date: Mon, 17 Feb 2003 11:03:16 -0800 (PST)
from: Eric Steig <steig@geophys.washington.edu>
subject: review of Holocne paper by Masson-Delmotte et al.
to: k.briffa@uea.ac.uk

Dear Keith,

Following is my review of Masson-Delmotte et al., "Common millennial-scale
variability ....".  My apologies for the delay.

I would prefer anononimity for this review.

I have reproduced the peer review form below.

Sincerely,

Eric Steig

---------------
Paper title: Common millennial scale variability of Antarctic and Southern
             Ocean temperatures during the past 5000 years reconstructed
             from EPICA Dome C ice core

Author(s): V. Masson-Delmotte, B. Stenni, and J. Jouzel

Checklist:

Title: Suitable.  I suggest inserting "the" before "EPICA".
Introduction: requires revision
Discussion: sufficient
Abstract: requires revision
Methods: better description and error bars required for spectral analysis
Conclusions: generally sound but spectral analysis overinterpreted
Scope: international interest/interdisciplinary interest/general
       significance is apparent
Length: appropriate
Language and style: English requires work of copy-editor;
       generally well organized
Referencing: some additional references required
Figures: not all are essential; some revision required

Recommendation:
(1)/(2)  Should be acceptable after modification and resubmission

Importance Ratin: Major Contribution

Detailed comments:

The authors have made significant contributions to the interpretation of
deuterium and deuterium excess from ice cores.  Especially in the last few
years they have demonstrated the utility of deuterium excess as a proxy
for sea surface temperatures.  The great advantage of their approach is
that both local (ice sheet)  temperatures and distant (sea surface)
temperatures are obtained from a single ice core record, largely
eliminating ambiguities about relative age.  This paper provides new
deuterium (dD) and deuterium excess (xs) data from the EPICA ice core at
Dome C, Antarctica.  An 2-D isotopic model is used to calculate
linear functions relating  dD and xs to site and source temperature,
allowing conversion of the isotope ratios to useful climate variables.
this approach has been used previously and shown to be very reasonble; it
is probably even more reasonable for the Holocene than the last glacial
period, since boundary conditions are changing less, especially on the
millennial timescales emphasized in the paper.

The paper is overall well organized, but there is too little description
of the deuterium and dueterium excess, which some readers will not be
familiar with.  I suggest adding a short paragraph, prior to the one that
begins "Here we focus..." which provides more reference to previous
theoretical work on deuterium excess.  Missing especially from the
references is the Kavenaugh and Cuffey paper from the Greenland IGS
meeting, and the Cuffey and Vimeux paper from Nature.  Also, reference is
made to the Stenni et al. paper from Science, but the reader has to guess
what was in that paper.  A brief description of that paper and its
conclusions would be appropriate.

The paper reaches three main conclusions.  First, that the early Holocene
optimum occurs early in Antarctica than at lower latitudes; second, that
site and source temperatures co-vary after about 5000 years ago (which is
tentatively attributed to an increase in ENSO-type variability); third,
that there is significant temperature variability on timescales of ~800
years at the Antarctic site, but not at lower latitudes.  Each of these
conclusions is important, if correct, because each provides insights into
how the climate system has evolved through the Holocene.  I find
particularly interesting the suggestion that the millennial-scale
variability in the Antarctic is probably of regional origin, since it does
not appear to occur at lower latitudes and is on a different timescale
than the often-discussed 1500-year cycles of the North Atlantic.

Overall, I think the conclusions are sound, but I am skeptical about
some aspects.

The interpretation of various "periodicities" in the data is overstated.
In fact, no confidence intervals are shown in the spectral analysis plots,
unless perhaps the dashed line shown is supposed to be the 95% confidence
(?).  The very different spectra obtained for the 0-3, 3-6, 6-9 and 9-12
ka intervals are interpreted as meaningful changes in the physical
processes involved (e.g. "some periodicities only appear in the last
thousdands of years").  A more conservative conclusion would be that none
of these "periodicities" are actually significant, relative to red noise.
Use of a more conservative spectral analysis routine, with a greater
number of degrees of freedom (MTM notoriously uses too few degrees of
freedom), would doubtless result in virtually no significant peaks being
detected.  I would advise against listing multiple "significant"
periodicities (e.g. on page 8, it says "220, 176, 150, 110....") and focus
instead on the much more interesting result that the deuterium is "redder"
than the deuterium excess, which implies (as the authors state) that the
millennial-scale power is of local rather than global origin.

Related to this, Vimeuex, Masson-Delmotte and others reported a 900-year
periodicity in deuterium excess from Taylor Dome.  Why does this show up
in xs (only) from Taylor Dome and dD (only) at Dome C?

The supposed connection with solar variability during the last 2000 years
is entirely unconvincing and I suggest deleting it.  No statistical work
is shown to suggest that this is significant.

Since the paper should be revised, I will not comment on specific
grammatical or typographical errors here.  Overall, the English is fine
but should be looked over by a good copy editor prior to publication.  Two
important things though:

1) "western" and "eastern Antarctica" is incorrect!  Use the terms "West
Antarctica" and "East Antarctica".  These are place names, not geographic
directions.

2) On page 10, the term "inversed" refers to the mathematical inversion,
but as written it implies that the graphs are upside down.  I would say
"calculated" instead of "inversed."

3) Figures: Figure 1 should show other ice core sites discussed in the
text (as well as well-known sites like Vostok).  Figure 5, 8a:
justification should be given for using the "reshaped" harmonic spectra;
otherwise (preferably) these should not be used, as they are appropriate
for electronic signal processing and are of dubious use in climate
research where narrow band signals would be a major discovery!  On both
figures confidennce intervals should be shown and explained.







