Description

Post-disaster damage mapping (DM) is mainly based on remote sensing data, often with data of different source and kind in terms of radiometric, geometric and temporal properties. However, some of these properties could be mutually complementary, while others make a simultaneous or integrated analysis more or less impossible or at least very complicated. Nevertheless, integrating different data types could be beneficial in terms of additional information and/or aspects of information quality enhancement. Object Based Image Analysis (OBIA) works with image objects/segments, instead of pixels, thus pixel-level matching is not necessary anymore. As long as the geometric relationships between corresponding image objects in different data sources are known and describable, they can be logically interconnected and can be analyzed as if they were one object. eCognition software allows multiple data sources to be used for individual image segmentations, leading to different object representations in the data used. The so-called maps-and-links-concept allows to logically link corresponding image objects the way as described above, and making an integrated image analysis possible in principle.

Objectives and Methodology

In this research the following major questions should be addressed:
- Which sensor combinations are most useful in the different contexts of DM (earthquake, landslides, floods and storms, wild fires etc.)?
- How can the different information gathered with different sensors best be used for damage assessment, object detection or monitoring?
- How can data gaps in time series be filled (e.g. RADAR during cloudy days vs. optical on cloud-free days)?
- What are quantifiable advantages and disadvantages of integrated OBIA in comparison to conventional (OB)IA in DM? What is best in which DM context?
- What are the limitations for integrated OBIA in DM?

Further reading