Earthquakes and sea-level change in Hokkaido, north-east Japan

Abstract

This thesis details the results of an investigation into the pattern of relative sea-level (RSL) changes in north-east Hokkaido, Japan. The aim of the research is to better understand the importance of seismic and non-seismic processes in controlling spatial patterns of vertical land motions over a range of timescales. The main focus is on using salt-marsh sediments as a source of data to reconstruct RSL change during the current interseismic period, since c. 300 calibrated years before present (cal. yr BP). Previous research on the Pacific coast of Hokkaido suggests that this period is characterised by subsidence caused by strain accumulation on the locked part of the Pacific/North American plates.

I apply foraminiferal-based methods of palaeoenvironmental reconstruction to develop, using transfer functions, quantitative reconstructions of RSL change at five sites in north-east Hokkaido. Contemporary foraminifera are zoned with respect to elevation and tidal inundation, and my preferred transfer function (a model that contains 87 samples and 24 taxa) has a prediction r2 of 0.75 and a root mean squared error of prediction of ± 0.32 m. I apply this transfer function to shallow fossil sediment sequences at five salt marshes and use a combination of 210Pb, 137Cs and tephra chronology to establish age models for the sequences. The reconstructions are consistent in demonstrating little net RSL change during the last 300-100 cal. yrs, with the exception of data from one site, Sarfutsu-toh, located on the northern tip of Hokkaido. Chronologies from two profiles developed on the Pacific coast record strong evidence for recent RSL rise since the mid-1980s, but during earlier periods of the 20th century reconstructed RSL was stable or falling.

I compare my reconstructions with other direct and proxy records of land and sea-level motions. Previously published GPS and repeat levelling data indicates subsidence in south-east Hokkaido during the 20th century, although the spatial patterns and rates of change have varied. An unknown amount of this subsidence at the Kushiro tide gauge likely reflects anthropogenic activities associated with sediment compaction as well as mining-induced subsidence. An analysis of the tide-gauge records from Hokkaido show a more varied pattern of land motions, although they also confirm subsidence on the Pacific coast, close to the Kuril trench. A database of Holocene sea-level index points provides insights into longer-term millennial-scale trends in RSL. Data from six regions of Hokkaido demonstrate stable RSL close to present during the mid- and late Holocene; only the northern tip of Hokkaido (around Sarubetsu) is there evidence for a small mid-Holocene highstand of 1-3 m above present.

Finally, a review of Pleistocene raised marine terrace data shows net uplift over the last c. 330 k yr, with two areas of particularly high uplift at Abashiri and on the Pacific coast near Kushiro.

The evidence presented in this research demonstrates that it is incorrect to infer that the current interseismic period is characterised by subsidence. Overall, RSL has changed little in the last 300-100 cal. yrs. The subsidence recorded in the mid- and late 20th century on the Pacific coast of Hokkaido is not typical of the full interseismic period, nor can it have been sustained over Holocene or Pleistocene timescales. Limited data from previous earthquake cycles indicate that RSL was stable, rising or falling during previous interseismic intervals. These observations suggest that a representative ‘Hokkaido earthquake deformation cycle’ may not exist. Future research should better understand the controls of Quaternary volcanic activity on regional deformation patterns, and apply microfossil-based techniques to multiple earthquake cycles at sites to help define the spatial extent of land motions associated with different events.