The amplification of ground motions due to complex topographies caused significant damage to infrastructure and triggered catastrophic landslides in past earthquakes. However, there is still a lack of reliable estimation of the topographic effect due to the sparseness of recordings on various topographies, and its influence on coseismic landslides remains yet uncertain. In this study, physics-based simulations of fault rupture and wave propagation are conducted to quantitatively evaluate these effects and develop prediction models for topographic amplification of ground motions. Physics-based and empirically-based frameworks are developed to predict regional-scale coseismic landslides. A well-documented earthquake case history is studied to validate these developed models.

This study...[ Read more ]

The amplification of ground motions due to complex topographies caused significant damage to infrastructure and triggered catastrophic landslides in past earthquakes. However, there is still a lack of reliable estimation of the topographic effect due to the sparseness of recordings on various topographies, and its influence on coseismic landslides remains yet uncertain. In this study, physics-based simulations of fault rupture and wave propagation are conducted to quantitatively evaluate these effects and develop prediction models for topographic amplification of ground motions. Physics-based and empirically-based frameworks are developed to predict regional-scale coseismic landslides. A well-documented earthquake case history is studied to validate these developed models.

This study begins with 3D spectral element simulations of wave propagation within Tuen Mun region (13 km × 10 km) in Hong Kong by using 14 selected ground motions based on the uniform hazard spectrum of Hong Kong. The physics-based simulations reveal that topographic amplification is frequency-dependent. The topographic amplifications of spectral accelerations are correlated with terrain features at different length scale associated with the period. A dimensionless parametric model is developed for predicting the topographic amplification factors (TAFs) by using relative heights of the terrain scaled by relevant wavelengths.

Another case study, the 2016 Mw 7.0 Kumamoto earthquake, is conducted to investigate the regional wave propagation, the near-fault wave characteristics, and the local topographic amplification effect, with an accuracy up to 5 Hz in a computational domain of 51 km × 43 km (horizontal) × 25 km (vertical). An equivalent linear procedure is implemented into the spectral element model (SEM) to account for strong soil nonlinearity near the earthquake source, and a finite fault system is employed to model the fault rupture process. The simulated waveforms are in good agreement with seismic recordings from KiK-net and K-NET. The near-fault effects such as the rupture directivity and the hanging wall effect are demonstrated. Prediction equations for TAFs of PGA and PGV in this region are developed, which is compatible with the parametric model developed for the Tuen Mun region.

Finally, a regional scale coseismic landslide framework is proposed that combines the physics-based wave propagation simulation at a regional scale, with a Newmark-type flexible sliding mass analysis at a site scale. The proposed framework is validated through the case study of the 2016 Mw 7.0 Kumamoto earthquake, which has a well-documented coseismic landslide inventory. The near-fault effect, topographic effect, and the soil nonlinearity on the distribution of coseismic landslides around the Aso caldera are studied. Overall, the physics-based simulation captures 49% of the observed landslides. To this end, the Kumamoto earthquake case history is further used to evaluate the effectiveness of empirically-based prediction of coseismic landslide. It is demonstrated that incorporation of the developed TAF prediction model can notably improve the empirically-based landslide prediction, which indicates that the developed parametric models can be well applied to engineering practice.