Joint Nuclear Regulatory Commission/U.S. Geological Survey Workshop on Probability of Landslide-Generated Tsunamis
Nuclear accidents caused by the massive tsunami that struck Japan in 2011 were grim reminders that nuclear power plants along U.S. coastlines must be assessed for both earthquake and tsunami hazards. Most tsunamis are generated by earthquakes, submarine landslides, or a combination of both. Currently, the U.S. Nuclear Regulatory Commission (NRC) evaluates earthquake hazards using probabilistic methods, that is, determining the probability that a certain level of ground motion will be met or exceeded in a given period of time. In contrast, the NRC evaluates tsunami hazards by determining the “worst-case” scenario, which is independent of time. Ongoing research is exploring whether tsunami hazards can be evaluated probabilistically at the design probabilities of interest to the NRC (typically, an annual probability of 1/10,000 or 1/100,000).
The significant obstacle in calculating tsunami severity at very low probabilities is the need to include submarine landslides as well as earthquakes in the analysis. Information on how often submarine landslides occur is uncertain or altogether lacking in some regions, resulting in high uncertainty in the probabilistic analysis of tsunami hazards in comparison with the probabilistic analysis of earthquakes.
For the past several years, the U.S. Geological Survey (USGS) has done research to identify and date significant submarine landslides along the U.S. Atlantic margin and in the Gulf of Mexico-Caribbean region. For example, see “Submarine Landslides as Potential Triggers of Tsunamis That Could Strike the U.S. East Coast,” Sound Waves, August 2009, and “Gravity Coring Offshore Puerto Rico and the U.S. Virgin Islands to Investigate the Timing of Submarine Landslides and Large Earthquakes,” Sound Waves, July 2008.
Recent efforts have focused on how submarine-landslide probability might be determined using available geological and geophysical information and on identifying the challenges of incorporating this information into the determination of tsunami probabilities. To address these issues, a joint “NRC/USGS Workshop on Landslide Probability” was held August 18–19, 2011, at the USGS Woods Hole Coastal and Marine Science Center in Woods Hole, Massachusetts. Academic, industry, and government participants provided an overview of topics including geological characterization of submarine landslides, geotechnical techniques and measurements of slope stability, hydrodynamic modeling of landslide-generated tsunamis, and probabilistic methods for hazard assessment. In attendance from the NRC were Annie Kammerer from the Office of Nuclear Regulatory Research and Henry Jones from the Office of New Reactors. The goal of the workshop was to bring together experts who study the geometry, geotechnical properties, and recurrence of submarine slope failures, the potential of these failures to generate tsunamis, and the probability of recurrence of extreme events, in order to answer the following questions:
Although probability is mentioned in all of these questions, a probabilistic assessment is only as good as the underlying data and assumptions; therefore, it is critical that we address the state of knowledge and the kinds of new data that will be needed to improve our ability to estimate the probability of landslides capable of generating tsunamis.
Workshop participants summarized the state of knowledge regarding the probability of landslide-generated tsunamis, particularly along the U.S. Atlantic and Gulf of Mexico coasts. Presentations were grouped into four sessions: (1) landslide geometry and recurrence, (2) landslide mechanics, (3) modeling landslide-generated tsunamis, and (4) probability of landslides and landslide-generated tsunamis.
The landslide geometry and recurrence session included an overview talk by USGS emeritus scientist Homa Lee on how submarine landslides are identified (including potential pitfalls) and an overview of mapped landslides in the Gulf of Mexico and U.S. Atlantic margins by USGS scientist David Twichell. Details of various techniques used to date submarine landslides and sediment-transport processes associated with landslides were given by USGS scientists Jason Chaytor and Daniel Brothers, respectively.
The first three talks in the landslide mechanics session focused on geotechnical characterization of slope stability, including a technique for determining the likelihood of landslide occurrence given information on earthquake ground-shaking probabilities and bathymetric slope, presented by Uri ten Brink, USGS. Don DeGroot, University of Massachusetts Amherst, described state-of-the-art instrumentation to provide geotechnical characterization of the seafloor. An overview of geotechnical considerations for landslide occurrence was given by Jacques Locat, Université Laval, Québec, followed by a presentation of a newly developed numerical model for simulating landslide dynamics by David George, USGS.
The session on modeling landslide-generated tsunamis focused on state-of-the-art hydrodynamic modeling of long- and intermediate-length waves associated with this unique type of tsunami. Juan Horrillo, Texas A&M University at Galveston, compared conventional two-horizontal-dimension (2HD) models with three-dimensional (3D) models. Laboratory experiments used to validate hydrodynamic models were described by Stéphan Grilli, University of Rhode Island. Patrick Lynett, University of Southern California, used nonlinear Boussinesq-type models to show how probabilistic variations in down-slope landslide length can be evaluated. For the final presentation in this session, Frank González, University of Washington, showed how numerical hydrodynamic models are used in probabilistic tsunami hazard analysis (PTHA), providing a transition to the final session of the workshop.
The session on probability of landslides and landslide-generated tsunamis began with a presentation by Hong-Kie Thio, URS Corporation, about how probabilistic tsunami hazard analysis has been applied in the Pacific Basin and how disaggregation (percentage contribution of different sources) can provide important information about which source regions dominate the probabilistic calculations. A version of probabilistic tsunami hazard analysis specific to submarine landslides was presented by Christopher Baxter, University of Rhode Island, in which geotechnical calculations were combined with knowledge of how tsunami amplitude correlates with the size of the landslide to estimate tsunami hazards for the U.S. east coast. A method to determine landslide probabilities over a broad region by using seafloor databases, such as usSEABED (see related Sound Waves article), was presented by Eugene Morgan, Duke University. Eric Geist, USGS, synthesized the results of the workshop into a presentation about the approaches and challenges of incorporating submarine landslides into probabilistic tsunami hazard analysis. Annie Kammerer, NRC, then talked about the challenges facing the NRC in providing guidelines to power companies for tsunami-hazard mitigation.
The workshop presentations provided a summary of current thinking related to submarine landslides, their probability, and the complex hydrodynamics of landslide-generated tsunamis. Consensus from the presentations and discussion was that acquisition of additional geological and geophysical data, particularly age-dates from past landslide events, is necessary before probabilistic tsunami hazard analysis can be conducted at NRC-relevant design probabilities. However, probabilistic and statistical methods using available site-specific and regional geologic data can yield estimates of, for example, the probability associated with the “worst-case” scenario.
in this issue: