|I have been asked to step in the editorial process of this article after the authors had provided the second version of their manuscript. |
The question which seems to be the main source of disagreement between the authors and the reviewers is the following: Can we use modern extreme events as a guide for the description of mechanisms at work in long-lasting events such as megadroughts? We do not know, and there are reasons for it not be the case. For example, at centennial scales, slow ocean dynamics may generate dynamical patterns with no analog at the decadal or interannual scale.
This is of course a difficult question, and perhaps the article has been a missed opportunity to further document the analogue hypothesis before applying it. But it is a tricky issue and at this stage it would be unreasonable (and counter-productive) to request new analyses. Therefore, let me mainly comment on two points raised by reviewer #1.
Should the authors have used CMIP experiments? The "PMIP" section of CMIP does indeed include a few millennium runs which might contain some long droughts. At least, I expect that modern GCMs should generate some patterns of decadal variability and perhaps decade-long droughts. There would thus be some scope for testing, in the GCM universes, the similarity of mechanisms determining droughts at the interannual and decadal scales, or perhaps even centennial scale. Of course this would only be helpful if indeed GCMs are good simulators of the dynamics of megadroughts, which no-one can guarantee because they have not been calibrated for this purpose. Hence, inspecting megadroughts in GCMs (if there are, any) is certainly a very useful task, but I concede that it would involve some massive work falling out of the scope of the present article.
The authors have somehow answered the concerns of reviewer #1 bearing on the identification of patterns associated with dry years in the modern climate record. They should however still be more explicit about the number of degrees of freedom used in the Student t-test. Statistics of composite records generated from an autocorrelated time-series can be treacherous and it would be helpful to have a bit more details here. Short of this the reader cannot be sure that the approach is valid.
One the other hand, the justification and language around the "modern analogue technique" still requires more work. The paragraph added after the first review did not quite do the job. Forget about uniformitarianism. This is a historical notion, originally opposed to catastrophism, that was meant to express the assumption that Earth's processes have remained the same through time. In the modern parlance the concept of "uniformitarianism" has been diluted into the broader and vaguer principle that geological, palaeontological and palaeoclimatological observations are helpful to understand present and future Earth dynamics.
What is needed to motivate and justify further work is simple (plain) language justifying the objective of the approach, its limitations, and what should be done to consolidate the working hypotheses.
- the column "Purpose of Climate Variable" in Table 1 is appallingly trivial.
- There is also a mistake in the numbering of sections and subsections starting at subsection 4.1.
- p. 12 l. 33: The citation " Using a complex numerical weather-predication model with data from May 1987 to May 1988, Palmer and Brankovic (1989) had significant skill in predicating an anomalous high pressure ridge over North America during the summer of 1988" is about a trained prediction model, and is irrelevant in the present context.
- Citation "By applying the modern climate analogue technique to paleoenvironnmental proxies from the mid-continent, Shinker et al. (2006) found that regional moisture influx and small-scale vertical motions in the atmosphere (i.e. subsidence or uplift) provide better information regarding precipitation than large-scale general circulation alone."
The wording is inaccurate. Shinker examined modern climatological data, to indeed observe that regional moisture influx and small-scale vertical motions in the atmosphere are a better predictor of precipitation than large-scale circulation in the modern climate. They then used this insight to help and contribute to the interpretation of palaeoclimate data. It is only at the latter stage that the hypothesis of an analogy applies.