GSV interface

A tool for accessing the GSV data from FDB and converting it into xarray.

Core idea

The current approach can be summarized in three steps:

1. Use pyfdb to request data from the FDB. This will return a pyfdb.Dataretriever file-like object from which you can read all GRIB messages matching a given request.

2. Use eccodes-python to decode the necessary information of each GRIB message (data, coordinates, attributes...). A custom iterator (MessageIterator) is used to iterate over each message in the pyfdb.DataRetriever object.

3. If requested by the user, interpolate the result to a regular LonLat grid of choice.

All requested messages are packed in a single xarray.Dataset object and returned to the user.

Installation instructionss

The gsv-interface package is available to install through pip. It can be installed with the folloeing command:

pip install gsv-interface

Checking installation

Top check the installation was succesfull run Python and try to import gsv.

>>> import gsv
>>> gsv.__version__

This should print the version number of the chosen release.

Dependencies

The followin non-Python libraries are required:

• ecCodes: https://github.com/ecmwf/eccodes
• fdb: https://github.com/ecmwf/fdb

This tool depends on the following Python modules, which are automatically installed though pip.

• "pyfdb==0.1.2",
• "numpy",
• "xarray",
• "eccodes",
• "healpy",
• "dask",
• "netcdf4",
• "cfgrib",
• "sparse",
• "cdo",
• "smmregrid==0.1.1",
• "polytope-client",
• "pyyaml",
• "jinja2",
• "packaging"

Documentation

Docs based on sphinx can be found in the docs/ directory of the repository.

Build dependencies for the docs can be installed with the following command (it's recommended to run it in a clean environment):

pip install -r requirements.txt

Then the docs can be built with:

make html

The resulting html docs can be then found in:

docs/build/html/index.html