In areas where volcanic activity has occurred and volcanic sediment has accumulated in rivers, significant caution should be taken in terms of the possible occurrence of disasters associated with debris flows. The use of a real]time detection and warning system, in combination with conventional erosion control measures, is an effective approach in mitigating such disasters. The advent of sensing technology that is able to detect debris flows using vibration sensors will contribute to the development of an automated real]time detection technique. Waveform analysis and Fourier spectral analysis techniques are adopted in developing this technology. We developed a detection method by investigating vibration records at the Hakodate erosion control project. For verification, the method was applied to data from the Nojiri River, which is under the administration of Osumi River and the Nationalway Office. For an event that occurred on 10 July 2004, a predominant frequency change was discovered in synchronization with wire]sensor cut times recorded by video]camera recorders located on check dams. In more precise terms, when the wire]sensors were cut the dominant frequency of the waveform shifted from the high]frequency range (above 100 Hz) to the low]frequency range (below 50Hz), and the spectrum intensity increased. This phenomenon depends on the conditions of stream flow.


Key wordsFautomatic detection, debris flows, landslides, vibration sensor, Nojiri River