Advancing Flood Risk Estimates Under Climate Change in Small Island Developing States

Abstract

Climate change is changing the characteristics of high intensity short duration rainfall - a key cause of rainfall-driven flooding, defined as fluvial (river) and pluvial (surface water) flooding. Some of the places most at risk to changes in rainfall-driven flooding are Small Island Developing States (SIDS). Thus far, scientific estimates of changes in rainfall-driven flood risk in SIDS are extremely limited due to a lack of suitable data. This has limited the capacity to implement proactive climate change adaptation, and thus there is a clear need to advance our understanding of rainfall-driven flooding in SIDS under current and future climate change. Depending on the scale of assessment, the state-of-the-art way to model flood hazard changes. As a result, in this thesis the focus is on rainfall-driven flooding in a SIDS context, exploring how at each scale - from local to global – hydrodynamic models can be applied using novel methods to investigate the impacts of changing rainfall characteristics under climate change on rainfall-driven flooding.
In Chapter 2, this thesis first explores the development of a data-rich local scale study in Bristol, United Kingdom (750km2). Although not a small island, this case study acts as a stepping-stone to develop a suitable rainfall-driven hydrodynamic modelling framework using high resolution Convection Permitting Model data so that it can be further explored in a SIDS context in Chapter 3. For the first time, an event set of ~13,500 Convection Permitting Model rainfall events are used to estimate rainfall-driven flooding. We find that when the spatiotemporal rainfall characteristics are represented using an event set approach, flood hazard estimates are higher than using a more traditional change factor approach (19-49% depending on return period). This suggests including the full spatiotemporal rainfall variability and its future change in rainfall-driven flood inundation modelling is critical for future flood risk assessment.
Chapter 3 scales the novel methods demonstrated in Chapter 2 up to the national scale in Puerto Rico (9100km2). An event set of synthetic tropical cyclone rainfall events are modelled spatiotemporally to estimate population exposure to flooding from hurricane rainfall in Puerto Rico for the present-day climate, which is approximately 8–10% of the current population for a 5-year return period, increasing by 2–15% and 1-20% under 1.5°C and 2°C futures. Validation of our model against High Water Mark data for Hurricane Maria demonstrates the suitability of this model when the rainfall spatial characteristics are considered, echoing the importance of its inclusion in the estimation of rainfall-driven flooding. This chapter also highlights a potential increase in flood risk in Puerto Rico, even under the 1.5°C Paris Agreement goal.
In Chapter 4, a high-resolution global flood model (~30m) is utilized to assess population exposure to flooding across all SIDS using a change factor approach. This has the advantage of facilitating the estimation of coastal flooding alongside rainfall-driven flooding in all SIDS – but using globally-available data means there is a trade-off in data availability and rainfall representation in comparison to the local and national scale methods in Chapters 2 and 3. Our analysis shows that present day population exposure to flooding in SIDS is high (19.5% total population: 100-year flood hazard), varies widely depending on the location (3-66%), and increases under all three climate scenarios - even if global temperatures remain below 2°C warming. We also demonstrate that population exposure and flood hazard are not strongly linked, and global vulnerability and risk metrics do not do a good job at representing the climate change impacts on flooding in SIDS.
Overall, this thesis advances our understanding of flood risk in Small Island Developing States from the local to global scale. It stresses the importance of representing rainfall spatiotemporally and using an event set approach when estimating rainfall-driven flood hazard. Although there are limitations and trade-offs, assessment from the local to global scale can be used to inform adaptation to flooding in different ways. Given the monumental task of adapting to a rapidly changing climate, improving flood risk estimates across all these scales is necessary to support proactive adaptation in Small Island Developing States.

Date of Award	10 Dec 2024
Original language	English
Awarding Institution	University of Bristol
Sponsors	GW4+ NERC DTP
Supervisor	Jeff Neal (Supervisor) & Paul D Bates (Supervisor)