This paper uses data on water vapour from two satellites to examine the connection between variability of water vapour in the upper troposphere and lower stratosphere (UTLS) and annular mode variations (AO or AAO) according to hemisphere. It is argued that in the upper troposphere a large fraction of of the non-seasonal variation on monthly time scales is related to the annular modes, but this fraction is smaller (or rather less statistically significant) in the lower stratosphere. Clearly extraction of robust water vapour variability signals from this type of satellite data is a significant task, and the authors have taken significant care over this. On the other hand I found some details of their analysis mysterious and complicated (e.g. sometimes using medians and sometimes means) and as a non-specialist in analysis of this sort of data analysis I worry whether these complications make the extract signal more or less certain.
The proposed connection between the annular modes and observed UTLS water vapour concentrations is interesting (the authors make the simple practical point that taking account of this connection might potentially improve trend estimates) and it is of course natural to consider possible mechanisms, of which the authors identify three — (i) variations in temperature changing saturation mixing ratios and therefore water vapour concentrations, (ii) effects of meridional transport and (iii) effects of tropopause variation.
The authors seem to rule out (iii), without providing any explicit evidence to support that. (They report that there is no significant correlation between high-latitude tropopause height or pressure and the annular modes — but do not provide any evidence or cite other any other publications on this topic.) They show some evidence for (ii) — though it is a little difficult to assess this because all quantitative information in the paper comes in the form of fractional change. To what extent is the magnitude of water vapour variations accounted for by the magnitude of variations in saturation mixing ratio? They argue for some role for (i), but this is based on the view (in my view incorrect) that Eulerian mean meridional circulation is an indication of meridional water vapour transport.
In summary, this paper identifies a scientifically interesting connection, but the assessment and discussion of possible mechanisms is weak (or even flawed) at present. In my view this assessment and discussion needs to be clarified and strengthened, and in addition several other parts of the text need to be clarified (see detailed comments below), if the paper is to be suitable for publication.
Detailed comments follow:
p3 l10: ‘In the middle stratosphere’ — I wasn’t sure why you qualified by ‘middle’. It would be more logical to emphasise the ‘stratospheric overworld’ — i.e. the part of the stratosphere that does not connect to the troposphere along isentropic surfaces (see e.g. Holton et al 1995) since it is this part of the stratosphere where water vapour concentrations will be controlled by entry at the tropical cold point and by methane oxidation. In the ‘lowermost stratosphere’ on the other hand, as you note, there is the possibility of rapid exchange with the troposphere along isentropic surface which means that water vapour concentrations are not set by the tropical cold point (e.g. see Dessler et al 1995 JGR).
p3 l23-30: It would be helpful to refer to the specific figure (actually Figure 7) in Thompson and Wallace (2000) that shows this signature in the mean meridional flow. But note that what is shown in that Figure is the Eulerian mean meridional circulation — and this is certainly not the whole story regarding the transport of trace species — see e.g. Plumb and Mahlman (1986 J. Atmos. Sci.). So if you want to associate a particular phase of the annular mode with transport of a trace species (in this part of the paper you are ignoring condensation/evaporation effects) then you need to justify carefully why the Eulerian mean circulation should be relevant.
p6 l24: Why do you use median rather than mean temperatures? Also if you are trying to calculate a monthly mean tropopause height then wouldn’t it make more sense to calculate tropopause heights on the basis of, say, daily data and then average those calculated heights?
p6 l27-29: Please clarify the definitions of the ‘thermal tropopause’ and the ‘lapse-rate tropopause’ — are they different? Also you have different tropopause definitions for NH and SH — using median temperatures for the former and maximum temperatures for the latter. If this is important — and I suppose that it is — then more explanation is needed. You say ‘the lapse rate tropopause concept has been used previously’ — I would have assumed that the lapse rate tropopause concept is the standard one (i.e. the ‘WMO definition’) used by almost all meteorologists — which makes me think that if you need to justify use of this it must actually be non-standard and therefore requires more explanation.
p7 l7-10: Again to me this use of medians for some purposes and means for others seems mysterious. If medians don’t work for the SH then why not use means for both NH and SH?
p7 l13: You say ‘Because ACE instruments sample southern high latitudes in only eight of twelve calendar months’ — but then you seem to imply that this is true in the NH as well (without saying it directly).
p7 l29 + Figure 1 caption: Why use ‘bias’ when this often has a specific technical meaning? For example in a subsequent sentence you say ACE-FTS has a high bias of 10% — which in this case (Hegglin et al 2013) means difference from a multi-instrument mean. It would be clearer to simply say ‘relative difference between MAESTRO and ACE-FTS’ when that is what you mean, both in the text and in the caption and indeed in the figure annotation.
p8 l4: ‘Accounting for an upper troposphere +10% wet bias in ACE-FTS, MAESTRO and ACE-FTS agree …’ — I found this confusing — do you mean ‘If we accept that ACE-FTS has a positive bias then the two agree’? Wouldn’t it be clearer to say that the two agree (by some criterion) apart from at levels where ACE-FTS has an acknowledged positive bias.
p8 l7: The ordering of the Figures seems odd — you mention Figure 4 and Figure 3 here, but Figure 2 has not yet been mentioned. In any case subtracting information in Figure 4 from that in Figure 3 would not make any sense because Figure 3 is for NH and Figure 4 is SH climatology. Personally I would find it clearer if the missing months (4 per year) were explicitly displayed in Figures 2 and 3 as ‘empty’ bands rather than somehow stretching 8 months to cover a year.
p9 l2: Since you have previously given Larson et al (2005) as a reference for the AO I assume that their definition is the same as that of Thompson and Wallace (2000)?
p9 l3-7: To me this (e.g. including annular mode index for trend uncertainty if it improves uncertainty without increasing bias) all sounds a bit complicated. I’m not convinced I (or any reader could repeat this calculation from the details given here).
p10 l3: ‘The southern high-latitude time series has slightly less water in the UTLS in late winter than at northern high-latitudes … due to the colder air temperatures.’ — this is pretty difficult to make out from your figures — not least because months are not shown (and which months are omitted is not clear from the Figures —- as noted previously).
p10 l10-15: There seems to be a strong implication here that upper tropospheric water vapour is explained by LOCAL temperatures — i.e. little role for effects of remote temperature variation being communicated by transport. Is that what you intend? Have other authors commented on this and indeed have there been previous modelling studies investigating this point?
p10 l18-19: You refer first to ‘weak seasonal variations in the lower stratosphere’ and then to ’The large seasonal cycle amplitude … in the lower stratosphere’ — I’m confused.
p11 l1-2: ‘the seasonal variation in water vapour concentration’ would be clearer.
p11 l11-13: ‘The use of medians is preferable … where the water vapour mixing ratios are not normally distributed’ — why should normally distributed or not determine whether medians are preferable?
p12 l12: This discussion of the radiative impact of AAO-related water vapour variations would seem better to me in the final Discussion section of the paper rather than here in the middle of the description of the variations themselves.
p15 l13-14: As noted previously, referring to the signature identified by Thompson and Wallace (2000) in the Eulerian mean meridional circulation is not a convincing argument for there to be a corresponding signature in meridional transport of water vapour.
p15 l24: ‘this response’ — do you mean ‘the response in the saturation VMR’?
p16 l6: ‘this isolated region’ — which isolated region?
p16 l18-21: You assert that there is no strong correlation between high-latitude tropopause height or pressure and the annular modes but present no explicit evidence, nor do you cite any other papers on this topic. A bit more concrete supporting evidence is needed. |