Step 1: Identical image definition

[whisperingpixel]

Sub-tasks: 1.) Find a definition of identical images 2.) Define requirements for what the id should reflect (version, acquisition time, ...), i.e., images should be traceable by their id, e.g, when they have the same footprint, but were captured at different times.

[whisperingpixel]

Characteristics for identification of identical images

I suggest to use the following characteristics:

• acquisition time +/- 5 minutes
• Granule identifier
• relative orbit

The granule identifier itself is not enough as they might have been acquired in different relative orbits. For example:

Granule identifier 33TUN with relative orbit 122:

Granule identifier 33TUN with relative orbit 22:

[augustinh22]

I would recommend granule identifier, date+timestamp +/- 5 minutes and S2A/S2B.

Then we have: where-when-satellite (the order like this since we are mostly interested in where and when, but it doesn't matter so much since we have all the metadata for searches)

Adding relative orbit would be fine but redundant, since a second orbit of the same granule cannot be acquired by the same satellite within 10 minutes.