REQUISITOS FUNCIONALES 126

Two statistic parameters – the maximum value and the variations all 24 simulations are shown for all the three components in Fig. 8.5to give a general image of the ground motion variations which contains the dominant features shown by the two previous examples, like

1

8.3 Effect due to Hypocentre Location 73

Figure 8.2: PGRR distributions for two example hypocentres. Left. H1 (Fig. 6.1). Right. H2 (Fig. 6.1). From top to bottom are x-,y-and z-components, respectively. The epicentres are indicated as red asterisks. The black dashed rectangle indicates the fault trace. Thin white lines are contours of the seismic velocity model. Note the color scale difference.

Figure 8.3: Rotation rate profile for the two example hypocentres: parallel to the fault trace. From top to bottom are thex-, y- and z-components, respectively. The shear wave velocity isosurface depth (at 2.0 km/s) is depicted in the bottom as the shadowed area. The maximum rotation rate amplitudes across this profile are shown with the inlet number.

8.3 Effect due to Hypocentre Location 75

Figure 8.4: Rotation rate profile for the two example hypocentres: perpendicular to fault trace. From top to bottom are the x-, y- and z-components, respectively. The shear wave velocity isosurface depth (at 2.0 km/s) is decorticated in the bottom as the shadowed area. The maximum rotation rate amplitudes across this profile are shown with the inlet number.

basin wide shaking, fault-distance dependent ground motion, and peak motion uplifted by the slip asperity. Two ratios are calculated and shown to characterize the variations: RSD – the one of the standard deviation relative to the mean value, and Rmax – the one between the maximum value and the mean value.

Figure 8.5: Properties of the PGRR distributions due to varying hypocentre. Left. Maxi- mum value. Middle. The ratio between the standard deviation and the mean PGRR value (combination of all 24 simulations) in percent. Right. The ratio between the maximum value and the mean value. From top to bottom arex-,y-andz-components, respectively. The black dashed rectangle is used to indicate the fault trace to avoid masking the high values on the fault trace. Thin white lines are contours of the seismic velocity model. Region A, B and C and station P1, P2 and P3 are picked up for more detailed discussion. Station R1 and R2 are used to analyze the hypocentre depth effect on the ground rotation rate (amplitudes and seismograms). Profile EF is used to illustrate the rotation rate variation across the fault trace. Note the color scale difference.

The maximum rotation rate of the z-component is about 8 times larger than the x-

component and 5 times larger than the y-component. Across the fault trace, the gradient of the z-component rotation rate is larger than the other two components. These two phe- nomena could be explained with the source mechanism – pure strike slip, and the fact that the rotation rate is the space derivative of the horizontal velocities.

8.3 Effect due to Hypocentre Location 77 the maximum PGRR distribution is symmetrical about the middle of the fault trace. That symmetry is not observed for the other two rotation rate components (Fig. 8.5 middle and bottom, left). For they-andz-components (Fig. 8.5middle and bottom, left), larger rotation rates are observed in the left part of region C (around the left tip of the fault trace) where the neighboring basin is deeper, than in the part right to the fault trace middle. This observation confirms that the fault parallel component of rotation rate receives more contributions from the directivity effect than the other two components.

The variation of the hypocentre-dependent ground motions is illustrated by showing the spatial distribution of two ratios RSD and Rmax (Fig. 8.5 middle and right). We take two example regions, A and B, where a dramatic basin depth variation is located. For the x-

component, there is no obviously high values of ratioRSD observed (Fig. 8.5top right) inside these two regions. However for they-component (Fig. 8.5middle right) and thez-component (Fig. 8.5 bottom right), large RSD is observed inside regions A and B, compared to their neighboring area. The largestRSD, 75% for the y-component and 72% for the z-component, of the whole study area are all located inside region A. Right on the fault trace, the ratio RSD for the x-component (Fig. 8.5 top, right) is compatible to its neighboring region, but not for the other two components (Fig. 8.5middle and bottom, right). For all the three components, especially for thez-component, there are large RSD found in the regions off the two tips of the fault trace.

The distributions in terms of ratio Rmax are calculated and shown in Fig. 8.5(right) for different component. The spatial distributions of this ratio for they-andz-components (Fig. 8.5middle and bottom, right) are quite similar to those of the ratioRSD (Fig. 8.5middle and bottom, middle): the largest value over the entire study area is observed at the same station (P1 fory-component and P2 forz-component); elevated rotation rate variations are observed in the small basin (region B); smaller variations are observed in the fault plane projection regions (region C) when compared to the neighboring area. But for the x-component, the difference between the distribution of Rmax and RSD is more obvious. The largest Rmax happens at station P3 whereas the largest RSD is at station P4. The spatial distributions of ratio Rmax are coarser than those of the ratio RSD. At the basin edge, about 2.8, 3.2 and 3.1 times larger than the mean ground rotation rate might be expected for the x-, y- and

z-components, respectively, if the mean ground rotation rate could be predicted.