You are on page 1of 11

Vol.10, No.

EARTHQUAKE ENGINEERING AND ENGINEERING VIBRATION

September, 2011

Earthq Eng & Eng Vib (2011) 10: 313-323

DOI: 10.1007/s11803-011-0068-y

A new attenuation model of near-fault ground motions with consideration of the hanging wall effect in the Wenchuan earthquake
Peng Lei1, Xie Lili1.2, Hu Jinjun1 and Wang Dong1*
1. Institute of Engineering Mechanics, China Earthquake Administration, Harbin 150080, China 2. School of Civil Engineering and Architecture, Harbin Institute of Technology, Harbin 150090, China

Abstract:

The hanging wall effect is an important factor that impacts the characteristics of strong ground motions in near-fault areas. Based on a residual analysis of ground motion parameters characterizing the hanging wall effect and in recognition of the nature of the effect, many models have been developed. In this study, after a comprehensive analysis of two existing models, a new model is proposed and used to model the hanging wall effect in horizontal peak ground acceleration (PGAH) and spectral acceleration (SAH) at a period of 0.1 s in the Wenchuan earthquake. Finally, comparisons between the modeling results of the hanging wall effect in the Wenchuan earthquake and the results predicted by using Abrahamson and Silvas NGA model (AS NGA) indicate that the AS NGA model predicts a much higher hanging wall effect than the model developed in this paper. Furthermore, the AS NGA model predicts a large hanging wall effect even at great distances, while the proposed model more accurately captures the trend of the effect.

Keywords: hanging wall effect; Wenchuan earthquake; attenuation relationship; NGA model

1 Introduction
The near-fault effect has been a topic of great interest in recent years to researchers who study ground motion characteristics. Due to the gradually extensive deployment of strong-motion instruments and the rapid development of technologies, increasingly high quality near-fault recordings data sets have been obtained, especially after recent major earthquakes, and an in-depth insight into the near-fault effect has become possible. The hanging wall effect is an important component of near-fault effects. Studies on attenuation relationship have identied this effect in the past (Campbell, 1993; Abrahamson and Somerville, 1996; Abrahamson and Silva, 1997). Because of closer proximity of sites on the hanging wall to faults as a whole than for sites at the same distance on the footwall (Somerville and Graves, 2003; Wang et al., 2008), the hanging wall effect is simply a geometric phenomenon causing ground motions to be larger at sites on the hanging wall than on the footwall. By the classical means of analyzing residuals, studies have been conducted to incorporate the hanging wall
Correspondence to: Peng Lei, Institute of Engineering Mechanics, China Earthquake Bureau, Harbin 150080, China E-mail: peng084@163.com Graduate Student; Professor; Associate professor; *Doctor Supported by: The National Basic Research Program of China or 973 Program, Under Grant No. 2007CB714200 and National Natural Science Foundation of China Under Grant No. 50808166. Received December 15, 2010; Accepted April 31, 2011

effect into the attenuation relationships, including both the general attenuation relationships such as the rst phase of NGA (next generation attenuation of ground motion) models and the single earthquake- specic attenuation relationships such as those in the analysis of Abrahamson and Somerville (1996) and Shabestari et al. (2003). The 2008 Wenchuan earthquake was the most devastating disaster to occur in the past few decades in China, causing huge economic losses and a heavy death toll. Meanwhile, it provided the most strong motion data since the beginning of strong ground motion observation in China. Based on the data set from this earthquake, this study attempts to better quantify the hanging wall effect exhibited in this earthquake. For the rst step, residuals on both hanging wall sites and footwall sites were obtained by regression of the Wenchuan-specic attenuation relationship and an obvious hanging wall effect was identied by the bias of the residual distribution. Then, a new model is developed based on modication of two models previously proposed to analyze the hanging wall effect in other earthquakes. Finally, a comparison of results predicted by both the present model and the Abrahamson and Silvas NGA model are presented and some discussions are provided.

2 Near-fault strong ground motion data set


In this study, the near-fault distance is dened as the closest distance to the fault plane from a given

314

EARTHQUAKE ENGINEERING AND ENGINEERING VIBRATION

Vol.10

site. The hanging wall effect is simply a geometrically related effect revealed by both empirical and numerical examinations. Such an effect can also be eliminated by dening a so called root-mean-square distance, by which weighted averages are applied to sites located on a hanging wall and a footwall (Wang et al., 2008). However, since the primary objective of this paper is to examine and model the hanging wall effect exhibited in the Wenchuan earthquake, the normally used closest distance is used to make the hanging wall effect emerge. Recordings within a near-fault distance of 100 km were used in this study. Furthermore, the closest distance from each site to the ruptured fault was calculated using the USGS fault plane solution (Chen and Gravin, 2008). Recordings off the ends of the ruptured fault are excluded from our data set. The recording stations are listed in Table 1 as well as site conditions, rupture distance, PGA, SA, etc. Since detailed site conditions were not accessible at the time of this article, a general idea of the site conditions for each site can only be derived by the modest information given in the recording data les. Therefore, site conditions are classied as simply soil or rock. Figure 1 shows the fault vertical projection on the ground surface, epicenter and the position of the stations used in this study with respect to the fault projection as well as their site conditions.

PGAN and PGAP (peak ground acceleration for geometric mean horizontal components, fault normal (FN) components, and fault parallel (FP) components), and corresponding acceleration response spectra SAH, SAN, SAP for 5% damping are derived by the same attenuation relationship. Note that even though many attenuation relationships are available, only the simplest ones are needed for this study; thus the one used in Abrahamson and Somerville (1996) is quoted. Then, the specic attenuation relationship for the Wenchuan earthquake considering the effects of site, geometric spreading and near-source saturation is given as y = b1 + b2 S + (b3 + b4 S ) ln(r + c) + (1)

where y is the natural logarithm of PGAH, PGAN, PGAP, SAH, SAN or SAP, S is a parameter introduced to model the site conditions using 0 for rock sites and 1 for soil sites, r denotes the distance in km used to represent the geometric spreading effect, is the regression error, and c is the near-source saturation term and assumed to be a constant as we focus on a single earthquake. By a nonlinear least square regression analysis, all the coefcients in Eq. (1) are obtained and are listed in Table 2 for PGAH, PGAN, PGAP and Table 3 for spectral accelerations SAH, SAN, SAP. Corresponding attenuation relationships are shown in Fig. 2 and Fig. 3.

Specic attenuation relationships for the Wenchuan earthquake

4 Modeling of hanging wall effect exhibited in PGA


The hanging wall effect represents a systematic difference between the ground motions at hanging wall sites and footwall sites, and it is commonly accepted that hanging wall sites experience more intense ground motion than footwall sites at the same distance to fault. Accordingly, the PGA residuals between the recordings on the hanging wall and the footwall and the mean attenuation relationship of all sites were examined. Then, by comparing the resulting hanging wall residuals against those of the footwall, it was found that residuals of the hanging wall are mostly positive while residuals of footwall are more likely to be negative when the hanging wall effect exists. The positive trend of hanging wall residuals means that the hanging wall ground motions are more likely to be greater than predicted by the mean attenuation relationship, while the negative trend of footwall residuals means that the footwall ground motions are more likely to be smaller relative to the values predicted by the mean attenuation relationship (Abrahamson & Somerville, 1996). 4.1 Residual distribution of PGAH Figure 4 shows residuals of the average PGAH; note that the residuals at the side of the hanging wall are mostly positive and the residuals at the side of the footwall are mostly negative. This is consistent

In this section, a simple form of attenuation relationship is derived from regression analysis performed on the recorded data with fault distances of less than 100 km from this earthquake. The PGAH,

Surface projection of fault

Fault surface trace Sichuan Province Rock Soil

Fig. 1 Distribution of stations used in this study with sites classication of rock (upend triangle) and soil (solid circle) and vertical projection of the fault solution by USGS, and also the epicenter marked by a star shown with respect to the fault. (Shadowed areas are excluded from the study)

Table 1 Stations and related parameters used in this study


SAH(cm/s2) 0.1s 316.69 67.579 122.64 113.85 171.33 58.45 107.6 98.402 65.523 589.73 211.31 363.49 87.373 375.47 914.26 840.03 596.65 209.41 236.6 704.86 255.45 453.31 466.69 760.42 247.23 244.07 120.22 234.63 46.842 55.401 107.56 88.821 127.46 130.38 217.16 99.54 102.46 67.342 87.225 526.88 428.64 500.32 1642.5 859.62 473.02 823.09 320.13 155.66 181.51 303.38 381.96 211.58 696.05 369.2 435.14 559.12 1542.8 289.17 430.89 120.5 283.35 200.35 270.89 674.11 235.59 390.89 577.38 956.94 303.19 303.53 128.78 252.05 43.086 66.709 96.771 77.436 137.21 112.67 274.81 90.506 122.41 41.183 100.68 596.81 474.86 457.4 1479.7 986.25 422.42 1159.1 294.28 105.66 169.15 242.87 304.14 204.91 646.85 327.75 462.55 512.48 1177.4 234.7 303.81 111.21 245.83 236.52 179.61 691.42 285.93 276.32 229.14 193.58 548.9 294.14 217.88 120.11 175.3 171.9 407.81 368.94 142.2 124.4 180.29 225.48 215.81 84.501 56.479 39.211 80.714 32.762 222 208.64 435.76 302.58 171.8 99.502 212.06 154.65 483.2 384.73 126.76 126.53 147.17 301.62 247.2 72.392 55.611 31.864 79.93 29.504 242.18 198.46 601.13 290.95 491.68 469.73 898.55 190.49 214.2 102.64 213.28 144.99 144.92 191.08 344.18 353.81 25.124 32.426 21.374 19.54 13.459 14.605 71.187 67.753 48.051 23.044 84.341 85.263 281 97.525 460.6 796.74 48.216 637.69 297.63 269.93 258.27 1155.7 734.69 957.99 1038.3 1014.6 1670.8 1386.6 327.18 69.449 212.98 175.97 302.53 92.077 135.33 84.658 80.927 461.39 334.11 732.78 117.33 659.93 810.1 850.23 1801.6 218.88 206.66 737.01 276.99 525.69 377.22 604.25 201.6 196.26 112.22 218.41 159.51 122.3 220.86 168.56 188.4 282.89 67.538 157.51 141.03 151.39 64.926 98.868 66.809 61.9 534.58 233.45 402.31 87.413 351.91 999.42 952.76 1074.3 42.792 28.537 23.439 20.717 32.166 20.135 15.835 13.194 13.06 85.884 96.527 46.214 29.122 85.517 97.689 303.91 171.96 42.362 25.845 22.981 315.98 171.96 698.88 334.58 310.55 554.87 411.48 735.83 46.674 429.94 363.61 340.3 360.79 852.45 620.97 648.61 746.83 1157.1 2109.8 1261.3 342.55 68.189 190.42 97.06 492.67 102.98 179.39 137.08 133.4 882.49 210.46 598.48 167.05 667.33 746.22 1078.5 719.51 354.53 67.62 95.481 91.914 193.91 52.621 117.1 144.93 69.359 650.57 191.26 328.42 87.332 400.61 836.36 740.64 331.38 41.93 23.40 22.53 30.46 32.68 22.68 24.11 13.72 16.33 59.00 47.55 49.96 18.23 83.18 74.41 259.8 55.31 50.92 46.00 119.5 101.8 118.39 150.88 171.6 109.47 85.76 110.12 75.56 98.63 57.36 48.25 81.50 36.37 0.5s 0.6s 2.0s 0.1s 2.0s 0.1s 2.0s PGAN (cm/s2) PGAP (cm/s2) SAN(cm/s2) 0.5s 0.6s SAP(cm/s2) 0.5s 0.6s

Station code

Lat (N)

Long (E)

Rrup (km)

Soil type

PGAH (cm/s2)

Hanging wall stations 062WUD 104.99 062WIX 104.48 051JZG 104.32 051JZW 104.21 051PWM 104.52 051SPT 103.6 051SPA 103.64 051HSD 102.98 051HSL 103.26 051MXD 103.68 051LXT 103.45 051LXM 103.34 051LXS 102.91 051MXT 103.85 051MXN 103.73 051MZQ 104.09 051WCW 103.18 164.27 116.04 187.1 123.78 228.38 31.861 137.06 88.488 98.344 165.18 325.97 334.55 225.76 359.39 476.65 849.45 753.18 293.54 227.37 269.04 449.51 368.57 295.32 524.42 111.55 129.55 60.1 157.85 326.55 150.46 281.77 404.72 588.22 203.51 652.97 128.06 81.529 76.996 103.29 560.75 451.16 478.38 1559 920.76 447 976.76 306.93 128.25 175.22 271.44 161.39 111.33 166.68 153.38 269.06 29.091 116.04 87.638 156.19 210.87 271.61 328.11 210.89 299.52 349.13 791.99 602.26 325.14 231.09 622.72 435.35 765.68 47.439 523.61 328.97 303.08 305.26 992.55 675.44 788.26 880.59 1083.5 1877.5 1322.5 334.77 68.816 201.39 130.69 386.07 97.376 155.81 107.72 103.9 638.1 265.17 662.24 140 663.62 777.5 957.57 1138.5

33.35 32.95 33.12 33.03 32.62 32.64 32.51 32.07 32.06 32.04 31.56 31.57 31.53 31.68 31.58 31.52 31.04

72.42 66.30 89.34 88.69 39.29 94.03 81.18 85.53 67.94 42.59 24.18 28.58 47.27 16.14 15.66 1.05 9.80

S R S S S R S S S S S S S R S S S

162.82 153.79 177.08 139.5 247.89 33.31 157.44 102.54 125.77 196.63 303.14 331.31 240.87 342.73 433.47 880.88 771.73

Foot wall stations 051GYS 105.84 051CXQ 105.93 051JYC 104.99 051JYH 104.63 051JYD 104.74 051AXT 104.3 051SFB 103.99 051DYB 104.46 051PXZ 103.76 051CDZ 104.09 051DXY 103.52 051XJL 103.8

32.15 31.74 31.9 31.78 31.78 31.54 31.28 31.29 30.91 30.55 30.59 30.38

55.93 95.86 24.09 11.74 18.59 11.28 13.74 42.16 30.41 81.31 42.36 77.56

S S S S S S S S R R S R

309.61 184.96 291.14 442.89 483.45 245.15 598.83 119.52 102.77 68.025 147.53

051PJW 051QLY 051PJD

103.63 103.26 103.41

30.29 30.42 30.25

74.47 40.41 64.14

S S S

No.3 Peng Lei et al.: A new attenuation model of near-fault ground motions with consideration of the hanging wall effect in the Wenchuan earthquake 315

051HYT

103.37

29.91

94.92

97.155 109.79 169.74 181.31 141.46

82.743 107.4 159.57 169 106.1

81.375 112.24 166.97 168.6 162.77

316

EARTHQUAKE ENGINEERING AND ENGINEERING VIBRATION Table 2 Values of coefcients of the Wenchuan-specic attenuation relationship for different components of PGA Component PGAH PGAN PGAP b1 7.9988 8.6887 7.3232 b2 -0.46955 -1.0401 0.084031 b3 -0.8552 -1.0309 -0.68251 b4 0.25216 0.39635 0.11166 c 1.8543 2.2206 1.4655 0.40592 0.39189 0.45574

Vol.10

Table 3 Values of coefcients of the Wenchuan-specic attenuation relationship for different components of SA at several periods Component SAH Period (s) 0.1 0.5 0.6 2.0 0.1 0.5 0.6 2.0 0.1 0.5 0.6 2.0 b1 8.8802 8.7459 8.0680 4.4587 9.5339 9.2772 8.0570 5.1994 8.2533 8.2715 8.0802 3.7413 b2 -0.94880 -0.77373 -0.22258 1.66290 -1.56710 -0.76037 0.044951 1.20190 -0.35547 -0.81422 -0.49123 2.10670 b3 -0.92116 -0.948000 -0.80906 -0.16787 -1.07990 -1.06640 -0.79881 -0.36784 -0.76789 -0.84117 -0.81954 0.027129 b4 0.43436 0.29963 0.12463 -0.46188 0.58915 0.29392 0.061037 -0.32000 0.28483 0.31222 0.18844 -0.60003 c 0.82994 3.01370 3.04810 0.91246 1.34520 5.13010 2.97330 1.33450 0.40300 1.05420 3.13710 0.51033 0.45529 0.63148 0.60202 0.63394 0.45361 0.66642 0.63420 0.63510 0.49233 0.63746 0.61145 0.67757

SAN

SAP

Peak acc. of different components (cm/s2)

103

102
PGA H Recorded PGAH PGAN Recorded PGAN PGAP Recorded PGAP

101 0 10

101 Rupture distance (km)

102

Fig. 2 Comparison of the Wenchuan-specic attenuation relationship for PGAH, PGAN and PGAP at soil site

with observations in previous analysis of the hanging wall effect on near-fault ground motions. In Fig. 4, the footwall residuals are placed at a negative rupture distance for purposes of comparison with hanging wall residuals. Note that in the 5 km to 40 km distance range, the hanging wall has the most signicant effect, and the residuals have a mean bias of 0.31470.1408. However, the scarcity of data in the short rupture distance area and especially the lack of available data renders the analysis of the Wenchuan-specic hanging wall effect inconclusive with regard to exactly how intensively the hanging wall effect impacts the near fault ground motions, even though it can be inferred from the current results that an obvious hanging wall effect did exist in this devastating earthquake. From an overview of the previous studies of hanging wall effect, it was found that in the Northridge earthquake, the mean bias of the

hanging wall sites at a distance range of 8 km to 30 km is 0.290.15 (Abrahamson and Somerville, 1996) and in the Chi-Chi earthquake, at a distance range of 3 km to 30 km, the hanging wall bias is 0.430.058 and the footwall bias is 0.030.006 (Shabestari and Yamazaki, 2003). Such a large variety among the results from different earthquakes may be a result of scarce nearfault data and mechanism of faults. The synthetic results (Wang et al., 2008 and Wang, 2010) suggested that as the dip of the fault increases from 0 to 90, the hanging wall effect immediately becomes increasingly obvious and then gradually weakens after a peak at around 20 dipping until it completely disappeared at 90 dipping. Furthermore, the hanging wall effect is larger for reverse and thrust earthquakes than for strike-slip earthquakes (Somerville and Graves, 2003).

No.3 Peng Lei et al.: A new attenuation model of near-fault ground motions with consideration of the hanging wall effect in the Wenchuan earthquake 317

Spectral acc. of differrent components (cm/s2)

104
T=0.1 s

Spectral acc. of differrent components (cm/s2)

104
T=0.5 s

103

103

102

SAH Recorded SAH SAN Recorded SAN SAP Recorded SAP

102

101 100 104

101 Rupture distance (km)

102

101 100 104

SAH Recorded SAH SAN Recorded SAN SAP Recorded SAP

101 Rupture distance (km)

102

Spectral acc. of differrent components (cm/s2)

T=0.6 s

Spectral acc. of differrent components (cm/s2)

T=2.0 s

103

103

102

101 100

SAH Recorded SAH SAN Recorded SAN SAP Recorded SAP

102

101 Rupture distance (km)

102

101 100

SAH Recorded SAH SAN Recorded SAN SAP Recorded SAP

101 Rupture distance (km)

102

Fig. 3 Comparison of the Wenchuan-specic attenuation relationship for SAH, SAN and SAP at soil site

1.5 1.0 PGAH residual 0.5 0 -0.5 -1.0

Soil site Rock site

-1.5 -100 -80 -60 -40 -20 0 20 40 60 80 100 Rupture distance (km)

Footwall

Hanging wall

Fig. 4 Residuals of PGAH at recording sites during Wenchuan earthquake with those at hanging wall placed at positive rupture distance and those at footwall placed at negative rupture distance

4.2 Empirical model of the hanging wall effect for PGAH As noted above, the hanging wall effect exhibits a trend, with residuals of hanging wall sites having more extensive ground motion as shown in the residuals of the PGAH in Fig. 4. It is possible and necessary to model the trend specically for the earthquake of concern to thoroughly analyze its hanging wall effect. Abrahamson and Somerville (1996) (AS1996) and Shabestari and Yamazaki (2003) (KS2003) both proposed models in

terms of piecewise function as shown below in Eqs. (2) and (3) that were used for the Northridge and ChiChi earthquakes, respectively. These two models have a similar form, and both have four boundary distances x1, x2, x3, x4 to be determined. Thus, the AS1996 model requires supplementation of data from other earthquakes due to limited data from Northridge earthquake. Then, data from the Northridge earthquake was used alone to regress the coefcient c of the model after boundary distances have been xed. Thus, an inter-event error (i.e., error originated from variability from one earthquake to another) was introduced (Abrahamson and Youngs, 1992). The model, therefore, to some extent deviates from the accurate modeling of the hanging wall effect in one earthquake. In addition, it was found that the process of obtaining the four boundary distances is quite intricate and it was difcult to determine how the authors could obtain the four boundary distances all being an exact integer, if the least square method was used to regress them. Of course, the nal results may be rounded, but such a process would also introduce subjectivity. The KS2003 model differs from the AS1996 only in that the constant value which was used to describe the lasting part of the most obvious hanging wall effect between distances of x2, x3 in the AS1996 model is changed into a function of Pr d (r ) Hanging wall Pr d (r ) All . This function is designed to move the general mean attenuation value based on data from all the sites away from the mean attenuation value based on the data from

318

EARTHQUAKE ENGINEERING AND ENGINEERING VIBRATION

Vol.10

only the hanging wall sites. Since there was a rich near fault data set from the Chi-Chi earthquake, the authors used only the Chi-Chi data. Nevertheless, the KS2003 model also seems to have a problem determining the four boundary distances, and the process of nding these distances is still intricate. In addition, the KS2003 model used the mean attenuation value only in the range of x2, x3 to obtain Pr d (r ) Hanging wall Pr d (r ) All , and the coefcients c1, c2 equal to the values of Pr d ( x2 ) Hanging wall Pr d ( x2 ) All respectively. and Pr d ( x3 ) Hanging wall Pr d ( x3 ) All , Consequently, the inuence of data falling in the distance range of less than x2 and larger than x3 was ignored, resulting in a bias to be generated. 0 c cos( (r x1 ) + ) + 1 2 x2 x1 (r ) = c HW c cos( (r x3 ) ) + 1 2 x4 x3 0 AS1996 model (r x1 ) c1 2 cos( x x + ) + 1 2 1 = Pr d(r )Hanging wall Pr d(r )All c 2 cos( (r x3 ) + 1) 2 x4 x3 KS2003 model for 0 r x1 for x1 < r < x2 for x2 r x3 for x3 < r < x4 for r x4 (2)

the hanging wall and the footwall (Abrahamson and Somerville, 1996). Therefore, x1 should be determined by the depth of the upper edge of a dipping fault. Then, to determine the rest of the six unknowns i.e., the x2 and c1 to c5, six conditions can be established, including: (1) four continuity conditions of R1 and R2 and their rst derivatives with respective to r at the joint point x2, and (2) two suite regression analyses using iterations of the boundary distance x2 to achieve a least total summation of the squared residuals, one on the recorded data over the region of x1 to x2 and one on the recorded data over the entire region. 4.3 Comparison of the proposed model with the AS1996 and KS2003 models Figure 5 shows a comparison of PGAH residuals obtained by using the Northridge earthquake data tted by the proposed model and by the AS1996. Note that the two models yield almost the same results. However, as stated previously, the AS1996 model introduced an inter-event error by mixing data from other earthquakes in determining the boundary distances, and its denition of the distance limits the effect of the disappearance; thus, the results depart from the actual hanging wall effect. However, due to the scarcity of data, and also the dominant inuence of intra-event errors compared to inter-event errors, this type of departure was not obvious and discernable. Thus, the proposed model is designed to avoid these aspects of bias, and the resulting curve is balanced among the data. A similar comparison is shown in Fig. 6 for the Chi-Chi earthquake data tted by the proposed model and by KS2003. Note that these two models predict a signicant difference, even though the KS2003 model involves no inter-event error terms. The major difference between the two models is in the KS2003s omission of the inuence of the data that falls in the distance range of less than x2 and larger than x3. The Chi-Chi earthquake provides much more data than the Northridge earthquake, and the data distributes more evenly over the distance range, thus the departure from

for x1 < r < x2 for x2 r x3 for x3 < r < x4 (3)

HWeffect

where HW (r ) and HWeffect denote residuals of PGAs on hanging walls used in AS1996 and KS2003, respectively. In this study, to improve the models of AS1996 and KS2003, it is suggested that: (1) the hanging wall effect should be reduced gradually instead of being restricted in a denite distance limit, and disappear immediately once the rupture distance limit is over, such as the distance limit x4 in AS1996 and KS2003; and (2) all data on hanging wall sites should be used in the regression analysis. Based on the analysis results, the model is revised as: R = R1 = c1r 2 + c2 r + c3 R = R2 = exp(c4 r + c5 ) for x1 r x2 for r x2 (4)

1.5 1.0 PGAH residual 0.5 0 -0.5 -1.0 -40


Footwall Hanging wall
Our model Model in AS1996 Northridge recording data

where the residual R may refer to both the PGA and SA on hanging wall sites; the boundary distances x1 and x2 and coefcient c1, c2, c3, c4 and c5 are all to be determined by residual distribution of hanging wall sites. In particular, because the hanging wall effect originated from the general proximity of the hanging wall sites to a fault plane, sites directly over the ground surface trace of the fault rupture or the vertical projection of the upper edge of a buried fault should be the division between

-30

-20 -10 0 10 20 Rupture distance (km)

30

40

Fig. 5

Comparison of the predictions by using the proposed model and the AS1996 when applied to the Northridge earthquake

No.3 Peng Lei et al.: A new attenuation model of near-fault ground motions with consideration of the hanging wall effect in the Wenchuan earthquake 319

the actual hanging wall effect due to the omission of some sections of data would become signicant. The modeling curve of KS2003 in Fig. 6 turns out to enclose almost all the data in the distance range of less than 40 km, where the most signicant hanging wall effect occurs, while the proposed model runs through all data in a more balanced way. The results indicate that the KS2003 model predicts a much higher hanging wall effect for the Chi-Chi earthquake. On the other hand, due to the abundance of near-fault data from the ChiChi earthquake, the proposed model can ensure that the hanging wall effect weakens smoothly at distances of around 50 km rather than being forced to disappear immediately, as expected when designing the model. Further, this type of slowly disappearing effect also can be captured in some NGA models, such as the model developed by Campbell and Bozorgnia (2007).
0.6 0.4 PGA resultant residual 0.2 0 -0.2 -0.4 -0.6 -0.8 -60
Footwall
Our model Model in KS2003 Rotated PGA of recording in Chi-Chi earthquake

Table 4 Boundary distances and coefcients of model for hanging wall effect of the PGAH in Wenchuan earthquake Parameter x1 (km) x2 (km) c1 c2 c3 c4 c5 Value 0 39.29 -0.0006943 0.0314 0 -0.1433 3.8093

Hanging wall

-40

-20 0 20 40 Rupture distance (km)

60

Fig. 6 Comparison of the predictions by using the proposed model and KS2003 when applied to the Chi-Chi earthquake

4.4 Application of the proposed model to the Wenchuan earthquake to model the hanging wall effect in PGAH The USGS fault solution for the Wenchuan earthquake used herein suggests that the upper edge of the fault is buried at a depth of merely 0.7411 km. However the eld investigation carried out by Chinese investigators (General Introduction to Engineering Damage During Wenchuan Earthquake, 2008; Li et al., 2008) observed many areas with large permanent ground displacement caused by the fault rupture reaching ground surface. So, it was assumed that x1 was located at 0 km. In this Wenchuan earthquake specic model, c3 = 0 since the model presumes no hanging wall effect at a distance of x1, which is nil here. Using the procedure described above results in: c4 = 2c1 x2 + c2 2 , c5 = ln(c1 x2 + c2 x2 ) c4 x2 2 c1 x2 + c2 x2

The values of the coefcients and boundary distances are listed in Table 4. A comparison of the residual

distribution is shown in Fig. 7, where the hanging wall effect term used in the AS NGA model with some modications to the original form used in Abrahamson and Silva (1997) is also plotted. The previous model lead to jumps in the hanging wall effect scaling for some cases and it was not clear how to apply the model for steeply dipping faults (Abrahamson and Silva, 2008). In fact, the previous model used in the ground motion attenuation relationship of Abrahamson and Silva (1997) has some similarities with the AS1996 model proposed by Abrahamson and Somerville to model the hanging wall effect in the Northridge earthquake exclusively. Then, as the proposed model is compared with the AS1996 model in the preceding context, it would be favorable to have a comparison of the hanging wall effect term in the AS NGA compared to the hanging wall effect term in other NGA models. Note that the AS NGA model predicts a much greater hanging wall effect than observed in almost the entire range of distance except the nal portion. The proposed model yields a much better estimation of this effect. In addition, it may be inferred that the hanging wall effect of PGAH in this event exists in a predictable level for the distance range with the most signicant effect. However, the AS NGA model presumes that an abrupt rise of values over the division of the hanging wall and footwall, which is different from the preference of Campbell and Bozorgnia (2007) and Chiou and Youngs (2006). Campbell and Bozorgnia (2007) suggests that a smooth transition between the hanging wall and the footwall exists, while the NGA models have no zero value over the division of the hanging wall and footwall. Differing from the above models, the proposed model denes a zero value on the division. Finally, the AS NGA model inherits the characteristics of the hanging wall term in the previous model that the hanging wall effect vanishes at a denite distance, which introduces more uncertainty and may be the cause of the signicant underestimation of the effect at the ending range of distance. Nevertheless, it is necessary to note that all the NGA models used a new measure of ground motion, GMRotI50, dened in Boore et al., (2006) instead of the traditional geometrical mean of two horizontal components used in this study. The traditional geometric mean is calculated as the squared foot of the product

320

EARTHQUAKE ENGINEERING AND ENGINEERING VIBRATION

Vol.10

of the peak values of the two as-recorded orthogonal horizontal components. This denition of ground motion depends on the orientation of the sensor as installed in the eld. This means that the ground motion parameter could differ for the same ground motion when the sensor has a different orientation. This dependence on the sensors orientation is most signicant for strongly correlated motions that often occur at periods of 1 sec or longer (Campbell and Bozorgnia, 2007). The new measure of ground motion, GMRotI50, removes the dependence on the sensor orientation, as it is computed based on a set of geometric means of the as-recorded orthogonal horizontal motions rotated through all possible non-redundant rotation angles. And, a single1.5 1.0 PGAH residual 0.5 0 -0.5 -1.0 -1.5
Footwall
Soil site Rock site Our model curve Hanging wall effects term in AS NGA

recognized that the rupture directivity has a signicant effect in long periods. Therefore, the residuals of SAH at several periods from short to long are examined in an attempt to identify the inuence of the directivity effect. Because of the large reverse slip in some areas of the fault plane (Chen and Gavin, 2008), it is assumed that the difference between the spectral acceleration residuals of the hanging wall and the footwall caused by the hanging wall effect would be inuenced by the rupture directivity effect at long periods. 5.1 Residual distribution of SAH The residuals of SAH at four different periods are plotted in Fig. 8 with the hanging wall sites at a positive rupture distance and the footwall sites at a negative rupture distance. Note that at the short period of 0.1s, residuals within 60 km on the hanging wall side are inclined to be mostly positive and those on the footwall side are inclined to be mostly negative. Thus, the distribution of residuals at 0.1 s means the hanging wall sites experienced much larger ground motion than the footwall sites. Several authors have indicated that the strong motions on the hanging wall are greater than on the footwall of a dipping fault, which are referred to as hanging wall effects (Abrahamson and Somerville (1996) for the 1994 Northridge earthquake, Chang et al. for the 1999 Chichi earthquake, and Si and Midorikawa for the 2004 Mid-Niigata Prefecture earthquake). Thus, at 0.1 s, the hanging wall effect is strong. At median periods of 0.5 s and 0.6 s, a neutral trend happens, i.e., the residuals on both the hanging wall and footwall sides seem to be evenly distributed on the either side of the zero line. Finally, at long period of 2.0 s, a completely reverse distribution occurs the hanging wall sites have their residuals mostly negative while the footwall sites have mostly positive residuals. Actually, the hanging wall effect decreases as the period increases. Without the existence of the hanging wall effect, residuals on the hanging wall side and footwall side should be evenly distributed on either side of the zero line like the pattern of 0.5 s and 0.6 s in Fig. 8. However, from the outcome of this study, the difference was reduced at rst, and then it continued to reverse over the balance ending with the situation that the footwall sites have mostly positive residuals. Therefore, it may be that the rupture directivity effect shows its inuence on ground motions at long periods, of 1.0 s longer, and the seemly extraordinary trend of residuals at periods of 2.0 s in Fig. 8 might be due to rupture directivity effect in this earthquake. From the distribution of stations around the fault shown in Fig. 1, note that on the side of the footwall, most stations were located close to the fault trace while stations on the hanging wall were mostly located far away from the fault trace, around which the rupture directivity effect is mostly concentrated for a reverse fault (Somerville et al., 1997; Shabestari and Yamazaki, 2003; Spudich and Chiou, 2008). Even from the simulation results of Spudich et al. (2008), the rupture directivity effect

Hanging wall

-100 -80 -60 -40 -20 0 20 40 60 80 100 Rupture distance (km)

Fig. 7 Comparisons of residuals of PGAH for all sites evaluated by the present model and the AS NGA model

period-independent rotation to minimize the spread of the rotation-dependent geometric mean over the usable range of oscillator periods is used (Boore et al., 2006). In effect, in this Wenchuan earthquake specic analysis, the use of a traditional geometric mean introduced no dependence on the sensors orientation. And, the comparison of the new measure, GMRotI50, with the traditional way shows that the GMRotI50 provides a theoretically larger ground motion, but only by a small amount (less than 3%). This means that the comparison of the proposed model of the hanging wall effect in the Wenchuan earthquake using traditional geometric mean with the AS NGA model is feasible.

5 Modeling of hanging wall effect exhibited in SAH in Wenchuan earthquake


Following the landmark paper on rupture directivity by Somerville et al. (1997), a great deal of empirical and numerical studies (Spudich and Chiou, 2008; Spudich et al., 2006; Hwang et al., 2001; Shabestari and Yamazaki, 2003; Bray and Adrian, 2004; Hu, 2009) have been performed. However, there are many inuencing factors involved in the rupture directivity effect and a small near-fault data set, thus the empirical analysis of the rupture directivity effect is not efcient and is not included in the rst phase of NGA. It is generally

No.3 Peng Lei et al.: A new attenuation model of near-fault ground motions with consideration of the hanging wall effect in the Wenchuan earthquake 321
1.5 1.0 SAH residual SAH residual 0.5 0 -0.5 -1.0 -1.5 -100 -80 -60 -40 -20 0 20 40 60 80 100 Rupture distance (km) 1.5 1.0 SAH residual SAH residual 0.5 0 -0.5 -1.0 -1.5 -100 -80 -60 -40 -20 0 20 40 60 80 100 Rupture distance (km)
Footwall Hanging wall
Soil site Rock site

Soil site Rock site

1.5 1.0 0.5 0 -0.5 -1.0

Soil site Rock site

Footwall

Hanging wall

-1.5 -100 -80 -60 -40 -20 0 20 40 60 80 100 Rupture distance (km) 1.5 1.0 0.5 0 -0.5 -1.0 -1.5 -100 -80 -60 -40 -20 0 20 40 60 80 100 Rupture distance (km)
Footwall Hanging wall
Soil site Rock site

Footwall

Hanging wall

Fig. 8 Residuals of averaged SAH of 5% damping at periods 0.1 s, 0.5 s, 0.6 s and 2.0 s in the Wenchuan earthquake

would exhibit in a broader area away from the fault trace on the footwall. Thus footwall stations will experience much more inuence from rupture directivity under the conditions of the rupture directivity effect in the Wenchuan earthquake, and thereby explaining the reverse trend of residual distribution. 5.2 Empirical modeling of the hanging wall effect for SAH Because all rst phase NGA models did not include any rupture directivity effect factors, the rupture directivity effect is mixed within the ground motion at long periods. When modeling the hanging wall effect for SAH and comparing it with the hanging wall function for spectral acceleration in the AS NGA model, it would be best to model the hanging wall effect for SAH only at 0.1s to avoid introducing the inuence of rupture directivity so no rupture directivity effect is included in the comparison. The same model as previously used as for PGAH is again used here. By using a similar procedure described in subsection 4.2 for PGAH, all coefcients and boundary distances of the model obtained for SAH of 0.1s based on the 0.1s residuals shown in Fig. 8 are listed in Table 5 and the SAH residual distribution for the Wenchuan earthquake is shown in Fig. 9. Note that just like the previous analysis for PGAH, the AS NGA once again predicts much larger SAH residuals. However, in this case, it predicts a very short distance of the hanging wall effect. In addition, in the distance range where the AS NGA model predicts that the hanging wall

effect disappears, the hanging wall effect on SAH is still very pronounced. Presumably, the scarcity of data and the resulting aleatory distribution pattern of data especially at large distances in this earthquake might be the explanation for this phenomenon. But as mentioned previously, it is possible that the AS NGA models estimation of the ending distance of no effect would introduce uncertainty and cause such a large difference. However, the AS NGA models prediction does include all data points on the hanging wall within the distance range with the largest effect in Fig. 9. From this phenomenon, it is possible that the hanging wall effect of SAH in this earthquake remains in the predictable level for a distance range with the most signicant effect as well. And, the AS NGA model achieved an acceptable prediction of the hanging wall effect to some extent in this earthquake.
Table 5 Boundary distances and coefcients of the model for hanging wall effect of the SAH in Wenchuan earthquake Parameter x1 (km) x2 (km) c1 c2 c3 c4 Value 0 24.17 -0.0012313 0.049363 0 -0.021441

c5

-0.22877

322

EARTHQUAKE ENGINEERING AND ENGINEERING VIBRATION

Vol.10

1.5 1.0 PGAH residual 0.5 0 -0.5 -1.0 -1.5


Footwall
Soil site Rock site Our model curve Hanging wall effects term in AS NGA

Wenchuan earthquake used in this study. The discussions with Mr. Gong Maosheng, Zhou Baofeng and Ren Yefei were extremely helpful.

References
Abrahamson NA and Silva WJ (1997), Empirical Response Spectral Attenuation Relations for Shallow Crustal Earthquakes, Seismological Research Letters, 68(1): 94109. Abrahamson NA and Silva WJ (2008), Summary of the Abrahamson & Silva NGA Ground-motion Relations, Earthquake Spectral, 24(1): 6797. Abrahamson NA and Somerville PG (1996), Effect of the Hanging Wall and Footwall on Ground Motions Recorded During the Northridge Earthquake, Bulletin of the Seismological Society of America, 86(1B): S93 S99. Abrahamson NA and Youngs RR (1992), A Stable Algorithm for Regression Analyses Using the Random Effect Model, Bulletin of the Seismological Society of America, 82(1): 505510. Boore DM, Jennie Watson-Lamprey and Abrahamson DA (2006), Orientation-independent Measures of Ground Motion, Bulletin of the seismological Society of America, 96(4A): 15021511. Bray JD and Adrian Rodriguez-Marek (2004), Characterization of Forward-directivity Ground Motions in the Near-fault Region, Soil Dynamics and Earthquake Engineering, 24: 815828. Campbell KW (1993), Empirical prediction of nearsource ground motion from large earthquakes, Proc. International Workshop on Earthquake Hazard and large Dam, New Delhi, India. Campbell KW and Bozorgnia Yousef (2007), Campbell-Bozorgnia NGA Ground Motion Relations for the Geometric Mean Horizontal Component of Peak and Spectral Ground Motion Parameters, PEER Report. 2007. Chen Ji and Gavin Hayes (2008), Finite Fault Modelpreliminary Result of the May 12, 2008 Mw 7.9 Eastern Sichuan, China Earthquake, Report by U.S. Geological Survey, (http://earthquake.usgs.gov/eqcenter/ eqinthenews/2008/us2008ryan/nitefault. php, lasted accessed 9 10th, 2009). Chiou and Youngs (2006), Chiou and Youngs NGA Empirical Ground Motion Model for the Average Horizontal Component of Peak Acceleration and Pseudo-spectral Acceleration for Spectral Periods of 0.01 to 10 Seconds, Interim Report for USGS Review. Hu Jinjun (2009), Directivity Effect of Near-fault Ground Motion and Super-shear Rupture, PhD Dissertation, Institute of Engineering Mechanics, China Earthquake Administration, Harbin, China. Hwang RD, Yu GK, and Wang JH (2001), Rupture Directivity and Source-process Time of the September

Hanging wall

-100 -80 -60 -40 -20 0 20 40 60 80 100 Rupture distance (km)

Fig. 9 Comparisons of residuals of SAH for all sites evaluated by the present model and the AS NGA model

6 Conclusions and discussions


In this paper, a new formula was developed to evaluate the hanging wall effect in the Wenchuan earthquake based on modications to two previous models, the AS1996 and KS2003. The proposed model in this paper offers some advantages over the AS1996 and KS2003 models. First, the proposed model make full use of all data on the hanging wall; second, it has only two boundary distances to x in the process, so it is easy to use; third, the proposed model suggests no denite distance limit to pinch the effect, even though it may have descended to be very small. The proposed model was applied to the Wenchuan earthquake and compared with the results from the hanging wall effect factor in the Abrahamson and Silva NGA model. From the comparison, the following conclusions can be made. (1) The AS NGA model provides a much larger prediction of the hanging wall effect in the Wenchuan earthquake. (2) The AS NGA model inherited a characteristic of the ending hanging wall effect at a denite distance, which is exhibited in both the AS1996 model and the hanging wall effect factor of Abrahamson and Silvas 1997 attenuation relationship. This characteristic might introduce uncertainty in determining the ending distance and might contribute to a large difference between the observed hanging wall effect in the Wenchuan earthquake and the values predicted by the model. (3) The reversed trend of residual distributions on both sides of the hanging wall and the footwall at long periods shows that the footwall stations have positive residuals, and the rupture directivity effect exerts its inuence on ground motions around the fault trace, as was presumed after analysis.

Acknowledgement
The authors are grateful to the National Strong Motion Network for providing recordings of the

No.3 Peng Lei et al.: A new attenuation model of near-fault ground motions with consideration of the hanging wall effect in the Wenchuan earthquake 323

20, 1999 Chi-Chi, Taiwan, Earthquake Estimated from Rayleigh-wave Phase Velocity, Earth Planets Space, 53: 11711176. Li Yong, Zhou Rongjun et al. (2008), Surface Rupture, Reverse-strike Slip in the Wenchuan Earthquake, Journal of Chengdu University of Technology (Science & Technology Edition), 35(4): 404413. (in Chinese) Shabestari KT and Yamazaki Fumio (2003), Nearfault Spatial Variation in Strong Ground Motion due to Rupture Directivity and Hanging Wall Effect from the Chi-Chi, Taiwan Earthquake, Earthquake Engng Struct. Dyn, 32: 21972219. Somerville PG and Graves RW (2003), Characterization of Earthquake Strong Ground Motion, Pure Appl. Geophys, 160(2003): 18111828. Somerville PG, Smith NF, Graves RW (1997), Modication of Empirical Strong Ground Motion Attenuation Relations to Include the Amplitude and

Duration Effect of Rupture Directivity, Seismological Research Letters, 68(1): 199222. Spudich Paul and Chiou BSJ (2008), Directivity in NGA Earthquake Ground Motions: Analysis Using Isochrone Theory, Earthquake Spectral, 24(1): 279298. Spudich P, Chiou BSJ, Graves R, Collins N and Somerville P (2006), A Formulation of Directivity for Earthquake Sources Using Isochrones Theory, U.S Geological Survey Open File Report 2004-1263, available at http://usgs.gov/of/2004/1268/. Wang Dong (2010), The Hanging Wall/Footwall Effect of Near-Fault Ground Motions, PhD Dissertation, Institute of Engineering Mechanics, China Earthquake Administration, Harbin, China. Wang Dong, Xie Lili and Hu Jinjun (2008), Geometric Effect Resulting from the Asymmetry of Dipping Fault: Hanging Wall/ Footwall Effect, Acta Seismologica Sinica, 30(3): 271278.

You might also like