L2A "tiles" do not always correspond to the same products as L1C


I’ve noticed that tiles for the same day may be different from L1C and L2A and even have different coverage.

I only stumbled upon one such tile, stored under the bucket key “tiles/33/X/XB/2018/9/13/0”, but there may be many more, so I decided to give a heads up.

In L1C its product is listed as “S2B_MSIL1C_20180913T105019_N0206_R051_T33XXB_20180913T125630”, while in L2A - “S2A_MSIL2A_20180913T113651_N0208_R123_T33XXB_20180913T141441” - products taken at different time by different sensors!

I decided to check if there is another tile stored for that date, and yup, there is a tile by the bucket key “tiles/33/X/XB/2018/9/13/1”, for which the situation is vice versa, so L1C’s “0” is “1” in L2A, and “1” is “0”.

So I wonder if there isn’t supposed to be exact correspondence in terms of tiles mapping between L1C and L2A buckets, or if it’s indeed a bug?

There is no exact mapping between L1C and L2A in terms of the pats. You should compare product identifiers.
The index (/0, /1,/2,…) is assigned based on order of how scenes are ingested. First scene from specific tile ingested on specific date gets /0 and so on.