Assessment of the state of a landslide slope based on the analysis of multidimensional time series of geotechnical monitoring data

Abstract

The purpose of the work is to develop a methodology for modeling and predicting the state of a landslide based on the results of observations obtained in real time using an automated monitoring system including geotechnical and geodetic equipment (surface tilt meters, electric three-dimensional fracture meters, soil pressure sensors, inclinometers). The object of the study is the real data of the results of geotechnical monitoring during the work on the “Construction and reconstruction of retaining walls on the A‑147 Dzhubga- Sochi-border with the Republic of Abkhazia, Krasnodar Territory. On the section km 195+310 – km 196+985”. Traditional methods of studying monitoring results are limited to visualization of observations and calculation of descriptive characteristics, meanwhile, observations represent a time sequence that can be considered from the point of view of the theory of stochastic time series. The aim of the study was to identify patterns of landslide processes in order to update the data-based approach. The obtained results of the analysis of landslide slope monitoring data allow us to conclude that observations can be described by models of multidimensional time series (autoregressive, moving average, moving average autoregression, pro-integrated moving average autoregression and vector regression), which can be used to predict the development of landslides and quantify risk.