~NA~T~u~R~E~v~o~L~- ~33~2~3~M~A~RC=H~l9~~~------NBNSANDVIBNS------~~7 Earthquakes mapping the extent of these suddenly buried soils, dating them and further research into their origin is important. Seismic potential of the The essential results of horizontal strain measurements in the subduction region Cascadia subduction zone are that wherever measurements have been attempted, strain appears to be Garry C. Rogers accumulating with a principal axis of shortening that is parallel to the modelled THE two main conclusions that emerged forward by B.F. Atwater, of the US direction of plate motion (M. Lisowski, from a special session at a recent meeting* Geological Survey (USGS), in Science USGS; H . Dragert, Geological Survey of are that the Cascadia subduction zone (236, 942-944; 1987), that sequences of Canada; N.A. Breen, Lamont-Doherty). appears to be storing strain, and that there buried wetland soils on the outer coast of Repeated horizontal control surveys are numerous examples of suddenly Washington State represent a series of comparing older surveys with modern buried wetland-soil horizons that seem to sudden down-drops generated by large laser ranging or global-positioning satel­ be best explained as examples of past subduction earthquakes, received con­ lite measurements show that shear strain great earthquakes. Although some siderable support. is accumulating at a rate of about 0.2 parts speakers suggested that there are special A search for such buried-soil horizons per million per year, comparable to rates conditions that could prevent great earth­ by several investigators reveals that they in other subduction zones such as Japan, quakes, such as the high temperatures of extend from southern Oregon (A.R. New Zealand and Alaska, where large the young subducting slab (J. Severing­ thrust earthquakes have occurred in haus, University of California, Santa BRITlSH COLUMBIA historic times. These measurements Barbara) or the abundant offshore support the hypothesis that the sub­ sediments (D.E. Byrne, Lamont-Doherty duction zone is locked and strain is cur­ Geological Observatory; L.R. Sykes, rently accumulating in the upper plate. Lamont-Doherty), none offered alter­ Measurements of vertical motion are native explanations for the data that also consistent with strain accumulation support the 'great earthquake' hypothesis. across a locked subduction zone. Changes The outcome of the debate has more than of sea level at tidal stations in the northern passing interest for the ten million people part of the zone reveal a characteristic who live in the vicinity of the zone and two-dimensional pattern that would result could be affected by a great earthquake. from the oceanic plate dragging down the The Cascadia subduction zone is a edge of the continental plate (my own small subduction zone that straddles the work). Integrating precise levelling data Canada-United States border on the west with tide gauge data shows a complex coast of North America (see figure) . It areal distribution of vertical motion but involves several small oceanic lithospheric one that is also consistent with strain plates (the last remnants of the large accumulation (S .R. Holdahl, US National Farallon plate) that are actively subduc­ Geodetic Survey). The patterns of vertical ting beneath North America. Low-angle and horizontal motion at the north end of thrust earthquakes, characteristic of most the subduction zone are complicated by subduction zones around the world, have deformation caused by large crustal not been observed there. Detailed moni­ The Cascadia subduction zone of western North earthquakes (Holdahl; Dragert). toring by seismic networks on both sides America. Oceanic lithosphere created at the A two-dimensional finite element of the border over the past decade shows nearby Juan de Fuca Ridge system subducts elastic model of the subduction zone with that, although small earthquakes are beneath the North American Plate. reasonable estimates for input parameters common in the overlying and descending Nelson, USGS) to southern Vancouver is found to fit observed strain measure­ plates, thrust earthquakes at the subduc­ Island (my own work). The most complete ments best when the subduction zone tion interface are absent even at the sequence has been found in large estuaries model is locked and close to an impen­ microearthquake level. The question thus on the outer coast of Washington ding earthquake (H.J. Melosh, University arises as to whether subduction is taking (Atwater; A .G. Hull, University of of Arizona). Current observations, how­ place aseismically or whether the region is California, Santa Barbara) and indicates ever, are too sparse to provide diagnostic in an interseismic period between great that submergence events occurred about constraints for the model. earthquakes. With this seismological 300, 1,000, 1,500, 1,700, 2,500, 2,800 and Before the arguments presented at the information, strain measurements and 3,500 years ago. The presence of sand meeting lead us to believe in a cycle of studies of the recent (the past 12,000 layers, interpreted by Atwater as tsunami great earthquakes on the Cascadia zone, years) geology of the region becoming deposits, help to mark the areal extent of the sudden subsidence events must be available, the question could soon be some events (M.A. Reinhart, University dated accurately and correlated over a answered. of Washington; W.C. Grant, USGS). large area. Similarly, strain must be shown The search for geological evidence of Investigations in smaller estuaries, partic­ to be accumulating over a large area. The past great earthquakes is focused on identi­ ularly those in Oregon, are confusing, as evidence seems to be sufficient, however, fying coastal features that indicate sudden buried soil layers are found , but not to place the onus on the sceptics of the changes of elevation relative to sea level. always (Nelson; Grant; C. Peterson, great-earthquake hypothesis to provide There are no recent marine terraces along Oregon State University). It is clear that convincing alternative explanations for the Cascadia subduction zone to suggest Erratum the observations. Until now, none has that sudden emergence events have IN the table of the article "How to live with emerged. 0 occurred (D.O. West, Golder radon" by Michael O'Riordan (Nature 331, 302; Associates). But the hypothesis put 1988), the term 'per cent' should have referred Garry C. Rogers is at the Geological Survey of • American Geophysical Union fall meeting, 7-11 December only to the absolute lifetime risk and not to the Canada, Pacific Geoscience Centre, Sidney, 1987, San Francisco. relative lifetime risk of exposure to radon. D British Columbia VBL 4B2, Canada.