Trends in the tropospheric general circulation from 1979 to 2022
- European Centre for Medium-Range Weather Forecasts, Shinfield Park, Reading RG2 9AX, UK
- European Centre for Medium-Range Weather Forecasts, Shinfield Park, Reading RG2 9AX, UK
Abstract. Atmospheric general circulation changes since 1979 are examined using the ERA5 reanalysis. Maps based on linear trends and time series for specific areas are presented. Attention is concentrated on monthly, seasonal and annual means, but shorter-timescale variability is also considered, including extremes. Changes in near-tropopause winds are the main focus, but related changes in temperature, wind and other variables throughout the troposphere are discussed.
Middle- and upper-tropospheric warming is larger in the subtropics and outer tropics than in the deep tropics, except over the Pacific. This is linked with a strengthening and meridional expansion of the tropical easterlies that has received little previous attention. Warming over several mid-latitude and subtropical land areas comes close to matching the large warming of the Arctic. Westerly upper-level winds in general weaken over the Arctic in winter, but strengthen in northern middle latitudes, contrary to arguments based on circulation changes due solely to amplified Arctic warming. The jet-stream region over the eastern North Atlantic and western Europe shifts southward. Westerlies strengthen in a band stretching south-eastwards from the tropical western Pacific to southern Australia, and in the polar-jet-stream region that surrounds Antarctica.
Extreme jet-stream winds increase over the North Atlantic. Net kinetic energy also increases, mostly associated with sub-monthly variability along the mid-latitude storm tracks and over the tropical Pacific. Available potential energy changes less. Geopotential height shows a distinct pattern of change in the stationary northern-hemispheric long-wave structure. There are increases in surface pressure over the North Pacific and southern mid-latitudes, and decreases over the Arctic Ocean and offshore of Antarctica.
Several comparisons are made between ERA5 and the JRA-55 reanalysis, and between ERA5 and the observations it assimilated. They show reassuring agreement, but some regional differences require further investigation.
Adrian John Simmons
Status: final response (author comments only)
-
AC1: 'Comment on wcd-2022-19', Adrian Simmons, 08 Apr 2022
A point missing in the submitted preprint is that although many of the trends examined in the paper appear to be quite uniform over the period of study, this is not the case for the strengthening and expansion of the tropical upper-tropospheric easterlies. These particular changes occur predominantly over the first 25 or so years of the period, with little change or even a slight reversal thereafter. This fits with arguments in the literature that some changes in tropical circulation observed in the past few decades have been manifestations of natural variability and accordingly not inconsistent with climate projections showing opposite changes. The point can be taken care of with minor changes to the text and to two or three figures.
-
RC1: 'Comment on wcd-2022-19', Theodore Shepherd, 02 May 2022
This paper is a very detailed and comprehensive analysis of the long-term trends in a number of variables representing the tropospheric general circulation over the period of the ERA5 reanalysis, 1979-2022. With over a 40-year timeseries, these trends are now long enough to become interesting in themselves. It is increasingly recognized that for documenting long-term changes in the tropospheric general circulation, direct observations tend to be problematical because of sampling and representativeness issues, and reanalyses ultimately provide the best way of obtaining a reliable record of the changes. It is of course necessary to be vigilant for potential sources of temporal inhomogeneity, as is exemplified in the present work which continually reminds the reader that nothing should be taken for granted. The uncertainty quantification presented here, which is a combination of standard statistical significance (essentially, a measure of the signal-to-noise ratio) to capture the effects of natural variability, and detailed comparisons (e.g. with other reanalyses, or with monitoring of the background fields) to assess potential systematic uncertainties, is interpretable and transparent, and far preferable to the oft-heard but meaningless mantra of a “full uncertainty budget”.
The paper is very descriptive but the overall philosophy is to carefully document what the data show, which I very much respect, and which will be useful for the community. It can be unhelpful to try to create a narrative where one does not really exist (the real world can be like that). Indeed, the paper questions some of the current narratives that exist around circulation changes, and raises some new questions to be explored. Both will be useful.
I can therefore recommend publication of the paper in essentially its present form.
My one scientific comment concerns the standard narrative of the poleward shift of the SH summertime (DJF) jet. The author criticizes its representation as a changing polarity of the SAM. In a similar vein, Byrne et al. (2017, doi: 10.1175/JCLI-D-17-0097.1) argue that the phenomenon is more usefully seen as a delay in the seasonal equatorward shift of the SH jet, induced by the delayed breakdown of the stratospheric polar vortex. That argument seems to me to be a temporal analogue of the argument made here concerning the latitudinal shift of the NH eastern Pacific jet, with time in place of longitude. The author might wish to make that connection, if he agrees with it, since it seems part of a more general point that looking at circulation changes from too local a perspective (either in seasonality, or in longitude) can miss the bigger picture.
Minor points:
Fig 1: Why is the black dot in panels a and b located at 2010 rather than 2005?
Fig 7 and discussion in text. Shouldn’t it be PV=-2 in the SH? And what happens at the equator, when PV changes sign?
Fig 10 and Table 3: I’m not sure what is the logic for discussing the annual-mean changes in upper tropospheric wind speed in such detail, when as Fig 11 makes clear, there is quite a seasonality to the changes.
-
AC2: 'Reply on RC1', Adrian Simmons, 08 May 2022
I thank the reviewer for both his general comments on the paper and the specific issues he raises. I am of course pleased that he recommends publication of the paper in essentially its present form.
As regards his specific scientific comment, I am grateful to be reminded of the paper by Byrne et al. (2017). It is indeed relevant to the discussion of the SAM given in the concluding discussion of the present paper, and I have drafted an additional paragraph to follow the existing one on the SAM.
Concerning the three points:
(i) I apologise for the error in the placing of the black dots in panels (a) and (b) of Figure 1. The original version of this figure was produced by computer code that had placed the dots in the right position. In a final revision of the paper before submission I considered the dots to be too small. I decided to increase their size by manually editing the postscript graphics file, and in panels (a) and (b) but not (c) and (d) I inadvertently increased the separation between the dots as well as their size. As it happens, Figure 1 has since been remade to include 120-month running means (as has Figure 9). This is one of the changes I mention in https://doi.org/10.5194/wcd-2022-19-AC1. The dots are correctly positioned, and of the larger size, in this new version of the figure.
(ii) The reviewer is of course right: I was careless in referring to the PV=2 surface rather than the |PV|=2 surface. I have corrected this. The reviewer asks what happens at the equator. For the formal record I note that a link to the documentation of the ECMWF forecasting system used to produce ERA5 is provided in the Hersbach et al. (2020) paper to which reference is made in the present paper. But to answer the reviewer directly, the calculations of variables on the |PV|=2 surface are made on the model’s Gaussian grid, which straddles the equator. Values are then fitted using spherical harmonics, from which values anywhere on the globe can be derived. This by itself does not avoid problems where PV is small through most if not all the atmospheric column, so the pressure of the nominal |PV|=2 surface to which variables are interpolated is not allowed to be lower than 89hPa. I have revised the text of the paper to provide the latter information, and took the opportunity to quote the average pressure and trend of the |PV|=2 surface. The trend is for a modest but statistically significant decrease in pressure (increase in tropopause height) over time of about 0.7hPa per decade. The global- and time-mean is 191hPa, close to the level of 200hPa for which the majority of results in the paper are presented.
(iii) I prefer to retain Figure 10 and Table 3. Yes, as the reviewer points out there is quite a seasonality to the changes, which is apparent from Figure 11 and its discussion. But there is also a fair deal of cancellation from one season to another, especially in the extratropical southern hemisphere, and it is difficult to assess quantitative aspects of this from Figure 11 alone. I think it is important to show that the net change over the years in jet-stream positions is quite small in many regions. Summaries such as presented in AR6 that mention only the largest seasonal changes, for which confidence is high, may give a misleading impression of net changes.
Figure 10 is also used to show the similarity between ERA5 and JRA-55, and that the same qualitative changes are found if one considers monthly-mean wind speeds rather than the speed of the monthly mean wind. Also, the “sanity check” provided by examining geopotential height (section 7) is illustrated for annual rather than seasonal trends (to prevent the paper becoming even longer). It links back to Table 3.
-
RC3: 'Reply on AC2', Theodore Shepherd, 16 May 2022
On point (iii), the author's response seems to suggest that the annual mean is the best indicator of "real" change, and that highlighting individual seasons might be seen as a form of cherry-picking. But for atmospheric circulation, everything we know about atmospheric dynamics -- including from much of the author's previous work -- tells us that the response to forcings will depend quite sensitively on the background state (even if the response is linear, it is only tangent linear), and the background state has a strong seasonal cycle. Thus, a seasonality in the forced response is to be expected and is not necessarily a sign of lack of robustness. Taking an annual mean beats down the noise but it also dilutes the signal. Zappa et al. (2015 J.Clim., DOI: 10.1175/JCLI-D-14-00823.1) examine this issue for the North Atlantic/European sector and suggest a form of optimal seasonal averaging (which actually is extended winter and extended summer), which better detects signals than the annual mean.
-
AC3: 'Reply on RC3', Adrian Simmons, 16 May 2022
I think my response was perhaps not as clear as intended. I do not wish to downplay the importance of the seasonal changes. Aside from the fact that Figure 10 is used to make some additional points - showing for instance a comparison between ERA5 and JRA-55 that would otherwise have to be either omitted or presented for each season - I do think it is important to point out that the changes from one season to another do tend to cancel in the annual mean for some features. It could (but might not) give a misleading picture if the season with the largest (most statistically clear cut) change is given undue emphasis in a situation when each of the other three seasons show a change of opposite sign but with magnitude smaller by a factor of 3. So I prefer to retain Figure 10 in addition to Figure 11. This does not mean I think Figure 10 provides a better indicator of "real" change than Figure 11. I agree with the point that for a particular feature a particular temporal averaging might be better than using the traditional four three-month seasons, but what is an optimal averaging period for one circulation feature is unlikely to be optimal for other circulation features, and I am trying in the paper to provide a global overview.
-
AC3: 'Reply on RC3', Adrian Simmons, 16 May 2022
-
RC3: 'Reply on AC2', Theodore Shepherd, 16 May 2022
-
AC2: 'Reply on RC1', Adrian Simmons, 08 May 2022
-
CC1: 'Comment on wcd-2022-19 by Gloria Manney & Michaela Hegglin', Gloria Manney, 13 May 2022
- AC5: 'Reply on CC1', Adrian Simmons, 21 May 2022
-
RC2: 'Comment on wcd-2022-19', Anonymous Referee #2, 14 May 2022
This is an interesting paper which is more in the forma of a somewhat-descriptive survey rather than necessarily developing new physical insights. It can be seen as an update on, or complementary to, a few aspects of the recently-released IPCC/UNEP Working Group I AR6 report. It has some focus on the comparison of two modern reanalysis sets (and makes remarks as to where and, in some instances why, these reanalysis products differ.
The manuscript has potential to be an important contribution to the literature. However, it requires revision as outlined below.
As an overarching comment on the paper some of the explanations resented are based on the basic dynamics and thermo-dynamics of the atmosphere. For example, from line 36 comment is made … ‘Changes in upper-tropospheric winds are linked to changes in surface flow and horizontal temperature gradients through the tendency of the atmosphere to remain close to thermal-wind balance. Where hydrostatic and geostrophic balance apply, the vertical shear of the wind is proportional to the temperature gradient across the direction of flow. The proportionality factor is larger at low than high latitudes’
Even though such statements are true, they don’t really belong is a ‘scientific’ paper. In many cases these remarks are obvious and the implications will be clear to the reader. I strongly suggest culling such remarks; this will make for a better paper, and also makes a valuable contribution to reducing the length of quite a long submission.
Lines 30-32: Reinforce this message by referencing the more recent paper of Screen, Bracegirdle, and co-authors (2018), Polar climate change as manifest in atmospheric circulation, Curr. Clim. Change Reps., 4, 383-395, doi: 10.1007/s40641-018-0111-4.
Lines 36-43: The remarks made a few lines earlier I the paper point to the compelc association and interactions between the thermodynamics and dynamics. In these introductory comments it would be very helpful to refer to the paper of
Theodore G. Shepherd, 2014: Atmospheric circulation as a source of uncertainty in climate change projections. Nature Geoscience, 7, 703-708, doi: 10.1038/ngeo2253
and his insights of ‘…
‘nearly everything we have any confidence in when it comes to climate change is related to global patterns of surface temperature, which are primarily controlled by thermodynamics. In contrast, we have much less confidence in atmospheric circulation aspects of climate change, which are primarily controlled by dynamics and exert a strong control on regional climate’
Lines 92-94: For easy reading, it would be beneficial if these trend values and corresponding p values were presented as inserts into the four boxes within Figure 1.
Line 100, caption of Figure 1: Even though it is perhaps obvious, it would be worth alerting the reader here that the y-axis ranges are different in the four time series. (This is mentioned in the text at lines 108-109, but should be emphasised here.)
Lines 117-120: Showing the (ERA5) trends that are significant (in part (b)) is interesting and helpful. It, of course, shows the same structure as in Figure 2a over the non-white areas. I appreciate that the author is showing these two plots (with much common information) for clarity. However, I’m wondering if the same purpose could not be achieved by presenting these two in ONE part of the Figure. For example, some subtle stippling could be added to Figure 2a indicating where the trends are NOT significant. This could save showing one map, and perhaps would be easy to absorb the information.
Incidentally, I presume that by ‘one-signed’ the author means ‘one-sided’ – please use this more conventional terminology. Also, a justification is required by using such a test, given that there are regions of cooling over the globe. More appropriate to use the two-sided test?
Line 158: ‘SST’ has already been defined (at line 142).
Lines 181-182: On this S/N issue worthwhile to reference recent paper of
Luke J. Harrington, 2021: Temperature emergence at decision-relevant scales. Environmental Research Letters, 16, 094018, doi: 10.1088/1748-9326/ac19dc.
Lines 209-211: This significant winter cooling over Eurasia and the northeast of the United States of America are important regional aspects of the complexity (and the consequences) of remote influences on the T2 trends. This warrants more attention than is presented here. Strongly suggest, for example, pointing out the role of teleconnections from the Arctic, high latitude blocking, Pacific SSTs etc. Making Reference to following will help on this:
Overland et al., 2019: Weakened potential vorticity barrier linked to recent winter Arctic sea ice loss and midlatitude cold extremes. J. Climate, 32, 4235-4261,
Luo, Xiao, and co-authors, 2016 - Impact of Ural blocking on winter Warm Arctic–Cold Eurasian anomalies. Part I: Blocking-induced amplification. J. Climate, 29, 3925-3947,
Dai, A. et al., 2020: ‘Combined influences on North American winter air temperature variability from North Pacific blocking and the North Atlantic Oscillation: Subseasonal and interannual time scales’. J. Climate, 33, 7101-7123, doi: 10.1175/JCLI-D-19-0327.1,
Rudeva, and coauthors, 2021. “Midlatitude winter extreme temperature events and connections with anomalies in the Arctic and tropics”. J. Climate, 34, 3733-3749.
Lines 265-270: Paper should make clear the physical/dynamic reasons why the near-tropopause winds are of great relevance here. Also, to make clear in the text here that Figure 6 is associated with the 200 hPa wind – the reader is finally told this at line 295, which is a bit late.
Lines 274-275: On this jet perspective consider citing the more recent works of …
Dong B, Sutton RT, Shaffrey L, Harvey B (2022) Recent decadal weakening of the summer Eurasian westerly jet attributable to anthropogenic aerosol emissions. Nature Comms. 13: 1148 doi: 10.1038/s41467-022-28816-5,
Hallam S, Josey SA, McCarthy GD, Hirschi JJM (2022) A regional (land–ocean) comparison of the seasonal to decadal variability of the Northern Hemisphere jet stream 1871–2011. Climate Dyn. doi: 10.1007/s00382-022-06185-5,
Liu X, Grise KM, Schmidt DF, Davis RE (2021) Regional Characteristics of Variability in the Northern Hemisphere Wintertime Polar Front Jet and Subtropical Jet in Observations and CMIP6 Models. J. Geophys. Res. 126: e2021JD034876 doi: 10.1029/2021JD034876.
Line 301: The term ‘variation’ is used thru the manuscript to mean different things, from the qualitative concept and also to the mathematical variance. This can become a little confusing for the reader. Here the author is referring to ‘… the total sub-seasonal variation, the sum of the variances of the zonal and meridional wind’. This is clearly defined, but it should be given a more precise and informative name, such as ‘summed variance’. In general use ‘variance’ when the statistical concept is being examined.
Lines 380-384: It would be helpful here (and other places where relevant) to relate these WC changes to the geographical distribution of tropical SST changes. This need not be comprehensive, and a few words will probably suffice. Worthwhile to not merely present a description of the changes, but also link them physically to other (driving?) processes.
Line 432: I am a bit confused by the use of the word ‘nominal’ here. Reference to Webster’s did not help me greatly (e.g., ‘existing or being something in name or form only’, ‘of, being, or relating to a designated or theoretical size that may vary from the actual’, …). Please to use more conventional terminology here and below (using, e.g., ‘de-trended’, ‘anomalies’, …)
Line 686: Just confirming the paper is referring to ‘surface pressure’ rather than ‘sea-level pressure’? If so, comment on why this choice was made.
Clearly space considerations have prevented the author from showing the SEASONAL trends in surface pressure, but the annual mean hides a lot of interesting seasonal behavior. To highlight this make reference in the paper to the study of Li et al. (2021 - Trends and variability in polar sea ice, global atmospheric circulations and baroclinicity, Ann. NY Acad. Sci., 1504, 167-186, doi: 10.1111/nyas.14673) who show that the strong midlatitude N Pacific increases are dominated by the DJF trends, while the annual deepening in the Amundsen-Bellingshausen Seas are predominantly due to large reductions in the intermediate seasons.
- AC4: 'Reply on RC2', Adrian Simmons, 20 May 2022
Adrian John Simmons
Adrian John Simmons
Viewed
HTML | XML | Total | BibTeX | EndNote | |
---|---|---|---|---|---|
366 | 143 | 21 | 530 | 4 | 3 |
- HTML: 366
- PDF: 143
- XML: 21
- Total: 530
- BibTeX: 4
- EndNote: 3
Viewed (geographical distribution)
Country | # | Views | % |
---|
Total: | 0 |
HTML: | 0 |
PDF: | 0 |
XML: | 0 |
- 1