Mosimann, Markus (2024). From Weather Forecasts to Impact-Based Flood Warning Systems - a Modelling Perspective. (Thesis). Universität Bern, Bern
|
Text
24mosimann_m.pdf - Thesis Available under License Creative Commons: Attribution (CC-BY 4.0). Download (19MB) | Preview |
Abstract
Climate change is exacerbating devastating flood events and has thus increased the critical need for effective flood risk management strategies. Traditional flood warnings often fall short in communicating specific impacts to communities, leading to poor preparedness and weak responses. This PhD thesis addresses this shortcoming by enhancing the early warning framework for fluvial floods in Switzerland. Translating weather forecasts into impact forecasts requires (i) the development of vulnerability models, (ii) the capability for near real-time computation of hazards and their impacts, and (iii) comprehensive understanding and integration of uncertainties. Finally, it needs (iv) cartographic and graphical visualizations to transfer the modelling perspective, namely the impact assessment derived from weather forecasts to decision-makers. The first paper in this thesis focuses on the often-overlooked aspect of household contents damage. This paper presents two regression models that analyze insurance claim records to estimate monetary damage and the degree of damage to household contents from that to building structure. These models are robust and enable the degree of damage model to be transferred as well. They are therefore valuable tools for improving the estimation of financial flood impacts. The findings underscore the significant contribution of contents damage to total building damage and challenge current flood damage assessments to broaden their scope beyond structural damage. The second paper introduces a novel library-based surrogate flood model designed for real-time impact-based warnings. The surrogate model was tested in northern Switzerland’s river and lake network and benchmarked against high-resolution transient flood simulations with reduced computational efficiency, and it demonstrates a high agreement with flood impacts assessed for buildings, people, and workplaces. In nine scenarios derived from hindcast archives, the surrogate model achieves a critical success index for area between 0.74 and 0.90 and for exposed people between 0.77 and 0.93 and thus proves its potential for real-time flood impact prediction at national scale. This optimization of the trade-off between spatial resolution and computational efficiency signifies a substantial advancement in impact-based flood forecasting. The third study explores the quantification of impact sensitivity to changes in flood magnitudes in floodplains. This paper introduces the floodplain sensitivity index (FSI), which integrates slope and curvature metrics, to provide a nuanced understanding of how floodplains respond to various flood scenarios. An analysis with the FSI demonstrates the nonlinearity and region-specific nature of the relationship between flood magnitude and impacts. The FSI aids in identifying significant magnitude thresholds both for short-term impact-based warnings and for and long-term climate-change-influenced flood risk mitigation. It defines ranges with high or low sensitivity of impacts to increases in magnitude and enhances the formulation and precision of impact-based warnings. Fourthly, by integrating and visualizing the models and methods developed in these studies into web tools, this thesis contributes to risk communication and shows the potential of web-based solutions for impact-based flood warning systems to bridge the gap between hydrometeorological forecasting and practical flood risk mitigation. In conclusion, this doctoral thesis introduces innovative modelling approaches that enhance impact-based forecast and warning services and provides valuable insights into critical aspects of flood risk management to help face the challenges of an anticipated increase of extreme flood events with severe impacts. The modelling perspective thus allows model chains to be extended from weather forecasts to impact forecasts. Impact forecasts can only be made with models for dynamic hazard and impact assessments that are coupled with weather forecast models. This perspective also facilitates the evaluation of simulation techniques for real-time applications. Besides addressing unresolved scientific questions, the focus on Switzerland and the web-based communication of its findings illustrates the practical applicability of this approaches , which aligns with the latest standards in web-based technology to expand the current early warning framework in Switzerland. Keywords: Impact-based flood warnings; flood vulnerability assessment; real-time flood forecasting; floodplain impact sensitivity; web-based flood impact visualization; flood risk management
| Item Type: | Thesis |
|---|---|
| Dissertation Type: | Cumulative |
| Date of Defense: | 28 May 2024 |
| Subjects: | 500 Science > 550 Earth sciences & geology 900 History > 910 Geography & travel |
| Institute / Center: | 10 Strategic Research Centers > Oeschger Centre for Climate Change Research (OCCR) 08 Faculty of Science > Institute of Geography |
| Depositing User: | Sarah Stalder |
| Date Deposited: | 25 Jul 2024 15:08 |
| Last Modified: | 12 Aug 2024 07:42 |
| URI: | https://boristheses.unibe.ch/id/eprint/5327 |
Actions (login required)
 |
View Item |