earthkit-data: BUFR - ECMWF internal earthkit training, March 2026

In this notebook you will see how to:

inspect BUFR data
extract BUFR data into a Pandas dataframe

Getting the data¶

First we read some BUFR data from disk with from_source().

import earthkit.data as ekd

d = ekd.from_source("file", "synop_10.bufr")
d

Featurelists and BUFR messages¶

To inspect BUFR data we need to convert it into a featureslist. It is a similar object to a fieldList, but it is an iterable of “features”, where a “feature” can be anything. In a BUFR featurelist each feature is a BUFR messag.

fl = d.to_featurelist()

len(fl)

fl.ls()

BUFR messages¶

# f is the first message in the featurelist
f = fl[0]
f

f.describe()

Subsetting¶

fl1 = fl.sel(dataSubCategory=1, ident=[60545, 48352])
fl1.ls()

Converting to pandas¶

BUFR data can be extracted into a Pandas dataframe using to_pandas(), which passes all the arguments to the read_bufr() <https://pdbufr.readthedocs.io/en/latest/read_bufr.html>_ method from pdbufr.

df = fl.to_pandas(columns=["latitude", "longitude",
                           "heightOfStation","airTemperatureAt2M"])
df

df = fl.to_pandas(reader="synop")
df

Please note it is also possible to call to_pandas() on the input data object “d”. It this case, first the data is converted to a featurelist under the hood, then to_pandas() is called on the featurelist.

df = d.to_pandas(columns=["latitude", "longitude",
                           "heightOfStation","airTemperatureAt2M"])
df