date: Fri, 30 Aug 2002 16:54:15 +0100 from: Tim Osborn subject: Re: PARCS High-Res Meeting to: Konrad Hughen ,Keith Briffa At 19:55 06/08/02, Konrad Hughen wrote: >Dear Keith and Tim, > >The PARCS High-Res Group is putting together a workshop in Boulder for Oct >31 - Nov 2. The focus of the meeting will be to present the latest >results of the Arctic high-resolution paleoclimate analysis, and to bring >the co-authors together to discuss issues such as analytical methods, >coordinating the addition of new records, identifying areas needing >coverage, future work, etc. The PARCS steering committee will send around >an email circular with more details soon, but I wanted to let you both >know personally and encourage you to come, if at all possible. Konrad, thanks for this. The email has now arrived about the PARCS meeting. It seemed that accommodation and shuttle to/from airport would be paid for from PARCS money, but it was a bit ambiguous about whether our international airfare would also be paid for. Could you clarify that please? Keith is away till Wednesday, but I'll talk with him then about the meeting - it would certainly seem worthwhile for either or both of us to attend. >On a research note, we've done a first-cut analysis of the records we have >so far, and several things come to mind. The spatial EOFs show strong >regional coherence, which of course is good. I'm worried about the >density network data though. They don't show as much long-term >variability as the other data sets, especially the late 20th century >warming. In general, do you feel that ring width data are more closely >correlated with temperature than density? At the local level, the density have a much stronger correlation with temperature than the ring-widths do. The optimal seasons are also different though - for density it is best correlated with an April-September mean temperature at most locations, though some central and eastern parts of northern Siberia seem to respond more strongly to a shorter high-summer seasonal-mean temperature. For ring-width, it is a shorter June-August seasonal-mean temperature that appears most strongly related. At the regional and circum-hemispheric level, the superiority of density over ring-width lessens - though not completely. I would still prefer the density, even at that spatial scale, knowing (i) that the local signal-to-noise ratio is better, and (ii) in the earlier period of the data set, when the number of chronologies with data is far fewer, it is likely that the ring-width would deteriorate more rapidly than the density. We are still working on the possibility of combining the ring-width and density data to produce an improved reconstruction, but for the present I think that - FOR THIS NETWORK - the density is certainly the variable to use. As for the long-term variability... (1) many, though not all, of the trees were sampled before 1985 and therefore miss much of the recent summer-time warming. (2) there is a relative decline of the tree-ring density in this network, in comparison with Apr-Sept temperature, that has been documented before though not yet explained. This is documented in Briffa et al. (1998) - the full reference was given in my email to you when I sent the data. There was other information in that email that you should look at too, in case you missed it first time round (I can re-send it if necessary). Anyway, because of this decline in recent decades, we often only use pre-1961 density data during calibration/reconstruction. Of course this shortens the period available for calibration and verification. The post-1960 data still produces a good high-frequency match to the temperatures, but doesn't have the same underlying trend. (3) there is a lack of multi-century variability because the way the time series were standardised (to remove tree-age effects) removed the low frequencies. At the regional scale, we had (see the Briffa et al., 2001, paper) applied an alternative technique (Age Band Decomposition, which is similar to RCS or Regional Curve Standardisation) that preserves this low frequency signal, but we had not applied it to the finer spatial scales of the data set that we sent you. So the data you have do not have the very low frequencies in them. We have now modified the finer spatial scale data so that it's low frequencies match our regional reconstructions, and I could send that data for you to use instead. That has the multi-century variability in it (though we still have the problem of the recent "decline" noted above). >You'd mentioned earlier that you were working on applying the same RGC >corrections to the ring-width records. Are these data available for this >project as well? Or do you think they might be soon? It might be good to >compare results using the ring-widths to density and see if there are >large differences. The density data we sent hadn't had the RGC corrections applied either - though see above where I explain that the density data can now be provided with low frequencies retained through the use of an age-banding technique (should be similar to a regional growth curve approach). Our work with the ring-widths is still progressing slowly. We don't yet have any data to make available, though Keith knows more about this than me (Keith, can you reply about ring widths). Best regards Tim Dr Timothy J Osborn | phone: +44 1603 592089 Senior Research Associate | fax: +44 1603 507784 Climatic Research Unit | e-mail: t.osborn@uea.ac.uk School of Environmental Sciences | web-site: University of East Anglia __________| http://www.cru.uea.ac.uk/~timo/ Norwich NR4 7TJ | sunclock: UK | http://www.cru.uea.ac.uk/~timo/sunclock.htm