Detecting social groups from space - Assessment of remote sensing-based mapped morphological slums using income data

Abstract

Over the last decades, massive urbanization processes lead to the emergence of large slum areas making them home to about a seventh of the global population. Although the variety of morphological characteristics varies significantly within as well as across cities, common determinants exist. Informal, or unplanned settlements in particular, do show similar morphologies over the world. They are characterized mostly by extremely high building densities and small building sizes, irregular arrangement of buildings and street network and are often located at exposed sites . Based on these characteristics, we deploy satellite images for a systematic mapping of morphological slum areas in the city of Rio de Janeiro, Brazil based solely on physical characteristics and analyse the mapping result with the official census data. Outcomes show first that morphological slums are a semantic and spatial sub-group of all slum areas contained by the Brazilian census and that remote sensing-based mapping yields accuracies of almost 94%. Second, analysis of census-based income data proofs that while almost 45% of all mapped slum blocks are characterized by incomes below the poverty line, as defined by the Organisation for Economic Co-operation and Development (OECD), this holds true for only about 6% of the formal urban neighbourhoods.