cc: Tim Osborn , Phil Jones date: Wed, 09 Feb 2005 11:09:57 -0500 from: "Michael E. Mann" subject: Re: CLIVAR abstract to: Keith Briffa Dear Keith, Phil, Tim: I'm attaching a preliminary result. Let me first explain what I've done. I've adapted to the Mann and Jones composite scheme and applied to pseudoproxies from Caspar's forced CSM1.4 coupled model experiment. The Pseudoproxy experiments and "Composite-Plus-Scale" scheme are described in more detail in the attached paper (provisionally accepted as letter in J. Climate, but likely to change a bit in response to reviewers). I've also attached the "supplementary information" from the paper. I've attached the result of an experiment using an average SNR=0.25, and I've compared the standard composite scheme (i.e., the Mann and Jones approach, but really the way we've all done this sorts of composites in the past, more or less), and the "Moberg" scheme. I've implemented the MOberg scheme in the following way. All pseudoproxies are initially decadally-smoothed since this is the typical starting point (e.g. Crowley and Lowery, Bradley and Jones, and Mann and Jones are all based on decadal resolution proxies) I choose half of the 12 pseudproxies for the low frequency calibration (>80 year periods), and the other half for the high-frequency calibration (<80 year periods). To emulate their use of low-resolution proxies, I first lowpass (at 40 year period) the proxies to be used in the low-frequency calibration (this approximates the smoothness of the sed cores, boreholes, etc.). This turns out to be the essential step. Because, Moberg et al standardized all proxies by their nominal standard deviation. This means that the low-res proxies (which have almost most of their variance at multidecadal timescales), get near unit weight in any compositing, while the high-res (i.e., "tree-ring" proxies), get downweighted because they have significant variance in both bands, so they get only a fraction of a unit weight. In short, the Moberg et al standardization procedure overweights the low-res proxies by design. Indeed, my experiment demonstrates this. I use their scheme, standardizing all proxies by nominal sd first, then separately composite the high f and low f components, recombine, and then standardize by the target (NH) series (i.e., I apply same mean and decadal standard deviation to composite as in the target true model NH series). The attached comparison shows a comparison of actual model NH series (red), Mann and Jones (e.g. standard) compositing approach w/ the 12 pseudoproxies (black), and the Moberg et al scheme (blue). As discussed in our J. Climate letter, CPS will underestimate the low f variance at low SNR, and that's clear here (red curve lies a bit colder than black). But what really jumps out here is the spurious low f variability (in this case, anomalous multidecadal cooling periods) w/ the Moberg scheme. This is just one example. I need to try some more, but the answer that is emerging is that their scheme imposes spurious low-frequency variability. Will update you when I have more complete results. Let me know what you think... Thanks, mike At 10:07 AM 2/9/2005, Keith Briffa wrote: thanks for this . I have to say that I do not think the Moberg paper advances the Science much. Need to see supplementary information , but my initial thought is that there are real problems with their treatment of the "low frequency data" . We are weighed down with trying to put a large European proposal together , but Tim and I are toying wit the idea of writing a response. I find it a bit irritating that they couch their piece as a specific criticism of the M+J series and virtually ignore the other reconstructions , hence setting up a straw Mann (ha ha) . Problem is that thsi juxtaposition of boreholes ,Echo-G and their series will create a new "state of the art" of (apparently) mutually reinforcing evidence. At 12:49 09/02/2005, you wrote: Hi Keith, Thanks for the clarification. I have seen the zero order draft and have looked over it, though not yet in full detail. On the whole appears quite fair, and balanced, and comprehensive. Very nice job! I still think the borehole discussion is a bit unbalanced. It sounds a lot like Henry Pollock to me. The Gonzalez-Rouco simulation is, as we know, somewhat questionable, and there is no discussion of the GISS simulation by Mann and Schmidt which suggests a likely low-frequency bias. So I'm likely to comment on this in my "official" remarks as a reviewer for the chapter (seems a bit odd that I'm both a contributor and reviewer, but as long as you all don't mind, I don't either). Other than that, I think my comments will probably be minor (you know how rare that is!). I see that the comments are due in early April, so I'll probably wait until it gets close so that I can provide an update on the status of in-review/in-press manuscripts. For example, I'm hoping to work w/ Phil on a sort-of response to Moberg et al, which should be quite clarifying (we can get the same result as them, basically, using the Mann and Jones series and a similar approach to what they did, but I think I can show that the approach does very poorly w/ pseudoproxies). By the way, the Mann et al J. Climate letter on the CCSM Pseudoproxy experiments in now provisionally accepted. Will keep you updated. Thanks again for keeping me updated, and congratulations on a wonderful job w/ the ZOD. Its an improvement on what we had in the TAR for sure, and I really like the fact that it gives justice to the large amount of work that has been done by the community since then... thanks again, mike (At 03:50 AM 2/9/2005, you wrote: Sorry !!!!!!!!!!!!!!!! You can see that I am in "headless chicken mode" - no idea how this message got reworked into my inbox! Have you seen the zero order draft ? Where are we with balancing views? Keith At 16:54 08/02/2005, you wrote: HI Keith, Thanks--this is a very old message. I think we provided this abstract many months ago already. SO, unless I misunderstand, nothing for us to do here, I think?? Mike At 11:33 AM 2/8/05, Keith Briffa wrote: Mike just seen this email - please forward their request again and I will look again cheers Keith At 22:43 12/11/2003, you wrote: Dear Keith, No doubt you got the reminder that the CLIVAR folks want a (400 word) abstract from us by Dec 15th for the June meeting. Did you want to take the first stab at this, then we can iterate back and forth a bit? let me know how you want to proceed w/ this... Thanks, mike ______________________________________________________________ Professor Michael E. Mann Department of Environmental Sciences, Clark Hall University of Virginia Charlottesville, VA 22903 _______________________________________________________________________ e-mail: mann@virginia.edu Phone: (434) 924-7770 FAX: (434) 982-2137 [1]http://www.evsc.virginia.edu/faculty/people/mann.shtml -- 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/ _______________________________________________________________________ Professor Michael E. Mann Department of Environmental Sciences, Clark Hall University of Virginia Charlottesville, VA 22903 _______________________________________________________________________ e-mail: mann@virginia.edu Phone: (434) 924-7770 FAX: (434) 982-2137 [3]http://www.evsc.virginia.edu/faculty/people/mann.shtml -- Professor Keith Briffa, Climatic Research Unit University of East Anglia Norwich, NR4 7TJ, U.K. Phone: +44-1603-593909 Fax: +44-1603-507784 [4]http://www.cru.uea.ac.uk/cru/people/briffa/ ______________________________________________________________ Professor Michael E. Mann Department of Environmental Sciences, Clark Hall University of Virginia Charlottesville, VA 22903 _______________________________________________________________________ e-mail: mann@virginia.edu Phone: (434) 924-7770 FAX: (434) 982-2137 [5]http://www.evsc.virginia.edu/faculty/people/mann.shtml -- Professor Keith Briffa, Climatic Research Unit University of East Anglia Norwich, NR4 7TJ, U.K. Phone: +44-1603-593909 Fax: +44-1603-507784 [6]http://www.cru.uea.ac.uk/cru/people/briffa/ ______________________________________________________________ Professor Michael E. Mann Department of Environmental Sciences, Clark Hall University of Virginia Charlottesville, VA 22903 _______________________________________________________________________ e-mail: mann@virginia.edu Phone: (434) 924-7770 FAX: (434) 982-2137 [7]http://www.evsc.virginia.edu/faculty/people/mann.shtml Attachment Converted: "c:\eudora\attach\pseudoproxy-jclimlett2.pdf" Attachment Converted: "c:\eudora\attach\supplementary2.pdf" Attachment Converted: "c:\eudora\attach\CompareSNR0pt25.pdf"