date: Mon Mar 9 17:26:51 2009 from: Keith Briffa subject: Re: citations and comments on draft to: Darrell Kaufman Darrell am only trying to play devil's advocate. As for the df problem (and I agree it is) , I think some direct reference to it in the text is preferable to letting referee use it as a stick to beat you. Yes, do try higher resolution regression , obviously for the proxies that allow this. To be honest, when I look at Figure 2 , I can see that decadally-smoothed data (or band-passed data) would show some inverse correlation between temp. and proxy average though - so you may be accused of cherry picking your time scale . Nevertheless, the logic of justifying how you are building from local high-resolution association, to regional lower-resolution interpretation should be attempted. We do not expect high-resolution association on Arctic average scale to be strong without very dense coverage of proxies, as is shown by instrumental analyses ( Briffa and Jones , 1993 ) - while at decadal scales coherence is greater (Jones and Briffa , 1996) so we can get away with less records - and presumably even less for long-term trend , provided it is captured in records . I find that I do not have pdfs of the papers I have mentioned - so will need to get them scanned and sent to you . I am not in the office tomorrow but will leave instructions to this effect cheers Keith Briffa, K.R. and Jones, P.D., 1993 Global surface air temperature variations during the twentieth century: Part 2, Implications for large-scale high-frequency palaeoclimatic studies. The Holocene 3, 77-88 Jones, P.D. and Briffa, K.R., 1996 What can the instrumental record tell us about longer timescale paleoclimatic reconstructions? pp.625-644 in: Climatic Variations and Forcing Mechanisms of the Last 2000 Years (Eds. P.D. Jones, R.S. Bradley and J. Jouzel), NATO ASI Series Vol. 141. Springer Verlag, Berlin, Heidelberg. (R) At 16:49 09/03/2009, you wrote: Keith: Thank you for your insights. I think I can deal with nearly all of your suggestions. You picked up on the plug for proxies from lakes at high latitudes. I'll try to tone that down, but do want to retain the message. I'll see whether our NCAR co-authors can address your suggestion about using an energy balance model to relate Milankovitch forcing to temperature. That will take some time, but I'd rather submit a manuscript with a high probability of success than to have to go back to the drawing board. Most importantly: You commented about the significance of the regression used to scale the proxy values to temperature. This is the weakest point of the paper, in my view. Because the focus is on the 20-year intervals, we only have 5 points for the regression, which is a serious limitation, and I doubt would hold up to statistical scrutiny. On the other hand, I am confident that the proxy values do correlate with temperature and the correlation is significant at even the annual scale (see Fig 2), and here I mean 'significant' even after accounting for autocorrelation effects on the df. Would it be valid to present the statistics for the annual correlation, then use this to support the scaling for the 20-year means, even though the n (= 5) for the 20-year means is too small to derive statistical significance? Or can you see another way around this issue? It's important to scale the proxy data to temperature, and I believe that our data can support this, I'm just not sure how best to make the case. I'll append the values for the five, 20-year intervals that I used to calculate the scaling in case you have some ideas to try. Also, could you please forward a pdf of your 1990 Nature paper? Our electronic subscription picks up at 2000 and I wanted to follow up on your suggestion that we place some probabilistic estimate on the significance of the recent warming. Thanks again; the paper will be much stronger with your input. Darrell JJA(C) Proxy 0.26 1.66 -0.05 1.35 0.09 1.54 0.11 1.32 -0.31 0.29 On Mar 9, 2009, at 8:44 AM, Keith Briffa wrote: Darell perhaps a short piece in the Supp. Inf. could give a little more detail on RCS , cite the original refs and show a Figure? but CAN NOT do for now , so let's go with it as is to make your deadline. This could be done later if referees say so. Attached is a tracked version with my comments (forwarded only to Ray also for hard-nosed comment!). Cheers Keith At 16:54 08/03/2009, you wrote: Also: Would it be important to cite the original sources for the tree-ring data, or do you think citing your new compilation would suffice? thanks. Darrell -- Professor Keith Briffa, Climatic Research Unit University of East Anglia Norwich, NR4 7TJ, U.K. Phone: +44-1603-593909 Fax: +44-1603-507784 [1]http://www.cru.uea.ac.uk/cru/people/briffa/ <2k synthesis v6-KRB.doc> -- Professor Keith Briffa, Climatic Research Unit University of East Anglia Norwich, NR4 7TJ, U.K. Phone: +44-1603-593909 Fax: +44-1603-507784 [2]http://www.cru.uea.ac.uk/cru/people/briffa/