the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 18 Jun 2026
60 years of global air-sea CO2 flux variability
Abstract. The ocean carbon sink has absorbed approximately 29 % of anthropogenic carbon dioxide (CO2) emissions in the last decade, mitigating climate change. Over time, the ocean carbon sink has grown nearly proportionally to atmospheric CO2 concentrations. Natural variability also modulates the ocean carbon sink, but the decadal-timescale mechanisms of this variability are not well-understood. Using LDEO-Hybrid Physics Data, a unique observation-based product that merges observations with hindcast models, we assess the decadal variability of global air-sea CO2 fluxes for 1959–2024, with a focus on the dominant regional contributions. The dominant mode of decadal air-sea CO2 flux variability exhibits strong synchronous signals between the tropical Pacific and the Southern Ocean. This synchronicity is modulated by Pacific Decadal Variability (PDV) and the interannually varying El Niño-Southern Oscillation (ENSO) as measured by the Multivariate ENSO Index. When PDV and MEI indices are positive, the Southern Ocean experiences stronger westerly winds and deeper mixed layers. However, fully explaining the weakening of the Southern Ocean carbon sink in recent decades, which had predominantly negative PDV and MEI, will require additional constraints, particularly for biological processes.
- Preprint
(15599 KB) - Metadata XML
- BibTeX
- EndNote
Status: final response (author comments only)
- RC1: 'Comment on egusphere-2026-3104', Anonymous Referee #1, 22 Jul 2026
-
RC2: 'Comment on egusphere-2026-3104', Anonymous Referee #2, 02 Sep 2026
Review of “60 years of global air-sea CO2 flux variability” by Wong et al.
Wong et al. analyze the global and regional variability of air-sea CO2 fluxes based on the LDEO-HPD data set. The data set integrates observations and global ocean biogeochemical models using machine learning techniques, allowing to extend the air-sea CO2 flux data set back to the 1960s. The authors extract the dominant modes of variability from the CO2 flux data and interpret them based on climate modes of variability and their associated physical imprints.
In my view, the study advances the research field because: 1) the observation-based data set provides an extended estimate going back to 1960s (while most CO2 data sets start in the 1980s), 2) the results show a synchronous variability between tropical Pacific and Southern Ocean air-sea CO2 fluxes, mediated by climate teleconnections between El Niño and Southern Ocean. This result has not been (to my knowledge) extensively addressed in past studies.
The analyses methods are to my understanding correctly applied and the figures are clear.
However, I believe that the story is not yet fully mature. I think the authors should do more to contextualize their results into the framework of past studies and to further analyze the data to explain the mechanisms at play. Especially the connection between tropical Pacific variability and Southern Ocean air-sea CO2 fluxes, which is a prominent result of the paper, could be better explored.
This a summary of my main comments:
Literature: there are a few studies on the role of ENSO for Southern Ocean air-sea CO2 fluxes (e.g. Verdy et al., 2007) and many more on the connection between ENSO and the Southern Ocean physics (e.g. Ciasto et al., 2011; Huguenin et al., 2024; Shi et al., 2024). Referring to these (or other) studies could already help explaining the processes at play in this study.
Analysis: the authors conclude that the reasons for which a La Niña phase causes CO2 outgassing anomalies in the Southern Ocean remain unresolved. Have the authors considered analyzing changes in upwelling or in sea ice cover? Once a clearer picture emerges, I always think it’s a good idea to include a schematic showing the mechanisms at play. This is of course just a suggestion.
Abstract: somewhat related to my comments above, I also find the abstract too vague. It could be more informative, concrete, and quantitative. What is the main goal? What are the main results? Maybe, what are the implications?
Novelty: I have the impression that the question of how the PDO/ENSO impacts the Southern Ocean carbon uptake has not been extensively explored in past studies. This offers a big opportunity for the current paper to fill this gap. This novelty could be highlighted, maybe also in the title?
Title: I find the title somewhat too general. Most of the paper focuses on the role of the PDO/ENSO for the global and regional variability of air-sea CO2 fluxes. Maybe the authors could consider a more informative title such as: “60 years of global air-sea CO2 flux observations reveal strong synchronous behavior of the tropical Pacific and the Southern Ocean”. If the authors want to keep the title more general, I would at least mention that the study is observation-based, such as “60 years of observed global air-sea CO2 flux variability”.
Specific comments
Abstract
Line 1: “Ocean” instead of “ocean carbon sink”?
Line 9: define MEI before
Line 10: “weakening of the Southern ocean carbon sink in the past decades.” This is not what your observations show (Fig. 2)
Introduction
Line 30: growth rate of atmospheric CO2?
Lines 35-36: is there evidence of a weakened sink post 2000? Or was prior 2000 meant?
Line 45: do you really mean “carbon-climate feedbacks” (implying a climate effect on atmospheric CO2 which then feeds back on climate) or is the meaning here rather “climate impacts”?
Line 50: maybe a matter of taste, but I would suggest using a different definition of fCO2 (e.g. closely related to pCO2)
Lines 62-64: I’m not sure I follow the meaning of this sentence. Does the divergence between data sets identify a lack of understanding? I would also add that it highlights the methodological issues related to data sparseness and model deficiencies.
Lines 68-69: “As the climatology of these misfits is the dominant component of the total misfit for all models”: in the sense that the model-data misfit remains relatively stable in time so that the corrections can be used also to extrapolate in the past? I suggest rephrasing.
Lines 70-72: since Bennington et al. (2022) also deals with variability of the ocean carbon sink since the 1950s, could you outline which are remaining knwoledge gaps and in which way your study marks a step forward?
In general, on the Introduction: since a large focus of the study is the impact of climate modes on air-sea CO2 fluxes, could you give some background on past studies dealing with this topic?
Methods
Lines 101-106: it’s not so clear to me which variables are used for the reconstuction of missing fCO2, which are used for FCO2 computation based on eq. 1 and which are used to study the connections between climate variability and FCO2. Could you please specify?
Line 125: “northern boundary at 40N” is certainly a typo.
Line 129: I suggest merging Section 2.3 and 2.4 in one single one (Variability analysis)
Line 131: comma too much after FCO2?
Line 134: in which way are the time series detrended? Do you remove also the intercept so that the time series fluctuate around zero or not? Referring to Fig.1b,c would be helpful.
Line 137: “allowing for fluxes to be compared between regions that have different areas.” I’m not sure I follow. One can compare also mean (instead of intergral) values for regions having different areas. Is your meaning rather: “to take into account the fact that different regions have different area sizes”.
Line 146: could you explain how you treat the time series before computing the EOF (detrending, smoothing, spatial weighting, normalization?)
Lines 158-162: Do I understand correctly that the primed quantities are “simply” anomalies computed with respect to the lon-term mean? If yes, maybe one could introduce them this way.
Lines 168-171: I am bit confused on the procedure. Long-term means of both ∆fCO2 and wind speed are always computed on the raw data or just for wind speed? Why are ∆fCO2 deviations computed on the detrended time series but the same is not for wind speed? I imagine this is because in the case of ∆fCO2 we wish to remove the external CO2-effect, but for wind speed we want to retain the internal climate effects. I suggest to explain the reasons for using this procedure.
Line 164: the quantity FOC2’ is called later “flux-proportionate variability”. Could you explain here that you call it this way and why?
Line 165-166: typically, figure numbers apprear in order in the text.
Line 174-175: To compute the composites, I understand that you select all months in which the PC1-FCO2 is positive or negative. In past studies, a threshold is often used, so that only the months/years when the anomaly is more prominent are considered. Could you justify this choice?
Also, are the spatial pattern of EOF1-FCO2 and the FCO2 pattern revealed from composites (positive minus negative) consistent? (consider mentioning this)
Line 181: consider using the wording “annual climatology” rather than “multiyear mean”?
Line 198: Why using “While”? The two parts of the sentence do not appear to be opposing each other.
Line 208-209: since you are showing three panels on FCO2-EOF2, I think it would be good to spend a few more words on it, also because it explains quite some part of the variability (19%). Is it connected to some other variability mode? Interestingly, it involves an east-west dipole in the tropical Pacific, and reinforces the Southern Ocean outgassing. PC2-FCO2 shows large positive values around 2000, when an overall global weakening of the ocean carbon sink was detected, most prominently in the Southern Ocean. Some thoughts on this?
Lines 223-224 and 258: I don’t understand what is meant with the 1.5 cycles.
Line 238: since here you are mostly focussing on the low-frequency variability, is it appropriate to talk about la Niña, which is a higher-frequency event? Maybe rephrase?
Line 240: in several parts of the subpolar Southern Ocean (where most of your positive FCO2 anomalies are located), the ocean is actually a source of CO2 (Fig. A1A, right).
Line 242: In the Pacific sector, it looks like an equatorward shift
Lines 242-246: I find the chain of mechanisms difficult to follow. Based on the literature La Niña is associated with a stronger Amundsen Low (e.g. Huguenin et 2024), which would then correspond to stronger -not weaker- winds in the Pacific sector. Also, I don’t understand which mechanism should cause the stronger outgassing in the Pacific sector. Since we see concomitant MLD shoaling and cooling (two processes which should not lead to increased outgassing) a process that comes to my mind is increased upwelling. Maybe wind stress curl composites could give some information on this.
Line 245-246: not clear why higher wind speeds in the Pacific lead to cooling, while stronger winds south of Australia to warming.
Discussion
Line 278: why should the damped variability in the 60s and 70s suggest that “the magnitude of the sink in the 1960s and 1970s may actually have been weaker during that period”?
Line 281-284: I think this is an important discussion point that should be expanded. The decadal variability of the global carbon sink shown in this study is very pronounced, especially in the Southern Ocean. But this is the region where - because of undersampling- pCO2 products are thought to overestimate the decadal variability of the carbon sink. Could the decadal variability of the LDEO-HPD product be put into the context of other available products?
Line 303: “and then using the index values of years to compute the PC1-FCO2-based composite.” Grammar doesn’t sound correct
Line 305: it is a bit confusing that the text refers to positive PDO phases (which in turn are associated with warmer SSTs in the tropical Pacific), but in Fig. 7 the SST anomalies in the tropical Pacific are negative even though the caption says: “values are regressed onto the PDO index”, so shouldn’t we be seeing the anomalies associated to positive phases of the PDO?
Lines 328-329: “as wind speeds weakened from the late 20th to early 21st century”. This statement is confusing. We know from multiple evidence that Southern Hemispehre westerlies strengthened in the past decades. Could you explain better? Is the trend removed here?
Lines 328-329: Sallee et al., (2021) find a deepening of the summer MLD. Is this relevant in this context?
Lines 330-332: So, the concomitant shoaling of the MLD cannot explain the stronger outgassing, correct? Interesting is that the largest MLD shoaling anomalies and the largest outgassing anomalies in the Pacific sector are not located in the same location (as seen in Fig. A1). Is this important? Did you consider increased upwelling as a possible mechanism causing cooling, MLD shoaling and higher CO2 outgassing? Since the CO2 outgassing anomalies are amplified at high latitudes of the Southern Ocean, did you consider changes in sea ice coverage?
Lines 336-344: the discussion on the role of the biological pump is not clear to me. Could you streamline your message so that it can better help to interpret the results found here?
Figures:
A1 (right): consider a tighter colobar axis to highlight the colors
I suggest to indicate in all captions which variables are detrended and which not.
References:
Ciasto, L. M., and M. H. England (2011), Observed ENSO teleconnections to Southern Ocean SST anomalies diagnosed from a surface mixed layer heat budget, Geophys. Res. Lett., 38, L09701, doi:10.1029/2011GL046895.
Huguenin, M. F., Holmes, R. M., Spence, P., & England, M. H. (2024). Subsurface warming of the West Antarctic continental shelf linked to El Niño-Southern Oscillation. Geophysical Research Letters, 51, e2023GL104518. https://doi.org/10.1029/2023GL104518
Sallée, JB., Pellichero, V., Akhoudas, C. et al. Summertime increases in upper-ocean stratification and mixed-layer depth. Nature 591, 592–598 (2021). https://doi.org/10.1038/s41586-021-03303-x
Shi, Y., Liu, H., Wang, X. et al. Responses of the Southern Ocean mixed layer depth to the eastern and central Pacific El Niño events during austral winter. Acta Oceanol. Sin. 43, 1–14 (2024). https://doi.org/10.1007/s13131-023-2228-0
Verdy, A., S. Dutkiewicz, M. J. Follows, J. Marshall, and A. Czaja (2007), Carbon dioxide and oxygen fluxes in the Southern Ocean: Mechanisms of interannual variability, Global Biogeochem. Cycles, 21, GB2020, doi:10.1029/2006GB002916.
Citation: https://doi.org/10.5194/egusphere-2026-3104-RC2
Viewed
| HTML | XML | Total | BibTeX | EndNote | |
|---|---|---|---|---|---|
| 78 | 54 | 17 | 149 | 16 | 12 |
- HTML: 78
- PDF: 54
- XML: 17
- Total: 149
- BibTeX: 16
- EndNote: 12
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1
This article provides an assessment of decadal variability of global air-sea carbon dioxide fluxes. The authors use the LDEO-HPD, a data product, in order to evaluate the dominant modes of decadal variability of global air-sea CO2 fluxes and compare regional contributions to this variability. They find strong correlations between dominant modes of decadal flux variability and strong synchronous signals between the tropical Pacific and the Southern Ocean.
The overall paper article is thorough and robust in analysis. A few comments I think would improve the overall clarity and analysis:
In figure 2, it looks like starting in 1980 when data availability increases, the variability increases as well. This is for all basins, including Atlantic and Indian. While there is a small mention of this in the paragraph starting line 275, it focuses on the components in Figure 4. Can this be addressed some? If the time period is shortened, does that impact some of the conclusions? The uncertainty is shown as 1 standard deviation—does this vary over time?
Speaking of uncertainty, some of the standard deviations seem quite large (for example, figure 5 part A shows the FCO2’ to be 64, which if that were shaded, would span from a positive to a negative at multiple points. The paper makes the point that the fCO2 is overwhelming the other components, which isn’t necessarily the case with uncertainty taken into account during all time periods. I think the paper could be stronger with addressing some of this uncertainty in the discussion.
Additionally, there are several points in the paper where the reader is asked to compare two different figures, that are often on different axes and different formats. For example, line 304 the comparison of Figure 7 to Figure 6 (left column) left me a little confused about if I was comparing the right things and flipping across multiple pages. While this can sometimes happen, I think it would be nice if a few figures were adjusted with either an overlay or some alternate method of easier comparison within one figure.
Overall, this paper is a highly technical evaluation that provides an important contribution to the development of gap-filling methods and improved understanding of the global carbon budget and ocean carbon dynamics.
Line-level edits: