ABSTRACT:

The Geotechnical Engineering Office keeps comprehensive rainfall and landslide data. A GIS database of all sizeable man-made slopes and retaining walls in Hong Kong is also available. Apart from basic information on slope geometry (height and gradient), the database contains other pertinent information. By means of statistical analysis of the data, good correlations between slope failure rate, maximum 24-hour rainfall and slope information have been established for soil cut slopes for a pilot study area. The correlations have been used to estimate the expected number of landslides that would have occurred in past rainstorms. The estimated numbers compare favourably with the recorded number of landslides in the rainstorms.

INTRODUCTION

Hong Kong has a long history of landslides, the majority of which were triggered by rainfall. In order to warn the public of landslide risk during periods of heavy rainfall, the Geotechnical Engineering (GEO) in conjunction with the Hong Kong Observatory (HKO) operates a Landslip Warning System. For operating the System, a good understanding of the relationship between rainfall and landslides and other factors affecting slope stability is essential. Various approaches have been adopted by investigators in the past to study correlation between rainfall and landslides (e.g. Brand et al, 1984 & Au, 1993). Constrained by the information that was readily available, the studies carried out in the past were all based on information on reported landslides, and data on the large group of slopes that did not fail in rainstorms were not considered. The results of the past studies therefore could not give a realistic assessment of the effects of rainstorms hitting different geographical areas which have different number of man-made slope features of different conditions. This paper describes a pilot study of the correlation between failure rate, rainfall and slope characteristics for soil cut slopes.

This content is only available via PDF.
You can access this article if you purchase or spend a download.