Two shallow landslides in a large-scale model slope were induced by artificial heavy rainfall (100mm per hour), in which the sand layer heights were set at 0.7 and 0.5m. A total of 2,931s rainfall triggered a relatively large landslide over a whole steep slope section in a sand layer 0.5m high. Conversely, only a small landslide was produced by a total of 4,000s rainfall in the lower half of the steep slope section in a sand layer 0.7m high. Shear deformation and seepage flow conditions were monitored in the experiments. By tracing the movements of the markers imbedded inside the sand layers, changes in strain were analysed and expressed as Mohr's circles. Also by approximating the equi-potential lines from data observed on pore-water pressure, changes in seepage flow directions were calculated. It was revealed that, although the sand layers were uniformly packed, the observed shear deformation and seepage flow conditions were not homogeneous. Just before the landslide was triggered, water tables were formed in almost all slope sections. However, a sliding surface was not necessarily formed below the water table and directions of maximum shear strain and seepage flow generally correlated to the slope-base direction in the parts inside the landslide, indicating their possible influence on the landslide.
In order to evaluate the mechanism and soil characteristics of slip surface of earthquake-induced and rainfall/melting snow-induced landslides, we examined soil properties such as grain size distribution, consistency index and shear rate dependency of residual strength. Experimental results revealed that slip surface soils of rainfall/snowmelt-induced landslides contain high amounts of clay fraction and high plasticity property, and positive rate dependencies of residual strength are commonly recognized on these soils. On the other hand, negative rate dependencies of residual strength are observed on many samples of earthquake-induced landslides. These soil characteristics reflect common properties of slow-moving reactivated landslides and long-travelling rapid landslides, respectively, and can be used as indicators to elucidate the site properties of landslides with particular patterns of movement