Any tools to regrid and crop the restart file on AWS in preparation for nested simulation?

[FeiYao-Edinburgh]

It seems that IDL has not been installed on ami-0ee8892ae47c31be1. Are there any other tools that can regrid and crop restart files in preparation for nested simulation?

[JiaweiZhuang]

Use xESMF! See this notebook for steps: https://github.com/geoschem/GEOSChem-python-tutorial/blob/master/Chapter03_regridding.ipynb

We are now doing everything in Python. IDL needs a license and is hard to install on EC2.

[FeiYao-Edinburgh]

Use xESMF! See this notebook for steps: https://github.com/geoschem/GEOSChem-python-tutorial/blob/master/Chapter03_regridding.ipynb

We are now doing everything in Python. IDL needs a license and is hard to install on EC2.

I am also a big fan of Python user. Will learn according to your tutorial!

[FeiYao-Edinburgh]

Use xESMF! See this notebook for steps: https://github.com/geoschem/GEOSChem-python-tutorial/blob/master/Chapter03_regridding.ipynb

We are now doing everything in Python. IDL needs a license and is hard to install on EC2.

Although this tutorial is very convenient to follow, it cannot be directly applied on initial_GEOSChem_rst.2x25_complexSOA_SVPOA.nc. This is because initial_GEOSChem_rst.2x25_complexSOA_SVPOA.nc contains some variables that do not have longitude or latitude like "hyam". So I just simply keep these variables when regridding but unsure if it is fine to do that way since I got error when run following nested GEOS-Chem, which indicates associations with restart files or met fields.

[JiaweiZhuang]

This is because initial_GEOSChem_rst.2x25_complexSOA_SVPOA.nc contains some variables that do not have longitude or latitude like "hyam".

hyam is simply a 1D vertical coordinate that does not need regridding. You should be able to just add it to the new nested grid file. Something like ds_nested['hyam'] = ds_global['hyam'] will do.

[FeiYao-Edinburgh]

This is because initial_GEOSChem_rst.2x25_complexSOA_SVPOA.nc contains some variables that do not have longitude or latitude like "hyam".

hyam is simply a 1D vertical coordinate that does not need regridding. You should be able to just add it to the new nested grid file. Something like ds_nested['hyam'] = ds_global['hyam'] will do.

OK. Something confirmed. Thanks.

BTW, when do conservative regridding, could you help me understand the following parameters? global_grid_with_bounds = {'lon': ds['lon'].values, 'lat': ds['lat'].values, 'lon_b': np.linspace(-180-5/2, 175+5/2, 73), 'lat_b': np.linspace(-92, 92, 47).clip(-90, 90), # fix half-polar cells } I know -180 and 175 are the start and end of the longitude, so by ±5/2, we got 72+1=73 grids. But this could produce longitude less than -180, is it OK?

How about latitude? Where do -92 and 92 come from (-90-4/2,90+4/2)? When comes to 2x25 global simulation, how? I see start and end latitude during these two grids are -89,89 and -89.5,89.5, respectively. I am not very sure about these calculations.

[JiaweiZhuang]

How about latitude? Where do -92 and 92 come from (-90-4/2,90+4/2)?

This comes from GEOS-Chem's (somewhat confusing) half-polar grids (grid boxes around the poles are half in latitude): http://wiki.seas.harvard.edu/geos-chem/index.php/GEOS-Chem_horizontal_grids

[FeiYao-Edinburgh]

How about latitude? Where do -92 and 92 come from (-90-4/2,90+4/2)?

This comes from GEOS-Chem's (somewhat confusing) half-polar grids (grid boxes around the poles are half in latitude): http://wiki.seas.harvard.edu/geos-chem/index.php/GEOS-Chem_horizontal_grids

Thanks for the hint included in the bracket. Although still confusing, I guess I have figured out where were -92 and 92 calculated from.

-89-4/2-4/2/2=-92
89+4/2+4/2/2=92
Similarly, if for 2x25 grid, it would be:
-89.5-2/2-2/2/2=-91
89.5+2/2+2/2/2=91

[FeiYao-Edinburgh]

Sorry for disturbing you again. I just wonder how to maintain attribute information with xESMF?

I have run Chapter03_regridding step by step but found the final regridded restart file missed attribute information contained in the original restart files (You can easily see this by running ds_result['lon'],ds['lon']). As such, the nested GEOS-Chem will complain with the following information.

HEMCO ERROR: illegal longitude unit in ./GEOSChem.Restart.20160701_0000z.nc4 - Must be `degrees_east`.

I believe it is must be very easy to maintain those attribute information by using Python. Would be greatly grateful if you could give some hints. Thanks in advance!

[JiaweiZhuang]

I believe it is must be very easy to maintain those attribute information by using Python.

ds['lon'].attrs['units'] = 'degrees_east' will do. Sorry for this unsmooth experience -- xESMF is a general-purpose tool that is not geared to GEOS-Chem restart files. If time allows we should add a high-level regridding function in https://github.com/geoschem/gcpy

[msulprizio]

Also, FYI, you can use open-source tools CDO and NCO to regrid and crop GEOS-Chem restart files. Please see these GEOS-Chem wiki pages for more information:

• http://wiki.seas.harvard.edu/geos-chem/index.php/Working_with_netCDF_data_files#Regridding_netCDF_files
• http://wiki.seas.harvard.edu/geos-chem/index.php/Working_with_netCDF_data_files#Cropping_netCDF_files

[JiaweiZhuang]

you can use open-source tools CDO and NCO to regrid and crop GEOS-Chem restart files.

We do have an discussion on implementing xarray-cli (pydata/xarray#2034) but most of the times writing Python code seems way easier 😄

[JiaweiZhuang]

Closing if no further questions