**3.3.2 Yushu earthquake area InSAR deformation extraction analysis result**

Using Doris InSAR data processing software and SRTM3 DEM data with 90 m resolution, the two-pass differential interferometry method was used to process the ALOS PALSAR data. We then get the seismic deformation interference phase image shown in Figure 5.

The radar interferogram clearly shows the spatial distribution of the surface deformation field caused by the Yushu earthquake. The coseismic deformation field within the image is about 82 km long and about 40 km wide along the fault. From the distribution of the interferometric fringes caused by the Yushu seismic deformation field, we can see that the distribution of the coseismic deformation is centered on the Garzê-Yushu fault zone, which is the triggered fault (Figure 4, the main fault I), and is parallel to this fault. From the distribution pattern of the interferometric fringes, we can see that the direction and density of the interferometric phase change are different for the two sides of the fault. From the southernmost point A to the fault direction, the interferometric fringe phase indicates an increasing trend from south to north. To the north of the fault, the interferometric fringe phase shows a decreasing trend from north to south. From the whole interferometric phase distribution, the change in the line of sight is left-lateral, revealing significant seismogenic fault sinistral strike-slip properties. It corresponds with the result of wide swath SAR image interpretation.

The seismogenic fault is in a northwest-southeast direction. Along the seismogenic fault zone, there are two major areas with large surface deformation, shown as ① in Figure 5(b) and ② in Figure 5(c). Position ① corresponds with the instrumental epicenter calculated by the National Earthquake Network, and ② corresponds with the macroscopic epicenter. From enlarged views of the interferogram of the instrumental epicenter area in Figure 5(b) and macroscopic epicenter regions in Figure 5(c), we can also see that the radar interferometric fringes change intensely around the instrumental epicenter, while the central region of the macroscopic epicenter has an apparent decorrelation due to the large surface deformation. Both of the two regional seismic fault slip dislocations are relatively large, but that of the latter region, which is close to the city of Yushu, will inevitably lead to stronger tremors for the city of Yushu and the surrounding area, where rupture has been the predominant cause of enormous casualties and economic losses.

PALSAR operates in the L-band, and a color change cycle in the interferogram represents 11.8 cm in the line of sight. According to the interferometric fringes analysis, on the north of the fault, the maximum sinking displacement in the line of sight is 11.8×3=35.4 cm. Since the surface near the epicenter was damaged during the earthquake, the coherence of the corresponding region in the two radar images is very low and cannot form effective interferometric fringes. Therefore, it is reasonable to conclude that one fringe remains on

Earth Observation for Earthquake Disaster Monitoring and Assessment 305

Fig. 6. Spatial distribution of collapsed buildings interpreted from airborne images and

From the RADARSAT-2 polarimetric data (FQ mode) and the polarimetric decomposition

to identify collapsed buildings. This method mainly utilizes three important polarimetric

of uncollapsed buildings is high while that of collapsed buildings is low. However,

Therefore, it is necessary to remove the disadvantaged influence of the bare soil surface before this parameter can be used for building identification( Ainsworth etc al.,2008 ). The circular polarization correlation coefficient ρ can be expressed as Eq. 6( F.Mattia, 1997),

three main land cover types were categorized: collapsed buildings, uncollapsed buildings,

is the circle polarization correlation coefficient which is very sensitive to artificial

<sup>∗</sup> >

�� |���|� >� |���|� >

parameters to extract the collapsed buildings. These parameters are *H*,

also related to the surface roughness. For a low roughness surface, the

� ���� − ���), ��� = ���� <sup>−</sup> �

buildings is as follows: 1) the extraction of the bare soil surface.

and bare soil surface. The basic process using

����� <sup>=</sup> � ������

entropy, representing the random level of target scattering,

are obtained by using Cloude's

method that uses only one post-earthquake SAR image was proposed

 and , Where

decomposition(Cloude, 1996,1997).The

� ���� − ���). For the Yushu urban area,

 and 

method to identify the collapsed

are obtained using

is the averaged scattering type

is

is

(6)

is also high.

sample images for typical regions.

**method** 

**3.4.1** *H*

and 

objects.

model, a new

 and 

Where ��� = ���� <sup>+</sup> �

each side of the fault in the low coherence area, and the cross-fault displacement in the line of sight should not be less than 11.8×8=94.4 cm.

Fig. 5. Coseismic deformation map from ALOS PALSAR data, (a) Differential interferometric phase map; (b) Differential interferometric phase of instrumental epicenter; (c) differential interferometric phase of macro epicenter. A, B, C1, C2, D1 and D2 in (a) represent the different positions and in (b) and in (c) are two large deformation areas (from Guo et al., 2010b).
