Portal de Periódicos da CAPES

Triggering and runaway processes of catastrophic Tsaoling landslide induced by the 1999 Taiwan Chi-Chi earthquake, as revealed by high-velocity friction experiments

2014; American Geophysical Union; Volume: 41; Issue: 6 Linguagem: Inglês

10.1002/2013gl059169

1944-8007

Tetsuhiro Togo, Toshihiko Shimamoto, Jia‐Jyun Dong, Chyi‐Tyi Lee, Che-Ming Yang,

Rock Mechanics and Modeling

Geophysical Research LettersVolume 41, Issue 6 p. 1907-1915 Research LetterFree Access Triggering and runaway processes of catastrophic Tsaoling landslide induced by the 1999 Taiwan Chi-Chi earthquake, as revealed by high-velocity friction experiments Tetsuhiro Togo, Corresponding Author Tetsuhiro Togo State Key Laboratory of Earthquake Dynamics, Institute of Geology, China Earthquake Administration, Beijing, China Correspondence to: T. Togo, duketogotetsu@gmail.comSearch for more papers by this authorToshihiko Shimamoto, Toshihiko Shimamoto State Key Laboratory of Earthquake Dynamics, Institute of Geology, China Earthquake Administration, Beijing, ChinaSearch for more papers by this authorJia-Jyun Dong, Jia-Jyun Dong Graduate Institute of Applied Geology, National Central University, Jungli, TaiwanSearch for more papers by this authorChyi-Tyi Lee, Chyi-Tyi Lee Graduate Institute of Applied Geology, National Central University, Jungli, TaiwanSearch for more papers by this authorChe-Ming Yang, Che-Ming Yang Graduate Institute of Applied Geology, National Central University, Jungli, TaiwanSearch for more papers by this author Tetsuhiro Togo, Corresponding Author Tetsuhiro Togo State Key Laboratory of Earthquake Dynamics, Institute of Geology, China Earthquake Administration, Beijing, China Correspondence to: T. Togo, duketogotetsu@gmail.comSearch for more papers by this authorToshihiko Shimamoto, Toshihiko Shimamoto State Key Laboratory of Earthquake Dynamics, Institute of Geology, China Earthquake Administration, Beijing, ChinaSearch for more papers by this authorJia-Jyun Dong, Jia-Jyun Dong Graduate Institute of Applied Geology, National Central University, Jungli, TaiwanSearch for more papers by this authorChyi-Tyi Lee, Chyi-Tyi Lee Graduate Institute of Applied Geology, National Central University, Jungli, TaiwanSearch for more papers by this authorChe-Ming Yang, Che-Ming Yang Graduate Institute of Applied Geology, National Central University, Jungli, TaiwanSearch for more papers by this author First published: 17 March 2014 https://doi.org/10.1002/2013GL059169Citations: 26AboutSectionsPDF ToolsRequest permissionExport citationAdd to favoritesTrack citation ShareShare Give accessShare full text accessShare full-text accessPlease review our Terms and Conditions of Use and check box below to share full-text version of article.I have read and accept the Wiley Online Library Terms and Conditions of UseShareable LinkUse the link below to share a full-text version of this article with your friends and colleagues. Learn more.Copy URL Share a linkShare onFacebookTwitterLinked InRedditWechat Abstract Pliocene sedimentary rocks of about 130 Mm3 in volume slid along bedding planes dipping by 14°, with an average speed of about 35 m/s, during the Tsaoling landslide. We conducted friction experiments to reproduce the initiation processes of this landslide, by idealizing landslide movements during the earthquake as accelerating/decelerating motion. Experiments were done on shale from the field, at 3 MPa normal stress corresponding to the overburden pressure. Results indicate that the accelerating/decelerating motion causes weakening and strengthening at each oscillation cycle and results in overall slip weakening which can be approximated as an exponential slip weakening. Behaviors during oscillatory slip are fairly similar to those during sliding at constant slip rates. Newmark analysis with measured frictional properties reveals that the landslide can be triggered with wet gouge properties, but the landslide motion stops with parameters for dry shale gouge. Delayed initiation of the landslide is consistent with a survivor's witness. Key Points High-velocity friction experiments reproduced earthquake-induced landslide 1 Introduction Tsaoling landslide (Figure 1a) is the largest among more than 20,000 landslides triggered by the 1999 Chi-Chi earthquake (Mw = 7.6), with its volume reaching about 130 Mm3 [Chen et al., 2003; Chigira et al., 2003; Chen et al., 2006; Tang et al., 2009]. A landslide mass of about 140 m in thickness on the southern slope of the Tsaoling Mountain failed and slid along flat bedding planes dipping by 14°, within alternated fine sandstone and shale beds in Pliocene Cholan formation near the boundary to the Chinshui formation (Figure 1a). Thirty-nine people living in a village on the landslide mass slide with the landslide, and seven of them survived [Chen et al., 2003]. A survivor narrated to us that the landslide started in about 10 s after he felt the Chi-Chi earthquake. The southern slope of the Tsaoling Mountain collapsed in 1862, 1941, and 1999 due to earthquakes and in 1898, 1942, 1951, and 1979 due to heavy rainfalls [Chen et al., 2006]. Figure 1Open in figure viewerPowerPoint (a) A cross section of the Tsaoling landslide area in the direction of sliding before and after the Chi-Chi earthquake (after Tang et al. [2009]). (b–d) Accelerations in vertical, NS, and EW directions, respectively, during the Chi-Chi earthquake recorded at the CHY080 station on the backside of the Tsaoling Mountain (positive upward, northward, and eastward). The time is set to be zero for the time of occurrence of the Chi-Chi earthquake at its epicenter. (e, f) Downdip accelerations ad (downward positive) and normal acceleration an (upward positive) on the sliding surface, respectively, as calculated using equation 3 in Huang et al. [2001]. (g) Frequency spectrum of ad as analyzed with the fast Fourier transform analysis. Modeling with the discrete element method [Tang et al., 2009] and an analysis using the Saint Venant equations [Kuo et al., 2009] reproduced the catastrophic Tsaoling landslide with the basal friction coefficients of 0.1 and 0.15, respectively. Moreover, results in Figures 13b and 13c in Tang et al. [2009] suggest that the Tsaoling landslide block slid nearly as an intact mass for about 1 km. We thus consider that early processes of the Tsaoling landslide can be studied by high-velocity friction experiments on the basal slip-zone materials. High-velocity friction experiments revealed dramatic weakening of simulated faults at seismic slip rates on the order of 1 m/s [e.g., Di Toro et al. [2011], papers quoted therein]. Similar experiments revealed very low steady-state friction coefficient of 0.05–0.2 for slip-zone materials from landslides (e.g., the Tsaoling landslide by Miyamoto et al. [2009] and the Vaiont landslide by Ferri et al. [2011]). A seismic record at a nearby station CHY080 and a survivor's witness quoted above indicate that the average speed of the landslide mass reached 22–42 m/s (see Supplementary Information S1). Analysis of landslide motion assuming a constant kinetic friction coefficient μk indicates that μk as low as 0.033–0.15 is needed for the landslide block to reach that range of velocity (Supplementary Figure FS01c). Thus, experimental results on the steady-state friction are consistent with the inferred range of μk. As for catastrophic landslides triggered by an earthquake, an important issue to be addressed is how landslide motion initiates during seismic ground motion. Early landslide motion during an earthquake will never be constant as in most previous experiments. Thus, in this paper, we conducted oscillatory velocity experiments by idealizing the initial movement as an oscillating accelerating/decelerating motion. We also consider that slip weakening or reduction of friction with increasing displacement is essential in triggering the catastrophic landslide. Indeed, Huang et al. [2001] performed an analysis of the Tsaoling landslide using the Newmark method [e.g., Newmark, 1965; Jibson, 1993], but they could not reproduce a runaway process of the landslide because the slip-weakening property of friction was not included in the analysis. Therefore, we attempt Newmark analysis of the Tsaoling landslide using measured properties, as attempted by Dong et al. [2013], but using a simplified friction law. We believe that an approach combining friction experiments and Newmark analysis will have many applications to catastrophic landslides along flat bedding or fracture planes in earthquake-prone areas [e.g., Xu et al., 2009]. 2 Experimental Procedures Friction experiments were conducted on crushed shale gouge between solid cylindrical specimens of Belfast dolerite of 25 mm in diameter with a Teflon sleeve outside to contain the gouge, using a low-velocity to high-velocity friction apparatus then at Hiroshima University [Togo et al., 2011; Togo and Shimamoto, 2012]. Shale is the weaker member in the alternated sandstone and shale beds, and therefore we used a shale sample collected from the Cholan formation immediately below the sliding surface of the Tsaoling landslide in our experiments. A rationale for using powdered gouge is that even a small slip along a fracture or bedding plane generates gouge and that the frictional behavior is controlled by gouge, rather than by host rock, once a slip surface is covered with gouge [Hirose et al., 2012; papers quoted therein]. We used dolerite because it is resistant to thermal fracturing caused by the frictional heating. Experiments were done at a normal stress of 3.0 MPa, corresponding to the overburden pressure on the basal slip plane due to the landslide mass of about 140 m in thickness (Figure 1a). Slip rates in our experiments were up to 1.3 m/s, not high enough to reproduce the speed of the Tsaoling landslide (about 35 m/s), owing to the limited capability of the apparatus. But we tried to extrapolate our experimental results to higher slip rates. Supplementary Information S2 details the experimental procedures and X-ray diffraction analysis of shale gouge. A new attempt in this paper is an oscillatory velocity experiment. While acceleration in a direction may favor for sliding of the landslide block, shaking back to the opposite direction will tend to stop the motion. Thus, we idealized the initial landslide motion as an oscillating movement with a linear acceleration from v = 0 to the maximum velocity Vmax followed by a linear deceleration back to zero, with a frequency f. Acceleration records in vertical, NS, and ES directions at CHY080 station (Figures 1b–1d) were used to calculate accelerations, ad and an, in the downdip and normal directions to the sliding surface, respectively (Figures 1e and 1f). Downdip acceleration ad is important in triggering a landslide, and it has frequencies f of 0.3–5 Hz with a maximum amplitude at 1.2 Hz (Figure 1g). We thus selected f = 0.3, 0.6, 0.9, and 1.2 Hz and the maximum slip rate Vmax = 0.33, 0.65, 1.0, and 1.3 m/s in our experiments to get an overview of the frictional behavior during oscillatory slip. Fluctuation of friction decreases with increasing f as we see later, and it was not necessary to cover higher frequencies. 3 Constant Slip-Rate Experiments We conducted experiments at constant slip rates v on dry gouge with room humidity and wet gouge with 25 wt% of water under drained conditions. Figures 2a and 2b exhibit friction coefficient versus displacement curves for eight dry runs at v = 0.0124–1.30 m/s and for four wet runs at v = 0.22–1.3 m/s. At slip rates greater than 0.161 m/s, the dry and wet shale gouges exhibit nearly exponential slip weakening from the peak friction coefficient μp toward the steady-state friction coefficient μss, as expressed by (1)where μ is a friction coefficient at a displacement d, and Dc is a characteristic slip-weakening distance [Mizoguchi et al., 2007]. Note that Dc is a displacement required for the friction coefficient to drop by 95% of (μp − μss) and is close to the slip-weakening distance in a linear slip-weakening law of friction. The value of μp for dry gouge is about 0.7, whereas wet gouge has a μp of about 0.6 which tends to decrease slightly with increasing v (Figure 2c). Values of μss for dry gouge decrease rapidly with increasing v at v greater than above about 0.1 m/s and reduces to about 0.15 at high slip rates, and μss of wet gouge decreases from 0.24 at v = 0.22 m/s to 0.07 at 1.3 m/s. Dc of dry gouge decreases from 7.5 m at v = 0.161 m/s down to 2.7–2.5 m at v = 1.0–1.3 m/s, whereas wet gouge has Dc decreasing from 3.2 m at v = 0.22 m/s to 1.8–1.3 m at 0.87–1.3 m/s (Figure 2d). Figure 2Open in figure viewerPowerPoint Friction coefficient versus displacement curves during the frictional sliding of (a) dry shale gouge with room humidity and (b) wet shale gouge with 25 wt% of H2O (data quoted from Miyamoto et al. [2009]), at a normal stress of 3.0 MPa and at constant slip rates as given in the diagrams. (c) The peak friction coefficient (open circles) and the steady-state friction coefficient μss (filled circles), plotted against the slip rate v in a logarithmic scale (dry gouge in red and wet gouge in blue). (d) The slip weakening distance Dc plotted against the slip rate for dry and wet gouges in red and blue, respectively. The peak and steady-state friction for dry gouge cannot be distinguished clearly at v less than 0.0436 m/s because the slip weakening is not pronounced (e.g., two curves in Figure 2a), and hence we plotted the average friction coefficient during the frictional sliding with red filled circles in Figure 2c (μ-d curves for the three datum points are not shown in Figure 2a since they are overlapped with the other two curves). The change in μss in Figure 2b can be described by an empirical equation: (2)where μss(v) is the steady-state friction at a slip rate v, μss|v = 0 and μss|v = ∞ are the steady-state friction coefficients at v = 0 and infinity, respectively, Vc is the critical velocity for dramatic high-velocity weakening, and A specifies how rapidly μss decreases with increasing v (A was added to equation 2 of Togo et al. [2011] to get better fitting). The solid curve in Figure 2b is a least squares fit to the data with constants, μss|v = 0 = 0.70 ± 0.010, μss|v = ∞ = 0.17 ± 0.011, Vc = 0.10 ± 0.031 m/s, and A = 1.6 ± 0.20, the error being one standard deviation. μss for wet gouge is slightly smaller than that for dry gouge, but we did not fit the data with an empirical law because of limited data. The change in Dc with v in Figure 2d can be fit with an equation [Togo et al., 2011]: (3)with constants, B = 2.74 ± 0.19 and C = 0.40 ± 0.05 for dry gouge, and B = 1.54 ± 0.32 and C = 0.39 ± 0.19 for wet gouge (solid and dashed curves, respectively, in Figure 2d). The meanings of those empirical equations are discussed in Yao et al. [2013a] and will not be repeated here. 4 Oscillatory Velocity Experiments This section reports results from oscillatory velocity experiments on dry shale gouge (no runs were made on wet gouge because it leaks easily from the interface between host rock and Teflon sleeve). In an example with f = 1.2 Hz and Vmax = 0.65 m/s (Figure 3a), μ decreases during an accelerating slip and increases during a decelerating slip at each oscillation cycle, while the overall friction decreases nearly exponentially with cumulative displacement. A rapid strength recovery with decelerating slip was reported in Sone and Shimamoto [2009], who conducted single accelerating/decelerating experiments. The overall weakening can be fit with equation 1 with μp = 0.80, μss = 0.26, and Dc = 4.2 m (solid curve in Figure 3b). Results for four different Vmax are shown in Figures 3c to 3f, using different colors for different frequencies f. In all cases, nearly exponential slip weakening is recognized, and f did not affect the overall weakening behaviors much as can be seen by slip weakening parameters below. Figure 3Open in figure viewerPowerPoint (a) A representative example of the oscillatory acceleration/deceleration experiment conducted with room humidity and at a normal stress of 3.0 MPa, with a frequency f of 1.2 Hz and a maximum slip rate Vmax of 0.65 m/s. (b) The same data for μ as in Figure 3a that are fit with an equation 1 for an exponential slip weakening (solid curve). (c–f) Friction coefficient plotted against displacement during oscillatory experiments with Vmax of 0.33, 0.65, 1.0, and 1.3 m/s, respectively, and with frequencies of 0.3, 0.6, 0.9, and 1.2 Hz as shown in different colors. Dashed lines in Figures 3b–3f show a friction coefficient of 0.25 that corresponds to a dip angle of 14° for the basal slip plane of the Tsaoling landslide. (g) Peak friction coefficient (open symbols) and the steady-state friction coefficient (solid symbols) and (h) the slip-weakening distance Dc for the oscillation tests, plotted against the maximum velocity Vmax (the lower horizontal axis) and against the average slip rate (the upper horizontal axis). Solid curves in Figures 3g and 3h are the relationships from constant slip-rate experiments in Figure 2. Slip weakening parameters are plotted against Vmax (lower horizontal axis) and average slip rate Vav (upper horizontal axis) in Figure 3g. Note Vav = Vmax/2 for linear acceleration and linear deceleration as in our experiments. Values of μp for the oscillation tests range from 0.66 to 0.80 and are about the same as those for the constant slip-rate tests, although μp tends to decrease slightly with increasing Vmax (open symbols in Figure 3g). Whereas μss from the oscillation tests tends to decrease from 0.36–0.39 at Vmax = 0.33 m/s to 0.11–0.17 at Vmax = 1.3 m/s (filled symbols in Figure 3g). If Vav for the oscillation tests is used in equation 1 for the constant slip-rate tests (solid curve in Figure 3g), μss values from both types of tests become about the same except at Vmax = 0.33 m/s or Vav = 0.165 m/s where μss from the constant slip-rate tests is greater than μss from the oscillation tests by about 0.1. Dc from the oscillation tests is fairly insensitive to Vmax and is somewhat variable depending on f (Figure 3h). Dc from the constant slip-rate tests using Vav (solid curve in Figure 4b) agrees on the whole with Dc from the oscillation tests although the latter is less sensitive to Vav than the former. Thus, the overall slip-weakening behavior during the oscillatory slip is roughly similar to that during constant slip-rate tests if Vav ~ v. Figure 4Open in figure viewerPowerPoint Results of Newmark analysis of Tsaoling landslide using slip-weakening parameters (a) and (b) for dry shale gouge and (c) and (d) for wet shale gouge. Acceleration in the downdip direction, friction coefficient, velocity, and displacement are plotted against time in gray, red, blue, and green curves, respectively. Figures 4b and 4d show the close-ups of the initiation portions of the landslide motion. 5 Newmark Analysis of the Tsaoling Landslide We now conduct Newmark analysis of the Tsaoling landslide incorporating the frictional properties reported in the previous two sections. No constitutive equation is proposed to describe frictional properties for arbitrary slip histories [e.g., Di Toro et al. [2011]], and precise analyses for the landslide motion are not possible at present. However, the overall behavior during the oscillatory slip is similar to the slip-weakening behavior at constant slip rates (cf. Figures 2 and 3). Thus, we use an exponential slip weakening, equation 1, as a simplified friction law with representative parameters: μp = 0.7, μss = 0.1, and Dc = 4.0 m for dry gouge and μp = 0.56, μss = 0.07, and Dc = 1.7 m for wet gouge (Figures 2 and 3). The friction coefficient μ for a friction angle of 14° is 0.25, so that the Tsaoling landslide along bedding planes dipping 14° could not be triggered unless μ drops below this value. This μ value is between the above μp and μss values, and hence a critical point for triggering the catastrophic landslide is whether μ drops below 0.25 or not during early landslide motion due to the seismic ground motion. Huang et al. [2001] derived an equation for the acceleration S of the landslide block during seismic ground motion: (4)where g is acceleration of gravity, θ is slope angle (14°), μ is friction coefficient, and ad and an are accelerations in the downdip and normal directions to the slip surface, respectively (Figures 1e and 1f). An upward normal acceleration or a positive an increases the normal stress on the sliding surface, thereby decreasing S (landslide motion suppressed). On the other hand, acceleration in the updip direction or a negative ad increases S and enhances the landslide motion. We replaced static friction coefficient with μ and neglected the cohesion term in the original equation. Slip weakening decreases μ and increases S, thereby promoting the landslide motion. Supplementary Information S3 reviews the derivation of equation 4 and reports results from the landslide analysis using a software by Jibson and Jibson [2003] without taking into account the normal acceleration an. Figure 4 exhibits S, μ, sliding velocity V, and displacement δ of the landslide block in gray, red, blue, and green curves, respectively. The landslide block moves when S > 0, and V and δ can be calculated by integrating S and V, respectively [e.g., Jibson, 1993]. Result for dry frictional properties in Figures 4a and 4b indicates that δ increased to 0.63 m, μ dropped to 0.59, and the maximum V reached about 1.5 m/s. The landslide motion stopped with decreasing seismic ground motion because μ = 0.59 > 0.25. Result for the time period 35 to 42 s in Figure 4b indicates that there were three pulses of accelerating/decelerating movements separated by two stop periods of about 0.5 s durations between 37 and 39.3 s. Whereas for wet frictional properties, μ decreases below a critical value of 0.25 at 39.3 s to trigger the landslide (Figure 4d). There were four small pulses of movements with S > 0 between 34 and 37 s, followed by three big pulses of accelerating/decelerating movements separated by stop periods of 0.2–0.5 s between 37 and 40 s. After the runaway of the landslide, μ dropped to μss (0.07) in a few seconds, and δ and V reached 1237 m and 65 m/s, respectively, at t = 76 s (time for the seismic record possibly for the collision of the landslide mass against the river bank; see Supplementary Figure FS01b). These δ and V values agree with the values estimated for the Tsaoling landslide (1000–1650 m and 57–66 m/s; Supplementary Information S1), and our analysis with wet gouge properties reproduced the landslide remarkably well. 6 Discussion and Conclusions We have conducted high-velocity friction experiments on shale gouge from Tsaoling landslide area at constant slip rates (Figure 2) and with oscillatory slip histories (Figure 3) and performed Newmark analysis of the Tsaoling landslide incorporating measured frictional properties (Figure 4). We consider implications of the results for future studies of catastrophic landslides (rockslides) triggered by earthquakes. We idealized early landslide motion as linear-accelerating and linear-decelerating motion with a frequency f of 0.3–1.2 Hz and a maximum slip rate Vmax of 0.33–1.3 m/s. Analysis of landslide motion revealed a few pulses of movements roughly with one pulse per second and Vmax of 1.5 m/s for dry gouge properties and 2.0 m/s for wet gouge properties (Figures 4b and 4d). The pulse-like motion is fairly similar to those of our oscillatory experiments in terms of f, Vmax, and linear accelerating/decelerating motion. A notable difference is that there were temporary stops for 0.2–0.5 s between the pulses of movements in the calculated landslide motion. Yao et al. [2013b] conducted slide-hold-slide tests on clayey gouge from the Longmenshan fault system after sliding at 1.4 m/s and with a hold time as short as 0.3 s. Their result indicates that the friction coefficient may increase by 0.15–0.25 for a hold of 0.5 s. This much of strength recovery takes place during decelerating slip with 0.3 Hz (blue curves in Figures 3e and 3f), and hence friction during accelerating/decelerating motion with a short stop is probably similar to the result from our 0.3 Hz oscillation experiment. We thus consider that our oscillating experiments were a reasonable proxy of early landslide motion although future experiments should cover oscillatory motion with temporary stops. Recently, Dong et al. [2013] conducted Newmark analysis of the Tsaoling landslide using equation 1, but they used μss and Dc as functions of slip rate, similar to our equations 2 and 3. That form of weakening law enforces to put μ back to a very high value at each deceleration, and it disagrees with our experimental results in Figure 3. We used a simple slip-weakening law in our modeling as a reasonable approximation of behaviors in Figure 3. However, an important future task is to establish a constitutive law that describes frictional properties for arbitrary slip histories for precise analyses of landslide motion. Such a law is not proposed in the field of fault mechanic either [e.g., Di Toro et al. [2011]]. Wet frictional properties lead to runaway of the landslide at 39.3 s after the occurrence of the Chi-Chi earthquake (Figure 4d), in about 19 and 12 s after the arrivals of P and S waves, respectively. A survivor's witness (10 s delay for the onset of landslide since he felt the earthquake) is in reasonable agreement with this result although he did not monitor the time interval under such chaotic situations. Our analysis clearly indicates that the landslide movement started at about 34 s since the generation of the Chi-Chi earthquake and that a few large acceleration pulses after 37 s during strongest ground motion was critical in triggering the landslide (Figure 4d). In their very recent paper, Chen et al. [2014] report high-frequency seismic signals recorded at a nearby seismic station at a time interval of 32–40 s and proposed that those are due to the initiation processes of the Tsaoling landslide. The oscillatory movement of the landslide block prior to its runaway in our analysis took place exactly in this period (Figure 4d). Chen et al. [2014] also report results from their Newmark analysis of the Tsaoling landslide with good overall agreement with the landslide movement, but they assumed frictional properties that do not agree with experimental results such as Figures 2 and 3 (see Supplementary Information S4). Analysis with wet gouge properties leads to the initiation of Tsaoling landslide, but that with dry gouge properties did not. The same conclusion was obtained in the analysis using a software by Jibson and Jibson [2003] neglecting the normal acceleration an although the landslide was slightly more stable without normal acceleration effect (cf. Figure 4 and Supplementary Figure FS06). Small differences in dry and wet frictional properties seem to be critical for triggering or not triggering the Tsaoling landslides. Clearly, smaller μp and Dc enhance triggering of landslides, and smaller μss increases the speed of landslide. It would be of interest to apply combined laboratory and Newmark analysis to many landslides (rockslides) triggered by earthquakes in the world. Acknowledgments We thank W.-L. Yu, Y. Miyamoto, K. Yano, and K. Oohashi for their assistance in the field and experiments. Present work was supported partially by a Grant-in-Aid for JSPS Fellows (201007605) to the first author (TT). The Editor thanks two anonymous reviewers for assistance in evaluating this manuscript. Supporting Information Filename Description revised_Supplementary-information_Togo et al_GRL_Tsaoling landslide.docxWord 2007 document , 4.6 MB Supplementary Information Fig.S1.pdfPDF document, 1,011.8 KB Figure S1 Fig.S2.pdfPDF document, 295.4 KB Figure S2 Fig.S3.pdfPDF document, 21 MB Figure S3 Fig.S4.pdfPDF document, 663.7 KB Figure S4 Fig.S5.pdfPDF document, 297.9 KB Figure S5 Fig.S6.pdfPDF document, 156.3 KB Figure S6 Please note: The publisher is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article. References Chen, T. C., M. L. Lin, and J. J. Hung (2003), Pseudostatic analysis of Tsao-Ling rockslide caused by Chi-Chi earthquake, Eng. Geol., 71, 31– 47. CrossrefWeb of Science®Google Scholar Chen, R. F., K. J. Chang, J. Angelier, Y. C. Chan, B. Deffontaines, C. T. Lee, and M. L. Lin (2006), Topographical changes revealed by high-resolution airborne LiDAR data: The 1999 Tsaoling landslide induced by the Chi-Chi earthquake, Eng. Geol., 88, 160– 172. CrossrefWeb of Science®Google Scholar Chen, T. C., M. L. Ling, and K. L. Wang (2014), Landslide seismic signal recognition and mobility for an earthquake-induced rockslide in Tsaoling, Taiwan, Eng. Geol., 171, 31– 44. CrossrefWeb of Science®Google Scholar Chigira, M., W. N. Wang, T. Furuya, and T. Kamai (2003), Geological causes and geomorphological precursors of the Tsaoling landslide triggered by the 1999 Chi-Chi earthquake, Taiwan, Eng. Geol., 68, 259– 273. CrossrefWeb of Science®Google Scholar Di Toro, G., R. Han, T. Hirose, N. De Paola, S. Nielsen, K. Mizoguchi, F. Ferri, M. Cocco, and T. Shimamoto (2011), Fault lubrication during earthquakes, Nature, 471, 494– 498, doi:10.1038/nature09838. CrossrefCASADSPubMedWeb of Science®Google Scholar Dong, J. J., C. M. Yang, W. L. Yu, C. T. Lee, Y. Miyamoto, and T. Shimamoto (2013), Velocity-displacement dependent friction coefficient and the kinematics of giant landslide, in Earthquake-Induced Landslides, edited by K. Ugai, H. Yagi, and A. Wakai, pp. 395– 401, Springer, Berlin, Heidelberg. Google Scholar Ferri, F., G. Di Toro, T. Hirose, R. Han, H. Noda, T. Shimamoto, M. Quaresimin, and N. de Rossi (2011), Low- to high-velocity frictional properties of the clay-rich gouges from the slipping zone of the 1963 Vaiont slide, northern Italy, J. Geophys. Res., 116, B09208, doi:10.1029/2011JB008338. Wiley Online LibraryWeb of Science®Google Scholar Hirose, T., K. Mizoguchi, and T. Shimamoto (2012), Wear processes in rocks at slow to high slip rates, J. Struct. Geol., 38, 102– 116. CrossrefADSWeb of Science®Google Scholar Huang, C. C., Y. H. Lee, H. P. Liu, D. K. Keefer, and R. W. Jibson (2001), Influence of surface-normal ground acceleration on the initiation of the Jih-Feng-Erh-Shan landslide during the 1999 Chi-Chi, Taiwan, earthquake, Bull. Seismol. Soc. Am., 91, 953– 958. CrossrefADSWeb of Science®Google Scholar Jibson, R. W. (1993), Predicting earthquake-induced landslide displacements using Newmark's sliding block analysis, Transp. Res. Board Transp. Res. Rec., 1411, 9– 17. Google Scholar Jibson, R. W., and M. W. Jibson (2003), JAVA programs for using Newark's method and simplified decoupled analysis to model slope performance during earthquakes, US Geological Survey, Open-File Report 03-005. Google Scholar Kuo, C. Y., Y. C. Tai, F. Bouchut, A. Mangeney, M. Pelanti, R. F. Chen, and K. J. Chang (2009), Simulation of Tsaoling landslide, Taiwan, based on Saint Venant equations over general topography, Eng. Geol., 104, 181– 189. CrossrefWeb of Science®Google Scholar Miyamoto, Y., T. Shimamoto, T. Togo, and J. J. Dong (2009), Dynamic weakening of shale and bedding-parallel fault gouge as a possible mechanism for Tsaoling landslide induced by 1999 Chi-Chi earthquake, in Proceedings of the Next Generation of Research on Earthquake-Induced Landslides: An International Conference in Commemoration of 10th Anniversary of the Chi-Chi Earthquake, edited by C. T. Lee, pp. 398– 401, National Central University, Jungli, Taiwan. Google Scholar Mizoguchi, K., T. Hirose, T. Shimamoto, and E. Fukuyama (2007), Reconstruction of seismic faulting by high-velocity friction experiments: an example of the 1995 Kobe earthquake, Geophys. Res. Lett., 34, L01308, doi:10.1029/2006GL027931. Wiley Online LibraryADSWeb of Science®Google Scholar Newmark, N. M. (1965), Effects of earthquakes on dams and embankments, Geotechnique, 15, 139– 160. CrossrefWeb of Science®Google Scholar Sone, T., and T. Shimamoto (2009), Frictional resistance of faults during accelerating and decelerating slip, Nat. Geosci., 2, 705– 708. CrossrefCASADSWeb of Science®Google Scholar Tang, C. L., J. C. Hu, M. L. Lin, J. Angelier, C. Y. Lu, Y. C. Chan, and H. T. Chu (2009), The Tsaoling landslide triggered by the Chi-Chi earthquake, Taiwan: Insights from a discrete element simulation, Eng. Geol., 106, 1– 19. CrossrefWeb of Science®Google Scholar Togo, T., and T. Shimamoto (2012), Energy partition for grain crushing in quartz gouge during subseismic to seismic fault motion: An experimental study, J. Struct. Geol., 38, 139– 155. CrossrefADSWeb of Science®Google Scholar Togo, T., T. Shimamoto, S. Ma, and T. Hirose (2011), High-velocity frictional behavior of Longmenshan fault gouge from Hongkou outcrop and its implications for dynamic weakening of fault during the 2008 Wenchuan earthquake, Earthquake Sci., 24, 267– 281. CrossrefADSGoogle Scholar Xu, Q., X. J. Pei, and R. Q. Huang (2009), Large-scale Landslides Induced by the Wenchuan Earthquake (in Chinese), Science Press, Beijing. Google Scholar Yao, L., S. Ma, T. Shimamoto, and T. Togo (2013a), Structures and high-velocity frictional properties of the Pingxi fault zone in the Longmenshan fault system, Sichuan, China, activated during the 2008 Wenchuan earthquake, Tectonophysics, 599, 135– 156. CrossrefADSWeb of Science®Google Scholar Yao, L., T. Shimamoto, S. Ma, R. Han, and K. Mizoguchi (2013b), Rapid postseismic strength recovery of Pingxi fault gouge from the Longmenshan fault system: experiments and implications for the mechanisms of high-velocity weakening of faults, J. Geophys. Res. Solid Earth, 118, 1– 17, doi:10.1002/jgrb.50308. Wiley Online LibraryADSWeb of Science®Google Scholar Citing Literature Volume41, Issue628 March 2014Pages 1907-1915 FiguresReferencesRelatedInformation