LANDWARD FINING IN LATE HOLOCENE TSUNAMI DEPOSITS FROM SOUTHEASTERN HOKKAIDO

Sand sheets that fine and thin landward in a sequence of marsh deposits along the Tokachi coast, southeastern Hokkaido, suggest that at least six large tsunamis have struck this coast within the last 3000 years. These tsunamis are unlikely to have been caused by volcanic eruptions, and are more likely the result of large earthquakes associated with the Kurile Subduction Zone.

The sand sheets appear in a 2 m thick sequence of marsh deposits that begin with soil formation on marine silts. The marsh deposits are fibrous peat, and are punctuated by tephra layers from several nearby volcanoes. The youngest of these (Tarumae-b and Usu-b) were deposited in 1667 and 1663, respectively, and appear about 20 cm below ground surface. A similar ashfall from the 947 AD eruption of Baitoushan (B-Tm) appears ~ 60 cm below ground surface. Last, Tarumae-c (~2700 ybp) lies ~1 m below ground surface. Deposits from the ~6500 ybp eruption of Komagatake (Ko-g) are not visible in the section, suggesting that soil formation postdates that eruption.

The sheets consist of medium to very fine sand, contain marine microfossils, are generally 2-15 cm thick, and thin and fine landward. They extend approximately 800-1000 m inland from the modern beach, and are locally laterally continuous for at least 400 m. Two sheets appear between Us-b and B-Tm, three between B-Tm and Ta-c, and one about 10 cm below Ta-c. Although this coastline is microtidal and does receive large storms, storm deposition appears to be restricted to a ~200 m wide stretch of dunes immediately behind the modern beach—certainly no historic storms have pushed material 1 km from the beach. We conclude, then, that these are most likely the deposits of tsunamis.

Although the sheets appear to be composed of similar material (and consistent with the modern shoreface material), the layers do not have similar grain sizes—the uppermost layer is consistently the coarsest of the units, and the layer immediately below B-Tm is consistently the finest. Similarly the uppermost layer is traceable the farthest inland, whereas the layer immediately below B-Tm pinches out the soonest. This correlation of grain size and inundation limit suggests that these waves were of different sizes, but does not rule out the possibility that the changes seen are related to changes in source material at the shore, and in shoreline position.