Optimising Sentinel-2 image selection in a big data context

Processing large amounts of image data such as the Sentinel-2 archive is a computationally demanding task. However, for most applications, many of the images in the archive are redundant and do not contribute to the quality of the final result. An optimisation scheme is presented here that selects a subset of the Sentinel-2 archive in order to reduce the amount of processing, while retaining the quality of the resulting output. As a case study, we focused on the creation of a cloud free composite, covering the global land mass and based on the images acquired in 2016. The total amount of available images was 635,096 with an average of 34 overlapping images per tile. The selection of the optimal subset was based on quicklooks, which correspond to a spatial and spectral subset of the original Sentinel-2 products and are lossy compressed. They typically represent only 0.05% of the data archive volume. The result of the proposed selection scheme was a reduced set of images, with an average size of 2.55 images per tile.

KEMPENEERS Pieter;  SOILLE Pierre; 

2017-12-19

Publications Office of the European Union

JRC108616

978-92-79-73527-1,   

1831-9424,   

https://ec.europa.eu/jrc/en/publication/proceedings-2017-conference-big-data-space,    https://publications.jrc.ec.europa.eu/repository/handle/JRC108616,   

10.2760/383579,