um-convpp shifts wind grids compared to mule-convpp?

[Chris Short (mo-cshort)]

Bit of a memory test this one.

We have been running some experiments with the UKCP18 2.2 km model (rotated pole, variable resolution).

Our suites have a postproc app which uses code from https://code.metoffice.gov.uk/trac/moci/wiki/app_postproc to convert output fieldsfiles into PP files, then archives them.

One of our suites uses postproc 2.1 and the other uses postproc 2.4. We have noticed that the u & v wind field grids are different in PP files produced by the two versions of postproc. Specifically, the longitude values differ by half a grid point. We suspect postproc2.4 is correct as the longitude values for wind fields in the PP files match those in atmospheric dumps.

Now, postproc 2.1 uses the UM utility $UMDIR/vn10.8/xc40/utilities/um-convpp to convert fieldsfiles to PP files, whereas postproc2.4 uses mule-convpp. Does anyone remember if a bug was fixed that could explain this, somewhere along the line between um-convpp at UM10.8 and mule-convpp as it is now?!

Thanks,

Chris