OPERA: Difference between time_start and acquisition time

[earthengine_urs]

We recently noticed slight differences between the value of the "time_start" field of a granule and the acquisition time encoded in the filenames for OPERA. For example:

Code: Select all

{
  "feed": {
    "updated": "2026-06-03T13:46:54.920Z",
    "id": "https://cmr.earthdata.nasa.gov:443/search/granules.json?online_only=true&collection_concept_id[]=C2949811996-POCLOUD&readable_granule_name[]=OPERA_L3_DSWx-S1_T45TYE_20241015T235952Z_20260415T225516Z_S1A_30_v1.0",
    "title": "ECHO granule metadata",
    "entry": [
      {
        "producer_granule_id": "OPERA_L3_DSWx-S1_T45TYE_20241015T235952Z_20260415T225516Z_S1A_30_v1.0",
        "boxes": [
          "39.6513564 89.2750399 40.6719183 90.6053959"
        ],
        "time_start": "2024-10-16T00:00:07.000Z",

The filename's 2024-10-15T23:59:52Z disagrees with the CMR record's 2024-10-16T00:00:07Z, which we only noticed because it crossed a day boundary and broke some assumptions that we made about the relationship between the two fields. We've since built a workaround but were wondering if this is expected, or if we've misunderstood any subtle difference between the time_start and acquisition time.