Alternate scanning directions not honored in NBM GRIB files

I was attempting to read some NBM (National Blend of Models) GRIB files using xarray and cfgrib: t_ds = xr.open_dataset('blend.t12z.core.f%03d.co.grib2'%12, filter_by_keys={'stepType': 'instant', 'typeOfLevel': 'surface','paramId':130}) The reading operation was successful in that it didn’t throw an error… but it also ended up looking really screwy when I went to make a very quick plot t_ds['t'].plot():

The twitter hive mind (thread for reference: https://twitter.com/JackSillin/status/1484617258852888582) quickly assessed that this was due to cfgrib not honoring some flag/option in the NBM GRIBs that reads the rows in the alternating order in which they were written. I’ve only ever had this issue with NBM gribs (to replicate the issue you may want to find yourself one or two https://noaa-nbm-grib2-pds.s3.amazonaws.com/index.html#blend.20220121/12/core/).

[shahramn]

ecCodes v2.25.0 has been released. See https://confluence.ecmwf.int/display/ECC/ecCodes+version+2.25.0+released

[iainrussell]

That’s great, thanks for letting me know!