Open willaguiar opened 2 months ago
I’m not sure I’m sold on A. The abyssal overturning isn’t driven by wind in the same way as the upper cell, i.e. the wind doesn’t drive the circulation. The wind is important for opening up polynyas and exporting sea ice off the shelf though.
B. Seems more plausible to me, since we see the response already within days at the start of the experiment.
If it was a change in surface residence time driving this, wouldn’t it take longer than days to see a difference between the experiments?
First, these are the Ekman Pumping mean on 1mtops, calculated for MOM5 and MOM6 for validation:
Land was masked after calculating the curl ( stresses on land were filled with 0s)
naive method on MOM5 I multiplied the Ekman pumping by surface area of each model (m3/s) and summed longitudinally. On the top left are values per latitude, on top right are differences between 5m and 1m cumsummed along the latitude ( so we can see the total signal) . I cumsummed from the shelf until 65S, north of which there is a significant upwelling on sigma surfaces ( bottom 4 plots), and defined that as downwelling region. The north of 65S is where I am defining as upwelling region.
This is obviously not a perfect calculation, as upwelling/downwelling won't be purely zonal, but gives a rough estimate.
We can see that downwelling does increase in 5m case ( top right, 5m-1m negative anomalies grow), and upwelling increase too (5m-1m positive anomalies also grow). It is hard to see changes in zonal mean isopycnals between 5m and 1m tho, on the shelf...
naive method on MOM6 In MOM6 however, we see mostly a change in upwelling, but not so much in downwelling. Also Density anomalies here appear mostly on the shelf, no so much offshore. Offshore ones seem almost opposite in signal when compared to MOM5
Better method: Ekman pumping binned by isobath depth
Downwelling through Ekman happen mostly South of the 1km isobath ( validation plots on top), so I binned the Pumping by the ocean depth (longs,isobath_depth). This we better separated the region south of the isobath, and I can then properly see how the coastal downwelling changed. We do see an increase in downwelling from 1m to 5m in both models. Although, the increase is much much smaller in MOM6. (perhaps there are differences in the way MOM6 calculates the wind stress, or ekman pumping?). Anyways, maybe all comes down to differences in Ekman pumping
Im binning the isopycnal along the same isobath space to see if we can see a shift in their slope as suggested by B
Have we yet ruled out the effect of ice-ocean stress on the sea ice?
Have we yet ruled out the effect of ice-ocean stress on the sea ice?
I think so. What would be that effect again? (Perhaps lower ice transport, lower sea ice production on polynyas?). Id expect that to show up as differences in freshwater fluxes from sea ice on the shelf, right? I did a test to see that by altering the SWMT to account for surface densities from 1 experiment, surface fluxes from another.
i.e., on the left is a (red) case of surface fluxes of 5mtop (slightly reduced sea ice formation) over the surface densities of 1mtop. On the right is the opposite in red too, i.e., 5mtop surface densities over 1mtop enhanced fluxes. Id expect that if that differences in sea ice production/export explain the DSW change, then we would have a significant decrease in the SWMT in the dense layers on the left and/or increase in the right plot, but that is not much of the case. Majority of the signal change when we alter the surface fluxes happens at lighter waters (~27.7 o n left, ~27.6 on the right). Which suggests me that it is likely something on the ability of surface cells of losing buoyancy given similar surface fluxes.
Unless Im missing something in this sort of test.
Below is both Ekman pumping and Isopycnal changes between models, binned into isobath distance( as well as density anomaly 5mtop - 1mtop). I can see the Ekman pumping kind of increasing stratification around the 1000m isobath in MOM5, but not so much in MOM6
These are all along the shelf. perhaps I need to redo this only for DSW formation regions
Just updating here another test we want to do regarding our discussions....
We concluded that is likely that differences in the way a thicker cell senses the wind can alter Ekman transport, and causes differences in surface densities over the shelf.....
To test that we can run 1mtop and 5mtop cases with absolute winds forcing the ocean.... If we have similar DSW formation in these cases than that should answer our questions....
According to issue #137 of OM2 GH, this can be done by setting absolute_wind=true
in namelist group surface_flux_nml
of ice/input_ice_gfdl.nml
And as far as I could see on fms coupler from MOM6 GH, the same approach of use speed differences to calculate tau is used on MOM6 line 453
I had a quick look and struggled to find any input parameters in MOM6/SIS2 to set relative or absolute wind stress. However if you look at the panan output on wind stress, you can see the imprint of surface ocean velocity, so it must be using relative wind stress also.
I had run now the 1m and 5m cases with absolute winds for 2 months [Jan and Feb]...... here is what I find.
$\tau_x$ and $\tau_y$
Below a the fields of $\tau_x$, relative wind case on top, absolute on bottom, 1mtop on the left, 5m-1m on the right.
We indeed find that $\tau_x$ is amplified in the 5mtop case when using relative winds (top right), but to a lower extent in the absolute winds(bottom right)
The same can be found below for $\tau_y$
Ekman Pumping fields
Below is the average Ekman pumping, relative wind case on top, absolute on bottom, 1mtop on the left, 5m-1m on the right.
The differences in Ekman Pumping with resolution are indeed bigger in the relative case (top right) than the absolute case (bottom right). But the differences in both case seem to be rather small tho (especially south of the isobath)
Ekman Pumping integrated along isobath Below is the Ekman pumping integrated along isobath, cumulative summed from shallow to deep isobath. Overall, it seems that the while the wind stress differences are stronger in the relative case, the change in the pumping itself isn't much from 5m to 1m, and smaller from relative to absolute curl . Even in the relative case, the difference of the pumping on the coast is about 4% of the total pumping. Besides since using absolute winds is not changing the pumping substantially.
Isopycnal displacement: absolute vs relative Below are the isopycnal profiles , separated for only the ross sea as an example, for the 1m (blue) and 5m (orange) case, in the relative (left) and absolute (right) wind case.... the displacement is fairly similar.... which I guess suggests that the correction for the absolute winds might not be what we are looking for....
Perhaps the change in pumping is just too small to explain the lightening of the shelf.
I was talking to Andrew Kiss about how two surface cells with similar wind stresses but different speeds might produce different velocities, which is due to how well it resolves the Ekman spiral. Based on that I have though on a different hypothesis ( new issue coming)
After talking to Kial, he told me that in previous OM2 experiments with different vertical resolutions, he noticed that the thicker the top cell, the bigger would be the wind stress sensed by the ocean ( it is a function of the velocity difference between ocean and atm Pacanowski 1987, Kial and Hogg 2019).
OM2 uses the stress curl to calculate the Ekman pumping, which means that the stronger curl in the 5m case would end up increasing the Ekman Pumping. This could alter the DSW formation ....
A- Since stronger Ekman pumping on the shelf would decrease the residence time of these waters at the surface of the shelf, not giving enough time to sea ice production to increase its density
B- By increasing stratification on the shelf, i.e., stronger wind driven downwelling pulls surface density layers towards the shelf (decreasing surface density), and pushes deeper shelf density layers offshore, creating a net increase in stratification as we see in issue 40. Here is some evaluation of those signals..