Debris flow modelling for enhanced geohazard evaluation along the Bailong River corridor, Gansu, China

Mass movements in rock and soil represent a major hazard in mountainous areas and may be triggered by tectonic activity, through gradual weathering and gravitational forces, or changes in pore-water pressures due to rainfall. Flow-like landslides such as debris flows form one of the most catastrophic mass movements in soil. Debris flows comprise a mixture of water and sediments and are characterised by long run-out distances, large magnitudes, and high velocities. In mountainous areas where habitable areas with low hazard exposure are few and far between debris flows pose a significant threat to lives and livelihoods.

The Zhouqu area of the Bailong River Basin (Z-BRB), located in the southern part of Gansu Province, China covers some 400km². It lies on the eastern margin of the Qinghai-Tibetan Plateau and is characterised by a high topographic relief and a neo-tectonically active environment where human activity and infrastructure are at risk from rockfalls, landslides, debris flows, and flooding. In 2010 a cloudburst event resulted in at least 77mm rainfall falling in 40 minutes over steep mountainous catchments east of Zhouqu. The event generated two massive debris flows that destroyed the central part of the town and resulted in 1756 fatalities. With ongoing climate change, the magnitude and frequency of extreme rainfall events may likely increase, potentially driven by northwards migration of monsoonal rainfall. It is thus important to understand how this will impact the landscape, and how this alters the potential hazards and risks. Thus, it is crucial to better understand the identification source areas of potential debris flow and landslide areas, as well develop better models to evaluate the severity of these events. While disaster risk cannot be eliminated through these activities alone, it is anticipated that the outcomes from this research can contribute to improved hazard and risk management that will enhance the opportunities for local communities to live safely in these dynamic landscapes.

This thesis aims to produce outputs that can be used to better understand the spatial and temporal patterns of debris flows in the Zhouqu-Bailong River Basin, which can contribute to improved risk management strategies. The first step was to identify potential landslide and debris flow catchments through GIS analyses. This includes geomorphological mapping and the use of the automated landscape classification tool geomorphons, which identifies the 10 most common landforms. The geomorphons can then be combined with the Distributed Melton Ruggedness number (DMRN) to provide a map highlighting areas and catchments where both the sediment storage potential and the sediment transport energy potentials are high. These highlighted areas can then form the focus for more detailed modelling and hazard assessments.

In addition to the identification of potential debris flow catchments, a new debris flow model was developed. This model is based on the 2D shallow water equations and presents a novel formulation for bed erosion and deposition. The model was validated against Takahashi et al. (1992) flume experiment and a sensitivity analysis of the key parameters of the model was performed. The outputs from the model were then calibrated against the mapped (and published) footprints of the 2010 Zhouqu debris flow disaster. Once confidence in the model performance was established, a series of debris flow scenarios were modelled comprising 10 different extreme events and 5 longer, low-intensity events. This enabled the investigation of the efficacy of the mitigation measures (dams, channels) and analysis of the changes in channel morphology and dam capacity over time. The modelling results show that the most efficient mitigation measure appears to be the concrete-lined debris channels on the deposition fans that provide efficient conduits for debris exiting the catchments without negatively affecting local communities under most modelled scenarios. Some residual hazards still remain though, and these are illustrated in the form of intensity maps derived from the modelled scenarios. Once additional aspects such as potential event return periods and local vulnerability are established these intensity maps can be converted to hazard maps and disaster risk assessments.