Support pyqg

[patnr]

This looks pretty good

https://github.com/pyqg/pyqg

See examples:

https://pyqg.readthedocs.io/en/latest/examples.html

[yumengch]

Hi Patrick,

This seems nice. I wonder if you have any interests in this project: https://github.com/Climdyn/qgs

I think it includes the MAOOAM model you used.

[patnr]

Ineed, I do have some interest in it: Climdyn/qgs#9
but I have not yet found the time to work with it.

I think qgs is quite different from pyqg, so support for both is desirable.

[patnr]

For reference, the reason I removed MAOOAM from DAPPER back in the day (here's the backup branch) was because I had yet to see any interesting DA experiments done with it. And at the time there was no way to output the fluid field plots, so all you got was time series. And the time series it produced either looked like white noise or just a constant; I guess that could just be me looking at it from an intermediate time scale (neither that of the atmosphere nor the ocean), but it made me question the predictability of it; seemed to me like unsophisticated/baseline DA methods would do as well as was possible.

As can be seen above, I'd still be interested in seeing experiments with it or qgs. It might be that @brajard has done something, but I'm not sure if he used DAPPER because he wanted different frequencies of ocean/atmos observations, meaning that the obs vector would vary in size, something that is not explicitly supported in DAPPER (though it can be somewhat "faked" using null padding)