Fundamentals of data assimilation applied to biogeochemistry

Rayner, Peter J.; Michalak, Anna M.; Chevallier, Frédéric

doi:https://doi.org/10.5194/acp-19-13911-2019

Articles | Volume 19, issue 22

https://doi.org/10.5194/acp-19-13911-2019

© Author(s) 2019. This work is distributed under
the Creative Commons Attribution 4.0 License.

Special issue:

Data assimilation in carbon/biogeochemical cycles: consistent...

https://doi.org/10.5194/acp-19-13911-2019

© Author(s) 2019. This work is distributed under
the Creative Commons Attribution 4.0 License.

Articles | Volume 19, issue 22

Review article

|

19 Nov 2019

Review article |

| 19 Nov 2019

Fundamentals of data assimilation applied to biogeochemistry

Peter J. Rayner, Anna M. Michalak, and Frédéric Chevallier

Download

Final revised paper (published on 19 Nov 2019)
Preprint (discussion started on 30 Oct 2018)

Interactive discussion

Status: closed

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

- Printer-friendly version

- Supplement

RC1: 'Review of "Fundamentals of Data Assimilation applied to biogeochemistry"', Anonymous Referee #1, 01 Dec 2018
RC2: 'Review of Rayner et al., Fundamentals of Data Assimilation applied to biogeochemistry', Anonymous Referee #2, 06 Dec 2018
RC3: 'Review comments on " Fundamentals of Data Assimilation applied to biogeochemistry"', Anonymous Referee #3, 22 Dec 2018
AC1: 'Response to comments/reviews', Peter Rayner, 08 Jun 2019

Peer-review completion

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

AR by Peter Rayner on behalf of the Authors (10 Jun 2019) Author's response Manuscript

ED: Referee Nomination & Report Request started (21 Jun 2019) by Marko Scholze

RR by Anonymous Referee #1 (11 Jul 2019)

Suggestions for revision or reasons for rejection

The authors have addressed all my concerns. I noticed some minor inconsistencies and formatting issues, which are noted below. I recommend the authors to carefully go through the manuscript to ensure the text is free from errors.

1. Page 5, Table 1: The definitions for the Jacobians of H and M (matrix H and matrix M) seem to be wrong (e.g. nabla H instead of nabla y).

2. Page 5, Table 1: Capitalize "Hyperparameters".

3. Page 6, Line 16: Not necessary to cite Krol et al. (2005) here.

4. Page 6, Line 28: Replace ";" with "." to be consistent.

5. Page 8, Lines 4-7: This paragraph seems to suggest that it is fine to over-fit the data, e.g. by including more target variables, as long as we post-treat the target variables (e.g. through aggregation). I am not sure I agree with this statement. For example, if there are more target variables than observations, the solution may be extremely sensitive to noise in the data, which could lead to biases even in the aggregated quantities.

If the authors disagree, it would be helpful to have some references that show that including too many target variables is not problematic, even for highly nonlinear problems.

6. Page 8, Line 14: Change "See Section 7.1" to "see Section 7.1".

7. Page 8, Line 15: Fix citation format.

8. Page 8, Line 19: "how to include the uncertainty in the choice in their assimilation". Consider adding "of target variables" after "in the choice".

9. Page 9, Lines 13-15: "Choosing a random positive s is far more likely to increase... Solving for log(s) as a target variable means that we can make increase or decrease equally likely". Maybe mention what distribution s is sampled from.

10. Page 9, Line 25: "study how the uncertainties of the fluxes change". Replace "fluxes" with "target variables".

11. Page 10, Line 12: "PDFS". Change to "PDFs".

12. Page 10, Line 20: Fix formatting for p(y|yo).

13. Page 12, Line 8: "(variances, correlation lengths etc) choices of dynamical ...". Add period and comma ("(variances, correlation lengths etc.), choices of dynamical ...)").

14. Page 12, Line 22: Fix citation.

15. Page 12, Line 25: "TRANSCOM". Change to "TransCom" to be consistent with the rest of the manuscript.

16. Page 15, Line 3: "section 4". Capitalize "Section".

17. Page 15, Line 9: "this generates a Gaussian". Consider capitalizing "This" or change to "which".

18. Page 15, Line 20: No need to define "TransCom" again.

19. Page 16, Line 18: "eg it does not". Change to "i.e., it does not".

20. Page 19, Line 9: "Particle filter". Capitalize "Filter".

21. Page 19, Line 21: "time window". Capitalize "Time Window".

22. Page 22, Line 2: "(section 7.1)". Capitalize "Section".

23. Page 23, Line 17: "this is true". Capitalize "This".

24. The authors mentioned in their reply that "We now argue in an expanded section on future developments that so-called “dual state” approaches that seek to constrain the state and parameters of the model should be more widely used in biogeochemical problems." However, I could not find such a statement in Section 9 (or elsewhere in the manuscript).

Hide

ED: Publish subject to technical corrections (29 Jul 2019) by Marko Scholze

AR by Peter Rayner on behalf of the Authors (06 Aug 2019) Author's response Manuscript

Short summary

This paper describes the methods for combining models and data to understand how nutrients and pollutants move through natural systems. The methods are analogous to the process of weather forecasting in which previous information is combined with new observations and a model to improve our knowledge of the internal state of the physical system. The methods appear highly diverse but the paper shows that they are all examples of a single underlying formalism.