Missing `conditions/meteorology` group in many cpm v2.6.2 Sentinel-2 L2A datasets

[emmanuelmathot]

Summary

Many Sentinel-2 L2A datasets produced with cpm_v262 around the 20 Feb are missing the /conditions/meteorology group (containing CAMS and ECMWF atmospheric data) that was present in earlier versions (e.g. cpm_v261). This breaks our validation model which expects conditions/meteorology as a required member of the conditions group.

Affected dataset

• Product: S2A_MSIL2A_20260220T090931_N0512_R007_T35SNA_20260220T115809.zarr
• Source URL: https://objects.eodc.eu:443/e05ab01a9d56408d82ac32d69a5aae2a:202602-s02msil2a-eu/20/products/cpm_v262/S2A_MSIL2A_20260220T090931_N0512_R007_T35SNA_20260220T115809.zarr
• CPM version: v2.6.2

Expected structure

Based on earlier cpm versions (cpm_v256 and earlier), the conditions group has the following structure:

conditions/
├── geometry/
│   ├── sun_zenith/
│   ├── sun_azimuth/
│   ├── viewing_zenith/
│   └── viewing_azimuth/
├── mask/
│   └── ...
└── meteorology/        ← MISSING in cpm_v262
    ├── cams/
    │   ├── total_ozone
    │   ├── relative_humidity
    │   └── ...
    └── ecmwf/
        ├── temperature
        ├── pressure
        └── ...

Actual structure (cpm_v262)

conditions/
├── geometry/
│   └── ...
└── mask/
    └── ...

The meteorology subgroup is entirely absent.

Questions for the data provider

1. Is the removal of conditions/meteorology intentional in cpm_v262? If so, should downstream consumers treat meteorology as permanently optional going forward?
2. Is this a temporary issue specific to certain datasets or processing runs, and will meteorology be restored in future deliveries?
3. Does this affect all cpm_v262 products, or only specific tiles/orbits?

[clausmichele]

Hi @emmanuelmathot, I just tried to convert that product locally with cpm 2.6.2 and 2.6.4. I see the conditions/meteorology group in both versions! So probably something went wrong during the automatic conversion pipeline, we need to investigate.

[senmao]

yes, I can confirm that the /condition/meteorology group exists in S2A_MSIL2A_20260220T090931_N0512_R007_T35SNA_20260220T115809 scene after reconversion with CPM 2.6.2. So this problem is temporary for certain scenes.

@christophreimer @SerRichard To resolve this issue, the problematic scenes need to be found and reprocessed. Maybe move this into a PMP issue.