Will Benfold

> Thanks @wjbenfold! Is there more to do here or can this be merged? I was just checking through the conversations to check they were resolved. It's now in!

Looks related to #4101

I like this. If we get the opportunity to overhaul merge/concatenate in the future, then I think this should be part of it, if we don't and this is still...

From a bit of investigation, I think that this is being caused mostly (if not completely) by the antialiasing in iris.plot.contourf. I'm not too surprised that this issue is present...

@mo-avdjturp, based on my (shallow) understanding of the initial issue, if anything @rcomer's iplt version now looks better than the plt one - should we keep digging into the differences...

Thanks for this, it's incredibly helpful to have a minimal broken example with a bug report. ### Diagnosis: We've (myself, @trexfeathers and @stephenworsley) discussed what's happening and are pretty sure...

After a bit more investigation, it looks like at least sometimes dask does manage to figure out that the return type should be `float64`, but then that belief is overridden...

https://cfconventions.org/Data/cf-conventions/cf-conventions-1.9/cf-conventions.html#use-of-the-crs-well-known-text-format It will change their output results, but not what their file means, so I think that's fine? The only bit of the CF conventions wrt. `crs_wkt` that slightly concerns...

> Do we have any way of testing that the WKT is representing the same thing as the `cs` object parameters or can we rely on cartopy for that? We're...

I can future flag it? There aren't users clamouring for this to be the default behaviour. It would mean a warning on pretty much every netCDF save though. We could...