City-scale Modelling of Factors Affecting Urban Pluvial Flood Hazard in Rapidly Developing Cities Using Global Data

Flooding is the most prevalent disaster worldwide accounting for 43% of all recorded global disaster events in the past 20 years leading up to 2018 Choy (2018). While migration from rural settlements to urban areas often mirrors economic advancement, it also presents socioeconomic and environmental challenges. Rapid urban growth strains existing infrastructure and also discourages the preservation of natural habitat in favour of building more developments causing urban flooding. Climate change and urbanisation have been reported as the major contributors to the increasing damaging effects of flooding to lives and livelihoods worldwide (Aerts et al., 2014). There is lack of adequate research focused on the dual impacts of climate change and urbanisation on urban flooding and water quality in rapidly developing urban areas of the world – a gap that will result in an increase in fluvial and pluvial flood risk, and further reduction in water quality (Miller and Hutchins, 2017).

This study highlights the importance of the use of free global datasets in the development of a city-scale 2D hydraulic model that assesses the impact of land use change and climate change on urban pluvial flooding in rapidly developing cities. This thesis presents three key results chapters assessing the ability of a simplified city-scale hydrological and hydraulic models to estimate urban pluvial flood inundation in a large catchment, before going on to establish the impacts of climate change and land use change on flood hazard. Topography has been identified as a key dataset of estimating flood extent (Horritt and Bates, 2001) and models of flood extent rely on DEMs in order to simulate paths of water flow, flood extent and depth. Errors in Digital Elevation Models (DEMs) can substantially affect the results of flood models (Stephens et al., 2012, Hawker et al., 2018).

Therefore, in order to increase the accuracy of the outputs from the hydrological and hydraulic models used in this study, a methodology for the correction of building error in DEMs was developed in Chapter 4 for removing building elevation artefacts from six global DEMs namely: (i) NASADEM, (ii) SRTM, (iii) MERIT, (iv) ALOS, (v) TanDEM-X 12 m, and (vi) TanDEM-X 90 m. The findings show that the removal of building elevation artefacts/error from global DEMs resulted in the improvement of the vertical height accuracy of global DEMs. The findings show that building density has an influence on vertical accuracy of global digital elevation models (DEMs).

This finding was a key step prior to research undertaken in subsequent chapters. In chapter 5, a city-scale hydrological and hydraulic model of the Nairobi catchment is built. The purpose of creating the models is to use the raw DEM and corrected global DEM derived in Chapter 4 to estimate the impact of land use change and climate change on urban pluvial flood hazard at city-scale level using global datasets.
The HEC-RAS software is used to create five categories of models for the extreme rainfall event of 1st to 13th of March 2018; a baseline model (S1-Baseline), S2-2000LU model, S3-CP4uplift model, S4-P25uplift model, and the S5-RawDEM model. The five sets of models are created in 2D and make use of the diffusive wave equation for simplification. The results showing a lot of promises by providing evidence for the hypothesis that urban flood models built at city-scale level using free global datasets have a good level of skill and are proficient enough to accurately estimate urban flood inundation depth and extent in rapidly developing cities characterised by sparsity of data.

Chapter 6 discusses the results of the flood inundations and flood hazard vulnerability maps from the HEC-RAS 2D hydraulic model under 5 different scenarios. It is found that topography plays an important role in flood inundation maps and that the accuracy of flood inundation maps can be improved simply by using urban corrected DEMs over raw DEMs as key input data when conducting both hydrological and hydraulic modelling. The findings also show that urban pluvial flooding is affected by both change in climate change and land use change, however, climate change is found to contribute significantly to surface water runoff and exacerbate the problems of urban flooding. The results also show that the baseline model using the urban corrected DEM as input data produced flood inundation and flood hazard vulnerability maps with better accuracy in comparison to a similar baseline model using the raw DEM as key input data.

Chapter 7 further explored the influence of climate change and land use change that is due to rapid urbanization on urban pluvial flood hazard. Chapter 7 focuses on the synthesis of the results obtained from the results of the 5 set of 2D hydraulic models discussed in chapter 6. Results demonstrated that climate change had more influence on urban pluvial flood hazard than land use change. It is found that climate change, rather than land use change is a bigger threat to urban area in terms of flood risk. Specifically, the effects induced by climate change under the CP4 and P25 climate rainfall models are much higher than the effects induced by land use change due to urbanisation in Nairobi from 2000 to 2020.

The findings show that the changes caused by current and future changes in rainfall intensities and frequencies are most likely to render most large urban areas vulnerable to extreme rainfall and pluvial flooding due to lack of resilience in existing drainage infrastructure and flood mitigation systems. Assessment of land use changes alone cannot fully account for hydrological and hydraulic alterations in the urban context and it is important for policy makers and people with responsibilities for managing urban flood risks to consider adaptation and mitigation strategies that considers increasing threat of urban flooding emanating from increased runoff from climate change rainfall.

This thesis has subsequently enhanced our understanding of the value of free global hydrological and hydraulic models developed at city-scale to model the impacts of climate change and land use change on urban pluvial flood hazard in data-sparse context of rapidly developing cities where availability of high-quality data for urban flood studies are a rarity. Finally, one of the key findings of this study is that in the context of conducting urban flood modelling in data sparse regions in rapidly developing cities across the world, it is possible to leverage the opportunities provided by the growing availability of free, global datasets to develop urban flood models. Traditional urban flood models rely on the use of high-quality datasets as key input data and require computers with high computational efficiency to run detailed flood inundation models. Most importantly, the study has demonstrated that it is possible to achieve a trade-off between complexity and resolution by the use of simplified 2D hydraulic flood models that use global dataset as key input data.