Baseline Validation Variables

When a step is run with a baseline, polaris compares the variables named in validate_vars for each output file against the same variables in the baseline’s copy of that file (see Validation). For MPAS mesh files that list is long, and it has to be the same everywhere or two steps that write the same kind of file would be validated to different standards. polaris.mesh.validate keeps the shared lists in one dependency-light module so that framework steps and component steps can both import them without pulling in anything else.

There are two lists:

• polaris.mesh.validate.MPAS_MESH_VALIDATE_VARS – the coordinate, area, index and connectivity variables of an MPAS mesh file; and

• polaris.mesh.validate.CELL_WIDTH_VALIDATE_VARS – just cellWidth, for the lon/lat cell-width file that drives JIGSAW.

Use them when adding the output file:

from polaris.mesh.validate import MPAS_MESH_VALIDATE_VARS


self.add_output_file(
    filename='culled_mesh.nc',
    validate_vars=MPAS_MESH_VALIDATE_VARS,
)

MPAS_MESH_VALIDATE_VARS deliberately contains only variables that a base mesh and a culled mesh both have, so the same list works for either without a step having to trim it. A step that writes extra fields should validate them by extending the list locally rather than adding them here, unless every MPAS mesh file in polaris will have them. Vector Reconstruction is the pattern to follow: the reconstruction weights get their own list from polaris.mesh.reconstruct.get_reconstruction_validate_vars(), because only some meshes carry them.