Modified Inverse-Velocity Method For Predicting The Failure Time Of Landslides

Predicting the time of failure is a topic of major concern in the field of geological risk management.Several approac hes,based on the analysis of displacement monitoring data,have been proposed to deal with the issue in recent years.The inverse velocity(INV)method,which is based on creep theory,is the most extensively employed method.In 1965,Saito simultaneously proposed a method to predict the failure time of landslides based on traditional creep theory.In terms of the previous study and large scale slope model tests,Fukuzono proposed that the logarithm of the surface displacement velocity is proportional to the logarithm of the surface displacement acceleration in the final stage of the slope movement just before failure.Voight proposed the inverse velocity method based on works by Fukuzono.In particular,this theoretical method was obtained from the assumption that mechanical properties are time-invariant and obtained from controlled laboratory conditions,which do not accord with reality.In natural landslides and in-situ conditions,there exist some limitations,including local slope motion,measurement er rors,and displacement rate changes triggered by periodic variation factors(e.g.human activities,rainfall and groundwater).These factors can substantially affect the reliability and accuracy of prediction results,and eventually hamper the use of the INV.In order to improve the practicality and reliability of the INV in prediction of failure time,the modified INV is proposed for continuous monitoring and discontinuous monitoring.In the case of continuous monitoring,based on the works by Voight and other scholars,a modified inverse velocity method is proposed by introducing a controllable variable into the traditional inverse velocity method.One landslide was detected in the certain open pit mine and four landslides were detected in the certain open pit mine by using ground-based radar.In the four landslides in the certain open pit mine,the acceleration obviously decreased a few hours before failure,which is inconsistent wit h the tranditional creep theory.The traditional inverse velocity method a nd the modified inverse velocity method were employed to analyze the se five landslides.The analysis results show that the modified inverse velocity method can improve the effectiveness of prediction by reducing the prediction error caused by the accelerat ion decrease before failure.The modified inverse velocity method proposed in this paper does not only improves the prediction accuracy,but also provides a study direction to improve the inverse velocity method,that is,controllable variables are introduced to improve the prediction accuracy.In the case of discontinuous monitoring,a modified inverse velocity method is proposed by combining the Bootstrap sampling method and the traditional inverse velocity method.Taking Beni mountain landslide(inverse velocity-time scatter as concave type)and Vajont landslide(inverse velocity-time scatter as upward convex type)as examples,the specific application of the modified inverse velocity method are introduced.Then,based on the monitoring data of 1-2 measuring points of Beni mountain landslide,a set of ideal velocity time series is obtained by fitting,a nd then random noise is introduced into the ideal velocity time series to construct a set of data of inverse velocity-time scatter distributed in a straight line.Finally,prediction accuracy of the modified inverse velocity method is studied.The results show that the modified inverse velocity method can improve the accuracy and reliability of landslide time prediction in a certain extent.