Science Service System
You are here : Home : Proposals_Summary

Summary of Proposal LAN1124

TitlePSInSAR deformation assessment using corner reflectors
Investigator Halounova, Lena - Czech Technical University, Civil Engineering, Mapping and Cartography
Team MemberNo team members defined
SummaryDue to the fact that the project is already running, and the artificial reflectors are installed, it is only neccessary to change their orientation with regard to a different satellite. The area is being leveled regularly, however, two different companies give different results, and the aim of the project is to give another independent source. We plan to acquire about 1 scene a month (33 days, does not have to be strictly regular), all from the same track, using the same image mode (STRIPMAP is preferred), same polarization (VV is preferred) and the same incidence angle (preferably about 25-30 degrees). In the system, we did not find any scene acquired for this area: we need to cover the area between 50-29 and 50-35 north (degrees-minutes) and 13-24 and 13-36 east (degress-minutes). In all scenes, the artificial reflectors will be found and the interferometric processing will be performed in order to estimate the deformations. Processing will be performed in the GAMMA software. Funding: Agreement about the research between the Czech Technical University in Prague and the Coal Services company.
Final ReportDEM used As an external DEM for topography subtraction, we used a combination of three available DEMs: a DEM from aerial photogrammetry acquired in 2013 for the outmined area, X-band SRTM from DLR for most of the area and ASTER GDEM 2 (for the area where SRTM-X data is missing), corrected to WGS-84 ellipsoid. This DEM was used for the processing in the GAMMA and StaMPS softwares. For processing only the reflector information, levelled heights were used. Artificial reflectors By mistake, the reflectors do not have the trihedral shape, but with regard to the radar ray, they are approximately dihedral, i.e. very sensitive to the orientation angle, especially in the horizontal direction. Their reflectivity (interpolated) ranges from 1e6 to 6e7 (for one reflector situated in a foreshortening/layover area, it is even 1e8). Standard deviation of the estimated (interpolated) reflector position is usually higher with smaller reflectivity, ranging from 0.08 pix (in range) and 0.01 pix (in azimuth) to 0.2 pix (in range) and 0.2 pix (in azimuth). In January 2013, the reflectors were re-oriented in order to get higher reflectivity and better quality of the deformation estimation. However, it seems that the sensitivity to orientation is higher than a accessible orientation accuracy: for some of the reflectors, the reflectivity went up, but for others, it went down (by even two orders). For the reflector situated in foreshortening/layover, the reflectivity went down by 2 orders and after the reorientation, some attempts to find its position were unsuccessful (the estimated position is different by approx. 2 pixels). For the interpolated position estimation, GAMMA ptarg script was used. For the processing of the reflector information, all pairs of scenes are used (pairs with highest residues are later excluded). A model for deformation estimation is constructed (with no height correction, as levelled heights are used) and the equations are solved, separately for each point (with reference to point 1). In order to avoid high residues due to ambiguity problems, the equations are solved iteratively, starting with 4 scenes (close to the reference one) and each time adding only one scene. We expect that no point is moving, so the solution with lowest ambiguities is preferred. Then, there is a net constructed out of the points and triangle sums are checked (and forced) to be 0 (operating only with 2pi multiples). The last step is to (mostly automatically, then manually) check if there are not too high phase jumps between the nearby dates, and correct them (also by 2pi multiples), and possibly to exclude some scenes if the phase jumps are high for more points. Most of the reflectors seem to be stable. However, reflector 7 placed on the most critical place (in a courtyard of a castle situated on a hill just above the open mine) seems to be going down during the first year and half of the period and then becoming stable. We attribute this to the fact that the reflector was dismounted due to a castle reconstruction and mounted again short before the first acquisition. Reflector 8, situated at the bottom side of the slope at the edge of the open mine, on the contrary, seems to move by 15 mm during the first 2 years "up" (i.e. the distance between the point and the satellite is getting shorter), but then it is also stabilizing. An uplift is possible in such an area and can be caused by deep landslides, or it can be a regular landslide where "down" still means closer to the satellite. The last moving reflector (11) is also uplifting (few kilometers away from the open mine), which is confirmed by levelling (in a period before this project). In the StaMPS processing, the interferograms (i.e. also time series) are smoothed in space, and therefore the time series are influenced by the nearby points. However, due to significant change in intensity due to reflector re-orientation, some reflectors were not selected as PS candidates. But here, it is confirmed that the area near the most critical reflector 7 (castle) is stable and also the area around reflector 8 is almost stable. However, it seems that the area aroun d reflector 11 (even if this reflector was not localized at all) is really uplifting. In addition, several other places of deformations were found in the area, which is partially undermined and contains several waste dumps. During processing in the GAMMA software, most of the reflectors were not detected at all, and the others were excluded during processing due to high residues. To conclude, we can say that TerraSAR-X is suitable for reflector movements monitoring, the reflectivity is significantly higher in comparison to Envisat (due to both shorter wavelength and better resolution) and are easier to find in the image. Also, the orbit quality and geocoding is much better. However, the shorter wavelength (and shorter 2pi multiple) has the disadvantage of higher atmospheric influence (it is more difficult to estimate the 2pi multiples, when the time series are influenced by the atmosphere).

Back to list of proposals

DLR 2004-2016