Recent variations in oceanic transports across the Greenland-Scotland-Ridge

By: Michael Mayer et al.

Understanding ongoing Arctic climate change is high on the international agenda, and the variability of ocean heat transport across the Greenland-Scotland-Ridge is important, as it is the main exchange between the Arctic (Nordic Seas) and the Atlantic. I find many aspects of the paper correct, and some of the conclusions drawn by the authors are well funded in my view. There is however one major issue that needs to be sorted, and there should be plenty of room for improvements in such a short paper.

The paper is a decent attempt at comparison of presently available ORA’s (Ocean Re-Analysis) and observations. The paper confirms three basic expectations and there are relevant citations given in these cases; a) AW volume inflow dominates GSR heat transport, and b) there is a larger seasonal volume/heat transport during winter, c) the AW heat transport dominates heat transport to the Arctic. There is also one new important change documented; that the large ocean-heat-transport pulse found by Tsubouchi et al (2021) for 2003-2012 has not continued.

The main problem with the paper that it is not comparing apples and apples. The ORA based values are (supposedly) net volume flux values calculated on the native grid and at the shortest distance between for example Iceland and the Faroes (Shown in Fig 3 a, but too small, and should be shown clearly in an early figure). This is not what is observed. The observations are drawn from a section running north from the Faroe Island (an OK map is found in Østerhus et al 2019 (Fig. 3)). I suspect that there is a large re-circulation of AW across the sill that is not observed. The observations document the flow north of the Faroes, and some of this water is likeley to cross the GSR again between the Faroes and Shetland. So the ORA based values are NET inflow, but the observations have an un-known portion of water that re-circulates and leaves shortly after entering the GSR.

There is actually one substantial paper that shows observations (and re-circulations) between the islands that is not cited; Rossby et al (2018). This is quite strange. In special as the net flow in Rossby et al (2018) is close to zero across the GSR. This is a substantial difference that needs to be explained, and one would think that the (fairly) high resolution ORA’s could provide an answer here. The net northward GSR volume transport found by Mayer et al of +0.7 Sv is comparable to what is found by others though. For example Smedsrud et al (2022) found +1.0 Sv as the centennial mean, and some net northward GSR flow must exist to Balance the net southward flow west of Greenland of about -1.7 Sv (Tsubouchi et al 2021). Bering Strait inflow is only about +0.8 Sv.

While the observations around the (Iceland, Faroes, Shetland) islands are subjected to fundamental problems regarding re-circulation, the ones along the Svinøy Section are not (Orvik 2022). Here the AW is well north of the GSR, and this would infact be a much better location for direct comparison. The Svinøy data appears consistent though; there are no trends in Volume or Temperature over the period 1995-2020, similar to Figure 1 e) f).  

The re-circulation of AW appears on my side to be the major explanation for the difference between the ORA’s and the observations. For example, there are no observations of southward AW flow east of Shetland. This was recently noted by Smedsrud et al (2022) where the simulated AW inflow mean (1900-2000) was as high as +9.5 Sv, but there is also a net AW outflow of -3.3 Sv. They had some hydrographic and current meter data to compare with, although it is a bit hard to understand those calculations.

The ‘high bias’ in volume transport from observations would carry onto the ocean heat transport. The Mayer et al (2022) estimate of a total surface heat loss of about 350 TW does not appear realistic. There are fundamental problems with atmospheric stratification and atmospheric boundary-layer parametrizations that need to be examined before such a high value can be accepted. If the ocean cannot transport this heat, it cannot leave the ocean surface. And the Barents Sea and the Arctic Ocean is the most rapidly warming ocean on the planet, so heat is accumulating too…

The main point is that the observations and ORA’s show different things, and you cannot compare them directly before you extract the flow where it is measured as well. With the increased resolution of the ORA’s it should now be possible to directly compare the observations with the simulated flow exactly where it is measured. Possibly this discrepancy can be solved, but until then I suspect that the ORA’s are just as correct as the «observations» and that the true OHT value is closer to 200 TW than to the «observed» 300 TW.

The paper was otherwise nicely written, and I have no detailed list of corrections. There is a proper map missing showing where the current meter and hydrographic sections are, and likewise where the grid cells for the ORA-based estimates are located. Most of the work done is fine, you just need to do a proper comparison, and discuss the other possibility that the ORA’s are more correct than you think.

Abstract: This is wrong: “The considered ocean reanalyses underestimate the observed 1993-2021 mean inflow of warm and saline Atlantic Waters of 8.0 ± 0.5 Sv by 7-23%”. What you have shown so far is that the simulated NET is lower than the observed inflow. There is an (unknown) volume of observed outflow, so the simulated NET could well be perfectly correct.   

Conclusion: This also needs to be modified; “The energy-budget based estimate from Mayer et al. (2022) suggests even higher net heat flux than oceanic observations, which confirms the underestimation of transports by the ocean reanalyses and indicates that observations possibly miss some influx of heat.”

It is good to have confidence in one’s own results, but as a scientists we must also be open for that they are wrong. In this case the observations are likely not showing what you state they do. They show inflow values, and does not give a proper estimate of the net flow.

Suggested new references:

Orvik, K. A. (2022). Long-term moored current and temperature measurements of the Atlantic inflow into the Nordic Seas in the Norwegian Atlantic Current; 1995–2020. Geophysical Research Letters, 49, e2021GL096427. https://doi.org/10.1029/2021GL096427            

Rossby, T., Flagg, C., Chafik, L., Harden, B., & Søiland, H. (2018). A direct estimate of volume, heat, and freshwater exchange across the Greenland-Iceland-Faroe-Scotland Ridge. Journal of Geophysical Research: Oceans, 123, 7139–7153. https://doi.org/10.1029/2018JC014250

Smedsrud et al (2022) Nordic Seas heat loss, Atlantic inflow, and Arctic sea ice cover over the last century. Rev. Geophysics.

This short paper presents an inter-comparison of transports in correspondence with the GSR section from several state-of-the-art reanalyses and simulations. It also includes a more detailed analysis of the recent decrease in heat fluxes after 2017, with some explanations and a proposal for sea level-based indexes. As such, it is an interesting piece of work, useful for the oceanic community, and as a reference for future validation and process-oriented studies. I recommend publishing this work after the authors address a few general comments.

1) Table 2 and the figures are not consistent: the figures report all the GREP members while the Table only the GREP ensemble mean. I recommend the authors report the values of all GREP members and their ensemble mean. Furthermore, the table should contain more diagnostics and include those already presented in the text (temporal standard deviation, correlation values, etc.). This could be very useful for future quantitative studies.

2) The inclusion of GLOB16 is not sound. This is a free-running model experiment, at a higher resolution than the other products, and a corresponding assimilation experiment does not exist. In practice, its inclusion poses more questions and does not help to answer any question. Using an ensemble of reanalyses helps to identify the envelope of uncertainty; using GLORYS12 helps address the impact of spatial resolution. But it is not clear whether results from GLOB16 can be ascribed to different resolutions? different model configuration? lack of observational constraints? different initial conditions? I miss the point for including it.

3) Section 3.3 is not much conclusive either. One suggestion from the authors is that SPG strength induces a delayed response in the AW temperature. However, lagged correlations and similar tools could be used to provide a quantitative answer about the fitness of the SLA-based indexes for capturing the AW inflow variability.

Minor point

In line 46: I suggest referring to "the so-called Arctic Mediterranean" as it might be non-obvious for a general readership