raphaeldussin / xoverturning

ocean overturning in xarray
19 stars 2 forks source link

Compute isopycnal overturning from z level output #11

Closed jetesdal closed 3 years ago

jetesdal commented 3 years ago

Very nice work! I would love to use this package for my analysis. I am looking for a way to compute the overturning in density space, but I only have z level data (ocean_annual_z). I see an example for density coordinates rho2. Is there a way to use the tool to get the isopycnal overturning (let's say in sigma1) from z level output?

To illustrate what my goal is, I am attaching a figure from Newsom et al. (2016). Here they present the isopycnal overturning in the Southern Hemisphere remapped to depth coordinates (panels A-F). Screen Shot 2021-05-11 at 9 32 44 PM

Currently, I can get the overturning in depth-level coordinates with moc = calcmoc(ds, dsgrid=grid)

And the output is equivalent to what is shown in panels G and H of the Newsom Figure 5 above: Screen Shot 2021-05-12 at 11 26 54 AM

However, I would like to get something as shown in panel A and B in the Newsom Figure. I think I can do this with xgcm transform, but I wanted to check if this might already be implemented in xoverturning.

raphaeldussin commented 3 years ago

Thanks Jan-Erik! Glad the tool works for you! I don't have a z -> sigma transformation built-in since xoverturning should be able to accept any vertical axis. I think xgcm is the way to go but some may argue that to do the computation properly you need the output in density space.

jetesdal commented 3 years ago

Hi @raphaeldussin! Thanks for clarifying.

I've been working on an example using xgcm. It uses the same sample data as in Computing Meridional Overturning. This might be out of scope for xoverturning (so feel free to close the issue), but maybe it would be interesting to include in MOM6-AnalysisCookbook?

Would it be useful to compare the results from ocean_annual_rho2 in Overturning.ipynb with my approach?

I assume Graeme (@gmacgilchrist) also has some insight and could comment on the validity of my current approach.

raphaeldussin commented 3 years ago

@jetesdal I think your analysis would be a great addition to the cookbook. We have runs with both rho2 and z output so you could make a comparison of both methods so we can estimate potential differences. Will close it here and please open a PR to the cookbook when you're ready. If you don't have access to data behind the firewall, write the analysis without running cells and I can run them for you.

gmacgilchrist commented 3 years ago

@jetesdal This looks great, and I think what you're doing is as good as we can do with offline data. Adding this to the cookbook would be great indeed, and likewise comparing this offline calculation with an online rho2 coordinate overturning would be valuable. I expect we have output for that in the runs that you're looking at.