Flood detection in urban areas is a major limitation of flood detection approaches using SAR backscatter. This is problematic for the disaster response community, as urban areas are also where most damage happens and the most people are exposed to the disaster, considering that the majority population lives nowadays in cities rather than rural areas, the trend increasing.
Input Data:
- Sentinel-1 SLC data
- Download recommended via the Alaska Satellite Facility
Sofware:
- SNAP
- Google Earth Engine
The proposed practice can be well applied in dense urban areas, like cities but also in areas with little to no vegetation cover. In these areas, mudflows, landslides, and flow paths of flash floods can be mapped, although this is not the objective of this practice.
Strengths:
- Independant of clouds
- Detection of changes at sub-pixel level, which is a major advantage in dense urban settlements
- Detection of floods, infrastructure damage and other surface changes
Limitations:
- The practice is limited in detecting floods in vegetated areas like forests and agricultural areas
- Short time span between the SAR image acquisitions essential
- No cloud computing options available, selection of ROI needed, no large-scale computations on local machines feasible.
- SLC scenes need a lot of storage capacity, i.e., Up to 5GB per scene.
Downloading the Scenes from the Alaska Satellite Facility
Processing the Scenes with a pre-defined workflow in SNAP
Exporting the Coherence GeoTiff
Importing the Coherence raster into Google Earth Engine
Change Detection in Urban Areas
Map of flooded/damaged areas
Pelich, Ramona; Chini, Marco; Hostache, Renaud; Matgen, Patrick; Pulvirenti, Luca; Pierdicca, Nazzareno (2022): Mapping Floods in Urban Areas from Dual-Polarization InSAR Coherence Data. In IEEE Geosci. Remote Sensing Lett. 19, pp. 1–5. DOI: 10.1109/LGRS.2021.3110132.
Attachment | Size |
---|---|
Slides: RP_COH.pdf (3 MB) | 3 MB |