Hello colleagues! I make WCS request (Sentinel2-L1C) for some area (bbox and geometry are set) and want to get available dates for given date range. And these dates looks strange for me:
2019-08-20 08:44:35+00
2019-08-20 08:44:50+00 (close to previous)
2019-08-23 08:54:32+00
2019-08-23 08:54:46+00 (close to previous)
2019-08-25 08:44:46+00 (only one result for this day!)
2019-08-28 08:54:28+00
2019-08-28 08:54:42+00 (close to previous)
(that’s only part of typical response)
As you can see, I have two results for almost each acquisition date (with just a few seconds between them). Can anyone explain me why it happens this way? And how should I choose the right one?
P. S. Just in case: I make requests with a sentinelhub Python library.
The swaths (roughly north-south) strips which Sentinel 2 records are split into sections. This happens when you have a bounding box which crosses the edge of these sections (I’m not sure the technical term).
For this reason, there is a “time difference” parameter you can set to merge the WCS results from different products so you don’t see the split.
@barrett is right.
What you see are the timestamps of “scenes” (granules), which is an artificial section of the orbit.
Sentinel-2 orbit takes approx 100 minutes. Unless you are close to the poles, you can easily assume that ±100 minutes is the same orbit. At poles, it is better to go with half of this time interval.