odhondt / eo_tools

A python toolbox for easily searching, downloading & processing remote sensing imagery from various public sources.

EO-Tools:

A pure python toolbox for easy programmatic processing of remote sensing imagery from various public sources.

New feature: standalone TOPS InSAR processor

• Bursts from Sentinel-1 TOPS InSAR pairs can now be processed and combined without using SNAP.
• A new processor for IW subswath can be used -- see s1-easy-tops-insar.ipynb in the notebook folder.
• Bursts can also be processed individually using the S1.core module for more flexibility. Results can be further processed as in-memory arrays or written as GeoTiff files. An example of such processing is found in s1-tops-core-demo.ipynb
• More features/improvements will be added in the near future such as:
  • Better memory and multi-core handling (probably using dask)
  • Full product (all subswaths and polarization) processing with optional AOI selection
  • Multi-temporal stacks
  • Radiometric terrain correction
• In a more distant future:
  • Tools for change detection, polarimetry, speckle filtering, DEM generation

Here are examples of amplitude, phase and coherence computed using this framework:

Overview

• Because this project is in active development the API is likely to change.
• Currently, the available features are:
  • Sentinel-1
    • New standalone InSAR processor (see previous section)
    • Legacy InSAR processor (running SNAP graphs through PyroSAR) computing the coherence, phase and intensities of an interferometric pair of SLC products
    • Write the result as a geocoded (terrain corrected) COG (Cloud Optimized GeoTIFF) file
    • Display these rasters on top of a folium map in a jupyter notebook
  • Sentinel-2
    • Tile merging and geocoding
    • Write any band to COG files
    • Visualization of color composites (Natural RGB, CIR, SWIR, etc) on a folium map
  • DEM
    • Automatically downloads and crops a DEM given a geometry
  • All products
    • Search catalog (using EODAG) and download products
    • Explore products by displaying their footprint on a folium map (custom function)
    • Show remote and local images on top of folium maps in the notebook
• Example notebooks can be found in the notebooks/ folder

Install & usage

• The package comes in two flavors
  • A conda package that contains the main functionality (Sentinel-1 InSAR, Sentinel-2 tile mosaic and DEM download)
  • A docker container (for more advanced users) that additonally works with a TiTiler server for interactive visualization in the notebooks
  • The legacy SNAP based processor is only available in the docker version.

Conda

• It is recommended to first create a conda environment to avoid package conflicts

conda env create -n eo_tools
conda activate eo_tools
conda install conda-forge::eo_tools

Docker

• It works as a dev container for VSCode.
  • Clone into the location of your choice.
  • Volumes paths can (and should) be changed in docker-compose.yml.
  • After opening the main directory, VSCode should detect the devcontainer file and ask to build the container. Once the container is running, the example notebooks can be used.
• Alternatively, it should also be possible to start the container from the main directory with docker-compose up -d in a terminal and attach to the container with any editor supporting docker.

Getting started

• Please make sure jupyter is installed in your environment
• Example jupyter notebooks demonstrate the different features
• For conda use the notebooks in the notebooks-cf directory
• For docker use the notebooks in the notebooks directory

Notice

• This project was originally forked from: https://github.com/eo2cube/s1_processor/, however since 99% of the code is now original, I have detached the fork.
• Visualization functions are using TiTiler https://developmentseed.org/titiler/
• Product discovery and download are using EODAG https://github.com/CS-SI/eodag
• The old S1 processor uses pyroSAR https://github.com/johntruckenbrodt/pyroSAR which executes graphs with ESA's SNAP software https://github.com/senbox-org