Brian F. Atwater

Unusually large earthquakes inferred from tsunami deposits along the Kuril trench

The Pacific plate converges with northeastern Eurasia at a rate of 8–9 m per century along the Kamchatka, Kuril and Japan trenches. Along the southern Kuril trench, which faces the Japanese island of Hokkaido, this fast subduction has recurrently generated earthquakes with magnitudes of up to ∼8 over the past two centuries. These historical events, on rupture segments 100–200 km long, have been considered characteristic of Hokkaidos plate-boundary earthquakes. But here we use deposits of prehistoric tsunamis to infer the infrequent occurrence of larger earthquakes generated from longer ruptures. Many of these tsunami deposits form sheets of sand that extend kilometres inland from the deposits of historical tsunamis. Stratigraphic series of extensive sand sheets, intercalated with dated volcanic-ash layers, show that such unusually large tsunamis occurred about every 500 years on average over the past 2,000–7,000 years, most recently ∼350 years ago. Numerical simulations of these tsunamis are best explained by earthquakes that individually rupture multiple segments along the southern Kuril trench. We infer that such multi-segment earthquakes persistently recur among a larger number of single-segment events.

Medieval forewarning of the 2004 Indian Ocean tsunami in Thailand

Recent centuries provide no precedent for the 2004 Indian Ocean tsunami, either on the coasts it devastated or within its source area. The tsunami claimed nearly all of its victims on shores that had gone 200 years or more without a tsunami disaster. The associated earthquake of magnitude 9.2 defied a Sumatra–Andaman catalogue that contains no nineteenth-century or twentieth-century earthquake larger than magnitude 7.9 (ref. 2). The tsunami and the earthquake together resulted from a fault rupture 1,500 km long that expended centuries’ worth of plate convergence. Here, using sedimentary evidence for tsunamis, we identify probable precedents for the 2004 tsunami at a grassy beach-ridge plain 125 km north of Phuket. The 2004 tsunami, running 2 km across this plain, coated the ridges and intervening swales with a sheet of sand commonly 5–20 cm thick. The peaty soils of two marshy swales preserve the remains of several earlier sand sheets less than 2,800 years old. If responsible for the youngest of these pre-2004 sand sheets, the most recent full-size predecessor to the 2004 tsunami occurred about 550–700 years ago.

Summary of Coastal Geologic Evidence for Past Great Earthquakes at the Cascadia Subduction Zone

Earthquakes in the past few thousand years have left signs of land-level change, tsunamis, and shaking along the Pacific coast at the Cascadia subduction zone. Sudden lowering of land accounts for many of the buried marsh and forest soils at estuaries between southern British Columbia and northern California. Sand layers on some of these soils imply that tsunamis were triggered by some of the events that lowered the land. Liquefaction features show that inland shaking accompanied sudden coastal subsidence at the Washington-Oregon border about 300 years ago. The combined evidence for subsidence, tsunamis, and shaking shows that earthquakes of magnitude 8 or larger have occurred on the boundary between the overriding North America plate and the downgoing Juan de Fuca and Gorda plates. Intervals between the earthquakes are poorly known because of uncertainties about the number and ages of the earthquakes. Current estimates for individual intervals at specific coastal sites range from a few centuries to about one thousand years.

A tsunami about 1000 years ago in Puget Sound, Washington

Water surged from Puget Sound sometime between 1000 and 1100 years ago, overrunning tidal marshes and mantling them with centimeters of sand. One overrun site is 10 kilometers northwest of downtown Seattle; another is on Whidbey Island, some 30 kilometers farther north. Neither site has been widely mantled with sand at any other time in the past 2000 years. Deposition of the sand coincided—to the year or less—with abrupt, probably tectonic subsidence at the Seattle site and with landsliding into nearby Lake Washington. These findings show that a tsunami was generated in Puget Sound, and they tend to confirm that a large shallow earthquake occurred in the Seattle area about 1000 years ago.

Geologic evidence for earthquakes during the past 2000 years along the Copalis River, southern coastal Washington

Evidence for several prehistoric earthquakes is present in deposits less than 2000 years old that crop out at the Copalis River estuary (47°07.2′N, 124°09.7′W). The deposits show that land subsided decimeters into the intertidal zone on as many as three occasions and that sand vented while the land underwent little or no subsidence on another occasion. The evidence for earthquake-induced subsidence consists of buried marsh and forest soils that evince sudden estuarine submergence attended on at least one occasion by a landward directed surge of sandy water. This combination of submergence and surge implies subsidence and tsunami. Exposed evidence for subsidence and tsunami along the Copalis River is strong for a time within a few decades of 300 years ago, moderate for a time 1400–1900 years ago, and weak (subsidence) or lacking (tsunami) for an intermediate time also about 1400–1900 years ago. The history of sudden submergence and tsunami in the past 2000 years at the Copalis River estuary resembles that shown by buried soils farther south in coastal Washington at Grays Harbor and Willapa Bay. The history can be explained most simply by great (Mw of 8 or 9) thrust earthquakes at the boundary between the Juan de Fuca and North America plates, with one or two earthquakes occurring 1400–1900 years ago and another close to 300 years ago. The evidence for earthquake-induced venting along the Copalis River consists of sand intrusions and vented-sand volcanoes. The sand rose through muddy estuarine deposits at least 3 m thick, entraining these deposits in fragments as much as 0.3 m long. Intrusion occurred at one or more times in the past 2000 years, and at least some of the venting occurred 900–1300 years ago. The sand bodies do not imply artesian flow of flood-pressured groundwater because the sand vented forcefully and infrequently through a lowland that probably lacked natural levees. The earthquake implied by the venting of sand 900–1300 years ago could have come from a source below, above, or at the boundary between the Juan de Fuca and North America plates. If from the subducted Juan de Fuca plate the implied earthquake was farther west or larger than the largest earthquakes within that plate during the past 100 years. Evidence permitting a source within the North America plate includes a fault scarp, an uplifted tideflat, and landslides near Puget Sound dating from the interval 700–1700 years ago. A plate boundary source would explain buried soils whose estuarine submergence and burial imply coseismic subsidence for parts of coastal Washington in the interval 600–1300 years ago. However, stratigraphy shows that little or no coseismic subsidence occurred along the Copalis River during the venting 900–1300 years ago. If a plate boundary earthquake struck coastal Washington 900–1300 years ago, the earthquake produced a different distribution of land level change than did the plate boundary earthquakes inferred for 1400–1900 and 300 years ago.

Sudden, probably coseismic submergence of Holocene trees and grass in coastal Washington State

Growth-position plant fossils in coastal Washington State imply a suddenness of Holocene submergence that is better explained coseismic lowering of the land than be decade- or century-long rise of the sea. These fossils include western red cedar and Sitka spruce whose death probably resulted from estuarine submergence close to 300 years ago. Rings in eroded, bark-free trunks of the red cedar show that growth remained normal within decades of death. Rings in buried, bark-bearing stumps of the spruce further show normal growth continuing until the year of death. Other growth-position fossils implying sudden submergence include the stems and leaves of salt-marsh grass entombed in tide-flat mud close to 300 years ago and roughly 1,700 and 3,100 years ago. The preservation of these stems and leaves shows that submergence and initial burial outpaced decomposition, which appears to take just a few years in modern salt marshes. In some places the stems and leaves close to 300 year old are surrounded by sand left by an extraordinary, landward-directed surge-probably a tsunami from a great thrust earthquake on the Cascadia subduction zone.

Earthquake recurrence inferred from paleoseismology

Publisher Summary This chapter describes three North American examples of earthquake history inferred from Quaternary geology and discusses earthquakes in the interior of the North America plate––in the New Madrid seismic zone of Missouri, Arkansas, and Tennessee. The study of prehistoric earthquakes––paleoseismology––provides long-term rates of earthquake occurrence to improve confidence in such forecasts. These earthquakes suggest the rates and patterns of recurrence that help define earthquake hazards. The eastern California shear zone, centered about 150 km northeast of Los Angeles, exhibits geologic evidence for prehistoric surface ruptures during episodes thousands of years apart. Typical intervals between the earthquakes span hundreds of years in the New Madrid and Cascadia examples and thousands of years in the eastern California example. Apart from enabling such estimates of recurrence intervals, paleoseismology can provide evidence for the regional clustering of earthquakes in seismic zones and for aperiodic rupture along the same part of a fault. Such findings have made paleoseismology an essential part of earthquake-hazard assessment in the United States.

A fan dam for Tulare Lake, California, and implications for the Wisconsin glacial history of the Sierra Nevada

Historic fluctuations and late Quaternary deposits of Tulare Lake, in the southern San Joaquin Valley, indicate that maximum lake size has depended chiefly on the height of a frequently overtopped spillway. This dependence gives Tulare Lake a double record of paleoclimate. Climate in the Tulare Lake region has influenced the degree to which the lake fills its basin during dry seasons and dry years: during the past 100,000–130,000 yr, incidence of desiccation of Tulare Lake (inferred from stiffness, mud cracks, and other hand-specimen properties) has been broadly consistent with the lake9s salinity and depth (inferred from diatoms and ostracodes) and with regional vegetation (inferred from pollen). Climate, however, also appears to control basin capacity itself: Tulare Lake becomes large as a consequence of glacial-outwash aggradation of its alluvial-fan dam. Late Wisconsin enlargement of Tulare Lake probably resulted from the last major glaciation of the Sierra Nevada. The lake9s spillway coincides with the axis of the glacial-outwash fan of a major Sierra Nevada stream; moreover, sediment deposited in the transgressive lake resembles glacial rock flour from the Sierra Nevada. Differential tectonic subsidence and deposition by a Coast Range creek facilitated the building of Tulare Lake9s fan dam during the late Wisconsin but were less important than deposition of Sierra Nevada outwash. Four stratigraphically consistent 14 C dates on peat and wood give an age of 26,000 yr B.P. for the start of Tulare Lake9s late Wisconsin transgression. The last major Sierra Nevada glaciation (Tioga glaciation) thus may have begun about 26,000 yr B.P., provided that vigorous glacial-outwash deposition began early in the glaciation. Onset of the Tioga glaciation about 26,000 yr B.P. is consistent with new stratigraphic and radiocarbon data from the northeastern San Joaquin Valley. These data suggest that the principal episode of glacial-outwash deposition of Wisconsin age began in the San Joaquin Valley after 32,000 yr B.P., rather than at least 40,000 yr B.P., as previously believed. An earlier enlargement of Tulare Lake probably resulted from a fan dam produced by the penultimate major (Tahoe) glaciation of the Sierra Nevada. Average sedimentation rates inferred from depths to a 600,000-yr-old clay and from radiocarbon dates indicate that this earlier lake originated no later than 100,000 yr B.P. The Tahoe glaciation therefore is probably pre-Wisconsin.

Periodic floods from glacial Lake Missoula into the Sanpoil arm of glacial Lake Columbia, northeastern Washington.

At least 15 floods ascended the Sanpoil arm of glacial Lake Columbia during a single glaciation. Varves between 14 of the flood beds indicate one backflooding every 35 to 55 yr. This regularity suggests that the floods came from an ice-dammed lake that was self-dumping. Probably the self-dumping lake was glacial Lake Missoula, Montana, because the floods accord with inferred emptyings of that lake in frequency and number, apparently entered Lake Columbia from the east, and produced beds resembling backflood deposits of Lake Missoula floods in southern Washington.

Rapid resetting of an estuarine recorder of the 1964 Alaska earthquake

Tides and plants have already restored much of a landscape that the 1964 Alaska earthquake destroyed. At the head of a macrotidal estuary near Anchorage, in the vicinity of Portage, subsidence during the earthquake changed meadows, thickets, and spruce groves into barren tidal flats. Tidal-flat silt and sand soon buried the pre- earthquake landscape while filling intertidal space that the subsidence had made. The flats supported new meadows and thickets by 1973 and new spruce by 1980. Three new findings confirm that the flats aggraded rapidly and that their vegetation is maturing. (1) Most of the postearthquake deposits at Portage date from the first decade after the 1964 earthquake. Their thickness of 23 sites in a 0.5 km 2 area was 1.4 ± 0.2 m in 1973, 1.6 ± 0.2 m in 1991, and 1.6 ± 0.3 m in 1998. (2) Many of the deposits probably date from the first months after the earthquake. The deposits contain sedimentary couplets in which coarse silt or very fine sand is capped by fine or medium silt. About 100 such couplets make up the lowest 0.5 m or more of the postearthquake deposits in two outcrops. These couplets thicken and thin rhythmically, both as groups of 5–20 couplets and as pairs of successive couplets. Probably, the groups of thick couplets represent the highest tides, the groups of thin couplets represent some of the lesser high tides, and the pairs record inequality between twice-daily high tides. (3) In the 1980s and 1990s, thickets expanded and spruce multiplied. The vegetation now resembles the fossil assemblage rooted in the buried landscape from 1964. Had the 1964 Alaska earthquake been repeated a decade later, the two earthquakes would now be recorded by two superposed, buried landscapes near Portage. Much more than a decade is probably needed to reset similar recorders at mesotidal estuaries of the Cascadia subduction zone.