Recurrence of intraplate earthquakes inferred from tsunami deposits during the past 7300 years in Beppu Bay, southwest Japan

Abstract

We studied onshore sediment cores obtained from a coastal marsh on the south coast of Beppu Bay, northern Kyushu Island, Japan, to ascertain whether prehistoric intraplate earthquakes in the bay had generated tsunamis, and to establish a chronology and recurrence intervals for these earthquakes. The latter had been inferred from investigations on shallow offshore faults, but the timing of the fault movements has large uncertainties. We identified five sand layers in the up to 8.8 m long sedimentary sequence, which is mainly composed of non-marine organic-rich mud and marine organic-poor mud. Most sand layers, 1–20 cm thick, exhibit sharp upper and lower contacts with the surrounding muds, implying that they were deposited by sudden events. The sand layers are geochemically distinct, with peaks in Si, K, Ca, Ti, Mn, Fe, and Sr, and diatom assemblages dominated by brackish–marine and marine species. This suggests that the sand came from the sea bottom and/or shore rather than from the freshwater marsh. Based on radiocarbon age modeling, the depositional ages of these sand layers are: 3270–3450 cal yr BP, 4250–4510 cal yr BP, 4970–5280 cal yr BP, 5750–6750 cal yr BP, and 6430–7070 cal yr BP. The oldest deposit is inferred to record a tsunami associated with the 7.3 ka caldera-forming eruption of the Kikai volcano, southern Japan. Thus, we exclude it from the calculation of the recurrence interval of tsunamigenic intraplate earthquakes in the bay. Recurrence is approximately 460–1850 years, including the 1596 CE Keicho Bungo earthquake and a fault rupture 1700–2200 years ago, which have been reported in studies on ruptures of shallow offshore strike-slip and normal faults. Brackish–marine and marine species dominate the fossil diatom assemblages in the upper muddy sand layer (younger than 2750–2870 cal yr BP) above the non-marine organic-rich mud, implying that the muddy sand was deposited in an inner bay or tidal flat environment. The facies change from non-marine to marine sediments after 2750–2870 cal yr BP cannot be explained by a change in the regional sea level because the latter has fallen in the last 6000 years. Therefore, the most likely interpretation for this environmental change is a local coseismic subsidence of the marsh, probably attributed to a rupture of the active fault located south of the marsh. This study is the first to report on recurrent tsunamigenic intraplate earthquakes based on research of onshore tsunami deposits, and suggests that investigating tsunami deposits can contribute to an assessment of tsunami risks for intraplate earthquakes elsewhere.