simulation of a magnitude 7.1 earthquake on January
24, 2016 in the Cook Inlet region of Alaska. The 3D
simulation demonstrates the complexity of the seismic
wavefield that arises from realistic models of Earth
structure. We see a striking effect due to the slow
wave speeds within Cook Inlet sedimentary basin. The
simulation was performed on the high-performance
computing cluster at the University of Alaska
Fairbanks, Research Computing Systems. The 3D wave
propagation code is called SPECFEM3D.
See Alaska Earthquake Center for other scientific content on this earthquake.
See also this amazing story about dog mushing at 1:30am on a frozen river during the earthquake.
Feel free to use or share the movies. If you do, please cite one of the following:
Tape, C., 2016, Magnitude 7.1 Alaska earthquake on January 24, 2016 (computer simulation), youtube movie at https://youtu.be/E4RYnpIOoPw
Tape, C., 2016, Magnitude 7.1 Alaska earthquake on January 24, 2016 (computer simulation, zoom), youtube movie at https://youtu.be/KdiETNfyaUo
A fancier animation, combining the full-scale and zoomed-in version, can be seen here (credit Hannah Foss):
notes from the youtube movie here
This is a 1200 km by 600 km oblique view of southern Alaska; some cities are labeled for reference: Kodiak (K), Homer (H), Kenai (K), Seward (S), Anchorage (A), Palmer (P), Cantwell (C), Nenana (N), and Fairbanks (F). This computer simulation shows a surface view of three-dimensional earthquake wave propagation of the magnitude 7.1 Iniskin earthquake on January 24, 2016; the epicenter is represented by the blue ball. The earthquake originated at about 120 km depth, within the subducting Pacific plate. The earthquake was felt over much of Alaska, from Anchorage to Fairbanks. The simulation demonstrates the complexity of the seismic wavefield that arises from realistic models of Earth structure. We see a striking effect due to the slow wave speeds within Cook Inlet sedimentary basin. The simulation was performed on the high-performance computing cluster at the University of Alaska Fairbanks, Research Computing Systems. The 3D wave propagation code is called SPECFEM3D. Credit: Carl Tape
A. MESH. The finite element mesh was generated with the GEOCUBIT software (Casarotti), an extension of CUBIT software to geological structures. The unstructured mesh is designed to optimally handle slow wave speeds near the surface and faster wave speeds in the upper mantle. The mesh contains 4,696,704 hexahedral elements that are largest at the bottom of the mesh, at 400 km below sea level, and smallest at the topographic surface. There are 312 million gridpoints in the mesh; the gridpoint spacing at the surface is about 200 m. The topographic detail that is visible in the movie is the actual top surface of the mesh. Denali is within the simulation region. (You can see places where the topography influences the wavefield.)
Credits: Ulrika Cahayani Miller and Emanuele Casarotti
B. STRUCTURE MODEL. The Earth structure model is that of Eberhart-Phillips et al. (2006), with an embedded model of Cook Inlet basin (Shellenbaum et al. 2010; http://www.dggs.alaska.gov/pubs/id/21961). The effects of Cook Inlet basin (west of Anchorage) on the wavefield are prominent. The ocean layer is ignored but is not expected to influence the wavefield at these periods. Active faults of Alaska, including the Denali fault (M7.9 on 11/2/2003), are plotted in white (Koehler et al., 2012, DGGS).
The model does not incorporate the shallowest structural models, such as the USGS Vs30 maps, which provide the shear-wave velocities in the uppermost 30 m (http://earthquake.usgs.gov/hazards/apps/vs30/predefined.php). The structure of Cook Inlet basin is modeled as the generic basin model of Brocher et al. (2008). As far as I know, there is no publicly available model of the 1D or 3D seismic structure of Cook Inlet basin. (If there is, please let me know!)
C. SOURCE MODEL. The earthquake is represented by a simple point-source model known as a moment tensor. The source-time function is approximated as a simple Gaussian function with a half duration of 2 seconds. At present (1/26/2016), seismologists are working to estimate a more complex and accurate source model for this earthquake; we anticipate using such a source model for future simulations.
Credits: Vipul Silwal and the Alaska Earthquake Center
D. COMPUTATION. The simulation used a time step of 0.006 s and iterated 44,000 time steps. The calculation took 10.0 hours on 240 cores (2400 CPU-hours) on the high-performance computing cluster at the University of Alaska Fairbanks, Research Computing Systems.
E. WHAT YOU SEE. The scalar field that is plotted is the simulated vertical component of ground velocity. The simulation keeps track of all three components of ground motion. (Alternatively one could plot ground displacement or acceleration.) The two pulses at the beginning correspond to the P wave and the S wave, which is followed by surface waves.