Mapping the Missing spaces and the role of Generic Slum Ontology

Urbanization in the Global South is often characterized by the proliferation of deprived neighborhoods (frequently referred to as slums). The reduction of the proportion of people living in slums is key for inclusiveness of urban areas and development and is specifically targeted by policies such as the SDG goal 11, which aims to “make cities and human settlements inclusive, safe, resilient and sustainable”. Consistent global information about the amount and spatial distribution of slums across cities in the Global South is needed to inform these policies and track progress. Unfortunately, such efforts are hampered by lacking, inaccessible, or outdated data. There is also often conceptual ambiguity of what is understood as a slum, informal settlement, or deprived neighborhood. There is a wide diversity in their appearance and perception within a single city, as well as at a global scale. To address this, we use the generic slum ontology (GSO) [1] and available spatial data to seek whether robust and transferable indicators with regional characteristics can be identified for global slum mapping efforts. The initial results of our analysis demonstrate that indicators such as building density and road characteristics in an image are potentially useful to describe differences between slum and non-slum built-up areas in the case-studies. This study highlights the opportunities of the GSO for the development of a global slum repository but also show the need of local adaptations and hence, the importance of the conceptualization of real-world features into image domain features. This understanding could be useful to upscale current algorithms. Further, we describe the gap between the geospatial data products developed in the remote sensing community and the information needed by policymakers and other user-groups. We discuss why an objective and transparent system for monitoring slums is needed to monitor global development goals as well as support local communities and NGOs. This becomes even more important in the time of global crisis like the spread of COVID-19 pandemic.
Keywords
urban planning, slums, informal settlements, poverty mapping, remote sensing, slum ontology; urban morphology
1. INTRODUCTION
Accelerated urbanization in many Global South regions and the low capacity of the housing market to provide affordable housing to low-income groups contribute to the growth of slums (SDG indicator 11.1.1). Accurate, comprehensive and up-to-date spatial information of such areas, as well as their evolution at city scale, are needed for local decision-making and to support pro-poor development strategies. However, data are often inconsistent, outdated, or unavailable. Slums can be considered as “missing spaces” as they are often not mapped or not included in official maps. In most countries, administrative definitions or income-based indicators are used to differentiate urban slum and non-slum areas. Remote sensing (RS) studies have shown the capability of satellite imagery to provide consistent and timely information on the location and physical dynamics of slums [2]. Developments of the last years indicate the development of object-based [3] as well as machine-learning-based methods, in particular deep-learning, to provide technological solutions [4]. For example, combining RS and local (non-official) data with deep learning models to extensively map, explore and understand the spatiotemporal dynamics of slums with a temporal granularity adapted to the local slum dynamics (ranging from a few months in cities of very high dynamics to one or two years). However, most studies fail to provide a clear definition and operationalization of such areas.