Make quality control plots a standard part of mesh generation for E3SM

[xylar]

As discussed here, it would be really helpful to have quality control plots (like here) generated automatically as part of creating meshes and initial conditions for E3SM.

This would presumably involve porting Matlab code to python as part of a new test case for the global ocean.

[xylar]

@proteanplanet, as a first step toward assessing how much work this would be, would you be willing to point me to your Matlab scripts so I can see how much code we're talking about and how much beyond my rusty Matlab skills this might be?

[xylar]

cc @vanroekel

[dengwirda]

Hi all, there are also a bunch of mesh quality routines available direct in jigsaw-python...

[xylar]

Thanks @dengwirda. Those would also be helpful to include. I think what is most critical in this context are some things about quality of coast lines related to specific features that @proteanplanet has picked out. So what I have in mind here are more quality control from a geopolitical and physical oceanography standpoint. I'm guessing what jigsaw-python provides are metrics that are more related to numerical method? If so, that would still be a useful addition. If you have specific suggestions or examples you'd like to point us to, please do!

[dengwirda]

@xylar, yes totally agree --- @proteanplanet's coastline/channel quality control is very important on its own (and makes very nice plots!). I think it could be of interest to add additional metrics though, to ensure geometric quality is consistent with what TRSK is happy to deal with. These are some standard things that I run in the jigsaw examples, with a little more detail here.

[xylar]

@dengwirda, got it. So these wouldn't be images saved out or fields added to the mesh necessarily but rather output to the log file when the mesh is being created, right?

[dengwirda]

@xylar, I think there could be a few possibilities. Printing to the log-file makes sense, but another option could be to use the metrics to identify additional cells/regions (that fall below some quality thresholds, say) to be passed to @proteanplanet's workflow for additional visual inspection?

[xylar]

Okay, so for that purpose, we would want to add these metrics to the mesh and pass them through at least to the culling stage.

[dengwirda]

@xylar, that would definitely work, but I'm not sure what the easiest path might be at this stage --- another option is that the MPAS mesh data-structure may (I think) have enough information in it to just compute these metrics on the fly within/before @proteanplanet's workflow.
Does it make sense for the three of us to discuss a little more first, and try to hash out a design?

[xylar]

The rest of this work is likely to be substantial. @proteanplanet's plotting scripts are, to the best of my knowledge, in Matlab. So porting them to Python will be a not insignificant task. I'm sure there will be time to discuss adding these metrics to that workflow. It likely won't happen in the immediate term, though.

[xylar]

@proteanplanet, could you point us to your Matlab code for making your quality control plots if it is publicly available?