Add OHC anomaly to timeSeriesOceanRegions

[xylar]

Checklist

• User's Guide has been updated
• Documentation has been built locally and changes look as expected
• Testing comment in the PR documents testing used to verify the changes

[xylar]

An example result from the test suite:

I will request reviews as soon as I have run this on a SORRM v2.1 simulation (with the southern hemisphere geometric feature MPAS-Dev/geometric_features#205 included).

[xylar]

I just noticed some glitches in the figure above. I think I've fixed them an will replace it if so.

[xylar]

Somewhat better but the title is still a mess.

[xylar]

Here's the SORRM v2.1 ensemble mean analysis over the historical:
https://web.lcrc.anl.gov/public/e3sm/diagnostic_output/ac.xasay-davis/analysis/southern_hemisphere_ohc/v2_1.SORRM.historical_ensmean/ocean/index.html
Here's the Southern Hemisphere:

[xylar]

@cbegeman and @darincomeau,

For the v2.1 paper, several details and associated questions:

• This a curve over a single range of depths (0 to 10,000 m shown here). We can run the analysis multiple times to analyze/plot different depth ranges. Is that sufficient for our needs?
• I am computing anomalies here from January of the reference year, and only for plotting purposes. The NetCDF files just have the OHC time series (not anomalies). Does that seem okay for our needs?
• I don't currently have a way to support (for plotting purposes) an anomaly reference year that isn't somewhere in the time series. MPAS-Analysis already isn't good at having an anomaly year from a different simulation. This means MPAS-Analysis isn't capable of computing the OHC anomaly during the SSP with respect to, say, 1950. I believe this is okay for our purposes but let me know.
• My thought is that we stitch together whichever OHC netcdf output we want to use in a separate analysis script, compute the anomaly and maybe a running mean, and fit a trend line. Does that sound right?

[xylar]

I won't merge this until we get MPAS-Dev/geometric_features#205 merge, do a release of geometric_features, and update the constraint on geometric_features here.

[xylar]

Further testing

I reran the test suite after updating to geometric_features 1.6.0. I needed to fix the name of Greenland Regions --> ISMIP6 Greenland Regions for things to work. With this fix, all analysis ran successfully as seen here:
https://web.lcrc.anl.gov/public/e3sm/diagnostic_output/ac.xasay-davis/analysis_testing/chrysalis/so-regional-ohc/

However, there was a glitch in the anomaly computation for the control run. Fixing this...

[xylar]

Okay, the main vs. control anomaly plots are fixed:

[cbegeman]

@xylar Great! Looking through the code, I don't see anything to change. Would you like me to do some testing, perhaps generating the v2.1 SORRM Southern Ocean OHC anomaly plot so it could be useful for the paper?

[xylar]

@cbegeman, it certainly would be fine if you want to test this on SORRM v2.1 I've done that on an earlier version above but it wouldn't hurt to run it again. You can use my config file in the link above as a starting point.

[xylar]

Here's the link to the main page with the config file.

https://web.lcrc.anl.gov/public/e3sm/diagnostic_output/ac.xasay-davis/analysis/southern_hemisphere_ohc/v2_1.SORRM.historical_ensmean/

[xylar]

Rebased, but this includes #1040 for now.

[xylar]

@cbegeman, if you don't have time to run new tests, I think that's fine. I'm pretty sure my previous tests with the SORRM v2.1 ensemble mean over the historical are sufficient. I wouldn't mind getting this merged by the end of this week.

[cbegeman]

Approving based on visual inspection of the code and @xylar's testing. Thanks for developing this feature, @xylar!

[xylar]

Thanks @cbegeman!