Overall impression

The paper investigates the physical and biogeochemical characteristics of mesoscale eddies at the surface of the Southern Ocean – a region of global importance for heat and carbon exchange and biogeochemical cycles, concurrently a region dominated by eddies. This study involves many novel aspects compared to previous studies and tackles relevant topics for the community. It distinguishes between warm-core anticyclonic eddies (AEs), cold-core AEs, cold-core cyclonic eddies (CEs), and warm-core CEs (termed here as ‘normal’ and ‘abnormal’ AEs and CEs). At the same time, the discussion lacks some depth and many aspects of the methods and results/discussion remain unclear. There are also many figures which are only discussed very briefly and could be moved to the Supporting Information. These issues should be addressed before publication.

General comments

1) From the introduction, it’s not entirely clear what’s new about this study (one finds the information eventually, but it’s quite hidden and only becomes apparent later). What’s new about this study compared to previous work should really be the focus of the introduction.

E.g., L86: Mention that the Frenger et al. studies investigated the vertical structure, while this study is only considering the surface. The same paragraph (from L86) also reads as if the only difference between the Frenger studies and this study is that this study differentiates between ‘normal’ and ‘abnormal’ eddies. However, there are many other differences (surface vs. interior; which parameters are considered; method of eddy detection…).

2) My biggest concern: McGillicuddy, Gaube, and others have pointed out that eddy-induced Ekman pumping results in the opposite signal compared to regular eddy pumping and that eddy-induced Ekman pumping is usually weaker, but can be significant, especially in regions with large wind stress. Thus, when seeing cold core AEs or warm core CEs, I would assume that there, eddy-induced Ekman pumping dominates. However, in this paper it is framed like a mystery that some AEs have cold cores, and some CEs have warm cores (termed ‘abnormal eddies’). Later, the study exactly finds this, at least in the analysis with SST (Section 5.1, esp. L278). Thus, I recommend rephrasing the storyline that cold-core AEs and warm-core CEs are likely to be dominated by eddy-induced Ekman pumping.  

3) The discussion should be deepened. Currently, Sections 4 and 5 mostly show the results with a lot of figures, and Section 6 (Conclusions) is mostly a summary of these findings. I’m missing a more in-depth discussion of what this now all means and how it matters. One thing to focus on especially is the surprising finding that when considering SST anomalies, eddy-induced Ekman pumping dominates in certain eddies, while for the other variables, different processes dominate. How can the same eddies pump DIC-rich water upwards without pumping cold water up? Once the discussion has been deepened, the abstract and conclusions can then also mention some more from the discussion. Right now, the abstract and conclusion sections are quite descriptive of the results but don’t tell us much about their significance.

4) Consider moving Fig.1 and 3 and Table 1 and 2 to the Supporting Information, they don’t add much new information. Fig. 5 is only discussed with one sentence (L192) and could also move to the SI or be discussed in more depth.  Similarly, Fig. 7 is only very briefly touched upon and can move to the SI.

Specific comments

It is not immediately clear that the study only focuses on surface properties. This could be added to the title and should be clearer in the abstract and introduction.

L10: add ‘horizontal surface’ before ‘composite’

L73: Are the signals also different when the seasonal signal has been removed? I.e., are the anomalies computed based on mean annual reference values, or on a monthly climatology? I have a feeling that if a monthly climatology is used as a reference, the eddy anomalies might not differ so much anymore by season. This should be mentioned/discussed.

L119: Be specific that the DIC is gap-filled in this step.

Section 2.2 and 2.3: be clearer that those datasets used were created by previous studies. It currently reads ambiguously if this was done during this study or if the data is from previous work.

L127: Is ‘This dataset’ referring to Landschuezter et al. 2014, or to the Liu et al. 2021 product that is used in this study?

• If it’s referring to Landschuezter: as that product has been used so widely, why did you not use that product? What’s the benefit of using Liu et al?
• If it’s referring to Liu et al: rephrase the sentence so that it’s clearer (but then some of the references are wrong as they were published before 2021…)

Section 2.3: Mention how the eddy detection method differs from other, more commonly used approaches, such as the AVISO eddy database (newest version: Pegliasco et al. 2021), and why it was preferred. One could have used the AVISO eddies and classified the eddies into normal and abnormal based on their SST signature (e.g., AE with cold SST anomaly is CAE...).

L144-149: I think this part of the paragraph still belongs to section 2.3.

L149-153: Needs a more in-depth description of how the composite eddies were made.

L171: Be explicit about how it differs from the method by Gaube et al. 2015.

Section 3.2: Add a reference for the methods to obtain the eddy-induced Ekman pumping.

L239: Discuss why we want to know how the pattern differs from the pCO2 pattern. I would have found it more interesting to see the pattern differences between normal and abnormal eddies, but there could be a reason why you chose this.

L255: Mention why stirring is not a process (we can see it in the plot, but it needs to be discussed).

Generally: Personally, I would not use the terms ‘normal’ and ‘abnormal’, as everything is normal and within the expected physics (when considering eddy-induced Ekman pumping), but this may be a personal choice. Maybe ‘regular’ and ‘unusual’ fits better, as warm-core AEs and cold-core CEs are a lot more common than cold-core AEs and warm-core CEs, but I’m nit-picking now.

Fig. 2: Consider using a sequential colormap. Specify the latitude where the white region starts (65S?). Most of the currents and topographic features are not referred to in the text and can be removed.  

Fig. 4 (and the following figures): Add in the caption what the magenta boxes are.

Fig. 6: Why are there some warm spots in cold eddies, and cold spots in warm eddies? By definition, the SST anomalies should be cold in cold eddies, and warm in warm eddies.

Fig. 8: Ensure all SSIMs have the same number of decimals.

Technical corrections

Throughout the document: change biochemical to biogeochemical.

It’s a good habit to discuss the findings of this paper in the present tense and refer to previous studies in the past tense. E.g., L9: change to ‘we analyze’ (instead of ‘we analyzed’); same throughout the whole document.

L12. I know many studies do this and it is a personal choice, but I dislike sentences with brackets for multiple things. Consider writing it out for each, e.g., ‘dominated by DIC anomalies in regions with larger magnitudes of DIC anomalies and dominated by SST anomalies in regions with smaller magnitudes’. Same throughout the whole document.

L22: existing (not exiting)

The font in some figures is very large.

The manuscript analyzed the biochemical influences of mesoscale eddies (including normal and abnormal eddies) in the Southern Ocean, by using machine learning and multi-source marine dataset. The manuscript estimated chlorophyll (Chl) and dissolved inorganic carbon (DIC) contributions to pCO₂, and found their seasonal variations. These results are interesting and are of vital importance on global biogeochemical cycles and the climate change. However, some description about methods/data are ambiguous and few conclusions need to be further discussed.

Major questions:

1. There are some other methods to identify abnormal eddies, such as using potential density and directions of geostrophic current. They are supposed to be introduced in the introduction, and point out why authors choose the method of SSTA.
2. The methods to derive pCO₂ from Chl, SST, DIC and other variables are supposed to be introduced with more descriptions or equations.
3. The method to define and identify abnormal eddies should be introduced in detail even if the authors cited the paper of Liu et al., 2021. Did they identify abnormal eddies according to SSTA>0/SSTA<0 within eddy boundaries/cores? How did they distinguish AEs and CEs just according to SSTA?
4. The descriptions about eddy dataset and identification are very poor. In line 139, the authors mentioned “the ground truth data set”. What’s the ground truth data set of eddies? Is it produced by the authors or a public dataset? That’s important to the verification.
5. How did authors match daily eddy dataset with monthly DIC and pCO2 temporally and spatially when doing composite analyses? Temporally, is eddy at JAN. 31st matched with DIC of JAN. or Feb. data? Spatially, is DIC data used within eddy boundaries or eddy cores?
6. Taking CEs for example, commonly, upwellings are thought to transport cold water to the sea surface, as well as richer nutrients at the same time. Therefore, CEs often show lower SST and higher chlorophyll. The conclusions from Figure 8 show SSTA within abnormal eddies are dominant by Ekman pumping. However, the chlorophyll anomalies of abnormal eddies are attributed to eddy pumping. The conclusions are contradictory to each other. If they are reliable, what’s the mechanism leading to contrasting vertical process on SST and chlorophyll respectively? Therefore, discussions of lines 279-280 and 295-296 need more explanations. Besides, line 279 should be “eddy-induced Ekman pumping”.
7. The manuscript is supposed to evaluate the accuracies of abnormal eddy identification method, which can combine with Argo profiles via temperature and potential density. At the same time, it should point out the method improvement in future

Minor questions:

1. Lines 31-32: Authors point out that eddies have influences on “biochemical parameters”. While, the listed references are both about chlorophyll, which is a biological parameter. References about chemical parameters should be introduced.
2. Lines 37-39: Rotations of eddies are related to the hemisphere. It should illustrate which hemisphere is talked about.
3. Line 56: How about eddy influence on chlorophyll during wintertime with deeper mixing?
4. Lines 107-108: The expression of OI-SST should be in agreement.
5. Line 166: JMA is suggested to be introduced as Japan Meteorological Agency.
6. Line 178: What are the denominators when calculating eddy frequencies? It should be expressed more clearly.
7. Lines 186-187: The conclusion is true in South America, but not evident in the south of Australia.
8. Line 285: How to understand “eddy trapping has little influence on Chl-a”? Please give more descriptions to explain it.
9. Lines 303-314: Are those conclusions for summertime still “dominant”? The magnitudes seem similar for summertime.
10. Figure 11 is suggested to be shown in wintertime and summer time respectively, based on which Figure 8, Figure 12, and Figure 13 can be better discussed.
11. In Figure 4, lines 558-599, the authors mean blue and red colors in the right column. However, blue and red colors are shown in each sub-figure, which is misleading.
12. Figures 4d and 4e show that abnormal eddies occur along fronts, where eddies are active, and along offshore areas where accuracies of altimeters are low. It is suggested to show ratios of abnormal eddies to normal eddies (WAEs/CAEs, CCEs/WCEs). Will the abnormal eddy signals offshore be amplified offshore? What are the mean depths of clustered abnormal eddies? It should be cautious with eddies shallower than 1000 m.
13. Figure 6. The abnormal eddies are identified from SST so the SSTA of Figure 6 is regular. The other three parameters are very noisy. The magnitudes of chlorophyll and pCO₂ signals induced by abnormal eddies are even higher than normal eddies, which are contrasting with eddy amplitude comparisons. Why?

This is an important manuscript. The Southern Ocean occupies a huge area of the global ocean, therefore, plays a key role in global ocean water cycle and global change. In the manuscript, the authors analyzed mesoscale eddy activities and their roles in re-distributions of ocean dynamic amd biologic factors, which are  important contributions to the Southern Ocean research. The manuscript was prepared. well. Thus, this reviewer recommends the submission to be acccepted for publication.