Characterizing the tropospheric water vapor spatial variation and trend using 2007–2018 COSMIC radio occultation and ECMWF reanalysis data

Shao, Xi; Ho, Shu-Peng; Jing, Xin; Zhou, Xinjia; Chen, Yong; Liu, Tung-Chang; Zhang, Bin; Dong, Jun

doi:https://doi.org/10.5194/acp-23-14187-2023

Articles | Volume 23, issue 22

https://doi.org/10.5194/acp-23-14187-2023

© Author(s) 2023. This work is distributed under
the Creative Commons Attribution 4.0 License.

Special issue:

Analysis of atmospheric water vapour observations and their...

https://doi.org/10.5194/acp-23-14187-2023

© Author(s) 2023. This work is distributed under
the Creative Commons Attribution 4.0 License.

Articles | Volume 23, issue 22

Research article

|

15 Nov 2023

Research article |

| 15 Nov 2023

Characterizing the tropospheric water vapor spatial variation and trend using 2007–2018 COSMIC radio occultation and ECMWF reanalysis data

Xi Shao, Shu-Peng Ho, Xin Jing, Xinjia Zhou, Yong Chen, Tung-Chang Liu, Bin Zhang, and Jun Dong

Download

Final revised paper (published on 15 Nov 2023)
Preprint (discussion started on 10 Feb 2023)

Interactive discussion

Status: closed

RC1:
'Comment on acp-2022-660', Anonymous Referee #1, 10 Feb 2023

In this study COSMIC radio occultation and ECMWF reanalyses data is used to estimate trends in tropospheric water vapour. The trend analyses is performed globally for specific latitude bands, separating the globe into latitude bins of 20° and on three pressure levels, namely 300, 500 and 850 hPa. Before the trends are estimated the data sets are inter-compared.
General comments:
The study itself is worth to be published, however needs major revisions before publication in ACP. Since in this study first a detailed inter-comparison is performed followed by a quite detailed trend analyses, the paper becomes very long and hard to follow. The current version of the manuscript gives the impression that actually two manuscripts have been combined. I would suggest to significantly shorten the paper, especially the inter-comparison part. The most important results of the inter-comparison should, however, be provided in an appendix or supplement to this manuscript because knowing the differences between data sets is important for interpreting the derived trends.
A motivation for why a separation into 20° latitude bins has been performed. Accordingly, a motivation for why the three pressure levels 300, 500 and 850 hPa have been used is missing. Why do you look at the seasonal cycle before calculating the trends?
Further, the manuscripts need significant improvements in writing and presentation of the results. Most of the figures and all tables need to be improved.
Some figures are not really concise and use to small fonts.
Instead of water vapour variation you should clearly state “seasonal cycle and trend”. The time period considered could also be mentioned in the title.
Specific comments:
P2, L38-41: This paragraph is too general and too broad and thus a bit out of the context of the study and thus not useful at all. The whole paragraph should be removed.
P2, L46-47: I would suggest to put the references at the end of sentence.
P2, L58: Please rephrase the sentence. “such as” is not correct here. It should rather read “and”.
P6, L196: Which selected months? Do you mean January and July? Why have these two months been selected?
P7, Figure 1: Why is the comparison done for ERA-5 and ERA-interim? Why not only ERA-5?
P9, Figure 2: Put the labels of the panels at the top left of each panel.
P9, Figure 2: I am surprised by the good agreement between the two data sets and was wondering if COSMIC data is assimilated into ERA-5. If yes, this needs to be considered in the interpretation and discussion of the results.
Figure 1-2 and according text could be moved to an appendix/supplement.
P10, L255ff: Is this shift in the NH/SH due to the ITCZ? If yes, then I assume this figure would look different for other months? Which month actually is shown here?
P11, Figure 3: The time period that has been considered should be added. Has an average over the 2007-2018 period been considered? For which month is shown in this figure? Or is here an average over all months/years shown?
Figure 4-6: I am not happy with these figures. In my opinion these are two overloaded and hardly readable. I am not yet convinced the figure 4b, 5b, 6b. These could be moved in an appendix/supplement. Or consider rearranging the results presented. See the following comment.
P12-14, Figure 4-6: My suggestion would be to completely change the way of presenting the results for the seasonal cycle. Wouldn’t it be better to show the NH and SH separately and then use one figure for each hemisphere showing the results for the three pressure levels. You then could have additionally one figure showing the differences for the three pressure levels (and as now with differences for both hemispheres).
P15, Figure 7: I am also not happy with this figure. Is it really worth to who three pressure levels? The results are quite similar and thus there is no need to show in all figures all three levels. Also I would suggest to improve the figure so that the hemispheres can be better compared. One way of doing this would be to add a vertical line in the middle of the plot separating the NH and SH bars.
P18, L451: Add references.
P19, L472: Which latitude bin? 0-20°, thus tropics? What trend do you derive for the other regions?
P22, Figure 10: Add a vertical line at 0° to visually separate NH and SH. You could also write in the plot SH and NH, respectively.
P23, Table 2: The table should be improved. In it’s current form it is hard to read and thus not really useful.
P26, Figure 12: I don’t understand this figure. What exactly has been done? Why is this kind of analyses useful? I think this analyses does not need to be shown in the main part of the paper and could be moved to the appendix.
P28, Figure 13: Why is here only 500 and 850 hPa shown and not 300 hPa?
P29, Figure 14 and corresponding text: It should be motivated how these sites have been selected.
P30-31, Tables 3-5: These tables are hard to read. You should find a way to present the results in a readable way. All additional information could be put into the appendix. E.g. in table 3 columns 2 and 3 could be combined.
P33, L762: ERA-5 is the latest version of reanalyses and has been significantly improved compared to ERA-interim. Thus, it is not astonishing that the agreement between COSMIC and ERA-5 is better than the agreement between COSMIC and ERA-interim. It would be maybe useful to check and discuss the results from the SPARC reananlyses comparison project (https://www.sparc-climate.org/activities/reanalysis-intercomparison/).
Technical corrections:
P1, L24: sub-tropics → sub-tropical
P1, L28: are around → “are found around” or “are observed around”
P1, L28: delete “at sites”
P3, L75: in → of
P5, L139: Put “in this study” at the begin of the sentence.
P5, L141: delete “for global environment and weather studies”.
P7, Figure 1: Write “ECMWF” in the x-label instead of “ERA”.
P8, L220: tropics → tropical
P11, Figure 3: Put labels at the top left of each panel.
P12, L294-195: same month …..same latitude zones → each month and each latitude zone
P15, L373: “RO” obsolete → delete
P17, Figure 8: Put labels at the top left of each panel (should be aligned).
P18, L425: “~ “should be “- “
P18, L433: Same here.
P19, Figure 9: Put labels at the top left of each panel.
P22, Figure 10: Put labels at the top left of each panel.
P22, Figure 10 caption: What do you mean with “zone—mean”? Zonal mean or do you mean the latitude bins?
P22, L535: add “tropical” after “northern” so that it reads “northern tropical 0° to 20° latitude bins”.
P24, L69: Write either “no data points” or “missing data points”.
P24, L584: [2020] → (2021)
P25, L589: [2020] → (2021)
P25, Figure 11 caption: either “no monthly data” or “ missing data”.
P26, Figure 12: Put labels at the top left of each panel.
P27, Figure 13: Put labels at the top left of each panel.
P32, L717: low latitude → tropical regions
P32, L720 and L727: trending → trending
P32, L739: tropics → tropical, subtropics → subtropical
P33, L751: trending → trends
P34, L778: from trending → from estimating the trend for 2007-2018 from
P34, L792: tropics and sub-tropics region → tropical and subtropical regions
P35, L803 and 810: degree sign misplaced.
P35, L805: difficulty → difficult
P35, L813: having → have
P35, L814: trending → trending
P36, L829: trending → trend estimation
P36, L846: [2021] → (2021)
P38, L912: Co-authors are missing.

Citation: https://doi.org/10.5194/acp-2022-660-RC1
- AC1: 'Reply on RC1', Xi Shao, 10 Mar 2023
  
  Please refer to the attached reply document.
  
  Citation: https://doi.org/10.5194/acp-2022-660-AC1
EC1:
'Review on acp-2022-660', Jayanarayanan Kuttippurath, 16 Mar 2023
Review of Characterizing the Tropospheric Water Vapor Variation using COSMIC Radio Occultation and ECMWF Reanalysis Data: Shao et al.
Title indicates the characterization of water vapour variability using measurements and reanalysis data, but the results and discussion are mostly on the comparison of COSMIC water vapour data with reanalysis at different spatial and temporal scales.
Major:
Its global comparison, why cannot use ground-based observations like radiosonde, GNSS, GPS, which are commonly used for validation and comparisons. This is very important as reanalysis data can have a relatively large bias in some regions (e.g. tropics, what we have found in our studies).

You have shown the bias and differences, but no valid reasons are given. Please discuss the reasons for the differences

Specific Comments:
Line 110-117: Please move these sentences to Data section, where COSMIC water vapour description is given.
Line 147: The ERA5 water vapour…….pressure levels. This sentence is about the availability of ERA5 water vapour at different pressure levels, so please move this to lines 140-145.
Line 182: Why only three pressure levels? 850, 500 and 300 hPa, and why these particular altitudes?
Line 186: Give references
Figure 1 and Line 185-197: What is the need of comparing COSMIC water vapour with both ERA5 and ERA-Interim, if it is already stated in Line 185-186 that the ERA5 water vapour retrieval is better than that of ERA-Interim? Also, why only January and July are considered here”?
Section 3.1 Global distribution of water vapour:
Why authors have shown the distribution of water vapour at 10-degree latitude and longitude grid not in the original resolution of COSMIC and ERA5? If bias is computed at a coarser spatial resolution, there might be a chance of large uncertainty and the regional variability will not be reflected in the bias estimates.
Line 232: Why COSMIC water vapour overestimates ERA5 in the upper troposphere?
Line 232-233: Since the water vapour concentration at 300 hPa is very small, its contribution to the total precipitable water would also be very small.
Section 3.3 Seasonal variability of COSMIC and ERA5 water vapour distribution: If you want to discuss the seasonal variability, discuss the seasonal changes and then present the bias. Also, why authors have divided the latitude in 20-degree interval here? Why not tropics, mid-latitude and Polar Regions then?
Line 393-341: It is already mentioned in the previous section “Decline in water vapour in southern hemisphere is faster than the northern hemisphere”
Line 364: This sampling error does not affect the bias discussed in the previous section? If it is, then how authors have addressed this issue?
Line 386-387: Sampling error for COSMIC ? Also, for ERA 5?
Line 408-409: “which is mainly due to the difference between the orbital-specific distribution of COSMIC RO observations and uniformly-distributed global ERA5 data”. Give references for this statement. Also, how orbital-specific distribution of COSMIC RO observations cause oscillations in the sampling error?
Line 411: Why COSMIC sampling decreases significantly after 2017?
Figure 8: Sampling is very small in 2011 as compared to that in 2007-2009. Its almost constant in 2011-2014, and then decreases until 2019. Why these disparities in the sample numbers?
Figure 9: Water vapour is increasing from 2008 to 2010, almost constant from 2011 to 2014, then again increased during the period 2014-2017, and finally it shows constant (i.e., no trends) at all three pressure levels. Why these particular distributions? Discuss
Line 480: How these results can be consistent or even comparable with Chen and Liu (2016)? They have computed the PWV trends (entire column of water vapour). Here only three pressure levels are taken. Please cite some other references, in which tropospheric water vapour trends are computed.
Line 488-491: Again, Chen and Liu (2016) is used here for the comparison.
Line 523-527: It can’t be directly attributed to the dry atmosphere.
Line 531: What do you mean by the most stable water vapour trend?
Line 622-623: This sentence about sea surface temperature has no meaning here. Better to write the trends in sea surface temperature, which can influence the water vapour trends.
Line 625: “Indo-Pacific warm pool region and increase in the equatorial region of the Pacific Ocean is what we here observe.” I do not see any analysis here for making this statement.
Line 626-630: How monsoon climate and precipitation affect the trends in water vapour in these regions? Precipitation is known for the sink of water vapour. Discuss this.
Figure 14: How these sites are selected?
Section 5.2: Without analysing cloud data how authors identified the regions of Stratocumulus clouds?
Line 661: “RO data can penetrate the cloud, and the water vapour retrieval from RO data is not affected by the stratocumulus cloud.” Reference for this statement.
Line 675: “The possible cause of smaller trends from ERA5 water vapour data over stratocumulus cloud-rich regions could be difficulty in accurately estimating water vapour at low height in ERA5 reanalysis data compared with COSMIC RO measurements”. Can you provide the reference for the statement?
Section 5.3: What is the basis for the selection of these sites?
Line 695: Where is the analysis of trends in ocean surface temperature?
Line 726-729: For site#17 ………..Pacific Ocean is on the west. The reasons stated for the decline in water vapour at site#17 are not convincing.
Line 729-730: Water vapour at 850 hPa is not a precipitable water vapour. Also, there is no “near-surface precipitable water vapour”.
Line 731: Again, precipitable water vapour, it just water vapour at 850 hPa.
Line 732: Earlier it is mentioned that COSMIC measurements are not affected by stratocumulus cloud, then how it becomes more challenging here?
Section 6: Most of the results and discussion are repeated here with the same references. Please rewrite this section and draw a solid conclusion.
Also, please crosscheck the citation Lui et al. (2016) in Line 838
----
Citation: https://doi.org/10.5194/acp-2022-660-EC1
- AC2: 'Reply on EC1', Xi Shao, 24 Mar 2023
  
  Please refer to the attached reply document.
  
  Citation: https://doi.org/10.5194/acp-2022-660-AC2

Peer review completion

AR: Author's response | RR: Referee report | ED: Editor decision | EF: Editorial file upload

AR by Xi Shao on behalf of the Authors (30 May 2023) Author's response Author's tracked changes Manuscript

ED: Referee Nomination & Report Request started (14 Jun 2023) by Jayanarayanan Kuttippurath

RR by Anonymous Referee #1 (20 Jun 2023)

ED: Reconsider after major revisions (12 Jul 2023) by Jayanarayanan Kuttippurath

Dear Authors:

Please find the review report of the referee (attached) and my comments here.
Please submit the revised MS at the earliest. Thank you.

------------------------------------------------------------------------------------------------------
Comments to consider for revision.

1. Check the spatial resolution of ERA5. Is it 0.2 or 0.25?

2. How the sampling error in COSMIC can affects trend estimation, but not bias?

3. “About the discussion on monsoon climate and water vapour trend”
Precipitation is generally known as a sink rather than a source of water vapour. The monsoon climate influences water vapour variability and trends through the moisture transport. Please modify the discussion to include a link between the water vapour trend and moisture transport, not precipitation.

4. “Representative months January and July for winter and summer in northern hemisphere, respectively”
I understand that these months can be representative of the seasons, but any significant changes in winter and summer cannot be directly linked to these months; as they could occur in other months of the respective seasons as well. In other words, any changes in these two months, such as trends and bias, may not represent the seasonal change. It should be specified in Methods.

5. Reanalysis data can be used as reference for the validation of satellite measurements? To support this argument, cite some references on the validation of ERA5 (to check the consistency of ERA5) and write about the errors, if any.

6. In reply to my previous comment “Why COSMIC water vapour overestimates ERA5 in the upper troposphere?” authors state that there is large uncertainty in retrieving water vapour in the reanalysis model in the upper troposphere. It should be clearly mentioned in the methodology, why there is a bias in ERA5 and how much, because ERA5 is used as the reference data for the validation of COSMIC here. It is better to check the consistency of ERA5 in other pressure levels also.
---------------------------------------------------------------------------------------------------------

Hide

AR by Xi Shao on behalf of the Authors (27 Aug 2023) Author's response Author's tracked changes Manuscript

ED: Referee Nomination & Report Request started (03 Sep 2023) by Jayanarayanan Kuttippurath

RR by Anonymous Referee #1 (12 Sep 2023)

Suggestions for revision or reasons for rejection

I really appreciate the effort the authors have put in the revision of their manuscript. However, there are still some technical/minor issues that should be considered before publication.

General comments:
P5, Section 2.2: This section is too long is not entirely describing the data. Here you have a mixture between data set description (L141-L167), method description (L169 to L202) and results (L204-227). The method part should be put into an extra subsection and the results part either there in a extra subsection or moved in an extra subsection to the results section.

P9, L264: Not clear why you consider reanalysis data for cloud-free scenes if there should be data available for cloud and cloud-free scenes.

P11, L292-293: “…….large uncertainties in retrieving water vapor in the reanalyses model……… “ Please rephrase. This is simply not correct. A model does not retrieve any data. In a model data is calculated.

P11, L295: You should formulate this more carefully. How can you be sure that if there is a bias this is a bias of the ERA5 data? What about the COSMIC RO data? Nowhere in the paper the quality of the RO data is discussed.

P12, L340-341:”……..affecting interhemispheric temperature difference, can affect the interhemispheric water vapor difference”. Sentence not clear. It seems in the latter part of this sentence something is missing. Please correct/rephrase the sentence.

Technical corrections:

P1, L23: upward -> increasing
P1, L24: downward -> decreasing
P1, L26: higher than ERA5 data -> higher than the ones derived from ERA5 data
P17, L453: trending -> trend
P17, L454: delete “than” and move “300 and 500 hPa at the end of the sentence and add “at” before so that the sentence reads: …….is lower by 1.44 and 1.22%/Decade at 300 and 500 hPa, respectively.
P20, L514: is -> are? Consider rephrasing sentence. It is difficult to understand what you want to say.
P21, L546: slope -> trend
P22, Figure 7 caption: replace 2 times “slope” with “trend”
P23, L584: add “the” -> in the affected regions
P23, L599: replace “degree” by the degree sign “°”

Hide

ED: Publish subject to minor revisions (review by editor) (13 Sep 2023) by Jayanarayanan Kuttippurath

AR by Xi Shao on behalf of the Authors (21 Sep 2023) Author's response Author's tracked changes Manuscript

ED: Publish as is (28 Sep 2023) by Jayanarayanan Kuttippurath

AR by Xi Shao on behalf of the Authors (03 Oct 2023) Manuscript

Short summary

Atmospheric water vapor plays an essential role in the global energy balance, hydrological cycle, and climate system. This paper characterizes and compares the global, latitudinal, and regional variabilities of COSMIC and ERA5 water vapor distribution, as well as the seasonality and long-term trends at selected pressure levels from 2007 to 2018. Evaluation of spatiotemporal variabilities of atmospheric water vapor ensures the qualities of COSMIC and reanalysis water vapor for climate studies.