This dasboard application based on the sepal-ui framework, provides a user-friendly interface for the use of the spatially cumulative sum algorithm for time-series change detection as proposed by Manogaran & Lopez. Its implementation has been taken from the SERVIR SAR handbook and has been optimised to run on GPU instances.
The module can be run with any single-band time-series (optical or SAR) generated through the SEPAL recipes menu. The module asks to provide the high-level folder.
The resulting output file can be found at:
~/module_results/cumsum/[name_of_output_folder]/cumsum_result.tif
The output file is a 3-band raster that contains the date of change (Band 1), its confidence (Band 2) and the magnitude (Band 3).
first download the repository to your own sepal account
git clone https://github.com/buddyvolly/cumsum_change.git
Then in the cumsum_change folder, launch the ui.ipynb notebook and run it with voila.
⚠️ If for some reason the sepal_ui module doesn't work on your instance, you can run theno_ui.ipynbfile as a notebook usingkernel->run all