Service Communautaire d'Information sur la Recherche et le Développement - CORDIS

Methodology for SAR interferometry studies with selection of coherent reflectors

Previous studies have shown that the presence of tropospheric contributions in the interferograms represents one of the limits for the detection of vertical movements. The capability of correctly removing of tropospheric effects from the interferograms requires a very refined knowledge of the permanent scatterers of the area under study that can be obtained only through the analysis of a large population of interferograms.

For that, using the CNES DIAPASON software, we derived 81 interferograms of the Gulf of Corinth from 38 ERS SAR raw images acquired between 1992 and 1999 on three different tracks. Thus, either the 1992 Ms=5.9 Galaxidi earthquake or the 1995 Ms=6.2 Aigion event have been covered in our database. We sampled very different time spans, ranging from 1 to 2710 days (7.42 years), to analyse the coherence of the region and the tropospheric effects at different time scales. A total number of 45 co-seismic interferograms were calculated in order to detect precisely the differences in the deformation field related to both the earthquakes occurred in the area.

These differences could be obtained by atmospheric changes in the area during the images acquisitions. Some pre-seismic (5) and post-seismic (31) interferograms were computed to detect the presence of local and global tropospheric effects also in the epicentral area. Using all interferograms, we produced a coherence map of the area, which is expected to be less sensitive to local or temporal effects, as ground deformation or tropospheric contributions.

That map collects all the pixels where the coherence is maintained over a long period of time, as towns or hard rock outcrops. It is used to mask the interferograms where the coherence is poor and the phase information lost. Almost all the southern side of the Gulf of Corinth is masked because of an important presence of vegetation and cultivated areas; only a few regions, in particular near the towns of Aigion and Xilokastro, remain coherent. Conversely, the northern side of the gulf has a fairly good coherence, especially the ~3-10km wide band located along the coast. In the field the area is rocky with little vegetation. Places with high coherence are Galaxidi area, the Ag. Pangalos peninsula south of Desfina and the Ag. Nikolaos peninsula north of the Alkionides islands.

We developed a methodological approach to minimise the tropospheric contributions in the interferograms at both global and local scales. Firstly, all the interferograms are filtered to reduce the noise present in the data and then masked using the collective coherence map to leave only the most coherent pixels. Secondly, assuming a horizontally homogeneous troposphere model we searched for any correlation between the phase of the most coherent pixels and their elevation.

On a large number of interferograms a simple linear correlation with the topography was observed. The resulting interferograms are then unwrapped in order to remove the phase ambiguity and a local-scale correction is performed. Comparing all the interferograms containing a given image, we are able to recognise all common anomalies. These local perturbations are then assigned to that image and subtracted from the interested interferograms.

Finally, the corrected interferograms are stacked and averaged to reduce the high-frequency noise still present after the treatment and to put in evidence a smoothed deformation field related to the earthquakes occurred in the area. The deformation concerns a N-S 16km wide by E-W 28km long band, reaching the maximum value of ~243+/-5mm on Psaromita cape, a limestone mesozoic platform located on the northern side of the gulf, in front of the town of Aigion. Within the uncertainty of our displacement map (5mm), the comparison between each corrected interferogram and the average deformation map reveals that no post-seismic deformation is observed in the data in the period 1995-1999.

Informations connexes

Reported by

IPGP - L'Institut de physique du globe
4 Place Jussieu
75252 Paris
France
See on map