date: Fri Sep 10 10:50:41 2004 from: Phil Jones subject: Re: [Fwd: 2004JD005306 Decision Letter] to: Adrian.Simmons@ecmwf.int, Vanda Da Costa Bechtold , Anton Beljaars , erp@ecmwf.int, Sami Saarinen , ers@ecmwf.int, Pedro Viterbo , Nils Wedi Adrian et al., Here are a few thoughts on the two reviews. First there should be 3 so let's be thankful there wasn't a third one. I hope it doesn't take you long to respond to these comments. The revised paper will likely go back to Reviewer 2. The editor is asking for major revisions, but I don't think they are in the major category as they only require changes to the text, not to any of the analyses, figures and tables. I'm away Sept 16 until Oct 31 at a series of meetings. I'll be here on Sept 24, Oct 1, 6-8, 14-15 and 22. Meetings are 3-4 days, so here all Fridays ! Reviewer 1 Seems to want 2 things, (1) clearer statement as to the purpose of the paper wrt detectability of long-term trends/interpret results and not saying whether ERA-40 is better than NCEP. I would have thought, though, that comparison papers are one way to advance atmospheric sciences. Reanalyses need to be compared with something to judge whether they are improving. Maybe need to add something about the need to spend time on improving the data input for the next generation (said in the GCOS plan). Hand in hand with this is improvement to the physics and its understanding. The second (2) is an additional comment about the SST fields. For the period in question (1958-) the SST fields are about as good as they are going to be. SST doesn't have biases in this period with buckets. If there is a problem it is with the lack of observations in the Southern Oceans and also with sea ice estimates around the Antarctic. Sea ice is likely to be an important factor and Nick Rayner (and Dick Reynolds) are working to improve the input. But 1958 onwards is a lot better than pre-WWII. I reckon any fictitious trends over this period are small. Improvements may come with full ocean reanalysis, but these may not go back to 1958. WRT CRU data used in ERA-40. This is impossible to assess. Over the 1958-2001 period we have about 1500-2000 stations going in. More like 1500 recently and 2000 in the 1960s. I'll explain why in response to Rev 2 later. The input temp data aren't exactly the same. CRU uses CLIMATs, you the 6-hr SYNOPs. CLIMATs are based on whatever method of calculation countries use. All are anomalies though, so this should be irrelevant. You should be able to use any SYNOPs (within reason) , but CRU must have the CLIMAT message (i.e. be based on full monthly averages). It would be good to assess how much we each use, but I don't think it is going to be that useful. Such a comparison would be a by-product of any concerted effort to go through what you do use as SYNOP input - with a view to adding in lots of missed data prior to 1967 and 1979. As for NCEP, this presumably just needs a sentence. I think your simulation was just one run? Reviewer 2 I thought some of these comments were being deliberately pedantic. Could the various models, forecasts/simulations/increments etc all be covered in one paragraph in section 2? I'm writing this as your email (with the response to Saki in) came through. We are in agreement about any changes being minor rather than major - perhaps Rev 2 said something additional in the confidential comments and the editor didn't look too closely. Anyway, it seems as though they don't fully understand the various components. I didn't when we started but I do now, so it must have been well written. Maybe bringing them altogether may help. I tend to concur with Saki that they seem more like Cai than Kalnay. Having heard Cai talk, I suspect sloppy and hand waving are words he might have got in reviews of his papers ! They are words I would not have used either. I will email NCDC again about their max/in datasets. A response to this should be that we are working on this and a comparable dataset to the mean temperature one (CRU) is not yet available. Some continents are good, but others are markedly poorer. Agree this should help, but comparison will be difficult because max/min will not always be when your 6-hr values are calculated. There will be longitudes that will be fine, others 3 hours out. Page 4, para 2 seems very clear to me. Page 5 last sentence. A rewording here. All you were doing was trying to portray agreement in the best light and not with offsets. In CRUTEM2v, the % of the NH (from land data) is 50% in the 1960s and 40% in the 1990s. I know that the NH is only ~30% land, but in these &s I count the whole box even if there is one island in it. The reduction is only partly caused by station closures. The real reason (well the main one) is that countries put out more data in the decade books (WWR) about 4-5 years after a decade. So the 1990s isn't yet out for some continents. The ones that add coverage are S. America, Africa, Asia and Australasia/Islands and they are not yet out. They make about 2/3rds of the difference, with the rest coming from closures. The GSN initiatives have halted the rate of decline and even improved it slightly. All the above is not relevant really as we only use CRU boxes which are near complete. If the 1990s were as good as the 1960s then we would get about 7% more boxes (mainly in South America, Africa and the tropics and maybe a few in the Russian Arctic. The corresponding numbers for SH land are 23% in the 1960s and 18% in the 1990s. Page 7 and 8 comments seem clear in the text to me. I reckon he's not read it that carefully. All I think you can do is spell it out a little more. Page 9 Add a reason why trends from 1979. The fact that everyone starts in 1979 is clearly one. Page 18 DTR will help - said already. ERA-40 and NCEP don't have greenhouse gas increases. It is by no means clear that their increases leads to downward DTR trends. The observational papers (like Easterling et al. ) Easterling, D.R., Horton, B., Jones, P.D., Peterson, T.C., Karl, T.R., Parker, D.E., Salinger, M.J., Razuvayev, V., Plummer, N., Jamason, P. and Folland, C.K., 1997: A new look at maximum and minimum temperature trends for the globe. Science 277, 364-367. only suggest this - there is no proof. GCM DTR changes are nowhere near as large as the observational changes. Also not all regions show DTR decreases. Europe, for example, barely shows this. In the TAR, there is a statement that observed changes in DTR are larger than simulated by models (coverage is less as well) and there is much greater uncertainty than with mean temperatures (p714 of the TAR, right-hand column). Most of the observed changes are due to changes in clouds. Models then didn't have indirect aerosol effect. I reckon that if we'd not analysed DTR and found the decrease, no modeller nor the detection people would be looking at it ! Cheers Phil Prof. Phil Jones Climatic Research Unit Telephone +44 (0) 1603 592090 School of Environmental Sciences Fax +44 (0) 1603 507784 University of East Anglia Norwich Email p.jones@uea.ac.uk NR4 7TJ UK ----------------------------------------------------------------------------