Evaluation of potential sources of a priori ozone profiles for TEMPO tropospheric ozone retrievals

Johnson, Matthew S.; Liu, Xiong; Zoogman, Peter; Sullivan, John; Newchurch, Michael J.; Kuang, Shi; Leblanc, Thierry; McGee, Thomas

doi:https://doi.org/10.5194/amt-11-3457-2018

Articles | Volume 11, issue 6

https://doi.org/10.5194/amt-11-3457-2018

© Author(s) 2018. This work is distributed under
the Creative Commons Attribution 4.0 License.

https://doi.org/10.5194/amt-11-3457-2018

© Author(s) 2018. This work is distributed under
the Creative Commons Attribution 4.0 License.

Articles | Volume 11, issue 6

Research article

|

19 Jun 2018

Research article |

| 19 Jun 2018

Evaluation of potential sources of a priori ozone profiles for TEMPO tropospheric ozone retrievals

Matthew S. Johnson, Xiong Liu, Peter Zoogman, John Sullivan, Michael J. Newchurch, Shi Kuang, Thierry Leblanc, and Thomas McGee

Download

Final revised paper (published on 19 Jun 2018)
Preprint (discussion started on 30 Jan 2018)

Interactive discussion

Status: closed

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

- Printer-friendly version

- Supplement

RC1: 'Reviewer comments', Anonymous Referee #2, 06 Feb 2018
- AC1: 'Response to Reviewer #1', Matthew S. Johnson, 19 Mar 2018
RC2: 'Referee comments', Anonymous Referee #1, 26 Feb 2018
- AC2: 'Response to Reviewer #2', Matthew S. Johnson, 19 Mar 2018

Peer-review completion

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

AR by Matthew S. Johnson on behalf of the Authors (28 Mar 2018) Author's response Manuscript

ED: Referee Nomination & Report Request started (09 Apr 2018) by Mark Weber

RR by Anonymous Referee #2 (20 Apr 2018)

Suggestions for revision or reasons for rejection

Reviewer #2 has read the answers from the authors on his original review comments, and has the following response:

The first section of the answers is related to the reasons for using Eq.1 for replacing the a priori with other a priori shapes, and is defended by the authors with references to a list of six papers showing that this method has been used before, '... as a good first order approximation of satellite retrievals'.
The text was also modified and now states: 'This linear approach is a good first order approximation of the non-linear satelltie retrievals...'.

It is indeed a _linear_ approximation and that's where the risk of using this method lies. Let's re-state Eq.1:

Xr = Xa + AK*(Xt - Xa) + G*e

The reason for Eq 1 to be a linear system stems from the fact that, following the notation of Rodgers (2000), the calculation of the AK depends on K, which is described in Rodgers (2000) as the sensitivity of the forward model [to the state vector] (see around Eq 3.7 for K and Eq 3.28 for the calculation of the AK, in Rodgers(2000)). This sensitivity is defined by the forward model, and is only linear in a very local range of the state vector, especially in non-linear ozone profile calculations.

My interpretation of Eq.1 in the paper is that the combination of the _original_ Xr, Xa, Xt and the AK, calculated in Zoogman (2017), are tied and _together_ form a linear system.

Replacing the original Xa (from TB-Clim) with other values is only allowed as long as the new Xa does not deviate too much from the original so the linearisation of the pre-calculated AK still holds.

In Figure 7 of the paper, the authors show various profiles. Among them the initial Xa (TB-Clim) and other a priori profiles. In 7b, the red solid line represents TB-Clim and the green solid line represents GEOS-5 FP. It is clearly visible that in certain locations in the vertical the differences between the red and green line are 100%. That is: one of the values is twice as large as the other.

This relates back on to the linearity issue: given that the ozone absorption is non-linear, the fundamental question is whether the assumption that Eq.1, as a linear system, can still be applied?

The second section of the answers is related to me assuming that the authors were doing actual retrievals, but instead they use Eq.1.

I have one comment, however, related to the answer from the authors: they refer us to Zoogman (2017, JQSRT) for more information on the setup of the calculation of the AK's and other elements in the comparison used in this paper. JQSRT is what I consider to be a 'closed source' journal. Its contents is not publicly accessible to reviewers without them paying for access with a credit card. The authors might as well have referred to an non-public internal report. The blunt referral is what I consider 'bad form'. Please consider more open access journals for publications.
I would like to encourage the authors to include the small sections of their explanations they have given in their answers into the main body of the paper instead of just refering the reader to a different paper behind a pay-wall (see also my question related to sect 2.2/2.2.1).

My other questions were more aimed at clarification and explanations.

In my opinion, the main remaining issue with the paper is the application / validity of the linearisation. I see two ways of solving this:

- The authors can approach this from a theoretical point of view and better describe the limitations of the validity / application of Eq.1 in the text of the paper, thereby warning the reader that they may (over-)stretch the linearity assumption

or

- The authors can approach this from an experimental point of view and show that the linearity of the pre-computed AK (based on the forward model) holds for a couple of cases. This could be done in an appendix or supplemental material.

Hide

ED: Reconsider after major revisions (23 Apr 2018) by Mark Weber

AR by Matthew S. Johnson on behalf of the Authors (24 May 2018) Manuscript

ED: Referee Nomination & Report Request started (01 Jun 2018) by Mark Weber

RR by Anonymous Referee #2 (04 Jun 2018)

ED: Publish as is (04 Jun 2018) by Mark Weber

AR by Matthew S. Johnson on behalf of the Authors (04 Jun 2018)

Short summary

This research was conducted to determine the impact of multiple a priori ozone (O₃) profile products on Tropospheric Emissions: Monitoring of Pollution (TEMPO) satellite retrievals. It was determined that non-climatological model predictions, in particular those from a chemical transport model, when applied as the a priori profile improved the accuracy of TEMPO tropospheric O₃ retrievals in comparison to the TB-Clim product that is currently suggested for use in the TEMPO retrieval algorithm.