University of Twente Student Theses
The Influence Of Urban Morphology On Flood Susceptibility In Slums In A Data Scarce Environment Using Machine Learning
Munyi, Jane-marie Muthoni (2024) The Influence Of Urban Morphology On Flood Susceptibility In Slums In A Data Scarce Environment Using Machine Learning.
|
PDF
8MB |
| Abstract: | The causes of flooding in Nairobi are multifaceted, with climate change and rapid urbanization making flooding of great concern, particularly in slums as they are the most vulnerable settlements. Climate change risks are heterogeneously distributed in cities such as Nairobi, with the urban poor and slum settlements experiencing higher flood exposure and aggravated impacts. Despite slums becoming potential locations for increased urbanization, their association with being ‘illegal’ have made them receive minimal attention from governments, making the problems they face, such as flooding invisible and barely understood since the risks they face are not quantified. The influence of slum morphologies on flooding such as density, spatial distribution and arrangement has been insufficiently studied, with key focus being placed on their locations in flood-prone areas. The overall aim of this research was to (i) investigate the influence of urban morphology by quantifying flood susceptibility and (ii) explore the distribution of flood susceptibility between slums and formal settlements. This aim was achieved following three objectives: identifying morphological flood factors alongside hydrological, environmental and geomorphological Flood Influencing Factors (FIFs), constructing a flood inventory (flood no flood locations) for Flood Susceptibility Mapping (FSM) using Machine Learning (ML) as the last objective, at a grid level of 100 meters by 100 meters. FIFs were derived from Remote Sensing (RS) while urban morphological flood factors were quantified into measurable characters (morphometrics). The flood inventory was generated by combining results from a flood simulation modelled using Fast Flood - a fast browser flood simulation tool, and Citizen Science (CS) flood information. Flood susceptibility was modelled using 2 Random Forest (RF) models by predicting the probability of susceptibility based on morphometric factors and FIFs (Model 1) and solely on FIFs (Model 2). Susceptibility values ranged from 0-1 with values near 1 indicating very high flood susceptibilities. Results from both models exposed disproportional distribution of flood susceptibility between slum and formal settlements. Slums are observed to be highly susceptible compared to formal settlements with median susceptibility values of 0.65 (in Model 1) and 0.5 (in Model 2) for slums and 0.3 in both models for formal settlements. Additionally, the results imply that urban morphology has a significant influence on flood susceptibility as the overall accuracy of Model 1 increased to 84.71% from 71.76% in Model 2 with the inclusion of morphometric factors. The findings suggested that distance to rivers was the most influential susceptibility factor, followed by building adjacency and mean inter-building distance morphometric characters respectively. The space and room left for water to flow and infiltrate as a result of the spatial arrangement in slums and their occupation in floodplains were discovered as the fundamental reasons as to why slums face more flood risks than formal settlements. Considering the bias towards focusing on the effects of floods on development (i.e. exposure), building development regulatory policies such as impact assessments and building arrangement guidelines ought to be formulated to evaluate the influence of development on flooding. Additionally, structural and non-structural measures were also provided as flood mitigation and adaptation measures. Some recommendations for future flood investigations lay stress on the use of CS information and high-resolution data for accurate flood mapping and spatial transferability of the model across geographically diverse regions. |
| Item Type: | Essay (Master) |
| Faculty: | ITC: Faculty of Geo-information Science and Earth Observation |
| Subject: | 10 humanities in general, 38 earth sciences, 74 (human) geography, cartography, town and country planning, demography |
| Programme: | Geoinformation Science and Earth Observation MSc (75014) |
| Link to this item: | https://purl.utwente.nl/essays/102398 |
| Export this item as: | BibTeX EndNote HTML Citation Reference Manager |
Show download statistics for this publication
Show download statistics for this publicationRepository Staff Only: item control page