0

I have a polygon shapefile that defines a very large ROI. In order to cover this ROI, I need to create a mosaic of rasters (Sentinel-1 images). The mosaic needs to be made with minimum number of rasters and should cover the ROI completely.

If I do a search for raster data based only on ROI, it gives multiple rasters that satisfy the ROI coverage but would lead to redundant data. It means that ROI that can be covered by a mosaic made of 3 images is covered by a mosaic made of 15 images.

The ROI is of continental scale and thus mosaic will be made from more than 1 raster.

I want to determine a workflow by which I can create a mosaic with minimum number of Sentinel-1 images that can overlap the ROI with minimum redundancy.

I think to progress in this scenario, I would do the following:

  1. Use the rasters with least overlap amongst themselves, but cover ROI shapefile. This would create a mosaic with minimum number of raster tiles.

Any suggestions in python, qgis, arcmap?

webapp
  • 710
  • 1
  • 6
  • 16
  • 1
    Just a bit more context: You could query via sentinelsat in Python (queries only scenes relevant for you aoi, no scenes that would be completely nodata within your aoi). This also returns the footprint polygon for your overlap calculation. Then use geopandas to merge these polygons itertively until you find the polygon combination with 100% aoi coverage but as little as possible polygons and area outside of the aoi. – Christoph Rieke May 18 '19 at 15:01
  • Hi @ChristophRieke, thanks for that info. Yeah I have been using sentinelsat for now to get the tiles in the ROI. But I will add the iterative merge step for optimal coverage. – webapp May 19 '19 at 20:01

0 Answers0