Sentinel1-Sentinel2-ARD

Creates analysis read data (ARD) from sentinel data. Creates multi-modal and/or multi-temporal Sentinel-1 and Sentinel-2 ARD.

Details can be found in the peer paper here:

A Flexible Multi-Temporal and Multi-Modal Framework for Sentinel-1 and Sentinel-2 Analysis Ready Data

The paper was published using v1.0.1 of the code, available here v1.0.1.

If you find this repository useful, please cite with:

@article{s1s2_ard,
  AUTHOR = {Upadhyay, Priti and Czerkawski, Mikolaj and Davison, Christopher and Cardona, Javier and Macdonald, Malcolm and Andonovic, Ivan and Michie, Craig and Atkinson, Robert and Papadopoulou, Nikela and Nikas, Konstantinos and Tachtatzis, Christos},
  TITLE = {A Flexible Multi-Temporal and Multi-Modal Framework for Sentinel-1 and Sentinel-2 Analysis Ready Data},
  JOURNAL = {Remote Sensing},
  VOLUME = {14},
  YEAR = {2022},
  NUMBER = {5},
  ARTICLE-NUMBER = {1120},
  URL = {https://www.mdpi.com/2072-4292/14/5/1120},
  ISSN = {2072-4292},
  DOI = {10.3390/rs14051120}
}

Running the pipeline

An example configuration is provided which will only use a single pair of (S1, S2) patches. This covers a Scottish city, Glasgow, as an example.

The code can either be ran using a python environment, or via a provided docker image.

The script needs some flags passed alongside the configration file, in order to specificy parameters for each specific run

• --config <config_filename>
  • a json file containing the region of interest and other region/dataset-specific parameters
• --credentials <sentinelsat_credentials_json>
  • json containing a username and password for access to sentinelsat.
• --njobs <integer>
  • how many parallel jobs to run
• --rebuild true|false
  • run all steps even if outputs already exist
• --outdir <path>
  • Where to save all outputs

The download_flow has different parameters

• --config <config_filename>
  • a json file containing the region of interest and other region/dataset-specific parameters
• --credentials <sentinelsat_credentials_json>
  • json containing a username and password for access to sentinelsat.
• --credentials_earthdata <earthdata_credentials_json>
  • json containing a username and password for access to noaa earthdata.
  • for downloading S2 products when product is archived in sentinelsat
• --credentials_google <earthdata_credentials_json>
  • json containing a username and password for access to google cloud.
  • for downloading S1 products when product is archived in sentinelsat

Running via python

If you already have the requirements installed locally, the script can be ran as follows:

python -m src.snap_flow --no-pylint run --config glasgow-s256x256-o0x0-20200801.json --credentials <sentinelsat_credentials_json> --njobs <how_many_jobs_in_parallel> --rebuild true|false --outdir <directory_to_save_output>

Running via python inside docker

A Dockerfile is supplied in the root of the repository. It's based on ubuntu:20.04 and python3.7. It installs GDAL and SNAP, among other python requirements.

From the root of the repository, the docker image can be built using:

docker build --network=host -f Dockerfile -t senprep:latest --force-rm=True

...which will create an image tagged senprep:latest. (You can also install via make senprep, which simply runs the above command).

The provided script in helpers/docker-snap allows you to run the command in various ways (download only, download and process, process only). If ran without any arguments, it will output help documentation.

To run the pipeline directly, you must provide two mounts to the docker image:

1. where the data will be exported
   • ex: to store in /temp/satellite, use -v /temp/satellite:/var/satellite-data/
2. the current working directory (for loading configurations etc).
   • -v $PWD/:/code/

sudo docker run --rm -it -v /temp/satellite:/var/satellite-data/ -v $PWD/:/code/ senprep:latest download_and_snap --credentials credentials.json --config configuration.json