Skip to main content

We are trying to assign points centered within the sentinel grid on systems that dont have internet access.


Therefore, wondering if there is a way to retrieve just the raw sentinel grid (of say the world) without pulling any data?

If you mean the Sentinel-2 tiling grid you can find it on ESA pages.


Direct link to kml file: https://sentinel.esa.int/documents/247904/1955685/S2A_OPER_GIP_TILPAR_MPC__20151209T095117_V20150622T000000_21000101T000000_B00.kml


If you want to know if it is possible to somehow calculate this grid from scratch, I do not know.


Im looking for something finer than that. And forgive me for i dont know exactly what im looking for.


For example:
0fbadd74-d3e5-4d2a-8e03-369c705c3c31_2021-05-01_2021-06-01_multiband.tiff (3.7 KB)


When we retrieve seperate images from sentinel the pixels all seem to fall within the same grid as it it does in the following image



That grid is what im looking for (20x20m) pixel grid


The Sentinel-2 10m bands’ pixels (from the same UTM zone) are all on the same 10m grid, which you can get just by rounding to nearest multiply of 10. So in the Sentinel-2 tile UTM zone, the pixels from 10m bands will always have a 0 as last digit.


For the rest of the bands it is not so straightforward to know where are they; e.g. for a given Sentinel-2 tile, the 20m bands could be on …00, …20, …40, or …10,…30,…50, …, and it is even worse for 60m bands. To know the appropriate 20/60m pixel grid, one has to know the specific tile offset: the upper left corner of the tile, which is the origin to start your 10/20/60m grid.


In principle, each Sentinel-2 (MSGR) tile (e.g 32VNJ) has the same offset, but ESA has already changed this in the past. You can get the offset from the (UTM) bounding box of the tile that you can retrieve with CatalogAPI (or WFS).


Hope that gives you an idea what you are after.

Matej


To clarify. It would be SUPER helpful to have a shapefile or KML of 60m, 20m, and 10m pixel grids that Sentinel uses precisely so that one can ensure alignment prior to fetching any tiles or pixels.

Do you all have that which you could share? I see it as a VERY useful companion to the tiling grid you all already post.

This would also help address other issues users have had in terms of aligning mappings between different pixel grids (e.g., 10 m/px resolution for all Sentinel-2 bands : how? - #10 by avrecko)


As said, the tile grid (and consequently the tile offsets) has already changed in the past, and might in the future and it is not up to us. Constructing the (10/20/60 m) pixel grids in advance is not on our to-do list, particularly as the solution to get the grid is possible (the gist of it is described in my reply above).


As it seems you would find such a grid very useful, I encourage you to use CatalogAPI to iteratively fetch Sentinel-2 tiles to find their offsets, and then construct the grid.


Reply