The product of rainfall intensity and effective accumulated rainfall is defined as a debris-flow rainfall triggering index (RTI). A new rainfall-based debris-flow warning model is developed in which a diagram is set up with the RTI-data on the ordinate and the time of rainfall on the abscissa in order to evaluate debris-flow occurrence probability. A method is proposed to determine the lower critical warning line (RTI=RTI10) and the upper critical warning line (RTI=RTI90), based on the RTI-values of historical rainfalls. The lower and upper critical warning lines divide the debris-flow occurrence probability into three parts. The RTI-values smaller than RTI10 suggest that the debris-flow occurrence probability is less than 10%, the RTI-values higher than RTI90 suggest that the occurrence probability is larger than 90%, and the RTI-values between RTI10 and RTI90 suggest that the debris-flow occurrence probability is between 10% and 90%. The proposed model was applied at Shueili, Nantou County, central Taiwan to evaluate the temporal variations of debris-flow occurrence probabilities caused by two rainfall events. The results show that the proposed model could effectively evaluate the temporal variations of debris-flow occurrence probability during a rainfall event.
The product of rainfall intensity and effective accumulated rainfall is defined as a debris-flow rainfall triggering index (RTI). A new rainfall-based debris-flow warning model is developed in which a diagram is set up with the RTI-data on the ordinate and the time of rainfall on the abscissa in order to evaluate debris-flow occurrence probability. A method is proposed to determine the lower critical warning line (RTI=RTI10) and the upper critical warning line (RTI=RTI90), based on the RTI-values of historical rainfalls. The lower and upper critical warning lines divide the debris-flow occurrence probability into three parts. The RTI-values smaller than RTI10 suggest that the debris-flow occurrence probability is less than 10%, the RTI-values higher than RTI90 suggest that the occurrence probability is larger than 90%, and the RTI-values between RTI10 and RTI90 suggest that the debris-flow occurrence probability is between 10% and 90%. The proposed model was applied at Shueili, Nantou County, central Taiwan to evaluate the temporal variations of debris-flow occurrence probabilities caused by two rainfall events. The results show that the proposed model could effectively evaluate the temporal variations of debris-flow occurrence probability during a rainfall event.