Upscaling coastal flood risk assessment - Insight into the uncertainty of a continental-scale model

Upscaling coastal flood risk assessment - Insight into the uncertainty of a continental-scale model

Bouziotas, D.

Kok, M. (mentor)
Bouwer, L. (mentor)
Giardino, A. (mentor)
Van Rooijen, A. (mentor)
Van Vuren, S. (mentor)
Vousdoukas, M. (mentor)

Civil Engineering and Geosciences

Hydraulic Engineering

2016-04-29

During the last years, flood risk assessments at continental and global scales are emerging as useful tools for risk hotspot identification, support for international policy-making and harmonization of strategies for climate change adaptation. The emergence of these models comes to cover the growing demand for integrated, large-scale risk assessments in fluvial and coastal settings, requested by actors such as international organizations and insurance companies. These stakeholders are alarmed by the increasing prospects of flood risk, driven by climatic and social variability. Despite their eminent usefulness and promising nature, these large-scale models are strongly limited by high levels of simplification in terms of process detail, hazard and inundation methodologies and outcome detail. This aspect of epistemic uncertainty (i.e. uncertainty due to lack of knowledge and modeling abstraction) in the flood risk assessment chain remains largely unaccounted for in continental and global scales. Uncertainty quantification is, therefore, needed at the large scale, as a means to communicate risk more effectively and increase model validity. In light of this, this study aims at providing insight on the underlying epistemic uncertainty of a continental-scale coastal flood risk model. Having LISCOAST, an integrated, coastal flood risk assessment framework for Europe currently in development at JRC, as a basis and following the definition of flood risk as a product of hazard, exposure and vulnerability, this thesis develops an analysis framework that examines multiple sources of epistemic uncertainty in the risk assessment chain. The developed algorithms examine four main sources of epistemic uncertainty, namely the inundation algorithms, the interaction of different hazard components, the way flood defenses are modeled and the assumptions behind the use of depth-damage functions. The impact of these sources to the modeled Expected Annual Damage (EAD), for present and future scenarios, is evaluated in a case study that expands upon two scales: a regional application in Faro, Algarve, Portugal and an international application in the Iberian Peninsula, covering the coastline of Portugal and Spain. The developed analysis framework is broad and readily able to be generalized in the native model scale, i.e. in the European coasts. The quantitative results of such an analysis serve a twofold purpose: not only can they lead to major improvements in future versions of LISCOAST, but they can also give general recommendations for improving the design approach of large-scale coastal flood risk models and communicate their potential and limitations more effectively. At the same time, they act as a point of departure to spark discussion in the risk modeling community on the large uncertainties that underpin large-scale flood risk assessments.

coastal flood risk
uncertainty
sensitivity analysis
risk assessment
direct damage
Europe

http://resolver.tudelft.nl/uuid:06e553ce-f491-4bf3-badd-58978f4fe7ac

Student theses

master thesis

(c) 2016 Bouziotas, D.