Sentinel-1 data management
IOTA² use every Sentinel-1 available dates to produce stacks without speckle. Actually, 4 stacks are generated considering mode acquisition
and polarization
: ASCENDING_vv, ASCENDING_vh, DESCENDING_vv, DESCENDING_vh.
In overlapping areas, inside orbit which belong to the same mode, this choice can lead to miss-computing data. It has been noticed than mixing orbits during speckle filtering and gapfilling highlighted some temporal patterns.
The purpose of this issue is to propose a new Sentinel-1 filtering work-flow in IOTA²
1 - as usual, orthorectify Sentinel-1 scenes.
2 - by orbits (relative orbit from filename), mode and polarization : produce the outcore raster Ex : outcore_DESC_vv_81.tif
and a stack without speckle.
3 - by mode and polarizations, chronologically merge filtered stacks
4 - to a given stack (Ex: filtered_DESC_vv.tif) get every masks in order to apply temporally re-sample it. By the way, erode them half the overlapping distance to remove overlapping between orbits.
5 - temporally re-sample stacks to a smart grid which stay to be discuss.
Step 3 is very time-consuming
, actually to be achieved a lot of otb's applications will be called (extractROI, concatenates...). An OTB application dedicated to merge time series could be created.
Step 4 is not easy to implement
: may the Sentinel-1 tracks could change ? acquisition segments