Recurrence Intervals for Great Earthquakes of the Past 3,500 Years at Northeastern Willapa Bay, Washington

Plate-boundary earthquakes have occurred repeatedly in the past several thousand years at the Cascadia subduction zone, where they are widely recorded by buried marsh and forest soils beneath estuarine wetlands. This report adds to previous accounts of such soils along the Pacific coast of southern Washington State. Our new evidence comes from outcrop surveys, diatom analyses, and radiocarbon dating of soils exposed about 10 km apart in banks of the Niawiakum and Willapa Rivers, tidal arms of Willapa Bay. This new evidence clarifies the timing of great (magnitude 8 or larger) earthquakes during the past 3,500 years at this part of the Cascadia subduction zone. All the surveyed outcrops display buried soils that probably record tectonic subsidence during earthquakes. Although in many cased we cannot rule out every alternative to such coseismic subsidence, we found no buried soil that is better explained by stream migration, storm, river flood, sea-level rise, barrier breaching, or sediment compaction. Each of the soils probably represents a marsh or forest that suddenly became a tidal flat and consequently was buried by tidal mud. In nearly every case the soils have too much lateral continuity and too little relief to record cutting and filling by tidal streams. Vascular plant fossils preserved within and above many of the soils show that storms or floods, if unaccompanied by lasting submergence, cannot account for burial of the soils. Where their remains are preserved, plants that had lived on the soils belong to species indicative of high parts of tidal marshes, or of tidal swamps or uplands. By contrast, the main vascular -plant species preserved in mud above the soils is Triglochin maritimum, a colonist of saltwater mudflats in southern coastal Washington. Assemblages of diatoms verify vascular-plant evidence for lasting submergence and show that such submergence is recorded also by buried soils with which vascular-plant fossils are not preserved. In an outcrop along the Niawiakum River, for example, diatoms show that each of six successive soils represents a high marsh or an upland, and that mud above each of these soils represents an intertidal or subtidal mudflat. Fossils further show that the change from high marsh or upland to mudflat probably happened too fast to have resulted from a gradual rise in sea level. For every soil studied for fossil diatoms, diatom assemblages imply that this change happened suddenly, without transition through low marsh. Gradual sea-level rise is further precluded for some soils by remains of vascular plants that had lived on buried soils. These remains include stems and leaves that were surrounded by mudflat deposits before they had time to decompose. Other alternatives to coseismic subsidence can be discounted as well. Submergence from breaching of a baymouth barrier is ruled out by soils of brackish marshes that require tidal connection with the sea. Localized settlement from earthquake-induced compaction of unconsolidated Holocene deposits does not explain the presence of buried soils directly above well-consolidated Pleistocene deposits. Evidence against alternative explanation at one outcrop can be extrapolated to other outcrops by radiocarbon and stratigraphic correlation of buried soils. Plate-boundary earthquakes probably account for all the subsidence events. The plate boundary is the only recognized fault common to all areas having evidence of coseismic subsidence in southern coastal Washington. Although some of these areas coincide with mapped late Cenozoic synclines, where coseismic subsidence might accompany earthquakes on faults in the North American plate, others are outside such synclines. The coseismically subsided areas include part of welt of Eocene basement rock 40 km long and 15 km wide, herein termed the South Bend antiform. This structural high, the largest in southern coastal Washington, has probably grown in late Cenozoic time. The South Bend Antiform should subside during plate-boundary earthquakes that flex the North American plate throughout southern coastal Washington. By contrast, the antiform might grow upward during an upper-plate earthquake that produces subsidence only in late Cenozoic synclines that flank the antiform. With these expectations in mind, we compared outcrops off the South Bend antiform (Niawiakum River) with outcrops on the antiform (Willapa River). We found no difference in sense of timing of earthquake-induced changes in land level. Like the surveyed outcrops along the Niawiakum River, the surveyed outcrops along the Willapa River contain buried soils indicative of earthquake-induced subsidence. the longest stratigraphic sequence exposed along both streams contain six or seven soils less than 3,500 years old, and these sequences correlate with one another on the basis of soil horizons, spruce roots, and radiocarbon ages. Though every buried soil identified in the surveyed outcrops probably records a plate-boundary earthquake is widely recorded by a buried soil. One buried soil was obliterated at many sites through centuries of decomposition, probably because shallow burial left the soil high in the weathering profile of a succeeding soil. Such decomposition also destroyed most of the organic matter associated with two other soils. In addition, a widespread soil might not be available for burial if an earthquake occurs too soon after its predecessor for much rebuilding of tidal marshes. Despite these limitations, the composite record from buried soils in large outcrops of northeastern Willapa Bay probably includes every earthquake in the area during the past 3,500 years that caused at least .5 meters of widespread coseismic subsidence and followed the preceding great earthquake by more than a century. Willapa Bay's earthquake history in the past 3,500 years probably includes seven events, each comprising a single rupture or multiple contiguous ruptures on the Washington part of the Cascadia plate boundary. Each event probably included at least one great earthquake, as judged from likely rupture widths inferred from modern geophysical evidence, likely rupture lengths inferred from coastwise correlation of buried soils, extensive sea-floor displacement inferred from a tsunami in Japan, and seismic-moment release deduced from plate motions and average recurrence intervals. The history begins with three events between 3,300-3,500 years ago and 2,400-2,800 years ago (ranges include estimated 95-percent confidence interval). The next recorded events occurred 1,500-1,700 and 1.130-1,350 years ago. They were followed by a poorly dated event that probably occurred before 900 years ago and may have been associated with rupture in inland faults in the North American plate 1,000-1,100 years ago. The most recent of the events happened close to 300 years ago, probably in January 1700. The six intervals between events in this inferred history average 500-540 years but range from about one to three centuries to about a millennium. The earliest two intervals sum to 540-1,100 years. The first of them may have been the longer, as judged from spruce roots that may indicate prolonged interseismic emergence. Next came an interval of 700-1,300 years, when spruce forests spread onto emerging tidal marshes and decomposition largely destroyed an underlying buried soil. This exceptionally long interval was followed by two short ones that together spanned no more than 800 years. The most recent complete interval, marked by another spreading of spruce forests and decomposition of earlier buried soils, lasted 600-1,000 years. This pattern of long and short recurrence intervals at Willapa Bay may match the pattern of intervals between turbidity currents in Cascadia Channel , on the abyssal sea floor 200 km off the central Oregon coast. Previous work showed that these currents largely originated at submarine canyon heads about 50 km west of Willapa Bay, and that great earthquakes may have generated 13 currents in the past 7,500 years. Although pelagic layers between the turbidites have been interpreted as evidence for recurrence intervals of fairly uniform duration, borrows in the turbidites suggest variability in recurrence intervals inferred from buried soils at Willapa Bay.