Research On Rainfall Thresholds For Forecasting Unsaturated Soil Slope Failure Based On Mechanical Model

Landslides have now become the second largest geological disasters, precededonly by earthquakes. As rainfall is one of the most recognized triggering factors for thisdisaster, the effect of rainfall on slope failure has attracted many scholars’ attention inrecent years. Currently research in this area is mainly divided into two categories: one isto establish empirical rainfall threshold based on rainfall-induced landslide data in thepast, which could be directly used for landslide warning; the other is to research theinfluence of rainfall on slope stability by tests or numerical methods, which is toprovide basic conclusions for landslide comprehensive warning. Plenty of results havebeen achieved in these researches and playing an important role in landslide warningsystem.The idea of this study could be regarded as a combination of those two types ofresearches mentioned above. This study established rainfall thresholds for landslidewarning by numerical methods, and furthermore modified the expression of rainfallthreshold to achieve better warning efficiency. Compared with previous studies, singleslope is taken as the research objective instead of slopes of certain areas. This wouldinevitably lead to more measurements and calculations. But with continuousimprovement of geological survey reports and rapid development of numerical methods,the additional cost for single slope research would definitely be covered by theincreased warning efficiency. Moreover, researches on single slope would be muchmore meaningful for landslide warning along transportation lifeline.The method used in this study was calculating the slope failure rainfall thresholddata in advanced mode of finite element program Plaxis2D, in which characteristics ofrainfall infiltration and unsaturated soil were considered. I-D curve is the most widelyused expression of rainfall threshold. In an attempt to modify the expression, rainfallintensity was averaged in separated periods (two periods in this study) instead of thewhole rainfall duration, which was used in I-D curve. Results showed that expressionwith averaged intensity in separated periods had made an obvious improvement. It wasmuch more realistic when describing the influence of rainfall on slope stability,compared with the previous expression. On this basis, this study redefined antecedent rainfall with a clearer physical meaning and proposed the calculation formula todescribe the first period of rainfall. When antecedent rainfall was added in I-D curve asthe third parameter, I-D curve would be developed to A-I-D rainfall threshold surfacewith higher warning accuracy. This study introduced the establishment and verificationof rainfall threshold surfaces with slope models of two representative geologicalstructures, homogeneous soil slope and slope on inclined bed rock. Lastly, this proposedrainfall threshold surface was applied in landslide warning system of Road No.1inRepublic of Congo. The results turned out to be persuasive.