Summary of November 2010 meeting to evaluate turbidite data for constraining the recurrence parameters of great Cascadia earthquakes for the update of national seismic hazard maps
This report summarizes a meeting of geologists, marine sedimentologists, geophysicists, and seismologists that was held on November 18–19, 2010 at Oregon State University in Corvallis, Oregon. The overall goal of the meeting was to evaluate observations of turbidite deposits to provide constraints on the recurrence time and rupture extent of great Cascadia subduction zone (CSZ) earthquakes for the next update of the U.S. national seismic hazard maps (NSHM). The meeting was convened at Oregon State University because this is the major center for collecting and evaluating turbidite evidence of great Cascadia earthquakes by Chris Goldfinger and his colleagues. We especially wanted the participants to see some of the numerous deep sea cores this group has collected that contain the turbidite deposits. Great earthquakes on the CSZ pose a major tsunami, ground-shaking, and ground-failure hazard to the Pacific Northwest. Figure 1 shows a map of the Pacific Northwest with a model for the rupture zone of a moment magnitude M w 9.0 earthquake on the CSZ and the ground shaking intensity (in ShakeMap format) expected from such an earthquake, based on empirical ground-motion prediction equations. The damaging effects of such an earthquake would occur over a wide swath of the Pacific Northwest and an accompanying tsunami would likely cause devastation along the Pacifc Northwest coast and possibly cause damage and loss of life in other areas of the Pacific. A magnitude 8 earthquake on the CSZ would cause damaging ground shaking and ground failure over a substantial area and could also generate a destructive tsunami. The recent tragic occurrence of the 2011 M w 9.0 Tohoku-Oki, Japan, earthquake highlights the importance of having accurate estimates of the recurrence times and magnitudes of great earthquakes on subduction zones. For the U.S. national seismic hazard maps, estimating the hazard from the Cascadia subduction zone has been based on coastal paleoseismic evidence of great earthquakes over the past 5,000 years. The instrumental catalog of earthquakes is of little use for constraining the hazard of the CSZ, because there are virtually no recorded earthquakes on most of the plate interface of the CSZ. There are no historical accounts in the past 150 years of large earthquakes on most of the CSZ. Until about 20 years ago, some interpreted this lack of recent and historical earthquakes as an indicator that the subduction zone was slipping aseismically and could not produce a great earthquake. The work of Brian Atwater and others, in the late 1980s and the 1990s (Atwater, 1987, 1992; Atwater and others, 1995; Nelson and others, 1996; Clague, 1997; Atwater and Hemphill-Haley, 1997; Atwater and others, 2004) demonstrated that submerged forests, buried soils, tsunami deposits, and liquefaction along and near the coast were compelling evidence of repeated great Cascadia earthquakes over at least the past 5,000 years. Atwater and Hemphill-Haley (1997) concluded from paleoseismic evidence at Willapa Bay, Washington, that great earthquakes ruptured the CSZ with an average recurrence time of about 500 years. The date of the last great CSZ earthquake, January 26, 1700, was established from historical records of the so-called orphan tsunami in Japan that is inferred to have been produced by this earthquake (Satake and others, 1996, 2003; Atwater and others, 2005) and is consistent with tree-ring data from drowned forests in Washington and Oregon. From modeling the observations of the tsunami, Satake and others (2003) estimated a moment magnitude of about 9.0 for this earthquake. Many other paleoseismic sites have been investigated along the Pacific Northwest coast from Vancouver Island to northern California and show evidence of great CSZ earthquakes. Nelson and others (2006) summarized the dates found from these studies and proposed correlations between sites indicating the extent of rupture for individual events. Dating of inferred tsunami deposits in Bradley Lake, Oregon by Kelsey and others (2005), as well as tsunami and subsidence evidence from Six Rivers, Oregon (Kelsey and others, 2002) and Coquille River (Witter and others, 2003), indicates that there were probably M w 8 ruptures in the southern portion of the CSZ in addition to the M w 9 events that rupture the whole length of the CSZ (Nelson and others, 2006). A parallel development over the past 20 years or more is the use of deep-sea turbidite deposits for identifying and dating great Cascadia earthquakes over the past 10,000 years (Adams, 1990; Goldfinger and others, 2003, 2008, in press; Goldfinger, 2011). Turbidites are sediment deposits in the deep ocean from turbidity currents, which are energetic flows of sediment and water along the continental shelf and slope. Adams (1990), using the counts of turbidites in deep-sea cores off the coast of Oregon and Washington collected and analyzed by Griggs (1969) and Griggs and others (1969), proposed that these turbidites were caused by the shaking of great Cascadia earthquakes. Part of his reasoning was that the number (13) of turbidite deposits that occurred since deposition of the Mazama Ash 7,000 years ago gave a recurrence time of about 500 years, consistent with that derived from the coastal submergence data. Adams (1990) also proposed the “confluence test” which evaluates the number of turbidites for submarine channels that form a confluence. He reported that the number of turbidites in the single downstream channel equaled the number in each of the tributary channels. He reasoned that this indicated that the turbidites in each tributary were simultaneously triggered and were, therefore, caused by a common forcing agent. He concluded that shaking from extended ruptures of great Cascadia earthquakes was the most likely cause of these turbidites. Based on the paleoseismic evidence of past great earthquakes, the hazard from the Cascadia subduction zone was included in the 1996 U.S. NSHM (Frankel and others, 1996), which were the basis for seismic provisions in the 2000 International Building Code. These hazard maps used the paleoseismic studies to constrain the recurrence rate of great CSZ earthquakes. Goldfinger and his colleagues have since collected many more deep ocean cores and done extensive analysis on the turbidite deposits that they identified in the cores (Goldfinger and others, 2003, 2008, in press; Goldfinger, 2011). Using their dating of the sediments and correlation of features in the logs of density and magnetic susceptibility between cores, they developed a detailed chronology of great earthquakes along the CSZ for the past 10,000 years (Goldfinger and others, in press). These correlations consist of attempting to match the peaks and valleys in logs of density and magnetic susceptibility between cores separated, in some cases, by hundreds of kilometers. Based on this work, Goldfinger and others (2003, 2008, in press) proposed that the turbidite evidence indicated the occurrence of great earthquakes (M w 8) that only ruptured the southern portion of the CSZ, as well as earthquakes with about M w 9 that ruptured the entire length of the CSZ. For the southernmost portion of the CSZ, Goldfinger and others (in press) proposed a recurrence time of M w 8 or larger earthquakes of about 230 years. This proposed recurrence time was shorter than the 500 year time that was incorporated in one scenario in the NSHM’s. It is important to note that the hazard maps of 1996 and later also included a scenario or set of scenarios with a shorter recurrence time for M w 8 earthquakes, using rupture zones that are distributed along the length of the CSZ (Frankel and others, 1996; Petersen and others, 2008). Originally, this scenario was meant to correspond to the idea that some of the 500-year averaged ruptures seen in the paleoseismic evidence could have been a series of M w 8 earthquakes that occurred over a short period of time (a few decades), rather than M w 9 earthquakes. Figure 2 shows the logic tree for the CSZ used in the 2008 NSHM’s (Petersen and others, 2008). This logic tree includes whole CSZ rupture earthquakes (M w 8.8–9.2) and partial CSZ rupture earthquakes (M w 8.0–8.7). In this latest version of the NSHM’s, the effective recurrence time of earthquakes on the CSZ with moment magnitudes greater than or equal to 8.0 over the various models is about 270 years (Petersen and others, 2008). This recurrence time applies to the entire CSZ, so that the hazard from great earthquakes was approximately equal along the whole zone, although the hazard estimates taper on the northern and southern ends of the CSZ, because of the way rupture zones of M w 8 earthquakes were distributed along the strike of the CSZ. The NSHM will be updated in 2013, as part of the standard update cycle that corresponds to the update cycle of the national model building codes that are based on the seismic hazard maps. A meeting was necessary to assemble a wide group of experts to hear Dr. Goldfinger explain his methodology for dating and correlating the turbidites and for developing the earthquake chronology. The overall goal of the workshop was to evaluate observations of turbidite deposits to provide constraints on the recurrence times and rupture extents of great Cascadia subduction zone earthquakes for the next update of the NSHM. Before the meeting, participants were supplied with the U.S. Geological Survey (USGS) Professional Paper of Goldfinger and others (in press), as well as material from Brian Atwater and Alan Nelson. The agenda of the meeting was developed by Art Frankel, with assistance from Chris Goldfinger, Brian Atwater, Alan Nelson, Mark Petersen, and Craig Weaver. The meeting was hosted by Chris Goldfinger of Oregon State University. We stress that it is difficult to evaluate in a two-day meeting the large amount of work that Goldfinger and his colleagues have done over the past 15 years or more. This meeting is the first step in a process that develops the inputs to the update of the national maps. The conclusions of this workshop will be discussed and possibly modified at the regional Pacific Northwest workshop for the hazard maps to be held in early 2012. Vetting new research results using informed expert opinion is an integral part of updating the national maps and does not reflect on the veracity of these results.