AI4Copernicus – Security Bootstrapping services
The Security Bootstrapping services and resources have been developed in order to support the AI4Copernicus project’s open calls. The development of these bootstrapping services aimed to reduce the time and resources of the bidders in the data preparation and allow them to focus on the development of innovative services based on AI. They include pipelines for the pre-processing and change detection of Sentinel-1 and Sentinel-2 imagery.
AI4Copernicus Security services are a collection of pipelines focused on the preprocessing and change detection computation of Sentinel-1 and Sentinel-2 products in order to facilitate the integration of Sentinel data in AI processing chains. All the pipelines, based on SNAP and GDAL, are available as dockerized applications.
The list of the services are:
- Sentinel-1 GRD pre-processing: This pipeline processes a S1 GRD product in native format to generate a terrain corrected image representing the calibrated backscatter in GeoTiff format. Several parameters are exposed to the users (e.g. final resolution, polarization, projection, land/sea mask).
- Sentinel-1 SLC pre-processing: This pipeline processes a S1 SLC product in native format to generate a terrain corrected image representing the calibrated backscatter in GeoTiff format. Several parameters are exposed (e.g. final resolution, polarization, projection, land/sea mask)
- Sentinel-2 pre-processing: This pipeline processes a S2 product in native format to generate a product with a common resolution for all the bands in GeoTiff format. The process allows to apply a land/sea mask and a cloud mask in order to have an output product ready for analysis. Several parameters are exposed (e.g. DEM, cloud mask type).
- Sentinel-1 Change detection: This pipeline processes pairs of S1 SLC products in native format to generate a series of products to assess the changes between both images. These products include:
- the coherence (the amplitude of correlation between the images),
- the ACD (Amplitude Change Detection), which is a RGB composite of the backscatter of the input images
- the MTC (Multi-Temporal Coherence), which is a RGB composite of the backscatters and the coherence
- binary mask of changes
- Sentinel-2 Change detection: This pipeline computes (and classifies) the changes using as input a pair of S2-L2A products by using the Change Vector Analysis approach. Several parameters are exposed (e.g. resolution, bands, number of change classes).
For more details, https://ai4copernicus-project.eu/wp-content/uploads/2022/11/AI4Copernicus_4th-OC_Technical-Documentation_Nov2022.pdf