Trouble splitting NICAS

JEDI
1

Hi,
I am trying to split nicas fields into more processors but kept getting the errors below. I wonder what is causing this to happen. My yaml file is also pasted below. Thanks in advance!

Info     : -------------------------------------------------------------------
Info     : --- Get C matrix from BUMP interface, ensemble 1
Info     :        Block air_temperature
Info     :           BUMP import for   1 components
Info     : -------------------------------------------------------------------
Info     : --- Copy namelist support radii into C matrix, ensemble 1
Info     : -------------------------------------------------------------------
Info     : --- Setup C matrix sampling, ensemble 1
Info     : -------------------------------------------------------------------
Info     : --- Run NICAS driver
Info     :        Ensemble 1
Info     : -------------------------------------------------------------------
Info     : --- Read global NICAS parameters
Info     :        Read NICAS global data
Info     :              Block: air_temperature
Info     :                 Component: 1
!!! ABORT in nicas_cmp_read_global on task #000001: amplitude should be loaded from
the BUMP interface
Abort(1) on node 0 (rank 0 in comm 0): application called MPI_Abort(MPI_COMM_WORLD,
1) - process 0
!!! ABORT in nicas_cmp_read_global on task #000004: amplitude should be loaded from
the BUMP interface
!!! ABORT in nicas_cmp_read_global on task #000005: amplitude should be loaded from
the BUMP interface
Abort(1) on node 4 (rank 4 in comm 0): application called MPI_Abort(MPI_COMM_WORLD,
1) - process 4
!!! ABORT in nicas_cmp_read_global on task #000006: amplitude should be loaded from
the BUMP interface
Abort(1) on node 5 (rank 5 in comm 0): application called MPI_Abort(MPI_COMM_WORLD,
1) - process 5
!!! ABORT in nicas_cmp_read_global on task #000007: amplitude should be loaded from
the BUMP interface
Abort(1) on node 6 (rank 6 in comm 0): application called MPI_Abort(MPI_COMM_WORLD,
1) - process 6
!!! ABORT in nicas_cmp_read_global on task #000008: amplitude should be loaded from
the BUMP interface
Abort(1) on node 7 (rank 7 in comm 0): application called MPI_Abort(MPI_COMM_WORLD,
1) - process 7
!!! ABORT in nicas_cmp_read_global on task #000009: amplitude should be loaded from
the BUMP interface

geometry:
  fms initialization:
    namelist filename: Data/fv3files/fmsmpp.nml
    field table filename: Data/fv3files/field_table_gfdl
  akbk: Data/fv3files/akbk64.nc4
  npx: 193
  npy: 193
  npz: 64
  layout: [4,4]
  field metadata override: Data/fieldmetadata/gfs-restart.yaml
background:
  datetime: 2020-03-01T00:00:00Z
  filetype: fms restart
  state variables: [stream_function,velocity_potential,air_temperature,surface_pressure,specific_humidity,cloud_liquid_water,ozone_mass_mixing_ratio]
  psinfile: true
  datapath: Data/2020022918/mem001
  filename_core: unbal.fv_core.res.nc
  filename_trcr: unbal.fv_tracer.res.nc
  filename_sfcd: unbal.sfc_data.nc
  filename_sfcw: unbal.fv_srf_wnd.res.nc
  filename_cplr: unbal.coupler.res
background error:
  covariance model: SABER
  saber central block:
    saber block name: BUMP_NICAS
    calibration:
      io:
        data directory: Data/c192-2020022900-2020022918
        files prefix: nicas_2020022900_2020022918/nicas_2020022900_2020022918
      drivers:
        multivariate strategy: univariate
        #read universe radius: true
        read global nicas: true
        write local nicas: true
      grids:
      - model:
          variables:
          - stream_function
          - velocity_potential
          - air_temperature
          - specific_humidity
          - cloud_liquid_water
          - ozone_mass_mixing_ratio
        #io:
        #  overriding universe radius file: cor_3D_gfs_universe_radius
      - model:
          variables:
          - surface_pressure
        #io:
        #  overriding universe radius file: cor_2D_gfs_universe_radius
      input model files:
      - parameter: nicas_norm
        file:
          set datetime on read: true
          datetime: 2020-03-01T00:00:00Z
          filetype: fms restart
          datapath: Data/c192-2020022900-2020022918/nicas_2020022900_2020022918/
          psinfile: true
          filename_core: nicas_norm.fv_core.res.nc
          filename_trcr: nicas_norm.fv_tracer.res.nc
          filename_sfcd: nicas_norm.sfc_data.nc
          filename_sfcw: nicas_norm.fv_srf_wnd.res.nc
          filename_cplr: nicas_norm.coupler.res
2

Hi @xtian15. It seems that your NICAS data has been computed in a previous run with more than one component (which means with more that one length-scale, in order to model different correlation shapes). In this case, each component has a given amplitude, which must be provided in the input model files section.

To confirm my assumption, it would be nice if you could:

  • either provide the yaml used to create the NICAS data,
  • or do a ncdump -h ${file} on one of the NICAS files and post it here.

Thanks!

3

Hi, @benjaminmenetrier . You are right that the system was looking for the amplitude files. The issue was solved after those files were specified in the input model files.

Thanks for looking into it!