|This study is interested in the influence of tropical Pacific SSTs on|
subseasonal variability in North American SAT during the winter season. The
authors use SVD, regression/correlation and composite analyses to investigate
how ENSO affects subseasonal variability through modulation of subseasonal
eddies - specifically, via changes to the vertical structure of the eddies
which have bearing on the amount of baroclinic energy conversion that occurs.
The revised manuscript has gone a long way towards addressing a number of
concerns about the clarity, interpretation and statistical significance of the
study, and I very much appreciate the work the authors have put in. The subject
is interesting and relevant to improving our understanding of climate dynamics,
as improving near-term climate predictions, and better understanding the
large-scale conditions for extreme events. It would be good to see a few final
issues resolved before publication.
1. Thanks for the explanation of why this SVD technique was used over
traditional ENSO indices. I agree it is worthwhile exploring the "flavour" of
tropical Pacific variability associated with SSV of temperature over North
America. As I understand, it turns out that the SST pattern from the SVD is
ENSO-like, so it doesn't really give us any new information on "flavours" -
other than perhaps the discussion of the residual in section 3.6. If so, then
it seems okay to leave the title as is, but in other places, this result and
implications should be clarified (e.g., ENSO should be replaced by "tropical
Pacific variability" in places like the first line of the abstract and title of
section 3.2; the abstract and summary should convey that the important thing
for North American SSV turns out to be pretty much ENSO). Otherwise, the nice
addition/explanation on L105 is completely disconnected from the rest of the
2. The original review included a comment about the portion of extratropical SSV
related to ENSO. I noted that it seemed important to establish this up-front,
since later on in Fig. 10, you show an SSV signal unrelated to your ENSO index
(SVD1) that is both substantial in amplitude and very similar to the
ENSO-related signal. The addition of Fig. 1 showing the climatology is very
useful, and I see now about 10% of the local interannual variance is related to
ENSO, compared to 50% unrelated. Can you explain a bit more then why you
conclude that ENSO plays a prominent role in modulating SSV over North America
(L356)? Is the idea that the 10% we get from ENSO is at least predictable,
compared to the rest of the SSV that is just internal atmospheric variability?
It seems important to mention this in the abstract also.
3. The results would be more compelling if the statistical tests were a bit
more systematic, to allow the reader (and the interpretation/discussion) to
focus on the robust signals.
- some figures use a 95% significance level, others use a 90% level
- also, I believe in some cases where "confidence level" is used, it should be
- some composites are defined with +/- 1 values of SVD1, others with +/- 0.5
- the correlations in Fig. 6 are quite noisy (spatially) and use a rather
generous significance level - it seems like field significance should be tested
- no significance indicated on Fig. 9, which explores the mechanism
4. Section 3.5 is clearer than it was previously. However, I'm not sure the
reasoning hangs together with the results as they're presented - it seems some
of the arguments would need to be backed up by more rigorous analysis of the
temperature distributions and how much they overlap. For example, I don't think
we can conclude whether the winter-mean sets the frequency of extremes by
shifting the distributions (L322), or whether the presence of a few extreme
days determines the winter-mean. Perhaps it would be better to make the
discussion more general overall - in some regions it seems the change in SSV
broadens the distribution towards one side or the other (cold or warm), and in
other regions, while in other regions, we see what may be more a shift. And
then show some histograms to bolster the discussion?
- L173: "North America" might be more straightforward than "conterminous..."
- L173-174: lower-case "northwest-southeast-tilted"
- L143 and L179: slightly different filter details for high-pass
- L231: seems Fig. 6 is mentioned in the text before Fig. 5
- L239: How was 1500 chosen as the optimal number of resamples?
- L292: "Differences in the Z500..." - where do we see this?
- Captions should include more details so that the reader need not go back to
the text to look up information, abbreviations, etc.
- Fig. 3: colour bar for energy conversion should probably be adjusted
- Fig. 8: nice with the crosses, but they don't show up well in this colour