Review comments on 10.5194/egusphere-2024-632

Summary:

This study investigated the superobservation methodology for satellite observations, especially for chemical tracers. The paper discussed how to construct superobservations and appropriately set their uncertainty. The authors took several aspects into account and visualized their contributions to the resulting superobservations. They also performed data assimilation experiments and showed that their superobservation methodology resulted in improved forecasts compared to simple thinning.

The paper appears to align with the scope of GMD and may have significant implications for data assimilation studies involving satellite observations. Unfortunately, certain points were not clear to me and could benefit from clarification prior to publication. In addition, the paper’s readability was somewhat challenging, possibly due to its unconventional structure. Therefore, I would like to recommend Major revisions.

Major comments:

1. Clarifications required.

Several statements in the manuscript were unclear to me. Most of them could be my misunderstanding, but I would like to ask the authors for enhancing the clarity.

Lines 196–197: The sentence beginning with “Given the same uncertainty,” is confusing. Could this be re-written?

Lines 322–323: Unfortunately, I could not understand this sentence. A rewrite may be necessary.

Lines 510–511: This sentence was confusing. I guess this sentence compares Figures 12a and 12c, yet I could not find clear differences between these figures.

Lines 561–562: Why does an increase of covariance inflation result in a larger O–B?

Line 641: What does correlation length mean? The cutoff radius of localization?

Line 703: I am not sure why adding one improves the results. I understood that N > Neff but does this sentence mean N–Neff=1? Could you explain? Furthermore, I could not understand that “This is not consistent with the experimental data” in Fig. 8c.

2. The structure of the paper

The manuscript does not have a typical structure that contains an introduction, method, result, discussion, and summary. Although the motivation of the study should be clearly noted in the first section, section 3 explains the motivation of superobservation as well. While section 5 discussed contributions of three aspects on uncertainty, section 6 revisits the topic of uncertainty. I might misunderstand, but I would like to ask the authors to re-consider the structure of the manuscript. This could make the paper more concise.

Minor comments:

Line 92: Remove an extra “observational?”

Line 238: Does the left-hand side correspond to ds?

Line 246: Does A_S mean the superkernel?

Line 299: What are Θ and Θ₀?

Line 531: I would recommend explaining the experimental settings a bit more. It would be better to include what observations were assimilated, how long the assimilation window, how localizations were set, and how covariance inflation was achieved.

Line 551: The Χ² metrics seem similar with the consistency ratio (Dowell and Wicker 2009, 10.1175/2008JTECHA1156.1). Are they the same?