long-term rate of 1.7–2.0 m/ka derived from the 6-ka and 120-ka strandlines (Yoshikawa et al., 1964; Kanaya, 1978). If the exceptionally high uplift rate produced by historical coseismic events is maintained, the next large earthquake should occur in 2040 (Shimazaki and Nakata, 1980). However, if the long-term rate prevails, which seems more likely, the next event, or a series of closely spaced events, should not occur for another 3 ka. Tide-gauge and geodetic data from central Japan, which reveal the spatial and temporal patterns of interearthquake crustal deformation between the 1855 and 1947 events, indicate that because of the nonlinearity of strain buildup and the significant permanent deformation, simple recurrence calculations from strandline data may overestimate the true interval between major earthquakes by a factor of 2–3 in the Nankai area (Thatcher, 1984). On the other hand, the three closely spaced historical uplift events on the Muroto Peninsula and the two closely spaced events on the Boso Peninsula suggest that the large vertical spacing between emergent Holocene strandlines in these and possibly other areas may represent multiple, not single, earthquakes. If this interpretation is correct, earthquakes at a particular locality on a convergent plate boundary may occur in relatively tight clusters separated by considerable periods of time (1–3 ka), in which case the date of the next event would be difficult if not impossible to predict.
On Kikai Island, the easternmost island in the Ryukyu archipelago of southern Japan, radiocarbon dates of fossil corals from four Holocene strandlines yield a uniform uplift rate of 1.8 m/ka over the past 6 ka and indicate that coseismic uplift followed a time-predictable pattern (Figures 6.24 and 6.26B) (Nakata et al., 1978, 1979; Shimazaki and Nakata, 1980). If the time-predictable model is correct, there should have been a large earthquake between A.D. 1400 and 1600. The lack of historical or geological evidence for an earthquake of this age suggests that Kikai Island, like the Oiso area near Tokyo, is overdue for a major earthquake (Shimazaki and Nakata, 1980).
Two very different earthquake recurrence intervals are recorded by emergent Holocene strandlines on Sado and Awashima Islands in the intensely faulted area off the west coast of central Honshu, the largest Japanese island (Figure 6.24). Seven emergent Pleistocene and Holocene strandlines occur on the Ogi Peninsula, the southernmost tip of Sado Island (Figure 6.15A) (Ota et al., 1976). The lowest (2 m) strandline was formed by uplift during a major earthquake in 1802 (Figure 6.13A). This strandline and the next higher (4 m) 6-ka strandline are both tilted northward about 2.5×10−2, which indicates that the 1802 event was the only tilt event in the past 6 ka. The uniform difference of 2 m between these two strandlines is ascribed to a 2-m highstand of sea level at about 6 ka BP (Ota et al., 1976). A more likely explanation is that the area was uplifted uniformly after 6 ka BP but before the 1802 tilt event. In any case, dividing the coseismic tilt produced in 1802 into the long-term tilt rate derived from the 120-ka strandline yields a recurrence interval for seismic events of 5–9 ka, which is consistent with the >6-ka interval documented by the two lowest strandlines.
At Awashima, which lies about 70 km northeast of Sado Island (Figure 6.24), tilted strandlines yield a much shorter recurrence interval (Nakamura et al., 1965). Awashima was uplifted a maximum of 1.8 m and tilted northward about 2.4×10−4 during the 1964 Niigata earthquake. The long-term tilt rate derived from the 82-ka strandline at 50–70 m above sea level is 4.3×10−4/ka. Dividing this tilt rate by the 1964 tilt yields an average recurrence interval of about 1.5 ka. The different earthquake recurrence intervals for the faults near the Ogi Peninsula and Awashima Island demonstrate that similar faults within the same tectonic province can have very different displacement histories.
The narrow zone comprising the Aleutian archipelago and the south coast of mainland Alaska lies along the overthrust margin of the North American tectonic plate in the north Pacific Ocean. At numerous localities in this seismically active area, great interplate megathrust earthquakes are recorded by emergent Holocene and historical strandlines.
On Middleton Island (Figure 6.18A) emergent Holocene strandlines record six coseismic uplift events over the past 4.5 ka (Figure 6.27A) (Plafker and Rubin, 1967, 1978). The lowest strandline at 3.5 m above sea level was produced by uplift associated with the great 1964 earthquake in southern Alaska. Radiocarbon dates on peat and wood from the wave-cut platforms of the five highest strandlines indicate that the time interval between uplift events increased gradually from about 0.5 ka to 1.4 ka and that the uplift rate decreased from 14 m/ka to 5.6 m/ka. If the island subsided slightly between major earthquakes, the uplift that accompanied each earthquake was greater than the vertical spacing between strandlines. It is not clear if the uplift events, which ranged from 7–12 m and averaged about 9 m, followed a displacement-predictable or a time-predictable pattern, but it is obvious that the 3.5-m uplift associated with the 1964 earthquake was not sufficiently large to maintain even the diminishing uplift rate (Figure 6.27A). In effect, at least half of the stress accumulated since the formation of the second lowest strandline at