Review of essd-2025-127

"An ensemble-based coupled reanalysis of the climate from 1860 to the present (CoRea1860+)"

by Yiguo Wang et al.

submitted to Earth System Science Data

https://doi.org/10.5194/essd-2025-127

corresponding data archive: https://doi.org/10.11582/2025.00009

General comments:

The authors present an analysis of their comprehensive reanalysis CoRea1860+, which is driven by SST observational products and covering almost the whole of the historical time period (in a climate science sense) from 1860 until the present. The manuscript in its present form is a rock solid data description manuscript, which should be published in ESSD. The corresponding dataset under https://doi.org/10.11582/2025.00009 is accessible to me.

I do have some comments below to improve or clarify some of the authors methodology and statements. I pledge the authors to consider them.

abstract l.15f

"Furthermore, CoRea1860+ aligns well with the other datasets for surface air temperature, precipitation, sea level pressure, and 500 hPa geopotential height, especially in the tropics where air-sea interactions are most pronounced."

This statement appears a bit too optimistic, given the coarse resolution of the stratosphere and the missing atmosphere observations. In particular the authors' results concerning 500 hPa geopotential height, as good as they may be given the fact that only "SST" was assimilated, could be well improved with just some "truly" coupled reanalysis incorporating the few atmosphere observations, which are available prior to 1950.

l.129f

"This version incorporates emissions and new aerosol-cloud interaction schemes..."

l.132f

"the model employs the version 7 coupler"

l.135

"The atmosphere component consists of 26 hybrid sigma–pressure levels, extending up to 3 hPa."

resolution of stratosphere?

QBO?

l.138

"The NorESM used in this study is forced by CMIP5 historical forcings before 2005,"

Why not CMIP6 external forcings?

Expected differences from CMIP6 forcing?

l.148ff

"Since 2011, the monthly SST data from the Optimum Interpolation SST version 2 (OISSTV2, Reynolds et al., 2002) are assimilated, because HadISST2.1 is only available until 2010."

Why not HadISST, which goes through until present?

l.161f

"The SST data in the regions covered by sea ice are not assimilated. These regions are identified using the sea ice mask in HadISST2.1 or OISSTV2."

How does the model's observed sea ice extend comply with the observed sea ice? Biases?

l.186f

"The climatology reference period is 1950-2009 covering a long observation-rich period."

This time period is strongly influenced by climate change signal. Would other time periods be a better choice?

I presume that only the satellite era represents an "observation-rich period" in terms of global SST coverage. Please comment on that.

Further below, the anomalies of OHC and AMOC are against 1950-2010.

l.243

"RAPID"

Please add reference Moat et al. (2024) in this paragraph.

l.299, Figure 1

Please make the legend smaller and moved up a bit to reveal all of the time series.

l.337ff

4.2.1 Ocean heat content

Within this subsection, it is a bit confusing to jump forth and back several times between 0-300m OHC and 0-2000m OHC. Please consider to re-order the paragraphs so that 0-300m comes first and 0-2000m OHC second.

l.384

Figure 3

Please consider to add the corresponding time periods as labels in plot, e.g. over Antarctica.

l.405

4.2.2 Atlantic meridional overturning circulation

Please add a statement on absolute values for the AMOC, or the mean biases of the systems, in particular in context with Figure 4.

Please consider to depict AMOC cells as well, e.g. time mean and yearly standard deviation 1950-2010.

l.454

Figure 6

Please consider to add the corresponding time periods as labels in plot, e.g. over Siberia.

l.532

Figure 9-11

Please consider to add the corresponding time periods as labels in plot, e.g. over Antarctica.

l.550ff

5 Conclusions

Please shortly discuss "shortcomings" or expectations in terms of missing atmosphere observations and the (I presume) rather coarse resolution in the stratosphere, which presumably inhibits QBO?

l.562

Figure 9-12

Please consider to add the corresponding time periods as labels in plot, e.g. over Antarctica.

l.571ff

"CoRea1860+ demonstrates a reasonable representation of the variability of OHC across different regions and depiction of AMOC variability."

Please comment on if the OHC characteristics could actually be expected to be well met by this assimilation approach, and on the absolute values of AMOC in context with the variability characteristics.

l.587ff

"CoRea1860+ demonstrates reasonable variability in sea ice concentration and extent (SIE) for both the Arctic and Antarctic regions."

Similar comment as before: could the well met sea ice characteristics be expected or not from this assimilation approach?

l.599f

"demonstrates reasonable atmospheric variability due to robust air-sea and air-sea ice interactions."

How does "reasonable" in this context compare with "reasonable" in the context above with OHC, AMOC, sea ice? The atmospheric characteristics are probably well worse than one would expect to get from a "truly" coupled RA including atmospheric observations?

This paper presents a new and interesting contribution to the ocean reanalysis landscape with the development of CoRea1860+, a long-term coupled ocean reanalysis that assimilates only sea surface temperature (SST) observations. By avoiding the assimilation of subsurface data, which is more sparse in space and time, the product is designed to maximize temporal consistency, making it particularly well suited for studies of climate variability on decadal and longer timescales. The manuscript provides a clear and thorough evaluation of CoRea1860+, demonstrating how it compares with both existing ocean reanalyses and observational datasets. The approach is novel and complementary to other reanalyses that rely on more comprehensive data assimilation strategies. 

I recommend minor revision, primarily to encourage the authors to expand their discussion and analysis in a few areas where the strengths of CoRea1860+ could be further leveraged—for example, by using its extended temporal coverage to explore decadally paced modes of variability or by more explicitly investigating the ocean-forced atmospheric variability. These two aspects are already well acknowledged in the paper, but could benefit from deeper exploration to further highlight the utility of the dataset.

Specific comments

• Line 10: Change “areas” to “aspects”.
• Line 27: Change “Retrospective analysis… is” to “Retrospective analyses… are”.
• Line 66: Rephrase “and the discard of the first two years of each streamline” to “from which the first two years are discarded”.
• Lines 76-78: It is not very clear what distinguishes semi-coupled from fully-coupled assimilation. Could you elaborate a bit more?
• Lines 87-89: The way it is introduced, it seems that the problems of full-field initialization (e.g. model drift) are specific to uncoupled reanalyses. This is misleading as I would actually expect uncoupled reanalyses to be less subject to model drift due to the anchoring effect of the boundary conditions. Since full-field initialization is also used in coupled reanalyses, I would introduce it as a standalone concept that is opposed to anomaly initialization. 
• Line 113: Change “reconstruction” to “reconstructions”.
• Lines 147-150: Could you comment what impact this discontinuity in the assimilated product is expected to have in your reanalyses? Could you also motivate the choice of this reference dataset and explain in particular why you didn’t consider HadISST1 which extends up to present?
• Lines 312-313: I wouldn’t say that the total error is stable as it shows a clear decreasing trend over time.
• Figure 2 and paragraphs in Lines 356-376: Can you motivate the choice of the two polar regions as regions that deserve specific validation? Is it because they have been more poorly observed before the satellite era?
• Lines 359-360: I don’t see why having more data available should traduce in higher variability. Isn’t the higher variance in this region with respect to the global ocean explained by the polar amplification phenomenon?
• Lines 362-363: The apparent good agreement after the 1950s might be artificial, as this is the period that you chose to compute the anomalies. If you choose a different one (the early 20th century or the whole dataset, the periods of agreement/disagreement might be very different.
• Lines 400-401: How much of this good agreement is because of the long-term trends. Have you recomputed the correlations with detrended timeseries?
• Section 4.2.2: It is important to point out that CoRea1860+ cannot describe the observed Ekman-driven component, as it doesn't include any wind forcing. That might explain some of the year-to-year discrepancies with the other reanalyses. It might be worth repeating the comparison but removing first the Ekman component (e.g. as in Baehr et al 2004).
• Line 425: Another relevant paper that compares the AMOC mean state and variability across different ocean reanalyses is Jackson et al (2019).
• Lines 429-432: I wouldn't say that CoRea1860+ compares well with the other reanalyses after 2005, as the year to year variations are completely off (for the reasons explained above). Interestingly, it does show a good agreement with RAPID.
• Line 440: Change to “Superimposed on this decline there is”
• Line 442: You should mention that the decline in the early 20th century is not captured in HadISST2. 
• Line 451: What is the agreement (i.e. ACC) when the trend is removed? Also, to me it doesn’t look like the datasets agree so much. The areas of significant correlation values change from one dataset to another, and the values are not so high. That’s an indication of high observational uncertainty across the different records. 
• Figure 6: Could you indicate what the gray hatched areas indicate? Is it a lack of statistical significance? And how is that significance computed? Note that to make a fair assessment the effect of the time-series autocorrelation on the effective sample size needs to be corrected (Bretherthon et al 1999). Also, the color scale assigning white colors to the -0.1-0.1 range is a bit misleading, as it doesn’t distinguish areas (e.g. the Okhost Sea in panel a) of low correlation from areas with no sea ice. 
• Line 509: To me the agreement over land areas is much less clear than for the ocean, with some areas like North America, Russia and Argentina with low and even negative correlations.
• Line 529-530: How do you explain this better agreement in winter than in the summer? In winter, the Arctic is largely covered by sea ice so the direct influence on evaporation (and through it on precipitation) is expected to be reduced.
• Line 533: The correlations are not high and significant everywhere, so I would directly say which regions show a good agreement and which ones don't. 
• Lines 556-558: Given the suitability of CoRea1860+ to look at both long-term climate changes (e.g. trends) and the slow modes of internal climate variability, I miss some analyses in the paper that assess both aspects. One example would be checking to what extent the low-frequency changes in the North Atlantic Oscillation have been forced by the ocean, by verifying if CoRea1860+ is able to reproduce them. 
• Line 571: Can you rephrase it? It’s not clear what you mean by depiction of AMOC variability in the context of the first part of the sentence.
• Line 576: What do you mean by open-water regions?
• Line 587: The authors repeatedly use the expression “reasonable variability” throughout the manuscript, which is ambiguous. I recommend using more precise terminology. Additionally, providing some quantification—such as the correlation coefficient and its p-value—would enhance clarity and support the interpretation.
• Line 610: It would be beneficial for the article to conclude with a prospective paragraph on the intended use of the CoRea1860+ at NERSC/UoB, also highlighting potential applications for other research groups to encourage its adoption.

References:

Baehr, J., J. Hirschi, J.-O. Beismann, and J. Marotzke (2004), Monitoring the meridional overturning circulation in the North Atlantic: A model-based array design study, J. Mar. Res., 62, 283–312, doi:10.1357/0022240041446191.

Bretherton, C. S., M. Widmann, V. P. Dymnikov, J. M. Wallace, and I. Bladé, 1999: The Effective Number of Spatial Degrees of Freedom of a Time-Varying Field. J. Climate, 12, 1990–2009, https://doi.org/10.1175/1520-0442(1999)012<1990:TENOSD>2.0.CO;2.

Jackson, L. C., Dubois, C., Forget, G., Haines, K., Harrison, M., Iovino, D., et al. (2019). The mean state and variability of the North Atlantic circulation: A perspective from ocean reanalyses. Journal of Geophysical Research: Oceans, 124, 9141–9170. https://doi.org/10.1029/2019JC015210.