Estimating flood risk under climate change

Abstract

Flooding is one of the most common natural hazards, causing disastrous impacts worldwide. Flood risk may significantly change under climate change, and understanding these potential future changes is critical to ensure adequate adaptation measures are implemented. Quantifying the uncertainties associated with model outputs and developing strategies to create valuable information from such uncertain data is crucial to advance our understanding and to make robust decisions. This thesis is centred around addressing three inter-related challenges associated with estimating flood risk under climate change: 1) understanding uncertainties in regional scale discharge estimates, 2) comparing different approaches to impact assessments (top-down vs. bottom-up), and 3) understanding drivers of future flood hazard under a hotter, drier climate with increasing intensity of extreme rainfall.

First, this thesis evaluates the uncertainties related to estimating extreme flood magnitudes using gauged-based methods and global hydrological models. We found that there are clear spatial patterns of bias associated with each method. Global hydrological models (GloFAS, PCR-GLOBWB, CaMa-Flood, WorldWideHYPE) were found to tend to underpredict extreme flow magnitude in catchments at low elevations, and overpredict in catchments at high elevations. The opposite pattern was found for the gauge-based methods. Second, we use the inundation extents from a global flood hazard model analyse the sensitivity of flooded areas and population exposure to variability in flood magnitudes. We found that topography and drainage areas correlated with flood sensitivities. We also found clear settlement patterns, in which floodplains most sensitive to flooding from frequently, low magnitude events had an even distribution of exposure across hazard zones. In contrast, floodplains most sensitive to flooding from extreme magnitude events had a tendency for populations to be most densely settled in these rarely flooded zones. Third, we explored climate change impact on high river flows using a high-resolution ensemble of convection-permitting climate projections to force a hydrological model and compared with operational guidance produced using a scenario-neutral, sensitivity approach. We found that estimates of future high river flows were very different depending on the approach used, in some cases as much as a factor of two.

Overall, this thesis advances our current understanding of uncertainties in estimating flood hazard and how to use this uncertain information for impact assessments. It stresses the importance of a probabilistic approach to flood hazard modelling and highlights the opportunities of new convection permitting models. Future work should focus on incorporating vulnerability into sensitivity analyses to improve understanding of how societies co-evolve in response to flood hazard.

Date of Award	23 Jan 2024
Original language	English
Awarding Institution	University of Bristol
Supervisor	Jeff Neal (Supervisor), Gemma Coxon (Supervisor) & Thorsten Wagener (Supervisor)