Dear Halldór,
Thank you very much for looking at the CARRA dataset and pointing to a specific problem.
We have recently realised that there are some slight inconsistencies between the data files produced before and after July 2022. The data from July 2022 was produced on our Atos computer, while the older ones on our previous Cray computer. The differences are coming from some differences in some auxiliary software, while the model is unchanged (so scientific consistency is ensured).
We have a similar Forum note at CARRA East - grid change in 2023 and after - #2 by Andras_Horanyi, which points to the same problem.
As a next step our intention is to fully understand the extent and details of the problem and then document it for the benefit of the users.
Thanks again, best regards
Andras Horanyi