Boundary layer cross section

[claireyung]

Plot of temperature structure, but focus on boundary layer cross section (y axis = distance from ice rather than ocean depth) -> can see temperature structure (to 5m grid resolution)

i.e. a remapping of #12

[claireyung]

In addition to the remapped temperature structure colour plot that Mike Dinniman suggested and is assigned to @dgwyther, @bkgf suggested doing some line plots of vertical profiles at a few select regions. These locations (just in a very draft plot for illustration) and vertical sections are shown below. I don't see much variation in Ocean2 (other than MOM6 which is currently the TYP simulation) but in Ocean1 you can really see the steppy-ness of the z level models and also some differences in BL temperature, as well as a much cooler BL in NEMO-UKESM1is, also visible in #12: any ideas why?

These are all averaged over the last year of the 20 year run.

Ocean1:

Ocean2:

[dgwyther]

That's a bit weird hey @claireyung . It's also odd that both POP and NEMO-UKESM seem to have identical temps near to the surface, although it could just look similar with that axis scaling. I think this is an instructive figure to show, although we must keep in mind that the raw data may have been at higher or lower resolution than the interpolated depths.

[dgwyther]

@claireyung @xylar @bkgf @mdinniman2
I've added some draft plots in a new notebook.

There's some interesting boundary features (or lackthereof in some models), but probably the most obvious things are the 'jaggedness' of the top layer, which I think results from:
a) the interpolated depth coordinate being at regular intervals and not quite aligning with the ice draft,
b) possibly an artefact of the plotting routine. If I change the y-limit to a few hundred m below the ice, for example, the jaggedness disappears (but you lose clarity ofthe boundary layer).
So I'm not exactly sure what to do about this...

Here are some example plots from the above notebook:

[claireyung]

Thank you @dgwyther! Wow, cool features!

I agree the jaggedness of the top points would have to do with the data output resolution, and probably the resolution and interpolation affects what the plot looks like too, but I can't think of a better way. You could probably plot "distance from the shallowest non-NaN value" as the y-axis to get a flat top of the plot but I think that would be less sensible because then it wouldn't really be the distance from the ice. Also, some models don't reach around zero depth periodically (e.g. FVCOM), does that mean the ice draft values are slightly off?

I think the stripes in the z-coordinate model are really interesting. My interpretation is it shows that in z-level models when you go up a "step" the point closest to the wall is coldest, and it gets progressively warmer as you move horizontally into the ocean, demonstrating a sort of discontinuity of horizontal heat transport and meltwater getting trapped near the step?

Also do you know what the white triangle in Ocean2 MITgcm-BAS is?

[dgwyther]

I added a couple of plots zooming in on the interface, with the interpolated coordinates and the ice draft (x,y).

It looks to me like the jaggedness is actually just missing values between the iceDraft point and the top value.

The triangle in MITgcm-BAS does not appear if you zoom in, so I presume it's an artefact of the mapping.