A real-time prediction method of slope failure using Bayesian approach based on slope surface tilting measurements

TY - JOUR

T1 - A real-time prediction method of slope failure using Bayesian approach based on slope surface tilting measurements

AU - Chen, Nuo

AU - He, Pengpeng

AU - Chen, Jiaxun

AU - Huang, Xiaocheng

AU - Fang, Kun

AU - Xie, Jiren

AU - Gorobtsov, Denis N.

AU - Novgorodova, Margarita A.

N1 - Publisher Copyright: © 2025 Elsevier B.V.

PY - 2025/11

Y1 - 2025/11

N2 - Abstract In recent years, slope failure prediction methods using surface tilting measurements have been developed and applied to landslide mitigation. However, it is still challenging to obtain the exact slope failure time because of the uncertainties in monitoring data. Existing methods are unable to provide a failure time interval with a defined confidence level, which is crucial for reliable field predictions by governments and researchers. In this paper, a novel landslide prediction method using the Bayesian approach and based on surface tilting measurements was proposed to estimate an accurate interval of slope failure time, which also allows dynamic updates as new monitoring data become available. Tests at different scales have been used to validate this method. Additionally, A comparison with the conventional prediction method indicates that the proposed prediction method provides more robust and accuracy failure time estimations. Furthermore, the relationship between prediction errors and input data quantity was also investigated using fifteen laboratory and field test cases. Regressive formulas with 95% confidence were obtained, which provides a valuable guidance for estimating prediction errors based on data availability in practice

AB - Abstract In recent years, slope failure prediction methods using surface tilting measurements have been developed and applied to landslide mitigation. However, it is still challenging to obtain the exact slope failure time because of the uncertainties in monitoring data. Existing methods are unable to provide a failure time interval with a defined confidence level, which is crucial for reliable field predictions by governments and researchers. In this paper, a novel landslide prediction method using the Bayesian approach and based on surface tilting measurements was proposed to estimate an accurate interval of slope failure time, which also allows dynamic updates as new monitoring data become available. Tests at different scales have been used to validate this method. Additionally, A comparison with the conventional prediction method indicates that the proposed prediction method provides more robust and accuracy failure time estimations. Furthermore, the relationship between prediction errors and input data quantity was also investigated using fifteen laboratory and field test cases. Regressive formulas with 95% confidence were obtained, which provides a valuable guidance for estimating prediction errors based on data availability in practice

KW - Tilting

KW - slope failure time

KW - Bayesian method

KW - landslides

KW - Landslides

KW - Slope failure time

UR - https://www.scopus.com/pages/publications/105018333250

U2 - 10.1016/j.enggeo.2025.108412

DO - 10.1016/j.enggeo.2025.108412

M3 - Article

SN - 0013-7952

VL - 358

JO - Engineering Geology

JF - Engineering Geology

M1 - 108412

ER -