Improvements in SAR image after gamma flattening with SarSen

Sarsen: Open source library for SAR satellite image processing

Project reference period: 2016 -­ on going
Open Source (partially sponsored by Microsoft)

Project Description

Sarsen is a collection of Open Source algorithms to process Synthetic Aperture Radar (SAR) satellite data in Python by B-Open. It currently supports geocoding and gamma flattening Sentinel-1 data on any DEM, and partially sponsored by Microsoft:

Sarsen GitHub repo

Sarsen comes with a command line tool that creates GTC and RTC images on the exact pixels of a reference DEM starting from Sentinel-1 GDR or SLC products. The quality of the geometric and the radiometric terrain-correction is very good and performance is acceptable, but will be improved in future releases.

The short-term aim is to support full cloud-native processing on Azure Planetary Computer straight off their upcoming Sentinel-1 GRD archive with user supplied DEMs.

Overall, Sarsen is currently in alpha, but mostly because we reserve the right to change the APIs. As far as we know the geocoding is already good enough for interferometric processing.

B-Open’s Role

B-Open has developed the foundations of the library as a side project in 2016 and has used it in a couple of minor projects. We have only managed to put it into real use in 2021 when Dan Morris (then AI4E at Microsoft) offered to sponsor porting the algorithms to the modern Python scientific stack, adding the gamma flattening algorithm and releasing it as Open Source.

Sarsen uses another Open Source package developed by B-Open and which parses Sentinel-1 using xarray (Xarray-sentinel GitHub repo, web page