Impact of Mesh and DEM Resolutions in SEM Simulation of 3D Seismic Response

Abstract

This study shows that the resolution of a digital elevation model (DEM) and model mesh strongly influences 3D simulations of seismic response. Topographic heterogeneity scatters seismic waves and causes variation in seismic response (amplification and deamplification of seismic amplitudes) at the Earth's surface. DEM resolution influences the accuracy and detail with which the Earth's surface can be represented and hence affects seismic simulation studies. Apart from the spatial resolution of a DEM, the mesh resolution, adopted in the creation of a 3D spectral element meshing, also changes the detailedness of surface topography. Working with high-resolution data is in most cases not possible on a regional scale because of its costliness in terms of time, money, and computation. In this study, we evaluate how low the resolution of DEM and mesh can become before the results are significantly affected. We simulated models with different combinations of DEM and mesh resolutions. The peak ground displacement (PGD) obtained from these simulations was compared with the PGD of the model with the finest mesh and DEM resolution. Our results show that any mesh or DEM resolution of 540 m or coarser will give unrealistic results. These results are valid for similar terrains as studied here and might not be directly applicable to regions with significantly different topography.

The impact of topography on seismic amplification during the 2005 Kashmir Earthquake

“…Previous study has explored at which resolution you can best model the topography in relation to mesh resolution (Khan et al, 2017). They analyzed the impact of data resolution (mesh and DEM) on seismic response using SPECFEM3D.…”

The Influence of Surface Topography on the Weak Ground Shaking in Kathmandu Valley during the 2015 Gorkha Earthquake, Nepal

“…Using a high-resolution DEM is accurate but generates a high number of spectral elements that make SEM computationally very expensive. On the other hand, a DEM resolution coarser than 540 m in spectral element simulations produces less accurate results [26]. Based on the available processing capacity, we used an SRTM DEM 9 arc-second grid spacing for incorporating realistic surface topography into an SEM 270 m hexahedral mesh.…”

“…where V min is the minimum seismic wave speed, and T 0 is the shortest period. A polynomial degree N = 4 was used to sample the wavefield [26]. To ensure a stable time scheme for the spectral-element simulation, the selection of time-step ∆t during the simulation of the seismic wave through the mesh was restricted by [33] ∆t ≤ C min_Ω(∆x/V).…”

“…We designed the mesh to resolve seismic waves with the shortest period of 0.57 s (up to 1.75 Hz frequency). The average GLL distance in the horizontal direction at the surface was 60-61 m, which is small enough to be resolved with 9 arc seconds grid spacing DEM data [26]. The mesh properties and simulation parameters are shown in Table 1.…”

“…The maximum energy release was 22 km north of Kathmandu city at relatively shallow depth [9,12,13,25]. Because of the intense surface topography between Kathmandu valley and the location of maximum seismic energy release (Figure 1), a seismic wave is expected to show constructive or destructive interference and subsequent amplification or de-amplification [26], respectively, which could significantly change the intensity of ground motion inside the valley. Here, we analyze the role of surface topography on ground motion distribution through a full elastic finite element seismic wave simulation.…”