cc: "Martin Juckes" , date: Fri, 25 Feb 2005 09:42:41 -0000 from: "Myles Allen" subject: RE: Millenial Temperature Reconstruction Intercomparison and to: "Keith Briffa" , "Martin Juckes" , I floated the idea of bringing Anders Moberg and Jan Esper in on the proposal (offering them both travel money), because one of the things I would want to do would be to get a better grasp of their error analysis, and it's always a lot easier to do this by talking friendlily to people than by reverse-engineering their papers. I tried and failed to understand Mann's error analysis using both approaches about 5 years ago, so I don't think it is worth trying again, particularly given his current level of sensitivity. I don't think anyone was particularly against the idea, but we haven't done anything about it. Would people like me to? Can we make a deliverable of this project a piece of public-domain IDL code (or matlab, if people prefer and someone else volunteers to write it) that takes temperature and proxy inputs and generates reconstructions using at least two methods (Moberg and Crowley et al, for example, plus ideally MBH, Juckes et al, Osborne & Briffa etc etc), providing a framework for comparison. This kind of exercise was hugely valuable in teasing out the origins of differences between different approaches to optimal fingerprinting prior to the TAR. If people provide me with inputs and exact specification of algorithms, I would be happy to produce this (if it's in IDL) as my contribution. Myles Climate Dynamics Group Atmospheric, Oceanic and Planetary Physics Department of Physics, University of Oxford Tel: 44-1865-272085/925 Fax: 44-1865-272923 E-mail: myles.allen@physics.oxford.ac.uk -----Original Message----- From: Keith Briffa [mailto:k.briffa@uea.ac.uk] Sent: 25 February 2005 07:59 To: Martin Juckes; hegerl@duke.edu Cc: Martin Juckes; Myles Allen; t.osborn@uea.ac.uk Subject: Re: Millenial Temperature Reconstruction Intercomparison and Evaluation (MITRIE) Hi all just for interest , Tim and I (and Philip Brohan at the Hadley Centre) have also been playing with "classic regression" and variants to avoid bias . I have also looked at the Esper stuff in detail and am convinced that that the application of the Regional Curve standardisation he employs introduces significant bias in his reconstruction , and coupled to the way he uses a cosign lat. weighting , another early bias . These act together to make the early part of his scaled series ( I do not call it a reconstruction , because some series included do not have a clear temperature signal ) too cold. So while the differences between reconstructions , and the amplitude of each can be varied by using similar scaling in each, there are real differences in represented signal (and spatial and processing questions) related to them also. There will be a significant job to do here just describing the underlying character and biases in these data . The outline looks good . At 16:40 24/02/2005, Martin Juckes wrote: >Thanks Gabi, > >I think that will fit in well, especially the systematic checking -- >which is precisely what they want to pay for, I think. > >I'll try to send out a slightly extended outline tomorrow. > >cheers, >Martin > >On Wed, 23 Feb 2005 hegerl@duke.edu wrote: > > > > > Hi all, > > > > WIthout having looked at the proposal (my home day today, so can > > 't look), I have already looked > > into scaling various reconstructions with a nonbiased technique (tls), and > > find that it makes some difference particularly to records that don't > > correlate magnificently with the target of reconstruction. So my hope is > > to sell this as part of the plan, and write it up. But there also seems a > > real difference, Mann and > > jones for example even if tls scaled has less variance than Esper et al. > > because in that reconstruction, the early 20th century is probably the > > strongest excursion, while in Esper et al earlier variability is stronger. > > So there still is real differences in record. > > so what I hope to add to the proposal is stuff I have already well > > started, but > > needs to be checked more systematically, and > > written up). Tom and I also have a new reconstruciton taht is submitted > > for publication (apart from one last iteration with Cos) with tls and is > > quite similar to Moberg et al > > > > Gabi > > > > On Wed, 23 Feb 2005, Martin Juckes wrote: > > > > > > > > Hello, > > > > > > I've finally got around to sketching a rough plan of objectives and > > > deliverables. Given the small amount of money and the specific aims > > > there is not much room for innovative thinking, but hopefully the > > > work done within the poroject will be helpful to work which people > > > are already doing. > > > > > > Essentially, I'm proposing to assemble/get hold of a representative > > > proxy database and run a representative set of reconstruction methods > > > on the same data. > > > > > > The proposal has to be submitted before Thursday, March 10, so we have > > > two weeks to come up with it. > > > > > > A rough outline is attached. > > > > > > Martin > > > > > > > > > > -------------------------------------------------------------------- > > Gabriele Hegerl > > Dept. of Earth and Ocean Sciences, Nicholas School of the Environment > > Duke University, Durham NC 27708 > > phone 919-684-6167, fax 919-684-5833 > > email: hegerl@duke.edu http://www.eos.duke.edu/Faculty/hegerl.html > > --------------------------------------------------------------------- > > > > > > > > -- Professor Keith Briffa, Climatic Research Unit University of East Anglia Norwich, NR4 7TJ, U.K. Phone: +44-1603-593909 Fax: +44-1603-507784 http://www.cru.uea.ac.uk/cru/people/briffa/