Controls of Site Selection, Slope Stability, and Rupture Pattern on the Sedimentary Record of Great Earthquakes

Like all forms of field geology, turbidite paleoseismology requires careful consideration of site context, temporal interval of interest, sediment supply, temporal change in the pathways, and flow dynamics before interpretations can be drawn. These factors are predicated on good bathymetric, backscatter and sub-bottom data and precise navigation so that the context of the core is well known. For example, in Cascadia, numerous channel systems exist and cover a range of time intervals since the Early Pleistocene. During high stands, many of these systems are relict, with limited terrigenous sediment supply, while some are indirectly recharged through cross-shelf transport from river systems. Holocene paleoseismic records therefore commonly depend on recycled materials from failure of local slopes to supply channels, slope basins, or fans. Local failures may serve to supply sediment at any point along a canyon system under expected shaking levels of ~ 1.0 g (saturation based on Mw 9.0) with or without recent sediment recharge. Recharge by active terrigenous sedimentation is apparently not required in Cascadia or Sumatra, where site locations without this recharge possibility have excellent records correlable to other paleoseismic sites. By comparison to Pleistocene fan-building currents in high latitude regions, Holocene currents are weak, rendering most areas of fan systems inactive. Even in low latitude areas, Pleistocene currents can be more robust due to exposure and erosion of the local shelf. Core and backscatter data show the Astoria and Nitinat Fans of Cascadia have little Holocene activity outside the main and uppermost distributary channels. Pleistocene channels are crosscut by active Holocene incisions and levees, restricting their role as depocenters. In the main channels, the most recent currents are largely confined closely within their levees, showing evidence of overbank extending typically 5-8 km from the channel, and consistent with turbidity current heights less than 150 m. Exceptions include channel bend areas where overbank may extend larger distances, depending on levee constraints. In a well-known example, the relict Juan de Fuca Channel on the Washington margin, the thickness of Holocene turbidite stratigraphic sequences in active channels ranges from 2-5 m and is modulated by along strike structural growth of anticlines, mud volcanoes, and frontal thrust landsliding. Recent proposals for alternate Holocene pathways in Cascadia attempt to integrate data from inactive fans, pose implausible pathways, use low resolution visual observations, and use other inactive channels. In Sumatra, no modern recharge of slope basins is possible due to the isolation of the outer forearc slope from terrestrial sediment sources by the unfilled inner forearc basin. There, canyons and channels are few, and most basin fill is derived from failure of local slopes. Transport distances are short in these unchannelized systems, thus sampling location near local slopes is critical to obtaining a paleoseismic record. A transect across Hydrate Ridge basin in Cascadia shows that runout distances can be as little as 5 km in such basins. Resolution of observations is equally critical for paleoseismology, as thick robust deposits may grade to thin silty or muddy ones over short distances. Therefore simple visual core logging is inadequate when compared to modern CT, XRF, and other geophysical data. In northern Washington, low turbidite counts reported in the 1970’s, and repeated recently, are the result of low-resolution visual logging. The northernmost Washington Canyons are relict, and have very low thalweg gradients, yielding finer grained turbidites that are not visually apparent in some cases. Thus for Holocene paleoseismology, cores must be collected from within main channels or near enough to local slopes (1-2 km) to ensure recording of earthquakes. This appears to hold true for both Sumatra and Cascadia. Margin sediments are repeatedly strengthened by many seismic cycles, this seismo-turbidites may be much less robust that one might expect based on slope stability considerations alone. The sources are commonly thin superficial failures of unconsolidated material, rather than large obvious rotational slumps. The outcome of paleoseismic investigations depends critically on these considerations, selection of core sites that address the question being asked, and collection of numerous cores to test sensitivities and hypotheses about sediment transport, seismic segmentation and other factors that may influence the sedimentary record.