Resolved issue in WFDE5 near-surface specific humidity

In December 2021, an issue was identified in the Near surface meteorological variables from 1979 to 2019 derived from bias-corrected reanalysis v2.0 dataset (aka WFDE5 v2.0).

We are pleased to inform our users that the Copernicus Climate Change Service (C3S) at ECMWF has now released an updated version of the data (WFDE5 v2.1). For details about the issue and the solution implemented, please refer to the 'Known Issues' section of the Product User Guide

For any enquiries about this, please contact us.

ECMWF Support

Dear Anabelle,

Thanks a lot for pointing out this particular issue with the near-surface-specific humidity variable since I think it might have an impact on some research we perform using the WFDE5 dataset. In particular, I used WFDE5 variables (v1.0; when that was still available) including near-surface-specific humidity, and downscaled/bias-adjusted selected CMIP6/SSP projections (for some research on climate change impacts).

Note that I used the "problematic" variable as an intermediate step to derive relative humidity. However, checking the provided source codes for v1.0 and v2.0, I see that there is no substantial difference between the methods used to calculate near-surface-specific humidity. Therefore, I assume that the abovementioned "error" was also present in WFDE5 v1.0. If this is true, then does the problem occur for all cases where T<0 or at particular regions/times? Is it possible to briefly mention some of the affected regions/time periods you have identified, please?

Many thanks!

Best regards,