The Effects of Differing Sequences of Earthquake Ground-Shaking on Coseismic Slope Stability

Abstract:

Studies of earthquake-induced landsliding typically consider slope stability during high-magnitude ground shaking events only. During such events, downslope movement of the landslide mass occurs when seismic ground accelerations are sufficient to overcome shear resistance at the landslide shear surface. This approach does not consider the potential effects that sequences of low-magnitude ground shaking events can have on material strength and, hence, coseismic slope stability. Since such events are more common in nature relative to high-magnitude shaking events, it is important to constrain their geomorphic effectiveness. Using an experimental laboratory approach, we present results that address this key issue. We used a bespoke geotechnical testing apparatus, the Dynamic Back-Pressured Shear Box, that permits realistic simulation of earthquake ground-shaking conditions within a hillslope. We tested both cohesive and granular materials that displayed ductile behaviour under standard strain-controlled monotonic shear tests. We applied dynamic stresses of varying amplitude, frequency and sequence, and monitored the resultant strain response to determine which factors, when combined, created notable deviations from standard monotonic shear behaviour. We observed that multiple dynamic stress/shaking events that are largely insufficient to cause large strains (and hence are conventionally deemed geomorphologically ineffective) can affect material stiffness such that the future behaviour of the sediment/landslide differs considerably from that observed in standard monotonic shear tests. In other words, low-magnitude ground shaking events can be effective precursory geomorphic processes. Critically, the sequence of ground-shaking events is an important control; where shaking conditions cause progressive densification of sediment, the frictional strength of the material subsequently increases. In turn, the resultant strain response to high-magnitude ground shaking events decreases. Our results have important implications for studies of long-term landscape evolution, in which modelled hillslopes are repeatedly subjected to multiple earthquake events but that currently lack appropriate empirically-constrained strength parameters.