Realistic Assessment of Slope Reliability for Effective Landslide Hazard Management: Fid-Bau Portal

Realistic Assessment of Slope Reliability for Effective Landslide Hazard Management

Cheung, R. W. M. / Tang, W. H.

Summary by Jie Zhang: Realistic assessment of slope reliability is essential to the mitigation of landslide hazards. Yet, accurate estimation of the slope reliability is inherently challenging due to the modeling errors involved in slope reliability estimation methods. The focus of this paper is on how to estimate the slope failure probability in a more realistic way. A statistical model based on survival analysis is first used to estimate the average reliability of a slope in Hong Kong utilizing the extensive landslide database in this region. While such a model provides useful information about the global reliability of slopes in Hong Kong, it is less relevant to an individual slope with detailed site-specific information. Therefore, a Bayesian method is employed to assess the failure probability of a slope calculated based on the structural reliability theory, in which the site-specific information of a slope can be readily considered. Through calibration with the observed performance data, the effect of model error on failure probability estimation is readily considered. In addition, with the Bayesian method suggested, the failure probability of a slope estimated from a statistical model can be merged with that calculated from a site-specific slope reliability analysis. Through systematical calibration and information combination, the procedure suggested in this paper can result in an improved estimation of slope reliability. The presented method is particularly useful for landslide risk management in regions with abundant historical landslide data.