Time series analysis of landslide prediction based on groundwater level variation

The loose deposit landslide was significantly influenced by the groundwater level which was easily monitored and changed before the slope surface moving. Based on the relationship between surface displacement and groundwater level, taking the monitoring of the groundwater level as landslide prediction or aided prediction had a broad prospect. The vector auto-regression model could be applied to analyze the data sequences. It synthesized the correlation and regression analysis of data sequences and could be used to analyze the time lag effect and the relation among several time series. Based on the concept that the downslide thrust of landslide would vary with groundwater level, the measured data of groundwater level could be converted into thrust which induced the change of slope acceleration. And then, the vector auto-regression model which included the measured acceleration of landslide and nominal acceleration stemming from thrust calculated by groundwater level was built. The effect law of groundwater level on surface displacement and the lag time could be estimated by the use of the model. The validity of built model in landslide displacement prediction was verified by application it to the No.6 landslide of Shangyu-Sanmen highway; and it provided reference for similar landslide prediction.