dhi-gras / majalicious

:lollipop: Orchestrate MAJA by creating a lot of symlinks

majalicious

Run MAJA V1 on Sentinel 2 data by creating a lot of symlinks

Dead

NB: MAJA V1 does not work for data past 6 November 2018, where Copernicus changed L1C format. 💥

Purpose

This little script automates the processing of Sentinel 2 timeseries with MAJA V1. It indexes by date all L1C and MAJA-produced L2A products in the folders you provide and creates symlinks to the ones MAJA needs to run in L2NOMINAL or L2BACKWARD mode.

This way, you e.g. always have the most recent L2A product for the current L1C product you are processing and you do not need to move any files around when starting to process a new scene.

Don't worry, it's not malicious. It's majalicious!

Building

docker build -t maja

Running

To see the full interface, run

docker run maja

For an actual run, you need to mount the input and auxillary data into the docker container. The default is that you have, inside the container, a folder /maja-aux with the following contents:

/maja-aux
  userconf
  GIPP
  DTM/{tile}

If you follow this pattern, you can omit the --src-userconf, --src-gipp, and --src-dtm flags.

Furthermore, you need to provide a --src-input directory that contains your .SAFE input files and a --dst-output directory for the L2A output.

The --dst-work-root directory must be inside the container (i.e. not on a mounted resource) for the symlinking to work.

A full example command is the following:

docker run -v /my/maja/aux:/maja-aux -v /my/maja/data:/maja-data maja --src-input /maja-data/input --dst-output /maja-data/output --tile 32UNG

Docs

This script is a re-implementation of the recipes at https://github.com/olivierhagolle/Start_maja

Please consult that repo for more information on why things are done this way.