Some of you might be wondering if there is any following to Reviewer Outs Himself. Jason Box revealed he was the reviewer of Frauenfeld, O.W., P.C. Knappenberger, and P.J. Michaels’s “A reconstruction of annual Greenland ice melt extent, 1785-2009” just published in Journal of Geophysical Research-Atmospheres and criticized the paper, its authors and if I understand correctly the associate editor whose identity he reveals. In the process Box posted the full review he sent to the editor of JGR.

On May 2, Michael’s posted the response to Box’s review, giving his reasons as follows:

Since Dr. Box has posted his verbatim review of our paper that he submitted to JGR, we will go ahead and post the verbatim response to his review that we submitted to JGR (a submission which also included our response to another reviewer, and a modified version of our paper).

As best as we can tell, Dr. Box chose not to evaluate our response to his review or even read the final version of our published paper before posting his original blog critique.

Reviewer #2 (Comments to Author):
I rank the paper: “Good” because the paper’s methods seem solid. Yet, depth with regard to examining causal factors is missing. Further, the paper’s main point, as it seems, that recent warming is not without precedent, may already be obsolete because 2010 was such an extreme melt year AND that more warming in Greenland is likely simply for Greenland to be in sync with the northern hemisphere. The paper thus, in the very least, requires a revision that includes consideration of 2010 data. Yet, consideration of causal factors of cooling and warming and treatment of the Box et al. (2009) prediction, which for 2008-2010 has been accurate, would give the paper the depth consistent with JGR’s standard.
We thank the reviewer for the relevant and helpful comments and suggestions which have greatly clarified and improved out manuscript! Our revisions/responses are described below.

Major Critique:

As is, the only depth of the paper is the statistical modeling, that is, the regressions to reconstruct melt area and comparison of the recent warming versus past warm episodes. There is theory to explain warming and cooling episodes in Greenland. Yet, the paper does include this important dimension. Therefore, to increase the depth or impact of the work, the paper should elaborate causal factors that explain the ups and downs in the reconstruction.

Response: In our revision, we have included in our Background section an acknowledgement that there has been much work done to identify the causal mechanisms of multi-scale climate variability in and around Greenland and we included a listing of some of the potential underlying causes, citing the recent work of Box et al. [2009] for a more thorough description. (see lines 53–56).
The primary objective of our work is to develop a proxy record of an index of total surface ice melt across Greenland such that recent direct observations could be better placed in historical perspective. As such, we, ourselves, did not independently seek to determine the specific causal mechanisms that may drive the variability in summer temperature or winter NAO which we show to be related to Greenland ice melt. As pointed out by the reviewer, there has already been a good deal of other work which has examined the potential underlying mechanisms. Relying on those findings, we have included a discussion of evidence of these influences on our findings (see the beginning of our Discussion section (lines 218–224).

The paper may already be obsolete without considering the extreme melting in 2010. I would therefore not recommend accepting the paper without a revision that included 2010. The numerous statements throughout the paper, like that in line: 19 “We find that the recent period of high melt extent is similar in magnitude but, thus far, shorter in duration, than a period of high melt lasting from the early 1920s through the early 1960s.”

Response: While we sympathize with the reviewer and would love to have our paper as updated as possible, practical issues get in the way.

We don’t collect any data ourselves, but instead, obtain the processed melt data from various research groups, each operating on their own time schedule. The final/official release of the SSM/I brightness temperatures also takes about 6 months before it is available to centers like NSIDC. The updates in past years were therefore usually not made available to us until 6 to 9 months after the end of the melt season. In fact, the timing (August) of our original submission coincided with us finally obtaining the 2009 updates. Thus, waiting for the 2010 melt data would push the submission of our revised paper back until late spring or summer of 2011, at which point we may find ourselves again experiencing an interesting melt season which reviewers might feel important to include.

We would like to note that waiting for one more year of data is not going to materially affect our analysis or conclusions. While the melt across Greenland has been elevated for the past 10 to 15 years (and continues in 2010), this period of time is still only about half as long in duration as the elevated (reconstructed) melt across Greenland from the 1920s through the early 1960s. So the addition of one more year of melt data (i.e., 2010), will not impact this comparison.

Considering the above, we hope the reviewer will understand that it is only practical for us to incorporate melt data through the 2009 melt season. Recognizing that 2010 may be exceptional, we have added discussion of the preliminary indications for this most recent melt season (see lines 243–244).

One thing different about the recent warming versus the 1920s warming is that Greenland climate continues to lag the northern hemisphere pattern… The work should therefore reflect on the prediction made in Box et al. (2009) that: simply to be in sync with the northern hemisphere pattern, Greenland climate must warm (after year 2007) by 1.0 – 1.5 C. In the years after 2007, that is, 2008-2010, this prediction has held true. And that still more warming should happen in Greenland in the coming few years is more likely than not. A major volcanic eruption, of course, see relevant literature, would cool Greenland’s climate for 1-3 years.

Response: This is indeed a very interesting hypothesis and we have added this possibility to our manuscript (lines 247–248).

The pre-1840 results should be abandoned because is cannot or at least it has not been demonstrated that there sufficient sampling to compare with the subsequent complete series.

Response: We include a reconstruction back to 1784 because in our opinion the statistics do permit such a reconstruction: we include the models that we use (Table 1) and we modify the error bars accordingly (Figure 2). While the pre-1840 model is not as strong as the post-1840 models, we feel that the information that it provides (including the uncertainty intervals) is scientifically valid and provides interesting and useful information to the scientific community. We therefore prefer to retain our pre-1840 results, but make a modification in the text to better draw attention to the fact that the uncertainty bands are wider and the model not as strong (see lines 213–215).

Title: A less ambiguous time frame should be included in the title than: “A Reconstruction of Annual Greenland Ice Melt Extent Going Back to 1784″ is needed…Something like: “A Reconstruction of Annual Greenland Ice Melt Extent 1784-2009″. Why? If the paper is published, some years down the line, the title would become ambiguous.

Response: Good point! We have changed the title according to the reviewer’s suggestion and made a slight modification to the abstract (line 17).

Minor Critique:
line 12 “three decades” instead of “several decades”

Response: We made the suggested change.

line 52: the following statement seems not accurate: “Such a comprehensive, annually resolved reconstruction has not previously been undertaken, and will better place current observations of melt extent in a longer-term historical perspective.” Box et al. (2009) modelled an annually resolved temperature reconstruction for the Greenland ice sheet.

Response: The Reviewer is correct in that Box et al. (2009) (and others as well) have modeled an annually resolved temperature reconstruction, but we were referring to our annually resolved ice melt reconstruction (not temperatures, specifically). We made this more explicit in the text (line 60).

line 103 define “closely match” quantitatively.

Response: This is something we have been attempting to do for quite a while, however our multiple requests to obtain the data from Steffen and Huff have so far remained unanswered. We have therefore clarified in the text (lines 120–121) that this is based on a graphical/visual comparison.

line 124 define “quite similar” quantitatively.

Response: We removed the sentence about other combination methods. For what it is worth, a PCA analysis of the three raw (unstandardized) ice-melt datasets produces a single significant PC that is correlated very highly (R = 0.997) with the standardized average.

line 140: Does this relationship account for sub-monthly melt frequency? “Our Greenland melt reconstruction therefore focuses on the relationship between monthly average temperatures” I suspect a reduced sensitivity to melt intensity for 2 reasons: 1.) summer variability is minimal; 2.) a summer average of e.g. 0 C still includes periods above melting.

Response: We did not investigate relationships between temperature and ice melt at a sub-monthly time scale. The melt data that we obtained (in 2 of the 3 datasets) did not allow for investigation at such a time scale. In our paper, we only used a single seasonal value for each year. We clarified this in the revised text (line 157).

line 162: explain “the direct measure of JJA temperature subsumes the summer NAO influence.”

Response: We meant that summer NAO primarily has a direct influence on the Greenland summer temperatures and that therefore, the direct observations of Greenland summer temperature already includes the impact of NAO variations. We added a brief bit of explanatory text (see lines 181–182).

line 166 “winter conditions act to pre-condition summer ice melt through a snow/albedo response” certainly because of thermal erosion of heat content. “snow/albedo response” is vague and does not mention important heat content issue.

Response: In the preceding paragraph (lines 164–172) in the text, we described the snow/albedo response a bit more thoroughly as “Hanna et al. [2008] suggest that enhanced winter snowfall results in increased summer albedo which decreases absorbed incoming radiation and reduces the amount of energy available for ice melt.”

We think this is a clear description, although, admittedly, we are unsure as to what the reviewer means by the “thermal erosion of heat content” so we don’t know whether that is covered in our description or not!

line 195: suggest “strong warming trend” instead of “strong positive trend”

Response: We made the suggested change.
line 195: “~1979-2009″ instead of “The last ~30 years”

Response: We made the suggested change.

line 211: By the same token as the arguments that the recent warming is not statistically unprecedented, the following statement need be substantiated using probabilities: “several sustained periods can be identified when a greater and/or more prolonged”

Response: In response to Reviewer 3, we have largely removed our descriptions of the combined probability that previous years/periods may have exceeded the melt extend in 2007. So perhaps this mollified Reviewer 2’s concerns about our claims in this section.

What we simply mean here is that visual inspection of the moving average in Figure 2 indicates a few periods: centered around 1935 (i.e. “the 1920s and 1930s”) and centered around 1950 (i.e. “the years around 1950) that have higher positive anomalies that stayed high longer than the current warming has.

line 221-223: a good point: “It is worth noting that the satellite observations of Greenland’s total ice melt, which begin in the late 1970s, start during a time that is characterized by the lowest sustained extent of melt during the past century (Figure 2).”

Response: Thanks.
line 248: remove “much”, overstatement
Response: We made the suggested change.

I added bold to the claim that I think Box disputes when claims “Their selective ‘findings’ were obsolete at the time the paper was submitted for publication. ” Even if Box doesn’t like the paper, or would prefer that one examine some a different metric than that chosen by FKM to diagnose the melt rate of Greenland Ice, what KFM claim seems plausible to me. That is: the duration of the recent melt period has not yet exceeded the duration of previous ice melt periods and one more year isn’t going to change conclusions of their paper using their metric.

My understanding based on comments is that Box intends to submit a response

Worry not. We’re preparing an extensive rebuttal to the Frauenfeld et al. publication, with attention to communicating science to the public. We also aim to address ALL questions posed in comments to this blog.

Note: It seems unlikely that Box’s rebuttal will address ALL questions posed in comments to his blog. For example, I would be very surprised if his rebuttal answers the question posed by Les Johnson, “Have you asked Santer the same question, about why all available data was not used in his Santer et al paper (2008)?”

In the meantime, we can all wait to see when Box’s rebuttal appears in JGR and which elements of his blog post are approved of by peer reviewers and the editor. I do hope reviewers will not permit the rebuttal to continue to make the inaccurate claim that KFM refused to use all data available at the time the paper was published. The melt data for 2010 were not available at the time of publication; the remain unavailable. By summer’s end they will be available. But this is hardly Santer-like behavior which involved leaving off years worth of widely available data and then reporting a “failed to reject” result.