Cascadia Great Earthquakes from the Turbidite Event History

Recent papers:
2003 Annali Di Geofisica Special Volume on Paleoseismology  PDF

2003 Annual Reviews  PDF

 The Holocene stratigraphy of submarine channels along the Cascadia margin has long been known to include a turbidite sequence.  L. D. Kulm and his students in the late 1960's and 1970's investigated the nature, distribution, and timing of these turbidite events in considerable detail using piston cores from some of the channels.  Although G. Griggs and L. Kulm thought that these events might represent earthquakes along the margin, the extremely low seismicity in Cascadia made such a hypothesis seem unlikely.  Following the discovery of the first buried marsh sequences on land,  Adams (1990) assessed the possibility that these cores contained a record of the Holocene great earthquake history of the Cascadia margin.  Adams examined core logs from cores in Cascadia, Astoria, and Rogue channels on the abyssal plain and their associated canyons and tributaries on the continental slope.

Mazama Ash

 The cores all contain a unique datable event, the ash layer from the eruption of Mount Mazama, at 6845 ± 50 radiocarbon yr BP (Bacon, 1983).  The ash was distributed to the channel system via the drainage basins of major rivers, similar to the distribution of Mt St. Helens ash following the 1980 eruption (Nelson et al., 1968).  Only channel cores contain the ash, indicating that airfall was not significant. Griggs and Kulm (1970) used the Mazama ash to calculate that the mean recurrence interval for the post-Mazama turbidites in Cascadia channel was 410-510 years.  Adams examined core logs for the Cascadia Basin cores, and determined that most of them had between 13 and 19 turbidites overlying the Mazama ash.  In particular, he noted that three cores all in the main Cascadia channel contain 13 turbidites, arguing that all 13 turbidites correlate along the channel.  Adams observed that Juan de Fuca canyon, and below the confluence of Willapa, Grays, and Quinault Canyons, cores contain 14-16 turbidites above the Mazama ash.  Below these two tributaries, cores in the main Cascadia channel contain 13 turbidites.  If these events had been independently triggered events with more than a few hours separation in time, the cores below the confluence should contain from 26-31 turbidites, not 13 as observed.  The importance of this simple observation is that it demonstrates synchronous triggering of turbidite events in tributaries the headwaters of which are separated by 50-150 km.  The extra events  in the upstream cores could be smaller interplate events or from other sources.

Synchroneity

Following the discovery of the first buried marsh sequences on land, Adams (1990) used existing cores to test the possibility that the Cascadia cores contained a record of Holocene great earthquakes of the Cascadia margin.  Fortunately, Oregon and Washington cores all contain a unique datable event, the ash layer from the eruption of Mount Mazama, at 7627 ± 150 cal yr BP (Zdanowicz et al., 1999).  The ash was distributed to the channel system via the drainage basins of major rivers, similar to the distribution of Mt St. Helens ash following the 1980 eruption (Nelson et al., 1968).  Only channel cores contain the ash, indicating that airfall offshore was not significant. 

Adams (1990) examined core logs for Cascadia Basin cores, and determined that nearly all of them had 13 turbidites overlying the Mazama ash. Following the discovery of the first buried marsh sequences on land Adams, (1990) assessed the possibility that turbidites in channels of Cascadia Basin contained a record of great earthquakes along the Cascadia margin. He found that cores along the length of Cascadia channel contain 13 turbidites and argued that these 13 turbidites correlate along the channel.  Adams observed that cores from Juan de Fuca Canyon, and below the confluence of Willapa, Grays, and Quinault Canyons, contain 14-16 turbidites above the Mazama ash.  The correlative turbidites in Cascadia channel lie downstream of the confluence of these channels.  If these events had been independently triggered events with more than a few hours separation in time, the channels below the confluence should contain the sum of the tributaries, from 26-31 turbidites, not 13 as observed (Figure 1).  The importance of this simple observation is that it demonstrates synchronous triggering of turbidite events in tributaries, the headwaters of which are separated by 50-150 km.  Similar inferences about regionally triggered synchronous turbidites in separate channels are reported in Pilkey (1988).  This elegant relative dating technique is used extensively in our Cascadia and SAF work.   

Using 54 new cores in Cascadia, we have confirmed and extended the event record temporally and spatially.  Thirteen post-Mazama and 18 Holocene events are found along~ 660 km of the margin in the Cascadia, Barclay, Willapa, Grays, and Rogue Canyon/Channel systems between latitudes 42N and 48N.  The most recent event took place in 1700 AD (Satake et al., 1996; Nelson et al., 1995), and an additional 12 turbidite events have occurred during the preceeding 7200 years, yielding a mean recurrence time of ~575 years.  

Triggering mechanisms: Are they Earthquakes?     

Are these events all triggered by earthquakes?  Common sense suggests that such a scenario is absurdly simplistic, yet our Cascadia work has led us to the unlikely conclusion that Adams (1990) was correct.  We now discuss the methods used to test the hypothesis and why it seems to work.  Adams (1990) suggested four plausible mechanisms for turbid flow triggering: 1) storm wave loading; 2) great earthquakes; 3) tsunamis; and 4) sediment loading. To these we add 5) crustal earthquakes, 6) slab earthquakes, 7)  hyperpycnal flow, and 8) gas hydrate destabilization. 

All of these mechanisms could trigger a turbid event, but how often do they actually occur, and can earthquake-triggered events be distinguished from other events?  Two basic techniques can be used to distinguish seismic from non seismic events:

1)  Sedimentological determination of individual event origin.

2)  Regional correlations that require synchronous triggering.  

Individual event determination can in some cases distinguish seismic turbidites from storm, tsunami, and other deposits using several methods.  Nakajima and Kanai (2000) and Shiki et al. (2000) report that seismo-turbidites can in some cases be distinguished sedimentologically.  They observe that historically known seismically derived turbidites in the Japan Sea and Lake Biwa are distinguished by wide areal extent, multiple coarse fraction pulses, variable provenance, and greater depositional volume than storm-generated events. These investigators traced known seismo-turbidites to multiple slump events in many parts of a canyon system, generating multiple pulses in an amalgamated turbidity current, some of which sampled different lithologies that are separable in the turbidite deposit.  These turbidites are complex, with reverse grading, cutouts, and multiple pulses.  Gorsline et al. (2000) make similar observations regarding areal extent and volume of seismoturbidites.  In general, these investigators observe that known storm triggered events are thinner, finer grained and have simple normally graded Bouma sequences, although complexity is also a function of proximity to the source, and some reports reach different conclusions (Mulder and Syvitski, 1996).  While there may yet be applicable global, regional or local criteria to make such distinctions, these are at present poorly developed and somewhat contradictory.

Thus far in Cascadia and the San Andreas, we have not attempted to distinguish between triggering mechanisms directly because the physiography, numerous tephra layers, and long historical records favorable to this method in Japan are not present on the US west coast.  Favorable factors that are present favor regional correlation and determination of synchronous triggering.  Determination of synchronous triggering can eliminate non-earthquake triggers with the possible exception of storm wave loading or multiple hyperpycnal flows for very large storms, and perhaps triggering by a tele-tsunami.  West Coast physiography favors filtering of non-seismic events from the record because a wide shelf separates river sources from canyon heads.  Hyperpycnal flow, or direct turbid injection from rivers, can produce turbid flows, and can even mimic earthquakes in that they may affect several rivers over a span of days.  We have found that while this certainly occurred during the Pleistocene when lowered sea-level resulted in direct river-canyon connections resulting in a major increase in turbidite frequency at the Holocene Pleistocene boundary.  Once sea level-rise in the Holocene isolated most west coast canyons from the rivers, sediment was spread across the shelf and not directly injected in the canyons (e.g. Sternberg, 1986).  Some exceptions are the Eel river record, which probably contains storm events, and the Viscaino channel along the northern San Andreas.  Both of these occur where the shelf is very narrow, and river injection is possible.  Deep canyon heads also prevent triggering by storm wave loading and distant tsunami, the last two non-earthquake sources.  For example, storm wave loading is an unlikely trigger in Cascadia, where, although deep water storm waves are large, the canyon heads where sediment accumulation occurs are at water depths of 150-400 m, too deep for disturbance by maximal storm waves of ~20 meters.   Tsunamis may also conceivably act as a regional trigger, however the tsunami from the 1964 Alaska Mw 9.0 event did not trigger a turbidite observed in any of the cores, although it did serious damage along the Pacific coast (Adams, 1990).  Crustal or slab earthquakes could also trigger slumps and turbid flows, though not regionally.  To test for this, we resampled the location of a 1986 box core in Mendocino channel, where the uppermost event is suspected to be the 1906 San Andreas event.  Since 1986, the Mw7.1 Petrolia earthquake occurred in 1992, with an epicentral distance of only a few km from the canyon head.  We found no turbidite in the 1999 box core, suggesting that triggering at that site may require earthquakes larger than Mw 7.2. Conversely, the Loma Prieta earthquake apparently did trigger a turbid flow event in Monterey Canyon at a greater epicentral distance (Garfield et al., 1994), though it is not known whether a discernable turbidite record exists from this event.  Japanese investigators have suggested a minimum magnitude of ~ Mw=7.2 for turbidite triggering, though we suspect that this minimum value is site and event specific.

Synchronous Triggering, Relative Dating, and Regional Correlations

Taking advantage of favorable physiography, we have used spatial and temporal patterns of event correlations that are unlikely be the result of triggers other than earthquakes.  We use multiple techniques to test for linkage between sites and thus test for synchroneity.  Typically, paleoseismologic investigations use radiocarbon constraints to establish these linkages, but often are unable to determine synchronous event chronology due to the inherent limits in dating techniques.  Relative dating techniques, if available, and if of sufficient resolution, are strongly preferred to test for synchroneity.  The “confluence test” of Adams (1990) is powerful in that it requires synchronous triggering within a few hours. Comparisons of the number of events between time markers is a somewhat less powerful technique that can be applied in some cases.  Recently we have begun to use direct physical property correlations, which are proving to be a powerful new method of testing for linkages (if present) between sites. 

Cascadia Results

            Using Adams’  “confluence test”, we concluded that the northern Cascadia margin contains 18 Holocene events, all of which pass this test of synchronous triggering.  The southern Oregon Rogue site, also contains a record of 18 Holocene and 13 post Mazama events, though Rogue channel has no confluence with the other systems.  It passes a weaker test by virtue of having an identical number of events to the northern margin (Goldfinger et al., 2003a,b). Subsequent to these results, we have found it possible to correlate the 18 Holocene events directly using physical properties of the turbidites themselves in the cores.  This method is discussed more fully below.  Using these correlations we believe it is possible to establish synchronous triggering for the northern and central parts of the margin between 42 and 48 degrees N for 18 events, with several “extra” event revealed at 2 southern sites.   In Cascadia, event synchroneity is established not only with radiocarbon ages, but with correlation techniques within a radiocarbon constrained framework.  Correlations have been able to link northern and southern sites, a connection that radiocarbon evidence alone cannot establish with either onshore or offshore records.  

Hyperpycnal flow

Hyperpycnal flow is the density driven underflow from storm discharge of rivers into marine or lacustrine systems, and has been proposed as a link to turbidity currents in a variety of settings. Documentation of hyperpycnal flows into lakes and shelf basins is abundant, however evidence of such flows entering canyons systems and moving into deep water is sparse. Most, if not all examples involve very short distances between the river mouth and canyon head, either during Pleistocene low-stand conditions, or in systems that have very narrow shelves during high stand conditions. Hyperpycnal flows extend further from river mouths with high discharge (Alexander and Mulder, 2002), but documentation is sparse. Wright et al. (2001) infer that hyperpycnal flow is strongly affected by ambient currents, and generally deliver sediment to the slope only upon relaxation of longshore currents. Most investigators cite Pleistocene examples when referring to flows reaching the abyssal plain or lower fan reaches (e.g. Mulder et al., 2003; Normark et al., 1998; Normark and Reid, 2002; Piper et al., 1999). Under low stand conditions, rivers and canyons are directly connected, and such flows are expected. These authors relate the deep water deposition of hyperpycnites to sea level control, or alternatively to climate shifts. An example of high stand hyperpycnal flow has been reported for the Var River, in which the canyon and river mouth are less than 1 km apart (Klaucke et al., 2000; Mulder et al., 1998a). Many large river systems deposit most of their load in river mouth bars, with lesser quantities making it past such bars in to a delta front slope (eg. Yellow River, Li et al., 1998). Whether hyperpycnal flows can reach the deep sea via canyon systems incised during the Pleistocene appears to be a function of shelf width, river peak storm discharge, Holocene aggradation of Pleistocene canyons, and the wave and current climate during peak storm discharge. However, the requirements for and evidence of hyperpycnal flows to the deep ocean under high-stand conditions (excepting very narrow shelves) remain unknown at best (Mulder et al., 2001).

Comparisons of hyperpycnal and earthquake generated turbidites

In some proximal settings such as large lakes, shelf basins, and fjords, records of both earthquakes and storm deposits are found. In one of the best comparisons, St.-Onge et al (2004) show that details of both seismic and hyperpycnal deposition in the Saguenay Fjord in eastern Canada are diagnostic, and argue that hyperpycnal deposits are distinguished by reverse grading at the base, followed by normal grading. The diagnostic reverse-then-normal grading for hyperpycnal deposits has been widely reported, and is the result of waxing, then waning flow associated with the storm. In the Saguenay Fjord, six events have normal grading, and are inferred to be earthquake generated. Four others have similar basal units, but are topped by a reverse graded unit, and then a normally graded unit, with no evidence of hemipelagic sediment between these units. These events are interpreted as an earthquake, followed by a hyperpycnite that resulted from the breaching of a landslide dam caused by the original earthquake. The dam breaching is a variant of the more common hyperpycnal scenario involving waxing and waning depletive flow (Kneller, 1995), but would likely produce a similar flow hydrograph (St-Onge et al., 2004).

In another well studied example, a major river flood in the1969 El Nino input ~25 million tons of sediment (20X the yearly Columbia sediment load) to the Santa Ana River in Southern California over a 24 hour period, in close proximity to nearby canyon heads. Sediment from this extreme flood did not continue down canyons as hyperpycnal flow, but deposited as a distinct yellow unit on the shelf and upper slope. Over the next 10 years it moved downslope as turbid layer transport caused by storm wave resuspension, similar to that observed by Puig et al. (2004) for the Eel system, and deposited as yellow layers between varves of the Santa Barbara Basin (Drake et al., 1972).

Other examples of reverse-then-normal grading for hyperpycnites have been reported and compared to normally graded failure deposits in the Var system (Mulder et al., 2001), Lake Biwa (Shiki et al, 2000a), and modeled by Mulder et al. (1998b), Felix (2002) and others. The dynamics of longitudinal and temporal variability and their effects have been discussed in detail by Kneller and McCaffrey, 2003, and Mulder et al., 2003.

Hyperpycnites are also commonly very organic rich as compared to seismic turbidites, having their sources in floods rather than in resuspension of older canyon wall material as in earthquake triggering (Mulder et al., 2001; Shiki et al., 2000; Nakajima et al, 2000). It has been suggested that this distinction may be used as a basis for distinguishing earthquake and storm deposits using OSL dating.

Cascadia Physiography

On the Cascadia margin, all of the canyon systems have Holocene shelf widths of 22-60 km, the exception is the Eel River, at which the shelf width is 12 km (Table 1). In the Pacific Northwest, large storms both increase discharge, and produce strong southerly winds that disperse sediment northward on the shelf via the Davidson current (Sternberg, 1986; Wheatcroft et al., 1997; Sommerfield and Nittrouer, 1999; Wolf et al., 1999) as well as hundreds of kilometers seaward as evidenced by satellite imagery (Wheatcroft et al., 1997). The Eel River system, with its narrow shelf is the best candidate in Cascadia for delivering hyperpycnal flows to the canyon head. However, Puig et al. (2004) saw no evidence of hyperpycnal flow into the Eel Canyon head in observations made during storms in the Canyon head at 120 m water depth. They did observe some cross-shelf sediment transport due to wave loading at this depth. Other observations made for depths of 280 m, and 900 m were also linked to storms, with wave loading the presumed mechanism, but no link to hyperpycnal discharge or to turbidity currents delivering material to the base of the slope and beyond was found (Puig et al., 2003, 2004). The width of the Cascadia shelf is somewhat exacerbated at the Astoria and Eel canyon systems where the canyon heads are somewhat to the south of the river mouths, inhibiting northward moving sediment flow associates with southerly storms (plumes or hyperpycnal) from entering the canyon. In all systems, the late Pleistocene transgression and subsequent aggradation have erased any topographic canyon or channel expression across the Cascadia shelf.

Cascadia turbidite characteristics

In our work, Cascadia turbidites show rare to no evidence of the characteristic reverse grading and high organic content observed for hyperpycnal deposits. We observe normally graded sequences with multiple amalgamated pulses, sharp erosional bases, and normal grading of individual pulses when present. The number and form of these multiple pulses persists over wide areas, and are the basis for individual event correlation. In further testing for event origin in our cores, we find no evidence of the1964 and 1995 extreme El Nino river flood events as hyperpycnal turbidite deposits in any core.

Kneller, B., 1995, Beyond the turbidite paradigm; physical models for deposition of turbidites and their implications for reservoir prediction, in Hartley, A. J., and Prosser, D. J., eds., Characterization of deep marine clastic systems: London, Geological Society of London Special Publication 94, p. 31-49.

Puig, P., Ogston, A. S., Mullenbach, B. L., Nittrouer, C. A., Parsons, J. D., and R. W. Sternberg, R. W., 2004, Storm-induced sediment gravity flows at the head of the Eel submarine canyon, northern California margin: Journal of Geophysical Research, v. v. 109, p. C03019.

Shiki, T., Cita, M. B., and Gorsline, D. S., 2000a, Sedimentary features of seismites, seismo-turbidites and tsunamiites—an introduction: Sedimentary Geology, vol. 135, p. vii-ix.

Wolf, S. C., Nelson, C. H., Hamer, M. R., Dunhill, G., and Phillips, R. L., 1999, The Washington and Oregon mid-shelf silt deposit and its relation to the late Holocene Columbia River sediment budget: U. S. Geological Survey Open File Report 99-173, 99-173.

  Back to the Active Tectonics Home Page