cc: k.briffa@uea, p.jones@uea
date: Tue Oct  5 16:18:29 1999
from: Tim Osborn <t.osborn@uea.ac.uk>
subject: Briffa et al. series for IPCC figure
to: mann@virginia.edu, imacadam@meto.gov.uk

Dear Mike and Ian

Keith has asked me to send you a timeseries for the IPCC multi-proxy reconstruction figure, to replace the one you currently have.  The data are attached to this e-mail.  They go from 1402 to 1995, although we usually stop the series in 1960 because of the recent non-temperature signal that is superimposed on the tree-ring data that we use.  I haven't put a 40-yr smoothing through them - I thought it best if you were to do this to ensure the same filter was used for all curves.

The raw data are the same as used in Briffa et al. (1998), the Nature paper that I think you have the reference for already.  They are analysed in a different way, to retain the low-frequency variations.  In this sense, it is one-step removed from Briffa et al. (1998).  It is not two-steps removed from Briffa et al. (1998), since the new series is simply a *replacement* for the one that you have been using, rather than being one-step further.

A new manuscript is in preparation describing this alternative analysis method, the calibration of the resulting series, and their comparison with other reconstructions.  We are consdering submitting this manuscript to J. Geophys. Res. when it is ready, but for now it is best cited as:
Briffa KR, Osborn TJ, Schweingruber FH, Harris IC and Jones PD (1999) Extracting low-frequency temperature variations from a northern tree-ring density network. In preparation.
Keith will be sending you a copy of the manuscript when it is nearer to completion.

I have also attached a PS file showing the original Briffa et al. (1998) curve, with annotation of cold years associated with known volcanic eruptions.  Overlain on this, you will see a green curve.  This is the new series with a 40-yr filter through it.  This is just so that you can see what it should look like (***ignore the temperature scale on this figure***, since the baseline is non-standard).

With regard to the baseline, the data I've sent are calibrated over the period 1881-1960 against the instrumental Apr-Sep tempratures averaged over all land grid boxes with observed data that are north of 20N.  As such, the mean of our reconstruction over 1881-1960 matches the mean of the observed target series over the same period.  Since the observed series consists of degrees C anomalies wrt to 1961-90, we say that the reconstructed series also represents degrees C anomalies wrt to 1961-90.  One could, of course, shift the mean of our reconstruction so that it matched the observed series over a different period - say 1931-60 - but I don't see that this improves things.  Indeed, if the non-temperature signal that causes the decline in tree-ring density begins before 1960, then a short 1931-60 period might yield a more biased result than using a longer 1881-1960 period.

If you have any queries regarding this replacement data, then please e-mail me and/or Keith.

Best regards

Tim