Research School for Socio-Economic and
Natural Sciences of the Environment
Research School for Socio-Economic and
Natural Sciences of the Environment

Bastian van den Bout

Date: 11 September 2020
Location: Enschede

Dissertation title: 

Integrated physically -based multi-hazard modelling

Group: Department of Geo-information Processing, University of Twente
Promotor: prof. dr. V.G. Jetten 
Co-promotor: prof. dr. C.J. van Westen

Natural hazard occur in many varieties throughout the world. Hydrometeorological hazards such as floods, landslides and debris flows are frequently induced by a common trigger such as extreme precipitation. When these process occur simultaneously, interactions can occur that can alter the dynamics of the event. Examples of such interactions are landslide dam formation in rivers, where deposition of mass movement material forms a natural dam that blocks a river, or increased mobilization of solids by merging of debris flows and floods. Both for understanding, reconstruction and prediction, physically-based modelling of hazards is a common and valuable tool used in hazard and risk assessment. Then, based on hazard intensities and probabilities, decisions to reduce disaster risk can be made. Single-hazard models, such as models for flooding, mass movements or storm surges are widely used for this purpose. These tool focus on isolated hazard and ignore interactions. For events with multi-hazard interactions, integrated multi-hazard models are required. In this doctoral dissertation, an integrated physically-based multi-hazard modelling tool is developed for use in hazard assessment. To do this, an existing opensource hydrology, runoff and erosion modelling tool is further developed to incorporate a variety of hazardous processes such as (flash) floods, slope failures, mass movements and entrainment of bed material. The first phase to the development of the multi-hazard modelling tool is a comparative implementation of three distinct flow models for catchment-integrated flood simulation. Both kinematic, diffusive and dynamic wave flow are calibrated on a series of events in three study areas in China, Spain and Italy. Results showed integration in catchment-scale hydrology allowed for new types of interactions between these processes that were valuable in reconstruction of the Fella flood event. The second phase developed a novel method for efficient, regional simulation of slope stability and failure volumes. The iterative failure method was developed that is able to link oneto-one with hydrology and flow aspects of the simulation. This allows for a variety of interactions and feedbacks between upslope hydrology and runoff patterns and slope stability. The third and fourth phase investigated the implementation of generalized adaptive mass movement equations, and entrainment by such flows. An existing set of generalized equations was adapted to cover shallow water flows. The equations describe flow in a two-phase manner, with distinct velocities and interfacial forces such as drag, virtual mass and viscous forces. The final set of equations automatically and continuously scales internal forces based on the consistency and properties of the flows materials. Using this approach, flow types are automatically estimated and can evolve during interactions. The model was calibrated and validated on the 2009 convective storm event in Southern Sicily, where hundreds of landslides and severe flash flooding interacted in the Scaletta area. Additionally, a reconstruction was made for the 2008-2010 Honchun co-seismic landslide process chain in central China. Both events highlight the additional insight that can be gained from integrated multi-hazard modeling, but stress the need to deal with uncertainties and parameterization complexities. The fifth phase of this work extended the set of generalized mass movement equations to include semi-structured material dynamics. By implementing a full stress-strain relationship combined with two-phase non-Newtonian generalized debris flow equations, an extensive new model was developed. In order to better support the complexities involved in solving the equations, a new depth-averaged variation on the material point method was implemented. Fluids were solved in a discrete Eulerian grid, while solids were solved using the smooth particle mathematical framework. Interactions between the two phases were solved using the gridding techniques used in the material point method. Comparison to flume experiments show high likeness of fracture patterns and final deposits for cohesive blocks of organic-rich clay. Finally the applicability of physically-based multi-hazard modelling tools to hazard assessment is investigated through even reconstruction, scenario exploration and ensemble analysis on the 2017 hurricane Maria event on Dominica. It is shown that, while integrated multi-hazard modelling provides several clear benefits to process understanding and hazard accuracy, model parameterization and uncertainties remain a hindrance to full application in hazard and risk assessment. Additionally, traditional assumptions considering probabilities of events and their Summary iii Integrated Physically-Based multi-hazard modelling relationship with trigger intensity can be broken due to the non-linear complexity of multi-hazard interactions. By either using ensemble analysis or smart scenario design, a proper framework must be made that can act as a foundation for application of integrated physically-based multi-hazard modelling.