Slums in Rio de Janeiro. Spatial and temporal analyses of slums derived from remote sensing data based on visual image interpretation

Kurzfassung

In developing countries the share of people living in slums is rising, therefore also the number of slum dwellings is rising. As this is an illegal process it is nearly impossible to monitor this process and also very difficult to generate accurate data about quantity and quality of housing. Moreover, the lack of a general definition of slums complicates the whole thematic. Due to the given reasons this study derived polygons of two different slum categories in the metropolitan region of Rio de Janeiro based on visual interpretation of the Esri World-Imagery-basemap. During this process a total of 725 individual slums covering an area of nearly 30 km² were digitized. 577 slums were classified as Favelas and Invasoes based on clear morphologic characteristics such as heterogeneity, cheap roof materials and very high building densities. The remaining 148 slums were allocated to the Loteamenton category, showing a more regular pattern and visible streets. Afterwards, the estimated time step of slum establishment and the average slope were derived from urban footprint data and the ASTER DEM. Data about the building density and average building height were generated during the visual interpretation of remote sensing data with the help of classifying the values into five and three categories. Afterwards, the collected data were analysed with the aim of confirming the classification of the two mapped slum categories based on quantitative data. For both slum categories applies that, in contrast to other Latin American mega cities, the majority of slums is not located at the peripheral regions of the metropolitan region of Rio de Janeiro. Both categories show a spatial concentration near the centre of the study area while the concentration of Favelas and Invasoes has a much higher extent. Furthermore, also the newer slums of the Favela/Invasoe category are located near the centre of Rio de Janeiro. In terms of their morphologic characteristics the two slum categories show very huge differences. On average, individual slums of the Loteamenton type are larger than Favelas and Invasoes while the Loteamenton blocks tend to consist out of higher buildings. Considering the remaining two characteristics, blocks of the first slum category are denser and definitely tend to be located at more disadvantaged locations regarding the average slope. Thus, it is the first category which is more vulnerable to natural hazards such as landslides. In summary, this study successfully confirmed the classification of the two mapped slum types based on quantitative data. It turned out that in terms of every analysed morphologic characteristic both slum categories show distinct features. Furthermore, the processes of densification and verticalization could be proved for all slums of the study area considering the temporal development of the building density and average building height. By reason of the above presented results, this study shows that slum areas can be derived from visual interpretation of remote sensing data. The resulting data set is very consistent probably only missing some very small slum blocks that were not capable of being differentiated from the surrounding formal settlements due to their very small size. But the data set can also be improved by analysing VHR data from new satellites such as Terra-SarX or Spot 7 which both have a spatial resolution of up to one metre. Furthermore, the given research can be extended by deriving morphologic patterns from textures. This was in part already been done for different cities using different approaches that are image based (Graesser et al. 2012: 1164ff., Kuffer et al. 2013: 1ff.), object based (Hofmann et al. 2008: 531ff.), lacunarity based (Kit/Lüdeke 2013: 130ff.) or based on texture analyses using Kennaugh-elements (Weigand 2014: 4ff.). If the same approach is used in different cities around the world, they will be comparable in terms of e.g. the spatial distribution of their slums or their morphologic characteristics. In addition this study has to be continued so that the temporal analyses are consistent. This provides the possibility to analyse whether the described trends are continuous or not.