Extreme scale multi-physics simulations of the tsunamigenic 2004 sumatra megathrust earthquake

Abstract

We present a high-resolution simulation of the 2004 Sumatra-Andaman earthquake, including non-linear frictional failure on a megathrustsplay fault system. Our method exploits unstructured meshes capturing the complicated geometries in subduction zones that are crucial to understand large earthquakes and tsunami generation. These up-to-date largest and longest dynamic rupture simulations enable analysis of dynamic source effects on the seafloor displacements.To tackle the extreme size of this scenario an end-to-end optimization of the simulation code SeisSol was necessary. We implemented a new cache-aware wave propagation scheme and optimized the dynamic rupture kernels using code generation. We established a novel clustered local-time-stepping scheme for dynamic rupture. In total, we achieved a speed-up of 13.6 compared to the previous implementation. For the Sumatra scenario with 221 million elements this reduced the time-to-solution to 13.9 hours on 86,016 Haswell cores. Furthermore, we used asynchronous output to overlap I/O and compute time.

Asthenosphere Flow Modulated by Megathrust Earthquake Cycles

“…This approach has brought much insight into the rheology of plate boundaries at subduction zones (e.g., Hu et al, 2014Hu et al, , 2016Klein et al, 2016;Kyriakopoulos & Newman, 2016;Masterlark, 2003;Wang, 2007), transform boundaries (e.g., Allison & Dunham, 2018;Freed & Bürgmann, 2004;Masterlark & Wang, 2002;Takeuchi & Fialko, 2013), and collision zones (Castaldo et al, 2017;Cattin & Avouac, 2000;C. While finite-element modeling has proven versatile for forward and inverse modeling, notably using the adjoint method (Agata et al, 2017;Crawford et al, 2016), resolving the details of fault dynamics with this technique requires out-of-the-ordinary numerical resources and methods (Agata et al, 2014;Ichimura et al, 2016;Uphoff et al, 2017). While finite-element modeling has proven versatile for forward and inverse modeling, notably using the adjoint method (Agata et al, 2017;Crawford et al, 2016), resolving the details of fault dynamics with this technique requires out-of-the-ordinary numerical resources and methods (Agata et al, 2014;Ichimura et al, 2016;Uphoff et al, 2017).…”

“…Liu et al, 2016;Wang & Fialko, 2018). While finite-element modeling has proven versatile for forward and inverse modeling, notably using the adjoint method (Agata et al, 2017;Crawford et al, 2016), resolving the details of fault dynamics with this technique requires out-of-the-ordinary numerical resources and methods (Agata et al, 2014;Ichimura et al, 2016;Uphoff et al, 2017). Viscoelastic simulations have also exploited the correspondence principle, whereby heterogeneous effective elastic properties are mapped into the viscoelastic properties by virtue of the Fourier or Laplace transforms (Chanard et al, 2018;Fukahata & Matsu'ura, 2006;Pollitz, 1992Pollitz, , 1997Smith & Sandwell, 2004).…”

Landers 1992 "reloaded": Integrative dynamic earthquake rupture modeling

“…Harris et al, 2018]. End-to-end optimization Heinecke et al, 2014;Breuer et al, 2015Breuer et al, , 2016Rettenberger and Bader , 2015;Rettenberger et al, 2016] targeting high efficiency on high-performance computing infrastructure includes a ten-fold speedup by an efficient local time-stepping algorithm [Uphoff et al, 2017]. Viscoelastic rheologies are incorporated using an offline code-generator to compute matrix products in a computationally highly efficient way.…”

“…Our dynamic source model incorporates new degree of realism by integrating a comprehensive set of geological and geophysical information such as high-resolution topography, rotating tectonic stresses, 3D velocity structure, depth-dependent bulk cohesion, and a complex intersecting fault geometry. Unifying aforementioned complexities is enabled by using SeisSol (www.seissol.org, Dumbser and ; Pelties et al [2014]), a software package specifically suited for handling complex geometries and for the efficient use on modern high-performance computing infrastructure [e.g., Heinecke et al, 2014;Uphoff et al, 2017]. This work extends recent models presented in Wollherr et al, 2018] which included complex fault geometries and off-fault plasticity but were restricted to 1D velocity structure, constantly oriented tectonic background stress and neglecting viscoelastic attenuation of the seismic wave field.…”

ExaHyPE: An engine for parallel dynamically adaptive simulations of wave problems

“…Hexahedral mesh generation can easily consume weeks to months and is limited for complex geometries of boundary and interface conditions, while form-fitted unstructured tetrahedral meshes allow for automatised meshing and complex geometries [56,57], however, pose numerical challenges, e.g. in form of misshaped sliver elements [58].…”