LogP coordinate errors in bump/NICAS

JEDI
1

Hi, All,

I’m trying to create a vertical localization nicas files for FV3.
It works fine with the default vert coordinate: sigma
When I tried to use:
vert coordinate: logP

It gave me errors of the following:

0: Info : Compute internal normalization
55: Abort(1) on node 55 (rank 55 in comm 0): application called MPI_Abort(MPI_COMM_WORLD, 1) - process 55
351: !!! ABORT in nicas_cmp_compute_internal_normalization on task #000352: missing convolution points

Does anyone know why this happens and what should I do to setup the vertical localization in logp instead of sigma?

Thanks,
Xu

2

@XuLu-NOAA from my understanding, for fv3jedi , it only use the same vertical coor as it is defined in general (like as in the background ). the user can’t specify/use a different unit. (not like mpas jedi)

3

Hi, @TingLei-daprediction!

In HAFS, we are using a hybrid coordinate with ak, bk. Is it reasonable to be vertically localized with sigma only?

Thanks,
Xu

4

@XuLu-NOAA From my understanding, as I mentioned previously, you can’t use a different unit for bump vertical scales. But, if you want to use logp, did you try use logp for the vertical coor in the fv3-geometry ? That actual specification is delegated to be done in ,say for fv3, set_and_fill_geometry_fields in fv3jedi_geom_mod.f90. If logp works for other part ( I never tried it), then, you should be able to use logp unit in bump. Or, as in mpas, you could add an extra “pre-processing” step to convert length scales in logp to the actual vertical coor before passed to bump, in which it is always the “vertical coor” is used to “explain” the passed vertical bump length scales. Hope this help!