Templeta generale

2.8.WP 1330 Application to inter-/post-seismic test-case 2 (Istanbul)

In order to evaluate the interseismic displacement relatad to the North Anatolian Fault (NAF) in the Istanbul area, we analized COSMO-Skymed SAR images from two different ascending tracks (Figure ).

Figure SRTM-3 DEM of the Istanbul area. Blue rectangles represent the coverage of the two COSMO-Skymed ascending tracks. Black lines show the North Anatolian Fault segments. Red triangles are GPS stations in the area.

We used 30 image strips (consisting of two standard image frames each) for the easternmost track (hereafter Track East) that cover the period between 05 June 2011 and 30 September 2013 (Figure ) and 68 image strips (of two standard image frames each) for the westernmost track (hereafter Track West) that cover the period between 24 May 2011 and 16 July 2013 (Figure ).

To evaluate the deformation time series we used the StaMPS technique [Hooper et al., 2007]. The raw data products were concatenated and focused with the GAMMA Remote Sensing and Consulting software suite. A set of scripts developed within the SiGRiS and MUSA projects were used to convert the output format of the GAMMA software focusing modules in one suitable for further processing with StaMPS. Using the DORIS software (http://doris.tudelft.nl), run through StaMPS processing scripts, we generated 28 interferograms related to a single master acquisition, namely 07 June 2012 for Track East, and 63 interferograms for Track West, with respect to a master acquisition on 08 April 2012. At the time of writing the deformation time series analysis for the Track East is complete, whereas that for Track West is ongoing. Our processing highlighted a bug in the GAMMA concatenation program, which was reported to GAMMA and corrected.

In Figure we show the mean velocity deformation map for Track East. The presence of mountains and vegetated areas reduce the number of coherent pixels in non-urban areas. The mean deformation rates vary between ±1 cm/yr and there is no obvious velocity gradient. It should be noted that previous C-band processing of this area yielded an even poorer coverage (Figure ).

The deformation rate map at each pixel is referred to a reference point (black triangle in Figure ) that showed a stable behavior during the analyzed period. In Figure we show the deformation time series at some areas that show a particular behaviour.

The following comments are in order:

• 
  Due to the right-lateral mechanism of the NAF, and the fact that the ascending flight path is almost perpendicular to the NAF, a gradient in the line-of-sight velocity component should be expected in the azimuth direction.
  
• 
  In Figure the expected gradient is not observed. This may be due to two reasons. The first concerns the impossibility of correctly unwrapping the interferometric phase across the NAF (entirely under water within the Track East in Figure . The velocity values on the coastal strip in the southern portion of Figure are thus to be considered unreliable. The second reason is that coverage is restricted to urbanized patches, within a 20 km strip extending northwards from the coastline. This short distance combined with the irregular coverage may cause the expected velocity gradient to be below the noise floor.
  

Figure Acquisition date vs perpendicular baseline for the 30 Track East images. Yellow triangle is the master image (20120607) used to evaluate interferograms.



Figure Acquisition data vs perpendicular baseline for the 68 Track West images. Yellow triangle is the master image (20120408) used to evaluate interferograms.



Figure Mean velocity deformation map for Track East. Colored triangles are points choose to analyze the deformation time series (Fig. 5).

Figure Mean velocity deformation map for a descending ERS/ENVISAT SBAS analysis by IREA-CNR. From http://supersites.earthobservations.org/istanbul.php

Figure Deformation time series for six points in the area of interest. The rectangles show the mean deformation of all the pixels in the corresponding circle (100 m radius). The dashed lines indicate the mean deformation velocity obtained with a linear regression.

2.9.Azimuth streak artefacts

This section shows the artefacts introduced by the azimuth streaking problem identified within MUSA on displacement maps generated with Multi Aperture Interferometry and offset-tracking. Figure shows MAI results on the Atacama desert dataset described in Table . Figure shows intensity tracking results on the 4 spotlight pairs in Table .

The Atacama dataset is extremely coherent and highlights the streak patterns parallel to the slant-range direction very clearly. In Figure bottom-left, a diagoanally-oriented discontinuity is also visible. We have reasons to believe however that this could be a software bug in the band-pass filtering procedures of the software of [Werner et al., 2001] and will report this observation. The range-parallel streak patterns have instead been observed with several sofware, and are therefore certainly not a bug of the [Werner et al., 2001] programs.

The spotlight dataset in Figure is interesting for several reasons:

• 
  The top-left dataset shows streak-like patterns of the typical spatial wavelength and intensity of the of those observed in several pairs of strip map images thorughout the project. This rules out the possibility that these artefacts are due to the periodic calibration tones, which are interleaved in the imagery data, causing missing data lines at periodic intervals (also within synthetic aperture formation time).
  
• 
  3 out of 4 datasets, one of which is extremely coherent, do not show the streak artefact. So far, these are the only coherent COSMO-SkyMed image pairs in which the artefacts were not observed. It should be noted that two of the three pairs (cfr. Table ) are temporally very close to the top-left image pair, which is affected by the artefact. The third pair was intead acquired several years later on a different area. The only common property among the three acquisitions seems to be the low incidence angle (>50°). This is opposed to the TD2.10 pair (top-left in Figure ) which has a very steep angle (20°). At the time of writing the dependence of these artefacts on the incidence angle has not been systematically explored however. It is therefore premature to draw any conclusion.
  
• 
  Finally, the absence of artefacts in at least some COSMO-Products, spanning years 2009 to 2013, suggests these are not due to data-format related issues, incorrectly handed by user or ground-segment software. If this were the case in fact, the artefacts should always be observed.
  

.

Figure Atacama desert datasets described in Table . Top-left: TD2.6. Top-right: TD2.7. Bottom-left: TD2.8. Bottom-left: TD2.9



Figure COSMO-SkyMed spotlight datasets described in Table . Top-left: TD2.10. Top-right: TD2.11. Bottom-left: TD2.12. Bottom-left: TD2.13