Using InSAR to detect railway deformations
Typ
Examensarbete för masterexamen
Program
Wireless, photonics and space engineering (MPWPS), MSc
Publicerad
2019
Författare
Bergevald, Carl
Modellbyggare
Tidskriftstitel
ISSN
Volymtitel
Utgivare
Sammanfattning
Railway deformations are issues that can possibly prevent an effective railway network.
Serious deformations may lead to decreased speed limits or in worst case the
need for closing parts of the railway for traffic until maintenance work has been
performed. The idea is that displacement maps from InSAR images can be used to
detect problematic areas so that the right actions could be taken before the deformations
become so large that they affect the traffic. Displacement images produced
by SkyGeo from the TerraSAR-X satellites have been used by Trafikverket in other
projects with good result. The issue is that TerraSAR-X data is expensive so the
idea is to use the freely available Sentinel-1 data instead. Displacement maps produced
by SkyGeo for both Sentinel-1 and TerraSAR-X, and a displacement map
produced by processing in the Sentinel Application Platform (SNAP) over an area
south of Gothenburg, Sweden are compared. It was found that using persistent scatterer
interferometry on a stack of images alone was not enough to remove enough
errors and the results were unreliable. However, when looking at the displacement
images provided by SkyGeo which has used models for error reduction, it seemed
like the results were more reliable. The results from the Sentinel-1 data had more
fluctuations and thus a slightly larger RMS error compared to the results from the
TerraSAR-X data but the trends were still clearly visible. A comparison of the coherence
and RMS errors showed that there is a connection between high coherence
values and low RMS errors which means that where there is high coherence there
should also be a low RMS error. The conclusion is that the use of Sentinel-1 data
to create displacement maps is plausible and seem to work well in the area used in
the project. However, further research on error sources and different nature types in
other areas is desired before reliable displacement maps can be made over the entire
land mass of Sweden.
Beskrivning
Ämne/nyckelord
InSAR, Persistent scatterer, Differential InSAR, Displacement, Sentinel-1, TerraSAR-X.