Orography variable name change

Question

Orography variable name change

Closed this issue 3 years ago · 6 comments

The variable name of orography has been changed by the CDS.

WARNING 2021-06-25: Variable "Orography" is now named "Geopotential". No change in the data themselves. Previous API requests asking for "Orography" will fail now. To download the corresponding data the API request should ask for "Geopotential".

era5cli already contains a variable called Geopotential which is not the correct one (multi pressure level instead of single level).

Answer 1 · 2021-09-08T12:03:04.000Z

Thanks for bringing this up. Can you link to the CDS source that explains this change?

Answer 2 · 2021-09-09T14:25:05.000Z

https://cds.climate.copernicus.eu/cdsapp#!/dataset/reanalysis-era5-single-levels?tab=overview

Answer 3 · 2021-09-13T09:30:48.000Z

geopotential is now available both on pressure levels and surface level. We need a way to tell era5cli which one is meant.

Answer 4 · 2021-09-14T09:51:37.000Z

One workaround would be to make "orography" work again for era5cli by redirecting it (under the hood) to geopotential at surface level. We could add a warning that this behaviour deviates due to changes in ERA5 in CDS.

Another option (perhaps in tandem) could be to make "geopotential" with specified level "surface" or something like that work.

We should be sure to document this carefully and perhaps also print an info message whenever the user requests geopotential, so as to avoid confusion and to keep the tool as intuitive as posssible.

Answer 5 · 2021-10-14T09:02:39.000Z

In PR #98, @Peter9192 suggested:

Would it make sense to have flags 2Dvars and 3Dvars (like era5cli info 2Dvars) instead, for consistency? If not given, let the tool automatically try to infer from the chosen variable(s) (like before). If ambiguous, raise and point to these flags?

I like this suggestion, as well as the consistency it provides.
I'm not sure if you mean this, but I would like the tool to raise a warning, not an error. The reason for this is backwards compatibility: a query that includes 'geopotential' without the --3Dvars flag then would not work, where in the past it did.

Answer 6 · 2021-10-14T10:00:10.000Z

Alternatively, we could add an option to the list of pressure levels (--levels surface ) to indicate that you want to take the variable from the single-level dataset instead of the pressure-level data.