I have accessed Anaconda Prompt for the environment that I will be working.
Once there, I have changed the directory to the place where I have the grib files that I will be working with, in order to install cfgrib, I have used the following command:
conda install -c conda-forge cfgrib
As the Python module depends on ECMWF ecCodes binary library, in the same Anaconda Prompt, I ran the following command:
conda install -c conda-forge eccodes
Then, opened the file in Spyder, and tried to run this simple script:
import xarray as xr
ds = xr.open_dataset('download.grib', engine='cfgrib')
ds
But got the following error:
Can't read index file 'C:\Users\name\Desktop\data\download.grib.90c91.idx' Traceback (most recent call last): File "C:\Users\name\anaconda3\envs\Weather\lib\site-packages\cfgrib\messages.py", line 351, in from_indexpath_or_filestream self = cls.from_indexpath(indexpath) File "C:\Users\name\anaconda3\envs\Weather\lib\site-packages\cfgrib\messages.py", line 323, in from_indexpath return pickle.load(file) EOFError: Ran out of input
Then,
An error ocurred while starting the kernel ECCODES ERROR : Unable to find boot.def. Context path=D:/bld/eccodes_1593015095851/_h_env/Library/share/eccodes/definitions
Possible causes: ‑ The software is not correctly installed ‑ The environment variable ECCODES_DEFINITION_PATH is defined but incorrect ecCodes assertion failed: `0' in D:\bld\eccodes_1593015095851\work\src\grib_context.c:226
What am I missing here?
Edit 1
The problem also happens if I use Visual Studio Code.
In the terminal, if I am inside the directory that the GRIB files are stored, and I activate the environment that I am working with:
conda activate Weather
And then:
& C:/Users/name/anaconda3/envs/Weather/python.exe c:/Users/name/Desktop/data/3_climate/2mtemp_open.py
It gives the same error as above.
As the error says,
ECCODES_DEFINITION_PATH
was not set.As I am using Anaconda on Windows 10, the path is the following
In order to set the environment variable, one can simply run
This should solve the problem.
However, a new one may emerge. As the .grib file may take a lot to run, consider using the data in the format NetCDF.