date: Tue, 3 Jun 2008 05:03:46 -0400 from: a.newton@nature.com subject: Nature Geoscience: Receipt of review for NGS-2008-05-00486 to: p.jones@uea.ac.uk MIME-Version: 1.0 Content-Transfer-Encoding: binary Content-Type: multipart/alternative; boundary="_----------=_12124838262352012" X-Mailer: MIME::Lite 3.021 (F2.74; T1.23; A2.02; B3.07; Q3.07) Date: Tue, 3 Jun 2008 05:03:46 -0400 Message-Id: <52121248382681@rhwww3.nature.com.nature.com> Dear Professor Jones This email is to acknowledge receipt of your review for the manuscript by Dr D'Arrigo and co-authors, entitled "Impact of volcanic forcing on tropical temperatures during the last four centuries". Thank you for your help in this matter. A copy of your review is attached below for your reference. Yours sincerely Hollie Cayzer Editorial Assistant Nature Geoscience For Dr Alicia Newton Your comments: Remarks to the Editor: As you'll be able to gather I think that such an analysis is speculative, as volcanic events are infrequent so the sample count is small. I've made some suggestions that the authors should consider. It will be difficult to do what I suggest and make it convincing, but I think there needs to more and extracting the ENSO signal is one way to highlight what's left. It certainly helped with the paper in Nature in last week's issue. Remarks to the Author: I read this paper fairly quickly shortly after receiving it and my initial thought was that it was a bit speculative. The problem was that volcanic events are fairly infrequent, and much of the 3-5 year timescale variability in the tropics is likely to be due to ENSO-related forcing/variability. About a week later I read the paper in the May 29 issue of Nature by Thompson et al. This had shown a neat way of removing the higher-frequency variability related to ENSO to help show the volcanic-related cooling events more clearly. This is exactly what this paper needs to do. The Thompson paper was only looking at the observational record and had monthly timescales to work with. What needs to be done here is to extract the ENSO (principally the La Nina events) signal from the series in Figure 1B. Thompson's approach using Nino-based SST is a possibility, but another alternative is a smoothed (say about 3yr timescale) version of the Southern Oscillation Index (SOI). The method could be tested on the observational SST series then applied to the reconstructed SST. The SOI extends back to the mid-1860s, but there are a number of tree-ring based reconstructions of the SOI that could then take ENSO out of what is essentially a coral-based SST reconstruction. In its present form, I don't think the paper gives that much more than speculation about the impact of past volcanic events in the tropics. To concentrate on the corals in the tropical SST reconstruction I would omit the tree-ring series from Nepal and the Quelccaya Ice Cap. These latter two can be used to assess reconstructions, but throwing them all in together doesn't seem to add that much and loses potential additional verification of reconstruction models. From Supplementary Table S2, the quality of the reconstruction is highly variable in quality, only seemingly useful back to about the early 1800s. I'd concentrate on this and the instrumental period, as the 18th century is pretty much devoid of major tropical volcanic eruptions. Unless you can separate out the major cooling events in the tropical SST series that cannot be related to La Nina, this contribution is premature. The volcanic influence on higher latitude trees (MXD) is clear from earlier work. Most of the extreme cooling events are related to tropical explosive volcanic events. These same events should be in your series, but they are likely masked by strong ENSO variability - which is negligible in the MXD series. My only other major comment is that I would only look at the 5 years following each volcanic eruption - and not extend the results out to 16 years. Even 5 years is a stretch from climate modelling of volcanic eruptions and observational evidence. None of the results are significant beyond 5 years. My one minor comment is that I would have used HadSST2 (Rayner et al., 2006) rather than the spatially infilled HADISST. It would be useful to compare the time series for these two datasets. In summary, I don't think there is enough new in this paper to warrant publication in its present form. I have made some suggestions as to how the exercise can be made more convincing. I would need convincing about much of a volcanic impact on tropical temperatures, because there isn't that much after the major events of the 20th century. This email has been sent through the NPG Manuscript Tracking System NY-610A-NPG&MTS