 emissions should be made available by > Morita-san and translated from the SRES to the MAGICC regions for the 15 > proposed cases. I understand that Steve Smith knows how to do this, but > am not sure he can do this at short notice. Steve?? > Second, we would have to find a solution for the emissions non-CO2 GHGs, > which are usually included in the SRES scenarios, but not in the > post-SRES cases. Tom's paper made it again abundantly clear that these > assumptions are crucial. Gases controlled by the Montreal Protocol can > be taken from SRES. I see several options for the other gases (notably > CH4, N2O, HFCs and ozone precursors), all of which are somewhat > arbitrary: (a) just take the non-CO2 emissions from the baseline > scenarios because more detail may not be needed at his stage (only > identification of interesting scenarios), (b) scale methane and nitrous > oxide proportional to the post-SRES CO2 emissions profile (slow > abatement) and ozone precursors according to the post-SRES SO2 profile > (more rapid abatement), (c) take the lowest non-CO2 GHG profiles from > the SRES set (probably B1) for all sabilisation runs, or (d) develop > some simple stylized set of non-CO2 emissions profiles. When it comes to > actual recommendations to the outside world after the Barbados meeting. > we would have to think more carefully about a consistent recommendation > for the non-CO2 gases. I hope that Morita-san can advice on an > acceptable solution for the short term. > It was suggested by Mike that Sarah could also calculate temperature > changes using various GCMs (according to Mike MAGICC can simulate 7 > different ones). This may further facilitate the selection process, but > even if so, the resources may be insufficient to do this for all 7*15 = > 105 cases. However, the results of the first 15 runs may allow us to > identify the most interesting cases for further analysis. I depend on > specialists to judge if this would be desirable or not. > > I will be in Wembley for the next 2 weeks for the IPCC Plenary meeting on > the Synthesis Report. John Mitchell will be there as well, and Martin will > attend as well coming Saturday, so I can discuss it with him again and I > hope to be able to read my email, so if there are major problems with the > proposed approach above, involving Sarah running MAGICC after having > received information on non-CO2 GHGs from Morita-san and translation from 4 > SRES to 3 MAGICC regions from Steve. I hope this works out. > > Kind regards, > > Rob > > (See attached file: Post-SRES data for TGCIA(rev.22Aug01).xls) > > ------------------------------------------------------------------------ > Name: Post-SRES data for TGCIA(rev.22Aug01).xls > Post-SRES data for TGCIA(rev.22Aug01).xls Type: EXCEL File (application/msexcel) > Encoding: base64