date: Fri Sep 2 10:21:40 2005 from: Phil Jones subject: Re: current status of the draft to: Anders Moberg Anders, Let's try and do a 6th region, so when you send an email to David tell him about the changes to the Black Sea region and also the Greater Alpine region. I would suggest the latter should include only Swiss and Austrian stations, unless there are any in the very south of Germany. If it is clear to David what he has to do, then it shouldn't be too onerous. We are taking off the key station, so it should be simpler. cc me on the email and I'll talk this through with David on Monday. As you say, just the necessary plots for the paper are all we need. Also give David titles for the plots, he'll have to remove the focus on.. As for WP4, show some of the cluster plots to indictae the difficulties. We'll have a number of the cluster experts in the group, so we'll get some feedback. Alex could show these. Either during or just after the meting we'll need to decide what additional work will be most profitable. Not much has happened on D16, so raise that not just in WP4, but also when WP3 is discussed. Also one or two ppt on the data problems would also be useful. Cheers Phil At 06:49 02/09/2005, you wrote: Phil, About Figure 11-13 (regional time series): Up to now, David has identified five regions. Five is an odd number and this makes it difficult to set the page layout for the plots (see attached). Four or six would be better. I have thought about adding a Greater Alpine region (there are quite many stations there) and changing the Blacksean region to a more NE-wards located SW Russian region (with more stations than the Blacksean). This, however, would mean more job for David to look through the time series and find a suitable selection of stations. What is your opinion about the option with 6 regions? "Presumably you'll talk about many of these analyses during WP4, but hopefully more on what is planned beyond this paper." Well, at the moment I have not many concrete ideas about what to do in the four extra months. I will need to see what the other groups are doing, and then discuss with you and others when we meet in Paris. We need to make sure that we deliver all the deliverables, and this will determine what will be the best things to do. I remember well that you and I talked about doing cluster analysis to define regions and then analyse trends in different regions. The preliminary work that Alex made on this tells us that this can be difficult. We should discuss if the cluster analysis is a useful way to proceed, or if we (me) should go in another direction. I can show some of the clustering results from Alex (or I can ask him to do it). I have a feeling that work on D16 (anthropogenic) has not proceeded far in any group. We'll see... Later today, I will specify exactly what I would like to get from David when he is back. He does not need to re-do all the plots etc. The important thing now is to make those calculations that are necessary for the paper, and also to make those plots that we will use. Cheers, Anders At 10:39 2005-09-01 +0100, you wrote: Anders, One other thing. I forgot a couple of comments I had on the figures. The current Fig 10 (precip) will need to say in the caption or the text, that the very low correlations for a few stations are mostly in the Med. I hope they are. These types of plots also are potentially useful as homogeneity tests as well. The related temperature plot is much better behaved. The other summary plots are all good. David will be able to revise the time series plots, omitting the key station. The maps look OK. Cheers Phil Anders, Here's a few comments on the current draft. The summary in the results sections is well written. Some bits will likely get repeated in the conclusions and the abstract. What might happen is that you might have to expand on them in the section of the paper and move what is there to the conclusions/abstract. Stick though with what you have for the moment. For the summary/abstract apart from the above, the points to stress are our attempts to produce consistency between the indices across Europe (expressing the precip indices as %s of 1961-90). Also the issue of trying to do as much homogeneity as practical to produce good results. I suspect this will get enhanced once David has produced the comparisons for precip, like you've done for temperature. Minor points There needs to be a sentence in the intro to say we're using standard software. I know it is later, but it could go in the 3rd paragraph. In the data collection section, there is mention of 4 principal sources, which is clear. This is then followed by the second category. I think you just need to say there are 3 categories, and maybe slightly better distinguish the two categories and sources. The first category is readily available sources, the second developed as part of EMULATE and the third data from individual scientists. Emphasize that none of these can be assumed to be OK (problem free) here or in the homogeneity section. Romania and Russia prove this. Could emphasize here and also in the introduction, that the more these series are used for work of this kind, the more any problems will be found. Different types of analyses highlight different sorts of problems. I think we've said some of these things in IMPROVE and also in the 2005 IJC paper. The issue is that homogeneity per se, without any form of analysis, is unlikely to create error-free datasets. Even with the CRU monthly data, we are still finding problems that have been there for ever. Several passes are needed through datasets to get rid of most of the serious errors. In basic quality controls, it would be better to briefly detail the QC steps here rather than refer to the EMULATE web site. Can do the latter, provided the basics are in the paper. Emphasize for Sweden that Tx and Tn were not homogenized separately. The climate indices section is good. David will be able to help with the Analysis Methods section. He should be back in on Monday. In the section on Spatial distribution of Trends, need to say first time that linear trends account for the numbers given, just so it is clear. Don't say precipitation day, but say wet day. The 9th word of the correlation between indices section is delaled. I think you meant dealed, but the proper word is dealt. This is just another English verb where for various tenses it is highly irregular. Para 2 say unsurprising rather than non-surprising. The second half of this paragraph took a while to read and understand fully what was being said. I look forward to the next draft. As I might have said I'll be away from Sept 8-17 inclusive. We are on target for a paper to be ready by the Paris meeting. Presumably you'll talk about many of these analyses during WP4, but hopefully more on what is planned beyond this paper. Hope all these are useful. Cheers Phil At 15:39 30/08/2005, you wrote: Phil, I attach the paper in its current status. As you see there is some rather complete text in most sections apart from the bit with regional time series and the final discussions and conclusions. Obviously, the text on precip will be changed once David has re-run his analyses with the new Russian data. If you have time, it would be good if you give me some indications of any big things that look bad or if you think something is missing. Also, any suggestions for points to stress in the discussion and conclusions are welcome. If you have too little time, then don't bother too much with this now. I will complete the paper anyway - hopefully towards next weekend or so. Cheers, Anders 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 ---------------------------------------------------------------------------- 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 ----------------------------------------------------------------------------