grib

To read all the GRIB from a file, use the following:

input:
  grib:
    path: /path/to/data.grib

You can also read specific GRIB messages by specifying them using the MARS language (excluding the keywords date, time, and step, as well as any post-processing options, such as grid or area):

input:
  grib:
    path: /path/to/data.grib
    param: [ u, v ]
    levelist: [ 1000, 850, 500 ]

You can also read a collection of GRIB files, using Unix shell wildcards:

input:
  grib:
    path: /path/to/*.grib

You can also use the requested date to build the filenames. For example, if the GRIB files containing the requested data are named according to the following pattern: /path/to/YYYY/MM/YYYYMMDDHH.grib with YYYY being the year, MM the month, DD the day, and HH the hour, you can use the following configuration:

input:
  grib:
    path: /path/to/{date:strftime(%Y)}/{date:strftime(%m)}/{date:strftime(%Y%m%d%H)}.grib

The patterns in between the curly brackets are replaced by the values of the date and formatted according to the Python strftime method.

See Create a dataset from GRIB data for more information.

Note

You can combine all the above options when selecting GRIB messages from a file.