Recent issues in three-hourly 2m temperature analysis data fields CARRA-East

Datasets and Usage C3S - Datasets and Usage
1

Hi all,

I found an issue with respect to recent (2025) three-hourly 2m temperature analysis data fields in CARRA-East and another issue common for many variables.

  1. In some months there appear from time to time bias fields in the form of circles with a radius of about 400 km. One type of these circles have increasing absolute biases to the center/pole of the circle. There is one stationary “plus pole” circle east of Svalbard. Another “negative pole” is moving with time. It seems to pop up the first time end of May 2025 with the center at the north coast of Scandinavia. During the following days it moves quickly to the west and later back and forth the same way until mid of June. One month later it appears for about 40 days in the area between Jan Mayen, west coast of Norway and Svalbard. This circle has some area in the east, which is not affected. Plots examples below are for August 16, 2025, 0 UT and June 3, 2025, 6 UT.

    carra-east_2mTemp_analysis_three-hourly_2025081600
    carra-east_2mTemp_analysis_three-hourly_2025081600398×518 121 KB

    carra-east_2mTemp_analysis_three-hourly_2025060306
    carra-east_2mTemp_analysis_three-hourly_2025060306401×522 115 KB

    Another type of this circle seems not to have that pole feature. It is stationary with the center in the southern part of Novaja Semlja. The plot below shows the data from March 1, 2025, 6 UT.

    carra-east_2mTemp_analysis_three-hourly_2025030106
    carra-east_2mTemp_analysis_three-hourly_2025030106399×521 110 KB

    Instead of the pole feature these circles exhibit sometimes some stripes as in this case, where stripes (or at least “straight” lines) are going from the center to the top, to the right, to the bottom, and to the north. Another common stripe pattern is also visible in the second plot above (June 3, 2025, 6 UT) with a narrow stripe going to the left and a broad one to the southeast. Here, the circle feature is hard to see. I haven’t investigated whether this issue appears in other years, in CARRA-West, or in other variables, too.

  2. Data of the border cells of the model should not be used. As can be seen in the plots above, they obviously exhibit some biases. This is also true for (many) other variables. I am sure this is an intrinsic feature of models with borders and unavoidable. However, it might be worthwhile to mention that in the known issues page (or to set these data points to undefined).

I don’t want to forget to thank everybody involved for this great data set (and for CARRA2, too)! This is very much appreciated!!

Best regards

Peter

2

Dear Peter,

Thank you very much for using the CARRA datasets and sorry for the problems found. I contacted the CARRA team to check the details of the issue and we aim to come back on this later.

Meanwhile, I just mention that indeed we can emphasise more the possible impacts of the lateral boundaries. It was not very clearly spelled out for CARRA1, but for the new CARRA2 (pan-Arctic regional reanalysis) we have already a Known issues item mentioning this issue:

I also use this opportunity to draw your attention to the new CARRA2 dataset Catalogue — Climate Data Store, where we have already published 24 years of data and the dataset will be completed until mid-2026 and published in the CDS (we also had a small article on this if you are interested in: First CARRA2 data release offers new insights into Arctic extremes | Copernicus ).

Thanks again and I hope to come back on this issue soon!

Best regards

                Andras Horanyi