I'm building a simple mosaic of Landsat-8 for a region of interest (a bounding box) for a time period, so that the latest data would be put on top of the mosaic. I've found 100 scenes with cloud cover from 0 to 5% that intersect with the ROI, so I have a list of 100 unique geometries, sorted by an acquisition date. Now I want to determine which scenes are "on top" of the mosaic in order not to download the scenes that won't be visible. I wonder what is the most efficient way to find the top-most geometries with the least number of intersections? The difficulty is that I've sorted images by cloud cover, so now some images in strips are missing

  • 1
    If you have an earth engine account, check example algorithms. There is an example doing exactly what you want to obtain (in ImageCollection section) – aldo_tapia Sep 22 '17 at 12:41
  • @aldo_tapia, I don't have an account yet, but I believe here's the example developers.google.com/earth-engine/ic_composite_mosaic . I'm interested in algorithm rather than implementation. Maybe there is an open source solution? – dr_times Sep 22 '17 at 14:49
  • you can do it in R for example, masking by QA band and sorting by relative area of this mask. But you need to download all scenes to select the best one, thats is why GEE is a great option. You can sort by cloud cover, mask by cloud score and fill cloud holes with other scenes. Finally, you can download the output mosaic. I my opinion, is the best plataform to perform your task – aldo_tapia Sep 22 '17 at 15:31
  • Also, check this paper. Contains the necessary code to reproduce it for your purposes – aldo_tapia Sep 22 '17 at 15:33

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.