The CyberShake Project: Better Seismic Hazard Information for Engineers

Downtown Los Angeles, from 110 Freeway Metro Platform / Photo Credit: Jason Ballmann, Southern California Earthquake Center

In earthquake country, safe buildings start with good science. The ground beneath our feet is complex, capable of earthquakes big and small – but to know exactly how the ground can move is crucial to building better structures. Every earthquake is unique, so experts need a solution that combines the best available knowledge. A multi-disciplinary group of SCEC scientists, engineers, and software developers are doing just that through CyberShake, a simulation project that incorporates cutting-edge science to model how earthquakes can affect specified sites and why. 

"As individual pieces of the science are improved, we bring them into CyberShake for more accurate results, showing the power of the SCEC collaboration in improving earthquake system science."

Traditionally, earthquake scientists and engineers depended on formulas known as Ground Motion Prediction Equations, or GMPEs, to estimate how much shaking could occur in an earthquake. GMPEs are developed from observed, historical earthquakes from around the world. While they can provide a good overall estimate of ground motions, GMPEs may not apply to a specific region's geological structure. Additionally, GMPEs only provide simplified measures of the ground shaking and not, for example, complete seismograms. These measures are the basis for building code design, but for critical facilities (hospitals, dams, power plants, for example), the use of seismograms in conjunction with structural simulations may be required to better illustrate site effects and directivity.

With CyberShake, SCEC scientists can incorporate a physics-based, system-level understanding of earthquakes through models of rupture mechanics, subsurface geology, and earthquake probabilities. They can produce results that include the complex regional earth structure and ultimately provide seismograms, in addition to the usual design metrics. These scientists draw from various parts of earthquake system science to paint a better picture of the seismic hazard.

Comparison of CyberShake Project data products in which the result on the left incorporated a better understanding of the crust at depth. One of CyberShake's inputs, the Uniform Community Velocity Model, directly leads to this clarification by expressing what types of rocks are where and at what speed seismic waves may travel through them. Thus, ground shaking may be higher or lower in certain areas than previously thought.

To get a better understanding of these elements, let’s look at a portion of the earthquake process. A fault is a fracture in the earth, where rocks on both sides are stressed and strained. These rocks cannot remain under such force forever, and when the rocks break and slip past each other at some point along the fault, with the rupture propagating along the fault, you have an earthquake. Why and how this process starts and stops depends on a variety of factors, all questions for the field of earthquake science known as rupture mechanics.

This point within the ground where this unzipping starts is referred to as the hypocenter (epicenter refers to the spot on Earth’s surface above the hypocenter), and the total area of the fault that ruptures is called the fault rupture area. The general, directional pattern in which a fault breaks can have a big impact on how way focus and propagate, known as directivity. You can see this directivity effect in many SCEC simulations, where shaking is more intense in certain areas, in direction of the rupture. This is due to the fault rupturing at a speed similar to that of the waves in the ground, creating a set of stacked waves in front of the rupture and resulting in larger ground shaking in the general direction of the rupture propagation.

Los Angeles is “like a bowl of jelly” because the city is settled on a basin full of soft sediments. When seismic waves go from stiff rocks to this kind of soft, loose material, they slow down and amplify, and may even reverberate within the basin, causing longer periods of shaking than normal. These effects are known as basin effects. These effects are considered in a simplified fashion within the GMPEs, but they can be directly taken into account with physics-based simulations, which consider the specific geometry of the rock/sediment interface in relation to the earthquake faults.

CyberShake simulations account for effects such as those and can provide improved seismic hazard estimates, relative to those from GMPEs:

  • The location of faults and the probability of them rupturing to produce earthquakes (UCERF)
  • The speed seismic waves travel through the ground, influenced by the kind of rock under our feet (UCVM)
  • Ground motion calculations that apply the physics of how faults break and how that influences the direction of seismic energy (AWP-ODC)

 

These pieces are combined together through complex software: as thousands of possible fault ruptures are simulated, waves are propagated through the earth model and ground motions are computed at specified sites of interest.

At the helm of the CyberShake effort is SCEC Director Tom Jordan, supported by SCEC’s research programming team working in the Community Modeling Environment (CME), led by Phil Maechling and Christine Goulet, along with Scott Callaghan, Kevin Milner, Fabio Silva, and John Yu. This group of full-time SCEC staff at USC supports the SCEC community with code and software development, simulation, validation, and high-performance computing. SCEC interns in the Undergraduate Studies in Earthquake Information Technology (UseIT) program also support CME efforts each summer.

Supercomputers are required to run and develop earthquake simulation projects like CyberShake, which require expensive customized scientific software development, and are very time-consuming to run. One high-performance computer may contain thousands of central processing units (CPUs) and/or graphics processing units (GPUs), unlike our desktop computers that may have only a few.

For example, Titan, a very large high-performance computer used by SCEC for earthquake simulations (including CyberShake) has almost 300,000 CPUs and 19,000 GPUs! This means Titan can solve trillions of equations in a matter of moments, faster than you scroll down to the bottom of this article.

"CyberShake brings together cutting-edge research from throughout the SCEC community to produce our best estimate of ground motion,” says SCEC Software Developer, Scott Callaghan. “As individual pieces of the science are improved, we bring them into CyberShake for more accurate results, showing the power of the SCEC collaboration in improving earthquake system science."

Each of these inputs draws on the extensive, multi-disciplinary expertise of the SCEC community developed over time. For example, we now have a unified model of the California crust and upper mantle and how seismic waves may propagate throughout it. The most recent update to California’s earthquake forecast is UCERF3, a joint effort between the U.S. Geological Survey, California Geological Survey (CGS), and SCEC, with partial support from the California Earthquake Authority, and was published in 2015. Advanced simulations for how seismic waves propagate are also incorporated, such as through AWP-ODC (Anelastic Wave Propagation - Olsen, Davis, and Cui; the names of 3 SCEC researchers).

While the CyberShake project’s inputs are exciting, they most importantly producing these outputs, or products:

  • Seismic hazard data products such as site-specific hazard curves and ground shaking maps
  • Simulation-based seismograms that provide intensity and duration information for all the scenario earthquakes estimated

The inputs and outputs of the CyberShake project, from left to right. Learn more about the CyberShake data flow.

The products generated by CyberShake provide details about how much the ground can shake at a given site, and how likely different amounts of shaking are. Researchers and design engineers can use this information to better understand the impact earthquakes have on buildings.

All buildings have a natural sway, or resonant frequency. Typically, tall buildings like skyscrapers have a lower resonant frequency than shorter buildings. If the seismic wave frequency matches a building’s resonant frequency, there may be serious damage to the building. It is like an opera singer producing a pitch so loudly and congruently with the natural resonant frequency of a nearby champagne flute that the glass shatters.

Of-course you don’t need to reach the resonant frequency to damage buildings. Engineers carefully consider potential ground shaking at the site, how tall the building can be, what materials to use, and how the building could sway and flex—all in order to accommodate earthquakes and other natural phenomena such as wind. There are many ways to design safer buildings, including the incorporation of high-tech but expensive strategies such as base isolators at the foundation level or dampers in the structure itself.

SCEC is working closely with earthquake and structural engineers to ensure that the seismic hazard information CyberShake produces is acceptible, defensible, and useful to building engineers in the Los Angeles area. The SCEC Committee for Utilization of Ground Motion Simulation (UGMS) is tasked with developing long-period response spectral acceleration maps for the Los Angeles region to be included in the National Earthquake Hazards Reduction Program (NEHRP) and American Society of Civil Engineers (ASCE) 7 Seismic Provisions and in Los Angeles City Building Code. These maps will be based on 3D numerical ground motion simulations produced by CyberShake, as well as the latest GMPEs developed by the Pacific Earthquake Engineering Research Center. Based on the promise of using simulations to produce better seismic hazard estimates, we can envision a future in which an engineer can identify a building plan tailored for the specific seismic hazards at the proposed site.

Further Resources: