Abstract
Early-warning systems for natural disasters are important tools for disaster risk reduction. In 2005, the
Japanese government initiated a new nationwide early-warning system for landslide disasters. The system
uses the RBFN method to set the criterion (Critical Line: CL) primarily based on rainfall data recorded as
not triggering disasters. Moreover, after large earthquakes, the thresholds (CL) are scaled down by 20–50%,
considering the seismic disturbances in the slopes. In this study, we examined the validity of these ratios by
analyzing disasters which occurred during the seven months after the Great East Japan Earthquake. As a
result, we found that the use of interim CL improved the landslide capture rate by 18.8%, indicating that the
interim standard is effective. The required minimum ratio of scaling down the interim CL is 20–30% in areas affected by more than level 5+ earthquakes without changes of landslide capture rates. The required minimum ratios are expected to reduce false predictions.
