date: Fri, 25 Jul 2008 07:39:52 -0400 from: k.ziemelis@nature.com subject: NATURE: Decision on Nature manuscript 2008-05-04767A to: K.Briffa@uea.ac.uk Content-Disposition: inline Content-Length: 10058 Content-Transfer-Encoding: binary Content-Type: text/plain Dear Colleague Thank you for your help with the manuscript entitled "Non-uniform interhemispheric temperature trends over the past 550 years." by Dr Duncan and colleagues. We have now received all of the referees' reports, which I have attached below for your information. In the light of these various comments, we have declined publication of this study. Thank you again for your help and I hope that we can call upon your advice in the future. Yours sincerely Karl Ziemelis Physical Sciences Editor, Nature Nature's author and policy information sites are at www.nature.com/nature/submit/. Reviewers' comments: Referee #1(Remarks to the Author) This paper reports on an important new reconstruction of temperature from the Southern Hemisphere covering the last 550 years. This is an important achievement, and clearly merits publication in the scientific literature. The authors are to be commended for this. However, I am unable to recommend this for publication in Nature. The analyses presented with that new record, and the conclusions drawn from that, are speculative and interesting, but are not firmly rooted enough to merit publication in a journal intended for such an interdisciplinary readership. These results need to be refined and further tested before they might warrant publication in Nature. I offer the comments below in the spirit of constructive criticism. My feeling is that this manuscript should be published in a more specialized journal after taking into account the comments below. Specific comments 1. The new record described is meant to represent an all-New Zealand temperature record. However, it is by no means clear what relationship this has to larger regions of the Southern Hemisphere. Is this record indicative of larger scale SH temperature variations? There is really no evaluation along these lines, and without this perspective the paper is severely hampered. For example, see the statement in lines 3-6 from bottom of page 4. What is the basis for this? 2. The analysis moves directly to an examination of a time series constructed by subtracting the NZ record from the NH record. This seems to miss the step of a thorough examination of the NZ and NH records by themselves. How much of the character of the difference time series comes from the NH series, and how much from the NZ series? Is there a 30-60 year enhanced variance in the NH record by itself? In the NZ record by itself? Along these lines it would have been quite useful to see a coherence analysis. Do the peaks in the wavelet spectrum come mainly from the NH time series? If so, then does the NZ time series really add anything? If not, is there power in the NZ time series at these time scales? What is the source? 3. Is there a physical basis for adding the IPO and AMO signals? The spatial pattern of the AMO is clear, in that it is a monopole and would be expected to project coherently onto NH temperature. However, it is unclear what the projection of the IPO would be onto NH temperature, since it is a far more structured pattern with positives and negatives. This needs some explanation and a physical discussion. The fact that the correlation appears better when the two time series are added is not a sufficient justification without a physical basis. 4. p. 3, top The phrase "... efforts to incorporat this variation into climate models ..." is a bit awkward. Models are constructed from the known physical laws of the climate system, plus paramaterizations of unresolved processes. The variability produced by such models is then what is seen in extended simulations - there is no "a priori" attempt to include a certain mode of variability, such as ENSO. 5. p. 6, lines 8-9 from top This statement is opaque to me. Please elaborate. Why that time scale? 6. p. 6, lines 1-7 These statements are merely another way to describe the temporal behavior of the time series, but really do not add any physical insights. 7. It is unclear if the authors are proposing that there is a physical link between NH and NZ temperatures (such that when the NH warms, NZ cools by some physical mechanism). If so, what is that link? If not, is the behavior seen merely dominated by the NH series? Referee #2(Remarks to the Author) Non-uniform interhemispheric temperature trends over the past 550 years By R. Duncan et al The basic message proposed by the authors in this manuscript is that the Northern Hemisphere mean annual temperature varies, largely out of phase, with annual average Southern Hemisphere temperatures on a characteristic timescale of about 30-60 years. One might query whether this is saying much of interest, but the authors couch this result in the context of statements about the roles of natural versus anthropogenically driven inter-hemispheric temperature differences. If the paper was to provide insight into the specific role of tropospheric aerosols and the validity of the way they have been included in IPCC AR4 models to achieve greater consistency between observed temperatures and forced GCM model experiments then this would provide a strong case for publication. However, I do not feel that the manuscript does anything like this . Nor does the manuscript really confirm that an understanding of the workings of the IPO or AMO are critical for "understanding and modelling future climate changes" to any greater degree than was already understood on the basis of instrumental analyses or model experiments. The primary mode of instrumental temperature variability across both the North and South Hemispheres over the 20th century, comprises the 3-stage wide-scale warming illustrated in Figures 1 and 2 of Parker et al. (2007). Understanding of regional and decadal-scale differences in temperature changes over land and ocean and between North and South are relevant for framing "uncertainties" in attribution studies and projections of likely warming rates, their significance on decadal scales should not be overstated. For example, the variance explained by the AMO and IPO in large-scale analyses of instrumental temperature variability in the 20th century is on the order of a few per cent (e.g. Parker et al., 2007). This paper is not able to advance our understanding of the causes or even quantify the changing amplitude of the IPO or AMO or the net effect of their combined influence over time. Looking at the new data the authors present, it is not clear to me that the evidence for "out of phase" cycling between the Northern Hemisphere reconstructions and the New Zealand reconstructions is particularly strong or constant over the whole length of the series. Any (even white-noise) bandpass filtered data series , when superimposed, would show periods of in phase and out of phase behaviour. The series shown by the authors are out of phase sometimes and not at others. We can deduce little about the past behaviour of the IPO or AMO for either circumstance. The characterisation of the dominant multi-decadal patterns of variability of the so-called IPO varies in spatial signature and temporal importance depending on whether SST or NMAT data are used as the basis data set (e.g. see Figure 3 of Parker et al., 2007 - quoted by authors). Regardless of which are used the relationships between New Zealand temperature variations and temperatures in other parts of the Southern Hemisphere, associated with changing IPO status are complex (and possibly seasonally dependent). This suggests that using New Zealand as a surrogate for mean Southern Hemisphere temperatures on the decadal timescale, may be unfounded. The authors imply that their data are being used to imply details of "contrasting temperature trends between the northern and southern hemispheres" and that these data are the only data available to compare against the northern hemisphere average data. While it is true that instrumental data show the AMO and IPO to be negatively correlated with New Zealand temperatures, temperatures in other areas of the Southern Hemisphere are not. The authors do not present a formal phase analysis or, for that matter, compare their New Zealand bandpassed data with those from temperature series representing other regions of the Southern Hemisphere that do exist (in South America, Tasmania and even New Zealand i.e. as published by Cook and colleagues). The work by Cook has specifically pointed out a characteristic peak in the variance spectrum at around 30-60 years in Tasmanian temperature and he carried out an analysis of its changing significance through time using singular spectrum analysi s. A similar detailed formal comparison of the covariance between the various available Southern Hemisphere data (and the Northern Hemisphere data) seems justified. Personally, I consider the plots of the Wavelet spectra shown in Supplementary Figure 1 of little value in this regard. It is true that Figure 2 of the manuscript, at first sight, appears to show an intriguing, even compelling, correlation between the series representing the sum of the IPO and AMO and the Northern Hemisphere/New Zealand temperature difference (specifically Figure 2c). However, previous analyses of instrumental data show New Zealand temperatures to be negatively correlated with each of (and so also the sum of ) the IPO and AMO. As these modes describe contrasting temperatures in the North and South Hemispheres it is to be expected that the Northern Hemisphere minus New Zealand temperature ( or estimated temperature fitted against it) will correlate positively . In summary I just can not see that there is enough new information or new insight provided in this manuscript to support publication. * Please see NPG's author and referees' website (www.nature.com/authors) for information about and links to policies, services and author benefits. See also http://blogs.nature.com/nautilus, our blog for authors, and http://blogs.nature.com/peer-to-peer, our blog about peer-review. This email has been sent through the NPG Manuscript Tracking System NY-610A-NPG&MTS