The project has been few years old already, and the underlying libraries are not easily reproducible for many people - from personal experience. It would be useful to develop a documentation for the library and publish in the github pages. The most viable solution seems to be mkdocs - https://mkdocs.readthedocs.io/

Due to phasing out of setuptools in preference of pyproject, the install does not install correctly using pip install .. To fix this problem, the build system needs an update.

The default 'fmt' parameter for watermask reader function is 'tiff', whereas the provided exemple is using 'tif' suffix, leading to an execution error.

Reference line:

The code currently uses print() function to print messages. The better option is to replace the print with logging module.

The Sentinel-2 data from scihub.copernicus are provided in .SAFE format, unlike the .zip format from Theia. This format however comes with a data descriptor .xml file, which allows direct reading of the underlying dataset without unzipping.

The current examples are designed to only work with the Theia dataset and for most part of the world, Theia does not provide any processed data, hence Copernicus is the way to go. Consequently, for analyzing the Copernicus data, a dedicated data reader is necessary.

I have just tested a functional prototype of data reader. It does the job of reading for now. But in light of this changes, perhaps a more harmonized reader is warranted.

Marching square can be used to extract a more smooth shoreline from the the original segmented imageries. The wikipedia page is here - https://en.wikipedia.org/wiki/Marching_squares. An example of the application of marching square is here - https://www.mdpi.com/2077-1312/11/3/627

Currently there are API for earthdata (USA), as well as scihub (Copernicus project) which can be programmatically accessed via python. It would useful to add a notebook demonstrating the download chain.

Earth data api: https://earthdata.nasa.gov/collaborate/open-data-services-and-software/api
Sentinel api: https://sentinelsat.readthedocs.io/en/master/index.html

Often the zone of interest is small, which means analysing the full image tile is not really necessary. Having a clipping feature would be a nice improvement of the toolbox.

The current environment.yml file does not install several important libraries - among them jupyter notebook and netCDF4. Although the toolbox is currently in update, it requires an update.

Lon-lat values are switched in shoreline output. The bug is found while analyzing T30TXR (+1).

Currently the example scripts needs the folders to be created by the user before hand. It should be handled by the script directly. Needs fixing.

Currently the plotting functionalities are handled using basemap library, which is currently deprecated. The ideal solution is to replace the basemap dependency with cartopy.

matplotlib has been changed drastically beginning v3.3. Currently, the environment.yml list matplotlib=2.2.3 and matplotlib-base=3.4, which conflicts with the imports in the basemap. The last version that works with matplotlib and matplotlib-base is v3.2. It is needed to be updated.