This package is a project by Marlin M. Mueller and Jonas Ziemer for the module GEO419 of the M.Sc. Geoinformatics at the Friedrich-Schiller-University Jena. It provides a framework for importing 3D-raster files in rasterio-compatible formats and provides basic filtering and statistical functionality. It was developed mainly for fire and agricultural breakpoint detection
Basic functionality includes:
- Importing time series data and preliminary light cleaning of data (specifically developed for Sentinel-1 stacks created in ENVI, but should work with all kinds of time series stacks)
- Applying different statistical functions along time axis of data to retrieve different metrics on a time scale
- Applying different filter functions along time axis of data to enable better detection of breakpoints
- Applying breakpoint detection functions based on filtered time series
- Creating new raster files to visualize applied functions
- Working in parallelized, multi-threaded way for efficient computation (6x faster with 16 threads vs 1 thread)
- Uses max. two times the amount of system memory of the input file size
- Easy expansion with new functions is possible
Developed in Python 3.8
In case you have git installed you can install the package as follows:
pip install git+https://github.com/marlinmm/BanDiTS.git
If not and you have trouble to find a way, please open an issue.
If you have trouble installing rasterio or the needed GDAL package on Windows, download and install the .whl files directly from here.
You can find an example notebook containing explanations for the general functionality here.
- For plotting of the result maps in the notebook the package matplotlib is required and the package skimage is recommended
- If you have trouble viewing the notebook on github, try viewing it here.