cc: "'michele'" , "'Maurizio Maugeri'" , "'Alexander Orlik'" , "'''Wolfgang Schöner' ''" date: Tue Jun 3 11:00:09 2008 from: Phil Jones subject: Re: WG: EI-news to: "Reinhard Boehm" Reinhard, Received OK this time and thanks. I had a look at the earlier spreadsheet last night. Looking at the new spreadsheet it looks as though the adjustments look good for the NW and NE regions. Basle and Geneva do not stand out as being different. All three areas show the recent warmth, but the warm bump in extended summers are comparable between the late 1940s/1950s and the 1790s/1800s, or the latter period was just a few tenths warmer. It does look as though for S the 1940s/1950s were clearly warmer than all earlier periods. You will be aware of a few of the Italian series drifting away from the others in the S region. I guess these may involve Milano and Padova - it is difficult to tell with the colouring of the lines. The same series drift away in the extended winter as well. These Italian series don't drift away in the difference plots (extended summer minus extended winter), so maybe it's a common problem. In the difference series (summer minus winter) the change is occurring prior to about 1890. This is much later than the problem decade of the 1860s when a lot of changes were occurring. I think in any paper it would good to also look at these difference plots for CET and De Bilt and maybe the Swedish series. I can send you the data to do this if you want. Some plots are in the attached paper from 2003. So, my suggestion would be to write up these adjustments. I assume you will need to explain the Kremsmunster results as a start. I presume the numbers I will need for the ALP-IMP WP9 paper are somewhere in this spreadsheet. I think all I'll need are the averages for the three regions NW, NE and S, but I won't start on this again until you've got an initial draft of your paper. Maurizo's point about clouds will work, but the cloud data will not be good enough to do this. Anders couldn't sort out the Swedish cloud data - even in the 20th century. I don't think the cloud data - even if digitized - would be a help. There are so many issues with observation times, and the fact that observers appeared to sometimes base the observation on the time they looked and sometimes over the whole day. The Swedish categories also reduced the further back in time. I don't want to slow down the paper, so this is only a suggestion, but I think the time series of rainday counts may be a better homogeneous proxy for cloudiness. Cloud observations will be a waste of time! Cheers Phil At 08:36 03/06/2008, Reinhard Boehm wrote: Phil, This yesterdays mail did not pass your file size limit. Therefore I try it once more, now with an attached slim version of the xls-file. I hope this will reach you now. Reinhard ___________________________________________________________________________________ Von: Reinhard Boehm [[1]mailto:kliboe@zamg.ac.at] Gesendet: Montag, 02. Juni 2008 11:04 An: 'Maurizio Maugeri'; 'Reinhard Boehm'; 'Phil Jones'; 'michele' Betreff: AW: EI-news Dear Maurizio, Michele and Phil, Thank You Maurizio for your mail. To initiate further proceeding, I attach here a completed file final-comp-HISTALP-ISAC.xls. It contains what Phil wanted (timeseries of summer minus winterhalfyears) and I have added also hom minus ori timeseries. And all already existing plots have been completed with the respective regional mean series (all based on the 20-yrs smoothed version). As I understood Maurizio argued to use the HISTALP version for further proceeding. Of course this is also my preferred option, but I still think we should mention, that independent homogenising activities in Italy produced a cooler EI-period, and maybe to use this as a kind of uncertainty measure. Whats your opinion? Reinhard ___________________________________________________________________________________ Von: Maurizio Maugeri [[2]mailto:maurizio.maugeri@unimi.it] Gesendet: Mittwoch, 28. Mai 2008 11:41 An: Reinhard Boehm; 'Phil Jones'; 'michele' Betreff: Re: EI-news Dear Reinhard, in the last months Michele and I spent a number of days on our early records with the aim of producing a revised and improved version of our homogenisation. We really tried to use all available information and we considered, beside the minimum and maximum temperature records, also the wide metadata availability that we have for some of our records. Moreover, after each new correction, we checked the homogenised records in order to verify whether the applied correction caused a new inhomogeneity in the daily temperature range records or it produced an anomalous picture of the yearly temperature cycle. According to our experience, such checks turn out to be very useful in order to avoid illegitimate corrections, which always constitute a remarkable risk in indirect homogenisation methods, especially in time periods characterised by low data availability and/or by a high number of inhomogeneities. So, we are really confident that we did our best with our records, even though we perfectly understand that the very low number of early records in our dataset is a strong limitation which reduces our results significance. Actually the records we used are not just Milano and Padova, but also Torino, Bologna, Genova (which is a very good record), Mantova, Udine and Alessandria. Moreover, we think that our data yielded interesting results that seem to be in good agreement with dendrochronological data, glacial data, etc. and that is also in good agreement with other datasets like the Swiss (partially) and Spanish ones. Nevertheless, I perfecly agree with you: if the vast majority of sites are warm in the EI period, we have to say that from the point of view of probability the warmer solution is more likely. Sciences are based on data and the final word pertains to the data themselves! So I agree with your proposal of applying the warm solution to the entire GAR, even though I think it is important to mention clearly in the publication that, beside the more likely warm solution, also a cold solution is possible though being less likely: the conclusion could be that, even though the data lead us to the warm solution, the question warm or cold for the early period is at present time still partially open. So, in the future, it could be very interesting to plan (and to propose for EU funding) a project aiming at reconsidering all EI European records with the aim of better understanding if the warm solution is really the most correct one. Such a project could also include other variables like e.g. cloud cover. Cloud cover can probably help to better decide between the warm and cold solutions because, if the 1860s strong summer warming showed by the cold solution is really present, that should be mirrored by a clear signal in the cloud cover records. At the same time, also the pressure records could be helpful (see e.g the EMULATE dataset and related projects). I have also read the answer by Michele from the Budapest meeting. I would suggest (for the Italian group) to stop here with the data analysis. Michele and I are completely aware of your records characteristics and the data you sent us are the same of the ones we subjected to a number of analyses during last autumn. On the other hand, you are perfectly aware of the characteristics of ours. So, I think that, after Phils comments, its time to start writing, trying and presenting the chosen solution and thoroughly discussing its possible limits. Ciao, Maurizio ----- Original Message ----- From: [3]Reinhard Boehm To: [4]'Phil Jones' Cc: [5]'Maurizio Maugeri' ; [6]'michele' Sent: Wednesday, May 28, 2008 10:08 AM Subject: AW: EI-news Phil, Maurizio, Michele Yes, the diverging seven Italian cold-series (those with the suffix ISAC) were independently homogenised by Michele using Italian series only. All the other series (also those seven ISAC-series) were homogenised by us, using our routine HOCLIS-procedure). So there seems to be a break in the 1860s rather than a real diverging climate signal north vs. south, and my point is, that I see no systematic reason to EI-correct the vast majority of series in order to fit the majority to the minority. The relation of the original series is something like 8 warm vs. 2 cold series in subregion NW, 8 warm vs. 0 cold ones in NE and 3 warm ones (Torino, Genova, Trieste) vs. 3 to 4 warm or partly warm (?) ones (Milano, Padova, Mantova?, Bologna?) and several with unclear to very low quality (Verona and Trento). So I think it makes no sense to adjust some 20 warm series to some 5 cold ones, if there are no strong reasons from station history to do so. Therefore I see only two remaining possibilities to deal with the problem: 1) adjust the 5 cold series to the majority (thats what we have already done) 2) to mention that there is also another solution based on Italian series only and to use this as kind of an uncertainty band for the early period As to the idea to produce AMJJAS minus ONDJFM plots: there are none so far, but they can easily be done (but not earlier than this week because I have to leave in an hour and have other things to be done this morning We will also produce (lowpass filtered) HOM minus ORI plots. We have done so for non filtered series, but there you dont see much because of all the highfrequent noise Best regards _______________________________________________________________________________ Von: Phil Jones [[7]mailto:p.jones@uea.ac.uk] Gesendet: Dienstag, 27. Mai 2008 17:08 An: Reinhard Boehm; 'michele'; 'Maurizio Maugeri' Cc: 'Alexander Orlik'; "'''Wolfgang Schöner' ''" Betreff: Re: EI-news Reinhard, Thanks. I need to look at the file in some detail. A quick look suggests to me that there are still problems with some series as the plots for the 4 groups diverge prior to about 1860. They remain quite tight except for one or two after this date. Have you plots of AMJJAS minus ONDJFM? Enjoy the dendro meeting. Cheers Phil At 14:45 27/05/2008, Reinhard Boehm wrote: Friends, This is to inform you about the latest EI-news. We have gone through all our longterm t-series once again and you see a summary of the result on the attached xls-file. On the first table sheet you see a comparison of all series for summer- and winter half years (AMJJAS and ONDJFM). Those with the suffix ISAC are the seven series which were sent by Michele and which, in the homogenised form (with Italian series only, if I remember well) are considerably colder in the warm season. Michele had just done once a comparison of the HISTALP-MEAN in Italy and the ISAC-mean. For a better understanding the second sheet shows the three subregions. In the attached file you get more information. You find the original and the homogenised series, for all single 32 long-term sites. All are lowpass filtered and without gaps and the originals are the outlier correctedoriginals, Alexander has produced in the last months. They are all anomalies to a long common 150 years period (1851-2000) The result in one sentence: the vast majority of sites are warm in the EI, so from the point of probability we think the decision is clear: the warmer solution is more likely. More in detail: It was (apart from the two strange series Verona and Trento) a game between Milano, Padova, Geneva and Basel (cold) and all the rest (also including some of the Italian sites, which are EI-warm). In my eyes the only real argument in favour of this group of cold EI sites is that the two of the three longest (starting in 1760) belong to them. But there are several (I think better) arguments for the warm solution: 1) the splitting happens quite suddenly in the 1860s, exactly at a time when in the Swiss weather service was founded and also in Italy there was a time of change from less organised to better organised service (your words Maurizio). This was not the case for southern Germany, Austria and Hungary. So the >Italian and Swiss sites should be more suspected for breaks especially in this decade 2) after the 1860s there were no regional differences in neither of the subperiods, therefore and also because of 1) it is extremely unlikely that we see a real splitting into subregionally different climate evolutions with a cold South and a warm north in the EI-period 3) we have also looked at the hom-ori series of each single site and we found no systematic accumulation of breaks in the 1860s in the warm series subset So for the moment we think it would be wise to use our solution for the entire GAR, but to mention in the publication, we want to start writing now, the existing difference between the warm and the cold solution and to use it as kind of uncertainty measure (whats your opinion Phil?). What would be the alternative: We should re-adjust more than 20 warm series based on the result of only four cold series. This really is the alternative, because we cannot describe it as a real subregional climate effect (as already argued). I am looking forward to your comments The rest of the week I am not at the ZAMG (EURODENDRO 2008) But maybe we could draw some decisions next week? Best regards Reinhard 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 ----------------------------------------------------------------------------