| 03 Sep 2026
Simulations of water transport over the Iceland–Scotland Ridge improved by local grid refinement around the Faroe Islands
Abstract. In the region between Iceland and Scotland, the warm water currents towards north and the cold water currents towards south are crossing a submarine ridge. Due to this complex bathymetry it is challenging for large scale ocean models to simulate the water flow realistically. High-resolution models perform better but are computationally more expensive. In this study, we investigate whether refining the model grid locally between Iceland and Scotland can improve the simulation of both the warm Atlantic water inflow and the cold deep overflow. The model grid is refined from 10 km to 3.3 km by using a two-way nest in a regional setup of the NEMO-TOP ocean model. A dye passive tracer is used to clarify the course of cold overflow water masses. Comparing the simulated water transports with in-situ observations around the Faroe Islands illustrates the benefit of the nest: 1) With the nest, more Atlantic water flows over the Iceland–Faroe Ridge and less through the Faroe–Shetland Channel, hence correcting the under-/overestimation seen in the coarse resolution model. 2) Overflow of deep water in the Faroe Bank Channel is increased by the nest to 2.14 Sv, which resembles well the observed value of 2.22 Sv. Additionally, the dye passive tracer simulation reveals that the nest enables the overflow water masses to decent to greater depths. These results indicate that the use of a local grid refinement by a two-way nest in a confined region is a promising and cost-efficient approach for improving the simulation of large scale water transports in the Atlantic.
This paper provides a nice demonstration of implementing a mesh refinement to better resolve complex small scale processes that are otherwise poorly represented in numerical models. It presents a clean comparison of the parent model resolution, the model with a factor 3 refinement local to the Faroe Bank Channel overflow region. The implementation of a dye tracer to track the water that overflows the Iceland-Scotland Ridge is a nice approach. However, here and throughout, the analysis and comparisons shown are rather superficial and I think there is a lot more potential for analysis that could quantify the improvements afforded by the nested domain. For example, volume census approaches could demonstrate the effect of improved resolution on the downstream water masses, total volume of overflow water vs the entrained volume - to what extent does the nested configuration preserve the overflow waters?
The authors have stated the code revision, bathymetry, initial and boundary conditions, but they do need to ensure that all files necessary to enable the reader to reproduce the results are made available. This could be through archiving the ancillary files etc. or by making available the scripts to produce them. The ocean and ice namelists for the simulation should also be archived.
Minor comments:
Lines 45-52: Discussion misses some important recent literature, for example that of VIKING20X (Biastoch et al., 2021) and its predecessor VIKING20.
Lines 54-55: Due to the way AGRIF works, very small nests are not computationally efficient. This is because the computational resource allocated to the parent domain must be distributed over the nest. If the nest is small compared with the parent this limits how effectively the model can scale.
Line 74: "resolution of 10 km". If I understand correctly, the parent model is based on a global 1/12 degree domain. In your regional configuration the largest grid cell is around 7 km, and in the region of interest (60-65N) the grid cells are 4-4.5 km. This would also make your nested mesh closer to 1.5 km. Please check.
Line 77: How was the nest bathymetry interpolated from ORCA12? Interpolation, rather than generation from raw bathymetry such as GEBCO, misses the opportunity to represent finer-scale bathymetric features that are potentially very important for overflows. Did you consider this?
Lines 81-82: Did you feed this bugfix back into the NEMO code? If not, please do so! Suspected bugs can be reported on https://nemo-ocean.discourse.group/.
Figure 2: It would be useful for comparison to show anomalies from the observations. Also, metrics such as RMSE could be calculated to quantify the improvement made by introducting the nested domain.
Figure 3: Please add some uncertainty estimates to the observations in 3b.
Figure 4: Please add some uncertainty estimates to the observations in 4b.
Section 3.5: Where you are able to identify cells which contain some overflow waters from the dye tracer, it would be really interesting to see how the inclusion of the nest has changed the watermass properties downstream of section S. For example, could you plot T-S volume census for the water south of the ridge, maybe for cells above a threshold concentration? Can you determine whether the refinement has resulted in quantifiably denser water, or water with different alpha/beta properties downstream of the overflow?
Lines 254-258: I'm not sure I follow this argument. You seem reassured that there is agreement between the model and obs, but this doesn't mean that either represent the truth.
Line 298: Rather than indicating that the nested model is more dynamic, you should calculate and plot the EKE for both.
Line 314: I disagree. Approaches such as excavating bathymetry can reduce biases arising from poor overflow representation, but I would not regard this the same as saying that they simulate the overflow well. There are overflow parameterisations that can be implemented (see e.g. Danabasoglu et al. 2010 and references therein).
Line 345: Rephrase/clarify. I think you mean that the volume of dye tracer in the overflow waters increases gradually.
Table A1: Please share your configuration, particularly the actual namelists used and any scripts for preparing the ancillary files. Where you list "Bulk formula" against the Surface boundary condition, it is important to know which, and whether you used e.g. relative or absolute winds.
Typos:
Abstract lines 1, 2: replace "towards north" and "towards south" with "northward" and "southward".
Abstract line 11: "decent" > "descend"
Line 27: "where it confines to" > "where it is confined to"
Line 268: remove "also"
Line 277: remove "anyway"
Line 323: "takes only" > "only takes"
Line 403: "layer like presented" > "layer like the one presented", or similar.
References:
Biastoch, A., Schwarzkopf, F. U., Getzlaff, K., Rühs, S., Martin, T., Scheinert, M., Schulzki, T., Handmann, P., Hummels, R., and Böning, C. W.: Regional imprints of changes in the Atlantic Meridional Overturning Circulation in the eddy-rich ocean model VIKING20X, Ocean Sci., 17, 1177–1211, https://doi.org/10.5194/os-17-1177-2021, 2021.
Danabasoglu, G., W. G. Large, and B. P. Briegleb (2010), Climate impacts of parameterized Nordic Sea overflows, J. Geophys. Res., 115, C11005, doi:10.1029/2010JC006243.