On December 17, 1987, an earthquake of magnitude 6.7 occured at a depth of 58 km, near Boso Peninsula, in Pacific Ocean. Many slope failures were induced by the earthquake shock, concerned with the loose sand (SimousaUpland, the northern part of peninsula) and “soft rock ” (Kazusa-Hills, the central part of peninsula). The geology of the latter is Plio-Pleistocene sedimentary rocks which consists mainly of mudstone, sandstone, and their alternations. The slope failures occured especially on valley cliff which consists of thick mudstone and the slope stability was controlled by tectonic joint system with north-northeast to south-southwest trend, high dip angles. The isoseismal was estimated from frequency of overturned tombstones. Both the local high seismic intensity and exposure of jointed mudstone are recognized as major factors in concentration of slope failure in Kazusa-Hills. In about 100 place, geometry of slope failures were measured by 5 m yard-stick. The frequency of failed volume conforms to log-normal distribution. The shape of failed source area seems affect to volume of slope failure. Motion of failing mass accelating on a slope can be explained approximately with the mass point model, introducing “Representative (ideal) length” of failing mass. Geometrical data show that larger mass can be transported more distantly as expected by the above-mentioned dynamic model.
For modeling groundwater flow and mechanical responses of fractured rock mass, to grasp the characteristics of fractures is essentially needed. Recently, the concept of fractal has been applied by some researchers to describe the fracture system. In this study, the authors report a statistical approach made to check the adaptability of the concept of fractal to the fracture system. The lengths of fracture-traces, apperently seen on the surface of rock mass, were measured by sketching and utilizing data of linearment analysis, while the widths of fracture-traces recognized on geological logs and maps were guaged. The measured data of fracture-traces were statistically processed to determine the distribution functions in large and small scales, then, cumulative-frequency curves were examined. Although log-normal distributions were commonly observed for any measured scales, detailed discussion on scale effect in measurement led us that the measurement error and loss of information became larger as the size of fracture-trace became smaller or larger relative to average value of the resolution of human eyes and the objective scales. Removing such effects, the distribution of measured data within a scale was deviated from log-normal, but was correlated approximately by power distribution, indicating that the fracturetraces possibly distribute obeying so-called “Fractal distribution”. The authors prospect that this fact might be useful for estimating reasonably the distribution and the characteristic factors of fractures of any scale from the measurements at available scales.