Standardised automatic cross-sensor change detection in semantic Earth observation data cubes

 Suggested by: Dirk Tiede, Martin Sudmanns, Hannah Augustin

Short description: The Sen2Cube.at system facilitates the first instances of semantic Earth observation (EO) data and information cubes, e.g. using Sentinel-2 data captured over Austria and Syria, and AVHRR and Sentinel-3 data captured over the Alps. All of these semantic EO data cubes can be analysed via Web-based interfaces (e.g. via Web-browser). An EO data cube is a way of organising EO data that abstracts data storage so that users can access EO data based on spatio-temporal coordinates rather than their file names or directory structures, which makes it a lot easier to access the data. An EO data cube is considered a multi-dimensional structure with at least one non-spatial dimension (e.g., time), where coordinate tuples of the dimensions are used for data access. A semantically enriched EO data cube provides for each observation at least one nominal (i.e., categorical) interpretation, which can be queried in the same instance. 

The Sen2Cube.at system uses the satellite image automated mapper (SIAM) as semantic enrichment engine, which provides spectral categories from reflectance values. Using a per-pixel physical spectral model-based decision tree, SIAM automatically categorises EO imagery based on reflectance values from multiple optical sensors (e.g., Sentinel-2, Landsat-8, AVHRR, VHR). The software is capable of producing different granularities (i.e. different number of colour names) from coarse (i.e., 18 colour names) to fine (i.e., 96 colour names), as well as additional data-derived information layers (e.g., multi-spectral greenness index, brightness), fully automated without any specific parameterization.

The goal of this master thesis is to develop and implement a change matrix based on the spectral categories into the Sen2Cube.at system (i.e. transfer of the matrix into Sen2Cube.at models and/or Juypter notebooks) to allow fully automated change analysis for any image combination in the cube and can support various applications (drought, flood, snow cover change etc,). Selected applications can be demonstrated using Sentinel-2 data in Austria and Syria and AVHRR/Sentinel-3 data in the Alps. 

References, suggested reading:

• Sudmanns, M., Augustin, H., van der Meer, L., Baraldi, A., & Tiede, D. (2021). The Austrian Semantic EO Data Cube Infrastructure. Remote Sensing, 13(23), 4807. https://www.mdpi.com/2072-4292/13/23/4807
  Augustin, H., Sudmanns, M., Tiede, D., Lang, S., & Baraldi, A. (2019). Semantic Earth observation data cubes. Data, 4(3), 102. https://www.mdpi.com/2306-5729/4/3/102 

Related to projects: https://sen2cube.at 

Start/finish: anytime

Prerequisites/qualifications: 

    + Remote Sensing

    + Programming