I have download ERA5 hourly data on single levels from the Climate Data Store for the entire globe (no subset applyed) and then I have downloaded the same variable applying a subset for Iceland region.

I realized the grid downloaded for the entire globe is different from the downloaded subset.

In the following picture we see the global grid in black hollow polygons, and the raster picture is the subsetted data. Since both grids are not alligned, for certain coordinates, we get a different value depending whether we are using the global file or the subsetted file.

Is the subset methodology explained in ERA5 documentation?

Do the grids reference differ according to the subset selected by the user? 

Is it just a spatial offset, or variable values change according to the selected subset?

Many thanks!

1 Comment

  1. Hi Taina,


    I had a similar question. I was able to find the following in the documentation on ERA5:

    "ERA5 data available from the CDS disks has been pre-interpolated to a regular latitude/longitude grid appropriate for that data." 1 ← this makes me think that there is some source of truth on existing lat/lon data points for the single level reanalysis data. However...

    "When you download ERA5 data, the grid geometry of your output data depends on the selected data format:

    • If you request the data in the native GRIB format it is delivered with the above grid geometry.
    • If you request the data in NetCDF format, it is automatically converted and interpolated from the above grids to a regular lat/lon grid." 2

    Yet I have only used GRIB format and I have found that the resulting lat/lon coordinates change with small perturbations in the given subset, always starting with the North & South points I give it. This seems to go against what is given in the documentation.


    Would be curious to hear anything you've learned on this topic since your original post!


    1 ERA5: data documentation#Spatialgrid

    2 ERA5: What is the spatial reference