- Caroline Gevaert, Divyani Kohli, Monika Kuffer, Karin Pfeffer – University of Twente, The Netherlands.
- Michael Wurm, Hannes Taubenböck – DLR, Germany.
Addressing the phenomenon of a “slum” is a major priority in urban planning in cities of the Global South. However, information on the location and dynamics of slums is often outdated, not available or highly uncertain (e.g., due to local politics – under-reporting). If the information is available, it is not in hand or shared with local communities. Remote sensing can support having updated and consistent information on the location and boundary of slums. Such information can support development plans and provide important basemaps for the design of upgrading projects and infrastructure interventions. Slums are often typified as consisting of small, densely built dwellings of poor construction material and separated only by narrow footpaths. Such physical features have typically been targeted by researchers to identify the differences between formal and informal areas. Methodological developments such as deep learning and better processing power may support this. Other developments include global initiatives to map urban areas and estimate population densities using machine learning methods; but also how to include non-traditional information such as tweets to incorporate the socio-economic dimension into slum mapping.
However, there are also some challenges and uncertainties regarding slum mapping. One persisting uncertainty regards the ambiguity regarding the term “slum” itself. The official UN definition includes socio-economic factors such as population density, access to safe drinking water, sanitation, and legal tenure in addition to the physical characteristics captured by remotely sensed imagery. Even when only focusing on the physical characteristics – models developed on slums in one area are not transferable to other areas due to the physical heterogeneity.
When successful strategies for slum mapping are developed, new questions can be raised. For example, how can these models be used to understand the temporal dynamics of slums? How do they change and can predictions be made to support urban planning? Also, who should have access to these maps? What are the ethical implications of distributing such potentially sensitive data?
In this special session on slum mapping at JURSE 2019, we invite authors to present their work regarding the following themes:
- State-of-the-art in slum mapping
– Current best practices
– Global mapping initiatives and data sources
– Non-image data sources: Twitter, night lights, household surveys
- Challenges and uncertainties
– Difficulty in defining a slum
– Technical difficulties (i.e., what do we need to improve the classification accuracy)
- Next steps
– Scaling up and model transferability
– Temporal aspects
- Ethical implications