An upper-branch Brewer–Dobson circulation index for attribution of stratospheric variability and improved ozone and temperature trend analysis

Ball, William T.; Kuchař, Aleš; Rozanov, Eugene V.; Staehelin, Johannes; Tummon, Fiona; Smith, Anne K.; Sukhodolov, Timofei; Stenke, Andrea; Revell, Laura; Coulon, Ancelin; Schmutz, Werner; Peter, Thomas

doi:https://doi.org/10.5194/acp-16-15485-2016

Articles | Volume 16, issue 24

Atmos. Chem. Phys., 16, 15485–15500, 2016

https://doi.org/10.5194/acp-16-15485-2016

© Author(s) 2016. This work is distributed under
the Creative Commons Attribution 3.0 License.

Atmos. Chem. Phys., 16, 15485–15500, 2016

https://doi.org/10.5194/acp-16-15485-2016

© Author(s) 2016. This work is distributed under
the Creative Commons Attribution 3.0 License.

Articles | Volume 16, issue 24

Research article 15 Dec 2016

Research article | 15 Dec 2016

An upper-branch Brewer–Dobson circulation index for attribution of stratospheric variability and improved ozone and temperature trend analysis

William T. Ball et al.

Download

Final revised paper (published on 15 Dec 2016)
Preprint (discussion started on 08 Jul 2016)

Interactive discussion

Status: closed

AC: Author comment | RC: Referee comment | SC: Short comment | EC: Editor comment

- Printer-friendly version

- Supplement

RC1: 'Review of Ball et al.', Alexey Karpechko, 01 Aug 2016
- AC1: 'Response to Referee #1: Alexey Karpechko', William Ball, 19 Oct 2016
RC2: 'Referee report', Lon Hood, 02 Aug 2016
- AC2: 'Response to Referee #2: Lon Hood', William Ball, 19 Oct 2016

Peer-review completion

AR: Author's response | RR: Referee report | ED: Editor decision

AR by William Ball on behalf of the Authors (20 Oct 2016) Author's response Manuscript

ED: Referee Nomination & Report Request started (20 Oct 2016) by Bernd Funke

RR by Alexey Karpechko (31 Oct 2016)

Suggestions for revision or reasons for rejection

The manuscript has improved compared to the previous version. I think the authors have made a good job. There are only few minor comments which I have on the revised version:
1. I believe the fact that your index does not correlate with the heat flux at 100 hPa (which represent the amount of wave activity entering the stratosphere in extratropics) is important and worth highlighting in conclusions and perhaps also in the abstract. The upper branch of BD is indeed driven by planetary waves and one would expect the 100hPa heat flux being an important proxy for it. But it is likely wave propagation and dissipation within the stratosphere that give raise the variability represented by your index, and this is not captured by the 100hPa heat flux.
2. P8L4 : ‘It seems surprising that these dynamical signals survive
other processes, such as chemistry and radiative effects, on a monthly timescale and we suggest 5 this warrants further
investigation, though this is beyond the scope of this paper.’
A dynamical event lasting just for a few days (or a sequence of such events) may influence the monthly mean statistics (temperatures, ozone, etc). So there is no need for an event to last for a month. Perhaps this is a sufficient explanation?
3. Figure 5 and Section 3.2:
How do you calculate TEMS and EPFD? References to the formulas would be helpful. Also do you use daily mean data or hourly (e.g. 6-hourly) data? Daily mean data may be not sufficient to represent forcing due to some tropical waves. This may be responsible for patchy patterns in the tropics, such as the negative anomalies surrounded by positive anomalies in Fig. 5a. Please give more details about your calculations
4. Section 4:
I realize that I do not understand how you construct the index, specifically how you merge the hemispheres. Is it so that June-October values come from SH and November-May from NH? Then why there are several different periods mentioned in Section 4.1?
5. Figure 6:
I believe positive values of EP-flux divergence indicate decreased wave activity, which is consistent with weakened meridional circulation and high temperatures in the tropics (high-T composite).
6. I think 1-sigma confidence interval is more commonly used wording than 68% confidence interval.
7. P17L5: ‘Nevertheless, use of AR0 or AR1, only appears to influence the uncertainties 5 on the trend
estimates in a clear way, and only the mean values to a small degree (see Fig. 13).’
I do not understand this sentence. In any case I think it is enough to consider AR1 results and just skip the AR0 results.
8. Figure 13 shows that the inclusion of the UBDC index makes the trends more negative, i.e. the trends are increased in the absolute values around 6-10 hPa, although within the uncertainty estimates, am I right? It is not absolutely clear from your text what do you mean by saying that UBDC index leads to a decrease of ozone trends: in terms of absolute values the trends increase.
9. P18L12: ‘temperature and ozone TREND estimates’

Hide

RR by Lon Hood (10 Nov 2016)

Suggestions for revision or reasons for rejection

As stated in my first review, this is overall a valuable and significant effort to improve statistical estimation of stratospheric ozone and temperature trends and natural variability. The presentation is generally excellent and the changes made in the new version of the manuscript are very responsive to the criticisms provided by myself and the other reviewer. However, some minor but important revisions are still needed in my opinion.

First, Figure 13 of the original manuscript was the key figure of the paper because it showed the improvement in trend estimates for the four ozone datasets and the three temperature datasets when the MLSD index was added to the regression model. In the new version of the manuscript, the old Figure 13 has been replaced with a new figure which shows the improvements for only one ozone dataset (SWOOSH) and one temperature dataset (SSU). The improvements are shown for cases with and without including a correction for autocorrelation of the residuals in the regression analysis. This is something of a step backwards because (a) there is no need to include the AR0 results because they do not correct for the autocorrelation; and (b) the improvements for the other three ozone datasets and the other two temperature datasets are not shown. Correction for autocorrelation is standard procedure in regression analyses so only the AR1 results are needed. One could put such a figure in an appendix or in supporting online information if it is desired to show the difference between the AR0 and AR1 results but it should not be in the main paper in my opinion. My suggestion is to correct the old Figure 13 using the AR1 results and put that in the final paper. Then readers can see the improvements for themselves for the various datasets. Adding the AR0 results only adds unnecessary complication to the figure in my opinion.

Second, in the abstract and conclusions, the improvements are stated as maximum improvements ("the index can account for up to 60% of the total variability ... the uncertainty on all multiple linear regression coefficients can be reduced by up to 30% and 25% in temperature and ozone, respectively ...". While these statements are true, it would be more accurate to give the range of the improvements as a function of pressure level and dataset. Looking at the new Figure 13, the improvement in the SWOOSH ozone error bar is indeed 25% at 2.5 hPa but it decreases to about zero by 1 hPa and by 10 hPa. This is a rather strong altitude dependence and readers should be made aware of this. Also, how do these results depend on the different datasets? Are the improvements less for the other datasets? This should be shown in the final version.

Thanks,
Lon Hood

Hide

ED: Reconsider after minor revisions (Editor review) (15 Nov 2016) by Bernd Funke

AR by William Ball on behalf of the Authors (28 Nov 2016) Author's response Manuscript

ED: Publish as is (28 Nov 2016) by Bernd Funke

AR by William Ball on behalf of the Authors (28 Nov 2016) Author's response Manuscript

Short summary

We find monthly, mid-latitude temperature changes above 40 km are related to ozone and temperature variations throughout the middle atmosphere. We develop an index to represent this atmospheric variability. In statistical analysis, the index can account for up to 60 % of variability in tropical temperature and ozone above 27 km. The uncertainties can be reduced by up to 35 % and 20 % in temperature and ozone, respectively. This index is an important tool to quantify current and future ozone recovery.