Mapping subsidence in an urban area with Differential InSAR and comparing with traditional methods
The main idea of this project is to investigate whether interferometry technique – more specifically Differential SAR Interferometry (DInSAR) - can be applied to urban settings where human activities such as deep excavations due to building works or mining causes land subsidence, for geotechnical conditions as those in Sweden, and to compare the result with results from traditional methods for settlement measurement.
Gothenburg was chosen as the testing ground as measurements of ongoing settlements in the central parts of Gothenburg has been carried out for a long time and vertical as well as horizontal movements has been measured at the start (before, during and after) construction of the Göta Tunnel. Metal pegs are installed in several places around buildings and precision leveling of the pegs are today carried out every fifth to seventh year by the City of Gothenburg using precision leveling instrument. The results of the measurements are noted in Excel sheets, one for each quarter. In this project, diagrams of the measured settlements have been drawn for chosen quarters, and the results have been visualized on an areal photography of the central parts of Gothenburg.
Initially, only the DInSAR technique would be tested and used for comparison. However, early on it became evident that due to the overwhelming presence of forests and urban green areas, simple DInSAR would not be able to provide adequate results for ground movement in the vicinity of Gothenburg. Therefore, the focus of the analysis was switched from DInSAR to PSInSAR. Initially, very conservative parameters were chosen with respect to minimum coherence between image pairs and the threshold for matching pixels in the image stack. Further refinement and an expansion to less restrictive input parameters increased the amount of measured points and in the central portion of the city the number of measurements increased from 36 to 774. In addition, the time span of images was reduced from 16 years to eight years. The natural land rise in Scandinavia was taken into account during the intermediate stages of the analysis – after the points with high coherence values were identified.
Comparisons between the precision leveling and the PSInSAR analysis from the ERS satellites were carried out in two steps. First a comparison was carried out for the PSInSAR analysis with coherence threshold of 0.75. It was clear that with this high threshold, and with no correction for land rise not all the ongoing settlements were registered. It was also noted that several points from the PSInSAR analysis registered heave. Thus, the threshold was lowered to 0.5 and a correction for land rise was carried out. Also the time span was reduced from 16 years (1992 – 2008) to eight years (1992 – 2000) as mentioned above.
Results and conclusion
The result from this PSInSAR analysis shows that the number of measuring points within the area of interest is more than the double and the points are mainly located in areas where ongoing settlements have been measured by the traditional leveling. But still, many of the quarters with ongoing settlements were not registered by the satellite analyses. In addition, the size of the ongoing subsidence registered by the PSInSAR analysis is now generally larger than the ongoing settlements measured by leveling. The results from the comparison can be summarized as follows:
with PSInSAR analysis from ERS satellites it is not possible detect all quarters with ongoing subsidence.
PSInSAR analysis from ERS satellites does not give the right picture in which quarters the largest settlements are occurring.
in some quarters with no clear trend of ongoing settlements during the measurement period of the leveling, the PSInSAR analyses have actually registered subsidence.
the size of the measured settlements from the nPSInSAR analysis are not in agreement with the size of the settlements measured with leveling.
Currently, satellite analyses as DInSAR and PSInSAR from ERS satellites are not accurate and reliable enough for use as follow-up on settlements of buildings. However, a new class of radar satellites as the TerraSAR-X, along with its sister satellite TanDEM-X, will provide high quality radar images for DInSAR and PSInSAR analyses, which will enhance the possibilities for use of this technique as follow up of settlements.
Swedish Geotechnical Institute (SGI):
Hjördis Löfroth, +46 13 20 18 54, firstname.lastname@example.org
Michael Ledwith, +46 8 57 99 72 98, email@example.com
20 maj 2009