No abstract

The Shumagin seismic gap, a segment of the plate boundary along the eastern Aleutian arc, has not ruptured during a great earthquake since at least 1899–1903. Because at least 77 years have elapsed since the Shumagin Gap last ruptured in a great earthquake and repeat times for the 1938 rupture zone and part of the Shumagin Gap are estimated to be 50 to 90 years, a high probability exists for a great earthquake to occur within the Shumagin Gap during the next one to two decades. Reconsideration of the rupture zones of the Aleutian earthquakes of 1938, 1946, and 1948 suggests that those events did not break the interplate boundary beneath the Shumagin Islands. Thus, the Shumagin seismic gap extends from the western end of the 1938 rupture zone to the eastern end of that of 1946. These boundaries also coincide with transverse structural features. At least the eastern half of the Shumagin Gap broke in great earthquakes in 1788 and 1847 and possibly in 1898–1903. The Shumagin Gap is probably not the result of aseismic slip; rather, plate motion is accommodated there seismically and episodically and can be expected to produce large earthquakes in the future. Although there is no definitive evidence of long‐term precursors of a possible future earthquake, several observations suggest that the Shumagin Gap is in an advanced stage of the earthquake cycle. Both teleseismic and local network data indicate a near absence of seismic activity (M ≥ 2) above a depth of 30 km along the main thrust zone within the gap; this is in strong contrast to adjacent portions of the arc where seismic activity is scattered across most of the main thrust zone. Two earthquakes with high stress drops (600–900 bars), which occurred at the base of the main thrust zone, may indicate the accumulation of a considerable amount of strain energy within the gap. A possible seismic gap at the eastern end of the aftershock zone of the Aleutian earthquake of 1957 has been identified near Unalaska Island. An earthquake that nucleates in the Shumagin Gap could also rupture the possible Unalaska Gap to the west, the 1938 aftershock zone to the east, or both, with resultant magnitude up to Mw = 9.0. Alternatively, the Shumagin Gap alone, or in one of the above combinations could rupture in a series of very large earthquakes instead of a single great shock. Past Alaska‐Aleutian earthquakes, including those of 1788, 1938, 1946, 1957, 1964, and 1965, have generated very large tsunamis. Future large earthquakes in the Shumagin Islands region could generate wave heights of several tens of meters along shorelines near the rupture areas. The Shumagin Gap is one of two major gaps along the United States portion of the Alaska‐Aleutian plate boundary and is one of the few areas in the United States where processes leading to a great earthquake are likely to be observed within a reasonable span of time.

The 5000 km long Queen Charlotte-Alaska-Aleutian seismic zone is subdivided into 17 unequally sized segments. Their boundaries are delineated based on the prior distribution of large and great earthquakes. The 17 segments are chosen to represent areas likely to be ruptured by "characteristic" earthquakes. This term usually implies repeated breakage of a plate boundary segment by either a large or great earthquake, whose source dimensions remain consistent from cycle to cycle. This definition does not exclude the possibility that occasionally adjacent characteristic earthquake segments may break together in a single "giant" event that is larger than the characteristic size outlined. Conversely, a segment can also sometimes break in a series of smaller ruptures. Formal computations of the conditional probabilities for future large and great earthquakes in the 17 segments of the Queen Charlotte-Alaska-Aleutian seismic zone are based on the following data sets and findings: (1) recurrence intervals from historic and geologic data; (2) direct recurrence time estimates based on rates of relative plate motion and the size or displacement of the most recent characteristic event in each segment; and (3) the application of a lognormal distribution of recurrence times for large and great earthquakes. Results of these computations indicate seven areas that have high (i.e., ->60%) conditional probabilities for the recurrence of either large or great earthquakes within the next 20 years (1988-2008). These areas include Cape St. James, Yakataga, the Shumagin Islands, Unimak Island, and the Fox, Delarof, and Near Islands segments of the Aleutian arc. When a shorter timeinterval is considered (1988-1998), those segments more likely to rupture in large (M s 7-7.7) rather than great earthquakes have a high conditional probability. These areas include the Unimak, Fox, and Delarof Islands segments. The largest uncertainties in these forecasts stem from the short historic record (providing a single recurrence time estimate for some segments, or widely varying estimates for others); from the unknown importance of aseismic slip; and from a vague definition of "characteristic" earthquake size. In fact, characteristic earthquake size may not be a time-invariant quantity. 1.$150 million dollars in damages throughout the circum-Pacific area were caused by tsunamis generated by three recent Alaska-Aleutian events (1946, Mr 9.3; 1957, Mr 9.0; and 1964, Mr 9.1; Lockridge and Smith, 1984). Hence in addition to the scientific interest in this seismic zone, there is a strong social and economic motivation to investigate and describe future earthquake and tsunami hazards for this region.Five tectonic regimes comprise the 5000 km-long zone of interaction between the Pacific and North American plates. These include 1) a predominantly strike-slip regime along the Queen Charlotte-Fairweather fault zone, 2) a zone of tran-sition between strike-slip and underthrust motion in eastern Gulf of Alaska, 3) a continental-type subduction regime in southern Ala...

For over thirty years, attempts have been made to gain information about sediment amplification during earthquakes from observations of ambient seismic noise. While the results of several feasibility studies have been encouraging, theoretical support for the technique is scant. We present a model for the response of sedimentary layers to ambient seismic noise. The noise sources are modeled as a random distribution (in time and space) of point forces located on the Earth's free surface. This model is applied to a site where observed noise spectral ratios, relative to a rock site, have previously been shown to reveal the fundamental resonant frequency of a soft clay layer. Approximating the sediment site as a single layer over a half‐space, the horizontal noise spectrum predicted by our model reveals the fundamental resonance and first harmonic of the layer. We also examine an estimate of site response proposed by Nakamura (1989), which is formed by dividing the horizontal‐component noise spectrum by that of the vertical component. Nakamura's estimate applied to both observed and predicted noise‐spectra was also successful in identifying the fundamental resonance, with a slight (<10%) shift toward lower frequencies. Future work is needed to determine the generality of our results, and to elucidate the influence of the simplifying assumptions.