NTHMP Landslide Tsunami Model Validation Workshop

Sponsor: National Tsunami Hazard Mitigation Program (NTHMP)

Steering Committee

1. Stephan Grilli, University of Rhode Island
2. James T. Kirby, University of Delaware
3. Juan Horrillo, Texas A&M University - Galveston
4. Dmitry Nicolsky, University of Alaska
5. Philip Liu, Cornell University

NTHMP is conducting a Landslide Tsunami Model Validation Workshop, to be help at Texas A&M University - Galveston, on January 9-11, 2017. Workshop participants will be conducting numerical studies of a portion or all of a set of proposed benchmarks. The set of candidate benchmarks will be posted on the workshop website (www.udel.edu/kirby/nthmp/landslide.html) around the beginning of October. Candidate benchmarks will be based on a subset of available laboratory date sets for solid slide experiments and deformible slide experiments, and will include both submarine and subaerial slides. A benchmark based on a historic field event will also be provided.

The expected outcome of the workshop are to develop: (i) a set of community accepted benchmark tests for validating numerical models used for landslide tsunami generation and propagation in NTHMP inundation mapping work; (ii) an extensive workshop documentation and a web-based repository, for benchmark data, model results, and workshop documentation, results and conclusions.

Workshop participants should go to NTHMP 2017 Landslide Workshop Page to register for the meeting.

Further details on venue and benchmarks will be posted here at the end of September.

Tsunamigenic Landslide Modeling Benchmark Development, Validation Workshop and Workshop Documentation
In its FY2009 Strategic Plan, the NTHMP required that all numerical tsunami inundation models be verified as accurate and consistent through a model benchmarking workshop/process. This was completed in FY2011, but only for seismic tsunami sources and in a limited manner for idealized solid underwater landslides. Recent work by various NTHMP states/areas, however, has shown that landslide tsunami hazard may be dominant along significant parts of the US coastline, as compared to hazards from other tsunamigenic sources.

1. Along the US east coast, a large volcanic subaerial landslide tsunami on La Palma (Canary Islands) is the dominant, albeit long return period, tsunami source (Abadie et al., 2012; Harris et al., 2012, Tehranirad et al., 2015). Many large underwater landslide scars, some a few 10Ka old, have been mapped by USGS (ten Brink, et al., 2008, 2014) along the continental shelf slope and Atlantic Margin; more recent USGS work (Chaytor et al., 2014) also shows large potential landslides in the Florida straight. Earlier work has shown that many of these landslides would have been strongly tsunamigenic (Grilli et al., 2009, 2014a). In 1929, in this broad geographic area and oceanic margin, a large landslide tsunami was actually triggered off of the Grand Banks by a M7.1 local earthquake (Fine et al., 2005).
2. In the Gulf of Mexico, the majority of the tsunami inundation mapping work that is being done as part of NTHMP is based on a few major underwater landslide sources (also mapped by USGS; ten Brink, et al., 2008) (Horrillo, et al., 2013).
3. In California, Oregon and Washington states, while both local and far-field seismic sources likely dominate tsunami hazard, many large historical landslides have been mapped which were likely to be strongly tsunamigenic. Notable examples in California include the Goleta slide off of Santa Barbara (Greene et al., 2006) and the Big Sur slide in the Monterey Canyon (Greene and Ward, 2003). Recent work on the Tohoku 2011 tsunami (e.g., Tappin et al., 2014) indicates that very large megathrust earthquakes (such as anticipated in the future for the Cascadia subduction zone), may also trigger very large landslides which could contribute significant additional wave activity in addition to the co-seismic tsunami. Hence, their study, siting and modeling should be done ahead of time, in preparation for such large seismic events.
4. In Alaska, numerous landslides, both underwater and subaerial, have been triggered by large earthquakes (e.g., the M9.2 1964 event) and artesian flows, and have caused tsunamis with significant local runup and inundation. In this region, the most notable such event is the Lituya Bay 1958 subaerial slide that triggered a tsunami in a narrow fjord, causing over 500 m runup on the other side of the fjord (Fritz et al., 2001, 2009; Weiss et al., 2009).
5. In Hawaii, larger landslides associated with volcano flank motion and collapse have occurred (e.g., Kalapana 1975; Day et al., 2005) causing significant runup. Such events will continue to occur in the future as a result of the continuous build up and weathering of volcanoes.
6. In Puerto Rico, a number of large historical landslide tsunami events have been mapped by USGS and modeled (e.g., in the Mona passage; Lopez-Venegas et al., 2008). In past years, there has been considerable model development and benchmarking activity for seismically induced tsunamis, in particular within the auspice of NTHMP (e.g., the Galveston benchmarking workshop for tsunami elevations in 2011, NTHMP, 2012; Horrillo et al. (2013) as well as the upcoming Portland workshop for tsunami velocities in 2015). For landslide tsunamis, however, both the model development and benchmarking efforts have been lagging. In 2003, the east coast NTHMP PIs were co-organizers of a NSF sponsored landslide tsunami workshop in Hawaii, and a similar follow-up workshop took place on Catalina island in 2006. Since then, to our knowledge, no similarly large and comprehensive benchmarking workshops have been organized. In 2011, following the NTHMP model benchmarking workshop, J. Horrillo organized a landslide tsunami workshop devoted to a review of the state-of-the-art in modeling. Later that year, the USGS Woods Hole group (U. ten Brink, J. Chaytor, and E. Geist) organized a similar workshop, during which the state-of-the-art in field work, geology, PTHA, and landslide tsunami modeling were reviewed.

A decade ago, investigators were satisfied with modeling solid block landslides (e.g., Grilli et al., 1999, 2002, 2005; Lynett and Liu, 2003; Watts et al., 2003, 2005; Liu et al., 2005) and benchmarking experiments were developed for those and used for tsunami model benchmarking (in particular as part of the Galveston workshop; Enet and Grilli, 2007). More recently developed models simulate deformable slides and solve both more complete sets of equations (dispersive, non-hydrostatic, Navier-Stokes) and consider subaerial or submarine slides as heavy fluids (e.g., Abadie et al., 2010, 2012; Horrillo et al., 2014), flows induced by sediment concentration (Ma et al., 2013), or granular flow layers (Ma et al., 2015). A number of recent laboratory experiments have modeled tsunamis generated by subaerial landslides made of gravel (Fritz et al., 2004; Heller and Hager, 2012; Mohammed and Fritz, 2012) or glass beads (Viroulet et al., 2014), but, to our knowledge, there are no experimental benchmark data for deforming underwater landslides that are initiated underwater, but some relevant experiments are in preparation to this effect.

Following discussions at a recent meeting, the NTHMP MMS subcommittee recommended that a landslide tsunami model benchmarking workshop be organized in the near future. In response to this recommendation, this proposal seeks funding from NTHMP to conduct such a workshop, on the model of the one organized in Galveston in 2011 and the upcoming workshop in Portland in 2015. To help with preparing the workshop scientific program, a small committee led by the two PIs will be formed in large part composed of NTHMP-MMS members. The committee will meet during upcoming MMS meetings and conduct conference calls to identify an agenda for the workshop and select a set of relevant landslide tsunami benchmarks (analytical, numerical, experimental, field). While these are not yet determined, recent experiments with granular/glass bead flows for subaerial landslide will be likely experimental benchmark candidates; the 1998 Papua New Guinea landslide tsunami (Tappin et al., 2008) could also serve as a field benchmark candidate. Some solid block underwater landslide experiments could also serve as benchmarks. A workshop webpage will be built, as a receptacle for the benchmark information and data, as well as other practical information regarding the workshop. The committee will help select a workshop date to coincide with an NTHMP meeting. To allow for enough preparation time and also for some ongoing experiments to be completed, the targeted date for the workshop is summer 2016, preceding or following the regular NTHMP/MMS summer meeting.

Once the workshop date is set, the benchmarking data will be posted a few months ahead of time, and, at the same time, potential participants will be invited to attend the workshop and given information on how to access the data. These participants will be in part NTHMP modelers and investigators (about 10), together with a group of selected experts and graduate students (about 15). Participants will be invited to simulate as many benchmarks as possible (with a minimum set to warrant financial support to attend the workshop), using their own model, and results will be compared during the workshop, and discrepancies with the benchmarking data discussed.

Although this may be more difficult to do than for seismic tsunamis, participants will be asked to attempt to reach a consensus at the end of the workshop, on both acceptable modeling approaches (and associated models) for various types of landslide tsunamis and acceptable levels of discrepancies with various types of benchmarking data. Finally, the scientific committee will prepare a workshop proceedings and a manuscript for publication in a peer-reviewed journal based on the workshop findings (as done for the Galveston workshop; Horrillo et al., 2014).

In order to provide an independent assessment of benchmark development and model evaluation, we will engage an outside reviewer to evaluate all stages of the process above. This step is provided for in the budget as an unspecified contract, with the external reviewer to be chosen early in the process in order to allow him or her to take part in every stage of the process. The bulk of benchmark development, organization of the workshop, and reporting of results will be handled by the proposing team.