Working with icosahedral data in python

[keksundso]

Hello,

it is not 100% fitting as a issue to this repo, still I hope you might be able to help.

I was wondering if it is possible to work directly with icosahedral-ICON-grib2 files in python, instead of first converting it to lat-lon-Data. I am currently using pygrib in python, but it can not handle "unstructured data".

So my questions are:

• How could i find the icosahedral coordinates which correspond to a given lat-long pair.
• Is there a python library which can work with icosahedral-ICON-grib2 files.

Thank you in advance

[bjoern-reetz]

Hello, and sorry that no one reacted to your issue. We currently do not have the resources to support this project adequately.

While I cannot offer advice regarding pygrib, I can give you some tips on how to process DWD NWP data in native icosahedral grid: As documented in chapter 3.1 of our ICON database reference, you can download so-called "grid files" at http://icon-downloads.mpimet.mpg.de/dwd_grids.xml. They are just netCDF files and contain information to reconstruct the geographic coordinates of points in the icon grid. But I would rather suggest a different approach.

As a convenience for you, we export the Lat/Lon coordinates of every point in the icosahedral grid as a seperate field: CLAT/CLON. To extract the data for a given lat/lon coordinate, find the index that has the smallest distance with respect to the CLAT/CLON fields and use that index to retrieve the data from other fields, e.g. TOT_PREC.

This of course assumes your GRIB libraries can handle unstructured data. When using Python, this can be done using python-eccodes. Remember that you need to use the DWD ecCodes definitions to use our custom GRIB short names.

[keksundso]

Thanks a lot for your suggestion. I just implemented your approach using CLAT/CLON. It works perfectly and significantly reduces the required computations.

I greatly appreciate your help and the data the DWD provides.