Reading GRIB2 files

The function open_dataset() creates a dataset from one or more files. Mandatory arguments are:

  • one (string) or more (iterable of strings) GRIB2 files.

  • template created by a call to make_template().

  • location of inventory files, if not collocated with data files. If not specified and the inventory cannot be found, it will be created.

Remaining optional arguments: chunks, preprocess, parallel and cache are the same as for xarray.open_mfdataset(). To continue the example from the previous section:

import glob
import pywgrib2_xr as pywgrib2

files = sorted(glob.glob('nam.t00z.afwahi??.tm00.grib2'))
tmpl = pywgrib2.make_template(files, lambda x: x.varname == 'APCP')
ds = pywgrib2.open_dataset(files, tmpl)
ds

<xarray.Dataset>
Dimensions:                   (latitude: 231, longitude: 278, time1: 1, time2: 2, time3: 2, time4: 2, time5: 8)
Coordinates:
  * longitude                 (longitude) float64 190.0 190.1 ... 219.8 219.9
  * latitude                  (latitude) float64 8.133 8.241 ... 32.86 32.97
  * time1                     (time1) timedelta64[ns] 00:00:00
  * time5                     (time5) timedelta64[ns] 03:00:00 ... 1 days 00:...
  * time2                     (time2) timedelta64[ns] 06:00:00 18:00:00
  * time3                     (time3) timedelta64[ns] 09:00:00 21:00:00
  * time4                     (time4) timedelta64[ns] 12:00:00 1 days
    reftime                   datetime64[ns] ...
    latitude_longitude        int64 ...
Data variables:
    APCP.surface              (time1, latitude, longitude) float32 ...
    APCP.surface.3_hour_acc   (time5, latitude, longitude) float32 ...
    APCP.surface.6_hour_acc   (time2, latitude, longitude) float32 ...
    APCP.surface.9_hour_acc   (time3, latitude, longitude) float32 ...
    APCP.surface.12_hour_acc  (time4, latitude, longitude) float32 ...
Attributes:
    Projection:             latitude_longitude
    Originating centre:     7 - US National Weather Service - NCEP (WMC)
    Originating subcentre:  0
    History:                Created by pywgrib2_xr-0.2.1

Note

The data variables are always in WE:SN order, regardless of the order in GRIB2 file. This follows wgrib2 default behaviour.