Project

General

Profile

New ECMWF CDS beta NETCDF formats and usage of NCO and CDO

Added by Howard Van Meer 3 months ago

Dear members,

I am a PhD candidate at Wageningen University, and I am currently working with the NETCDF files available through the ECMWF CDS Beta platform for daily single values seasonal forecasts (https://cds-beta.climate.copernicus.eu/). I've noticed that these files are now using CF 1.7 convention formats, specifically with the forecast reference time included as both a dimension and a variable. Additionally, the time units in these files are specified as "seconds since 1970" rather than the more commonly used "days since 1900."
Unfortunately, these changes have caused compatibility issues with the NCO and CDO commands that are critical for my work. For example, CDO commands like shifttime and deltat , which I typically use to obtain daily precipitation or radiation data, are failing to process these files correctly under the CF 1.7 format.
Given these challenges, I would appreciate your guidance on the following points:
1. Is there an anticipated update for CDO that will support CF 1.7 conventions in the near future?
2. Are there any recommended workarounds for dealing with these files in the meantime, Since I rely heavily on NCO and CDO for data manipulation?
I understand that this is a Beta version, and such issues are to be expected. However, any advice or solutions you could offer would be greatly appreciated.
Thank you for your attention to this matter, and I look forward to your response.
Best regards,
Howard van Meer
PhD Candidate, Wageningen University


Replies (1)

RE: New ECMWF CDS beta NETCDF formats and usage of NCO and CDO - Added by Uwe Schulzweida 3 months ago

Dear Howard,

Unfortunately there is no way to read these NetCDF files with CDO. And this will not change in the future, sorry.
The problem is that these NetCDF files are not compliant with the CF convention. The first dimension is the number of ensemble members. Ensembles are not supported in the CF standard.
And coordinate variables such as “forecast_reference_time” are only provided in addition to a mandatory time coordinate. So these are only local extensions to the CF standard.

Best regards,
Uwe

    (1-1/1)