cc: "k.briffa@uea.ac.uk" <k.briffa@uea.ac.uk>
date: Wed, 3 Apr 2002 07:01:58 -0500
from: "drdendro@ldeo.columbia.edu" <drdendro@ldeo.columbia.edu>
subject: RE: Fwd: RE: Recent NH reconstruction
to: "t.osborn@uea.ac.uk" <t.osborn@uea.ac.uk>

Hi Tim,

Thanks for the data. It is good to have it all in one file like that. I honestly 
hadn't read your "breezy" article (that Mike hates so much) closely 
enough to realize the extent to which you recalibrated everything. That is 
fine, of course, to make all comparisons comparable. What surprised 
me the most was that you were able to use Mann et al. data from only 
land areas north of 20N. Mike's big argument has been that his series 
would gain some amplitude of multi-centennial low-frequency 
variability is the tropics were not included. Yet your results indicate that 
that is not the case, at least by much. I go down to Mike's place to give a 
seminar tomorrow. I am sure that he will fill my ears with more of his 
complaints.

Cheers,

Ed

Original Message:
-----------------
From: Tim Osborn t.osborn@uea.ac.uk
Date: Wed, 03 Apr 2002 09:19:22 +0100
To: shaopeng@umich.edu, k.briffa@uea, 
drdendro@ldeo.columbia.edu
Subject: Fwd: RE: Recent NH reconstruction



>>From: "Shaopeng Huang" <shaopeng@umich.edu>
>>To: <k.briffa@uea.ac.uk>
>>Cc: <drdendro@ldeo.columbia.edu>
>>Subject: RE: Recent NH reconstruction
>>Date: Thu, 28 Mar 2002 08:09:52 -0800
>>
>>Hi Keith,
>>
>>I guess both your data sets and the recent Esper et al reconstruction 
are
>>not available for distribution at this stage. I would appreciate if you
>>would let me know when they are available.

Shaopeng,

in fact they *are* available and Keith forwarded your e-mail to me - but I 
was too busy last week to reply.  A data file is attached that contains 
most of the time series that appear in our figure.  The file is ASCII and 
should be self explanatory (I hope!).

>>I have a question regarding your discussion on the borehole 
reconstruction.
>>You showed one simple (unweighted) average curve and one 
gridded average
>>curve. What is the size of a grid in your gridded averaging, 5x5 or 
1x1? I
>>agree that some weighting scheme might be helpful to reduce bias. 
But in an
>>earlier testing, we found that the difference between weighted and
>>unweighted averages within the range of uncertainty.

We tried several grid definitions and found some sensitivity to the 
choice.  The curve we plotted was based on a 10x10 grid.  We are 
planning 
to write up something short about the sensitivity of the NH estimate to 
this and other choices, but haven't decided where to include it or in 
what 
form yet.  Keith and I have a comment in press with GRL (.pdf file is 
attached) that you might find of interest (it was accepted in September 
and 
it seems that no progress has been made since - I'm really annoyed 
with GRL).

Best regards

Tim


--------------------------------------------------------------------
mail2web - Check your email from the web at
http://mail2web.com/ .

