Theoretical simulations of debris flow and their applications to hazard mapping using GIS

Abstract

Debris flow is a dangerous form of natural landslides because of its unpredictable nature of occurrence. Recently, the risk of debris flow in Hong Kong increases tremendously due to the rapid development of rural areas next to steep terrains. For example, the steep terrains along northwestern boundary of Tuen Mun, a suburban area in the New Territories of Hong Kong, are covered by a layer of residual soil consisting of boulders or by colluvium. Within this area, the largest reported debris flow in Hong Kong occurred on September 11, 1990. About 2.5km north of the 1990 Tsing Shan debris flow, two rainfall-induced debris flows occurred on April 14, 2000 reaching the nearby Leung King Estate. Such events threaten lives and properties of the citizen of Hong Kong. This thesis, therefore, intends to propose a method to determine the potential hazard zone of debris flow by incorporating numerical simulations with geographic information system (GIS). This method is termed flow-dynamics-based approach. For illustrative purposes, Leung King Estate is studied in this thesis. The GIS facilitates to handle massive information for hazard mapping. Debris flow mechanisms are included implicitly in the debris-flow hazard mapping through the numerical simulations of debris flow. This thesis adopts the flow dynamics model proposed by Takahashi et al. (1992) for the aforementioned purpose. However, the model is found not resulting in satisfactory runout distance of the 1990 Tsing Shan debris flow. Thus, this thesis proposes a new onset criterion of erosion by using the minimum stream power concept to modify the model and also makes modifications on the stopping criterion of the model. The modified model gives much favorable results comparing to field observation. The parameters of the modified model for simulating debris flows in Hong Kong are calibrated primarily by using the record of the 1990 Tsing Shan debris flow as well as supplemented with parameters observed or adopted elsewhere. Finally, this thesis proposes a new method for generating a more reliable flow-dynamics-based GIS hazard map by incorporating debris flow mechanisms. Thus, no so-called expert opinions, which are bounded to be subjective and biased, are needed. In addition, the application of the present study to the design of protective barrier against debris flow is also illustrated. Although various improvements still can be made, this new approach should open up a direction for further research.