Impact of atmospheric transport on CO2 flux estimates derived from the atmospheric tracer inversions
- 1Department of Biogeochemical Signals, Max-Planck Institute for Biogeochemistry, Jena, Germany
- 2Department of Physics, Faculty of Sciences, Ibb University, Ibb, Yemen
- 3Department of Physical Geography and Ecosystem Science, Lund University, Lund, Sweden
- 4ICOS Carbon Portal at Lund University, Lund, Sweden
- 5Meteorological Observatory Hohenpeissenberg, Deutscher Wetterdienst, Hohenpeißenberg, Germany
- 6Institute of Geoscience, Friedrich Schiller University, Jena, Germany
- 1Department of Biogeochemical Signals, Max-Planck Institute for Biogeochemistry, Jena, Germany
- 2Department of Physics, Faculty of Sciences, Ibb University, Ibb, Yemen
- 3Department of Physical Geography and Ecosystem Science, Lund University, Lund, Sweden
- 4ICOS Carbon Portal at Lund University, Lund, Sweden
- 5Meteorological Observatory Hohenpeissenberg, Deutscher Wetterdienst, Hohenpeißenberg, Germany
- 6Institute of Geoscience, Friedrich Schiller University, Jena, Germany
Abstract. We present an analysis of atmospheric transport impact on estimating CO2 fluxes using two atmospheric inversion systems (CarboScope Regional (CSR) and LUMIA) over Europe for 2018. The main focus of this study is to quantify the dominant drivers of spread amid CO2 estimates derived from atmospheric tracer inversions. The Lagrangian transport models STILT and FLEXPART were used to assess the impact of mesoscale transport. The impact of lateral boundary conditions for CO2 was assessed by applying the global transport models TM3 and TM5. CO2 estimates calculated with an ensemble of eight inversions differing in the regional and global transport models, as well as the inversion systems show a relatively large spread for the annual domain wide flux ranging between -0.72 and 0.20 PgC yr-1 with a mean estimate of -0.29 PgC. The largest discrepancies resulted from varying the mesoscale transport model, which amounted to a difference of 0.51 (PgC yr-1), in comparison with 0.23 and 0.10 (PgC yr-1) that resulted from the far-field contributions and the inversion systems, respectively. Additionally, varying the mesoscale transport caused large discrepancies in spatial and temporal patterns, while changing the lateral boundary conditions lead to more homogeneous spatial and temporal impact. We further investigated the origin of the discrepancies between transport models. The meteorological forcing parameters (forecasts versus reanalysis obtained from ECMWF data products) used to drive the transport models are responsible for a small part of the differences in CO2 estimates, but the largest impact seems to come from the models themselves. Although a good convergence in the differences between the inversion systems was achieved by applying a strict protocol of using identical priors, and atmospheric datasets, there was a non-negligible impact arising from applying a different inversion system. Specifically, the choice of prior error structure accounted for a large part of system-to-system differences.
- Preprint
(4088 KB) -
Supplement
(216 KB) - BibTeX
- EndNote
Saqr Munassar et al.
Status: final response (author comments only)
-
RC1: 'Comment on acp-2022-510', Anonymous Referee #1, 24 Aug 2022
The authors have put together a study examining the impacts of different inverse modeling setups on estimated CO2 fluxes across Europe. I think the authors have put together a nice study that will contribute to the literature on inverse modeling and atmospheric transport uncertainties. I have a few ideas and suggestions for the authors to consider as they revise the manuscript:
Holistic suggestions:
- (1) The manuscript text has a few English grammatical issues scattered throughout. I would do a close read for grammar before uploading the revised manuscript.
- (2) The authors test out two different choices for atmospheric models (STILT and FLEXPART), two different choices for the boundary condition, and two choices for the prior uncertainty. In several cases, the choices made in each category have some similarities. For example, both STILT and FLEXPART are driven with ECMWF meteorology (albeit different ECMWF products), and the boundary condition is generated using different versions of the TM model. I suspect that the results and uncertainties in the fluxes could look quite different if the authors had made different choices. For example, the differences due to meteorology (e.g., Fig. 7) might look very different if the authors included simulations using non-ECMWF meteorology (e.g., using a product like GEOS, MERRA-2, WRF, etc.). Similarly, differences due to the background might look very different if the authors had used a different approach, like model simulations based on GEOS-Chem or an empirical boundary condition estimate like that developed by Arlyn Andrews for CarbonTracker-Lagrange. I would be careful about generalizing the results throughout the paper and/or be vigilant about bringing in results from existing studies that may have used a wider variety of products or methodological choices.
- (3) I would strongly recommend re-organizing the results and discussions sections (sections 3 and 4). I felt that these sections often duplicated information. E.g., information presented in Sect. 3 is often repeated in Sect. 4. In addition, the text sometimes hops around from topic to topic without a strong sense of direction or flow (e.g., I felt this way about Sect. 3.1 and the beginning of Sect. 3.2.).
- (a) I would specifically re-think the purpose of sub-sections 3.1 and the beginning of Sect. 3.2. These two sub-sections present information on a smattering of different topics, information that is often revisited or repeated later in Sects. 3 and 4. I think the authors intended to give a broad overview of the results in Sects. 3.1 and 3.2, but the resulting text felt scattered to me and lacked a direction. For example, the authors mention forward model simulations briefly at the beginning of Sect. 3.2 and then repeat similar information about the forward simulations later in Sects. 3 and 4. I would also avoid generic phrases like "large differences" or "smaller differences" when there are quantitative metrics that could be used instead.
- (b) In addition, I would make sure to use strong topic sentences at the beginning of each paragraph in Sects. 3.2.1, 3.2.2, and 3.2.3. Doing so would help give each section a stronger sense of flow and direction.
- (c) Furthermore, I thought it odd that Sect. 4 repeated each of the topics presented in Sects. 3.2.1, 3.2.2, and 3.2.3, often with duplicate information. I would merge the paragraphs of Sects. 3 and 4 that discuss similar topics (i.e., merge the separate discussions of transport in Sects. 3 and 4).
Specific suggestions:
- Line 18: "wide flux ranging between -0.72 and 0.20 PgC yr-1". Can you provide any context on how large or small this range is? I.e., what do these numbers mean to someone who isn't intimately familiar with CO2 budgets? Also, what domain are you referring to here?
- Line 25 "models themselves": Can you be more precise about what you mean here?
- Introduction: the introduction feels a bit like a laundry list of different papers, and some themes are mentioned multiple times at different points in the introduction (e.g., vertical mixing and PBL heights). There are definitely a lot of papers in the literature on atmospheric transport errors. It could be more effective to organize the different paragraphs of the introduction around specific themes. E.g., have one paragraph focused on vertical mixing, another on advection, etc.
- Introduction: What knowledge gaps are there in the existing literature that this study attempts to fill? The introduction doesn't really address this question at present; instead the authors frame this study as yet another study on top of the existing studies they mention.
- Introduction: I recommend taking a look at Karion et al. (2019, https://acp.copernicus.org/articles/19/2561/2019/). That study is focused on methane, not CO2, but I think it might be relevant for the themes in the present manuscript.
- Lines 179-180: How were these model-data mismatch errors chosen?
- Line 211 "However" -- I don't think this is quite the word you want here. Maybe "With that said ...."
- Lines 240-245: Do you have any numbers or quantitative metrics that show the relative impacts of these different model parameters on the fluxes? At present, it's not entirely clear what kind of differences the text is referring to.
- Line 250 "quite large differences": Can you be more specific?
- Section 3.2.2: Differences in background estimates yield smaller uncertainties in the posterior fluxes relative to differences in atmospheric transport models. I wonder if that result could be due, in part, to the fact that both background estimates are generated using relatively similar models (e.g., TM3 vs TM5).
- Lines 326-332: Most of this information repeats information provided at the beginning of Sect. 3.
- Lines 332-342: This material feels like it belongs in 3.2.1. I.e., this information on atmospheric transport seems like it belongs in the sub-section on atmospheric transport.
- Lines 349-350: This is the third or fourth location where the manuscript mentions forward runs. I would either (1) Consolidate all results/discussion of the forward runs within a single section, or (2) Divide this discussion into the sub-sections of Sect. 3 corresponding to different model parameters.
- Lines 3540-364: This text feels like it belongs in Sect. 3.2.1 with the results on atmospheric transport.
- Lines 368-370: I wasn't able to follow this sentence.
- Lines 402-425: The text feels like it belongs in Sect. 3.2.2.
- Lines 407-409: I'm not sure that I follow here. Who would apply this correction and in what circumstances?
- Lines 413-425: This text appears to repeat information in Sect. 3.2.3. I think the text in this paragraph belongs in Sect. 3.2.3.
- Figure 7: I suspect the differences due to meteorology could be much larger if using a non-ECMWF product. E.g., if using GEOS, MERRA-2, WRF, etc.
-
RC2: 'Comment on acp-2022-510', Anonymous Referee #2, 23 Sep 2022
In this study, the authors have examined the leading drivers of CO2 flux inversion by setting up an ensemble of eight inversions with different regional transport (STILT and FLEXPART), global transport (TM3 and TM5), and inversion systems (CSR and LUMIA) over Europe in 2018. The surface measurement of CO2 dry mole fraction is used to estimate the posterior. The results exhibit a large spread of CO2 flux estimation, majorly driven by the mesoscale transport model, rather than two other contributions. Overall, this study is interesting and worthy of investigation; however, some queries need to be addressed.
Major comments:
- I don’t think the title could well represent the entire manuscript. This study quantified the impacts of regional transport, lateral boundary conditions, and inversion systems on estimating biogenic CO2 flux. Even though the major contribution of regional transport was found and more results laid out in order to dig into details in context; still, the main part of the study is not only about atmospheric transport. Otherwise, you may want to use more regional transport models and focus majorly on the impact of atmospheric transport.
- Similar to 1), I recommend re-organizing the introduction part as well in order to give a better bridge to readers. The current introduction is mostly focusing on atmospheric transport (Line 30 – 57).
- The number of ensemble members for each suite representing mesoscale transport, global transport, and inversion system is too limited (two of each) with overlapped characteristics. For instance, TM3 and TM5 should have similar characteristics. Also, I doubt the result of differences of “meteo” shown in Figure 7. Those are all the ECMWF datasets. It would be better to use ECMWF and non-ECMWF to present the impact of using different types of meteorological data.
- Line 181-182: Why do two inversion systems use different uncertainty for the observation?
Minor comments:
- Line 25: “the largest impact seems to come from the models themselves” is too vague.
- Line 46 – 53: The accuracy is the meteorological data is not addressed in the main text.
- Equations 1-3: It’s better to add equations containing a sensitivity term (calculated by STILT and FLEXPART)
- Line 228 – 231: From figure 2, major corrections appeared over western and southern Europe, not around the observational sites.
-
AC1: 'ACs on RC 1 and RC2', Saqr Munassar, 21 Nov 2022
Dear Handling editor Susannah Burrows,
Dear Anonymous Referees,
first, we would like to thank you very much for your constructive commnets on our manuscript acp-2022-510. In the attachement, you will find our response to your comments. We have addressed the comments in a chronological order starting with Anonymous RC1 and Anonymous RC2.
We, indeed, thank you and appreciate your time you take to revise our manuscript.
Best regards,
Saqr Munassar, on behalf of the authors
Saqr Munassar et al.
Saqr Munassar et al.
Viewed
HTML | XML | Total | Supplement | BibTeX | EndNote | |
---|---|---|---|---|---|---|
307 | 112 | 11 | 430 | 21 | 2 | 3 |
- HTML: 307
- PDF: 112
- XML: 11
- Total: 430
- Supplement: 21
- BibTeX: 2
- EndNote: 3
Viewed (geographical distribution)
Country | # | Views | % |
---|
Total: | 0 |
HTML: | 0 |
PDF: | 0 |
XML: | 0 |
- 1