According to the Sentinel-2 user guide here (https://sentinel.esa.int/web/sentinel/user-guides/sentinel-2-msi/naming-convention), the compact naming convention is arranged as follows:
- MMM_MSIXXX_YYYYMMDDHHMMSS_Nxxyy_ROOO_Txxxxx_.SAFE
The second date is the "" field, which is 15 characters in length, and is used to distinguish between different end user products from the same datatake.
For a single tile, there could be multiple product discriminators. Here I have these two different product discriminators for the same tile 51TYN:
- S2A_MSIL2A_20190509T023551_N0212_R089_T51TYN_20190509T045217.SAFE
- S2A_MSIL2A_20190509T023551_N0212_R089_T51TYN_20190515T163551.SAFE
I checked out the pixel values in ENVI and the corresponding pixel pairs have slightly different values.
To do a montly median composite using the timeseries images, the same location in different image in such a case should be only counted once. Since these two images are taken from the same tile, which one should I use if I want to do a montly composite?
Hello
In my opinion, if the DN images are very slightly different from each other, you can combine it with the composite command or use the Cell Statistics tool in GIS software.
The various statistical calculations that can be performed with this tool are:
Majority
Maximum
Minimum
Mean
Median
Minority
Range
Standard Deviation
Sum
Variety
Dear Haijun Li please consider as valid only the following product: S2A_MSIL2A_20190509T023551_N0212_R089_T51TYN_20190515T163551.SAFE
The other one was wrongly processed and published but then replaced by the corrected one. Have you found any other example like that? Thanks
Nasir Hameed the reference for the correct published Sentinel-2 core products is the Copernicus Open Data Access Hub (https://scihub.copernicus.eu/). Here you can see that only the second one is published.
Could indeed happen that products generated with wrong processing parameters are temporarily published and then replaced by the corrected ones. The problem is that the data access relays such as Google, Amazon, etc. do not always delete the old (wrong) ones but keep them all online.
Dear Enrico Giuseppe Cadau ,
I have found more examples like this:
- 51UYQ - 20190506
- S2A_MSIL2A_20190506T022551_N0212_R046_T51UYQ_20190506T043625.SAFE
- S2A_MSIL2A_20190506T022551_N0212_R046_T51UYQ_20190506T044236.SAFE
- S2A_MSIL2A_20190506T022551_N0212_R046_T51UYQ_20190514T184421.SAFE
- S2A_MSIL2A_20190506T022551_N0212_R046_T51UYQ_20190526T102533.SAFE
- 51UYQ - 20190509
- S2A_MSIL2A_20190509T023551_N0212_R089_T51UYQ_20190509T045217.SAFE
- S2A_MSIL2A_20190509T023551_N0212_R089_T51UYQ_20190515T163551.SAFE
There are four different product discriminators for the same tile (S2A_MSIL2A_20190506T022551_N0212_R046_T51UYQ). I have no idea which one should be the corrected product.
I downloaded these L2A products from Google Cloud Platform instead of ESA. You may find these images for the tile on the website (https://console.cloud.google.com/storage/browser/gcp-public-data-sentinel-2/L2/tiles/51/U/YQ?pageState=(%22StorageObjectListTable%22:(%22f%22:%22%255B%255D%22))&prefix=&forceOnObjectsSortingFiltering=false). And I have listed more such examples in the attached file ' Sentinel-2-product-discriminators-examples.txt'
I could not find the person who is in charge of the L2A products release on Google Cloud Platform to report these issues. How could we find the corrected product? Are there any conventions and rules to determine which one should be used?
Thanks.
Dear Enrico Giuseppe Cadau ,
I found both two products released on the Copernicus Open Data Access Hub (https://scihub.copernicus.eu/dhus/#/home):
- S2A_MSIL2A_20190506T022551_N0212_R046_T51UYQ_20190506T043625.SAFE
- S2A_MSIL2A_20190506T022551_N0212_R046_T51UYQ_20190506T044236.SAFE
- S2A_MSIL2A_20190506T022551_N0212_R046_T51UYQ_20190514T184421.SAFE (released on SCIHUB)
- S2A_MSIL2A_20190506T022551_N0212_R046_T51UYQ_20190526T102533.SAFE (released on SCIHUB)
However, the thumbnails show different coverages. Does that mean we should use both products for this tile 51UYQ-20190506?
Thanks,
Dear Haijun Li the last example provided is a mix of different cases.
Sometimes a Sentinel-2 tile can be dumped onto two different ground stations, in this specific case two different complementary products are independently generated and published. This is a nominal behaviour of the Sentinel-2 ground processing centre. You can indeed see that they have different generation date but same sensing time. So yes, for such a case you should "merge" the two products to get the complete one.
Anyhow if you find a mismatch in the product discriminator (generation date) between google/amazon and the SCIHUB to be 100% sure of the correctness of the product and for the reason mentioned in my former post, please refer to what is published on the SCIHUB.
Dear Nasir Hameed , for instance please check this news: https://scihub.copernicus.eu/news/News00523
The above is just an example which can lead to synchronisation error between the SCIHUB and the third-parties S2 data products relays.
Best
- Badges
- Science topic
- Similar topics
- Computer Science
- Images