University of Twente Student Theses
Using Urban Morphology and Citizen Science Methods for Mapping Slums Across Different Cities in Africa
Umar, A.R. and Belgiu, M. and Dijkstra, A.M. and Campomanes, F.C. (2024) Using Urban Morphology and Citizen Science Methods for Mapping Slums Across Different Cities in Africa.
|
PDF
7MB |
| Abstract: | The major challenge of rapid urbanisation is the proliferation of slums especially in Low- and Middle-Income Countries (LMICs). Previous studies have demonstrated the ability to map the location and extent of slums using Earth Observation (EO) data and Machine Learning (ML). However, these methods are still limited in spatial transferability partly due to the inter and intra-diverse characteristics of slums. This thesis aims to improve the accuracy and spatial transferability of slum mapping models using an urban morphology-based framework and citizen science methods. Focusing on Nairobi, Kenya, and Accra, Ghana, the study leverages Random Forest (RF) models to distinguish slums and formal areas based on a comprehensive set of morphological features and open geospatial data. Results show that features such as building density, road conditions, and size of buildings are relevant parameters for the identification of slums. Validation through expert evaluation and quantitative analysis indicated high accuracy and precision, with accuracies exceeding 80% in transferring models trained in Nairobi to Accra. Key findings demonstrate that density-related metrics were significant indicators. However, over-reliance on density metrics can lead to misclassifications, especially in high-density formal areas. This limitation highlights the need for incorporating additional data, such as building heights, to improve model performance. We conclude that our proposed methodology is effective for slum mapping and offers a cost-effective and scalable solution for urban planners and city authorities to map slums, contributing towards the achievement of Sustainable Development Goal 11 (SDG 11). This study recommends further testing of the approach in other cities, especially in Africa, with careful consideration of open dataset quality to ensure model reliability. |
| Item Type: | Essay (Master) |
| Faculty: | ITC: Faculty of Geo-information Science and Earth Observation |
| Subject: | 10 humanities in general, 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/102079 |
| 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