The sediment]related disaster warning system as one of the non]structural measures for the prevention of disasters has been operated since 2005. But the critical line of the sediment]related disasters to predict the occurrence of the disasters is not well understood. Also the information to judge the issuance of the evacuation orders is limited. As a result it is difficult to warn in a timely manner, residents to evacuate, at present.
To solve these problems, the utilization of the logistic regression model is investigated in this paper. It is experimented to express the indices which are utilized to judge the issuance of the sediment]related warning information or the evacuation orders, by the easily understood probabilistic values using the logistic regression model. It is confirmed that the critical line of the sediment]related disasters can be decided straightly from the calculated disaster occurrence probability and the hourly probability of disaster occurrence can also be obtained. It becomes clear that the logistic regression model can be applied to areas where the data of disasters are scarce and that the calculated exceeding probability can be utilized as the warning line.

Key wordsFlogistic regression model, sediment]related disasters, critical line, warning information