The ht and hu variables for bathymetry use the actual vertical water column height in ACCESS-OM2-01, while variables like temp have fixed st_ocean_edges. To average along isobaths, I currently define a new bathymetric field, called ht_coarse or hu_coarse, that uses the NaNs in variables like temp and u to bin in isobaths.

We need to think about how/whether to incorporate the actual bathymetry in the binning here, and consider whether the current workaround is reasonable.