The Shumagin Islands, Alaska, earthquake of 31 May 1917

Abstract

Abstract Although macroseismic reports indicate that great earthquakes have occurred in the vicinity of the Shumagin Gap of the eastern Aleutian arc, the largest, well-documented event within the gap has been an M S 7.5 event that occurred in 1948; this earthquake apparently ruptured only a 75-km-long segment of the 200-km-long gap and as such is not considered a gap-filling event. Relocations of turn-of-the-century seismicity by Boyd and Lerner-Lam (1988) suggest, however, that another, more poorly documented event nucleated near the Shumagin Islands on 31 May 1917. An M S of 7.9, determined from a single station bulletin, has been assigned to this event; although consistent with its apparent aftershock distribution, macroseismic reports appear to be too small for such a large event. We use surface waves to estimate the moment and rupture bounds and body waves to constrain the focal mechanism and depth of the 1917 event. The average M S at 13 azimuthally well-distributed stations is 7.4 ± 0.3, much smaller than the previous estimate of 7.9. The spatial variation in individual magnitude estimates is azimuthally dependent (7.9 to the west and 6.8 to the east), implying source directivity. Assuming a thrust fault mechanism, body-wave inversion yields a source depth of 29 km and M W of 7.4. In addition, assuming rupture velocity and rupture propagation along-arc to the west (consistent with the aftershock distribution), a rupture length of 75 km and M W of 7.5 are required to match long-period surface waves recorded at Uppsala, Sweden, and Gottingen, Germany. These results suggest that the average slip of the 1917 event was only about 75 to 150 cm and was restricted to the easternmost portion of the gap. These results, when combined with rupture zones and recurrence intervals of other eastern Aleutian events, imply a temporally heterogeneous rupture style where the rupture bounds, and consequently the sizes of earthquakes, vary from cycle to cycle. Estimating the magnitudes or rupture extents of the next series of events based on the previous series appears to be unjustified.