Julius Busecke

> Is this right? Seems like the simplest (finite distance) implementation to me. I personally would want to make the finite vol way of calculating div the default? It will...

This would allow the user to e.g. weight velocities in the vertical beforehand (if you are dealing with a z* or other variable grid) and should thus be pretty general?...

> grid.diff(V, axis='X', metric_weighted=['X']) I am not sure this gives the desired result. This does weight the variable, diffs it and then *unweights* it again with the same type of...

I meant 2. from the above. `metric_weighted` will return the same unit for the output as the input (because it first multiplies but then divides again, but on different grid...

Hi @navidcy, thanks for using xgcm. Short answer: Yes. I have worked with several models with xgcm sucessfully. Do you have a concrete problem or example that you are concerned...

What does pop use? Happy to make a n example with MOM5/6, but I just have to say that I am quite backed up with stuff rn. In principal it...

Hmm maybe I have been thinking about this a bit to simplified (forgive me I stay as far away from the Arctic as possible hahaha). So you are asking what...

Wouldnt we have to invert velocities as well? Is #117 relevant here?

Yes. After the boundary padding refactor that will just be a new padding function!

Ok, now that v0.7.0 is released, I am very curious to pick this back up. I think the actual code implementation will be quite easy now. I suggest the code...