The Blind Spot – Reducing Knowledge Gaps In Urban Poverty With Earth Observation Interacting With Structured And Unstructured Geodata
We live in a period of one of the largest human migrations ever: the shift from rural to urban environments, resulting in an urbanized world (UN, 2018). We also live in times where more data is available than ever before in human history. So should we not have the necessary data available to understand these processes of global urbanization?
How is it still possible that the World Migration Report (2015) states in 2015 that we still face major knowledge gaps on probably the largest group of people pushing into cities – the poor? Is this based on a real lack of data, or are we less interested in this phenomenon?
First of all, yes we face a real lack of data. Common data sources on the urban poor such as censuses are often outdated, inconsistent, not available with corresponding spatial accuracy or thematic detail, or they are just not available at all. Moreover, the credibility of these data sources is challenged by researchers, stating that especially the poorest often remain invisible in these data (e.g. Tacoli, McGranahan & Satterthwaite, 2015).
Earth Observation data for approaching urban poverty
So, we need new data and new methods for providing new perspectives to gain a more comprehensive knowledge of dimensions, spatial patterns, and processes related to the social group of the urban poor. “Distance brings clarity” is a common saying, and it perfectly fits the distant view as the Earth is monitored by remote sensing satellites. The built environment captured in these data can contain an expression of inequality in cities and, by this means, socio-economic disparities even become visible from space. Organic, amorphous, complex, and dense seas of makeshift shelters have significantly different physical appearances than formal, planned parts in cities (Taubenböck & Kraff, 2014).
Figure 1 illustrates this appearance of a slum area in Mumbai India in direct comparison to formal settlements. Beyond, a remote sensing based classification in the level of detail-1 reconstructing the buildings and their patterns is visualized.
We have proven that these specifics of the built environment derived from remote sensing data spatially capture the living environments of the social group of the urban poor to a certain degree (Fig. 2; Wurm & Taubenböck, 2018). We find the people living in morphologic slums show a comparatively low variability, and many of the households located there are living below the poverty line. Thus, the morphologic proxy developed here becomes a legitimate one for approaching phenomena related to urban poverty with an explicitly spatial view.
Remaining solely in the domain of Earth Observation (EO) from space, often qualitatively described or observed disadvantages for the underprivileged within the urban landscape can be proven. Their settlement areas are located more often in highly exposed areas. As an example, using the steepness of terrain derived from a digital surface model as a proxy, we can show that landslide risk for slum areas is significantly higher. With respect to infrastructure, night-time lights captured from data from the International Space Station (ISS) reveal lower light emission in slums. This testifies to an undersupply of infrastructure and reveals political neglect (Kuffer et al., 2018).
The combination of Earth Observation with other geodata data for approaching urban poverty
The combination of EO data with other data sets considerably expands the field of application. Based on in-situ surveys in combination with 3D models of the built environment, we find a high probability that official census data underestimate population figures significantly (Taubenböck & Wurm, 2015). This insight reveals political denial of the urban poor in many cities across the globe. As detailed survey data are mostly absent, social network data have the potential to lower knowledge gaps. The interaction of EO-data with geotagged tweets reveals that in many cities across the globe slums belong often to “digital deserts” – areas where online activity on Twitter is less than the city averages (Taubenböck et al., 2018).
However, the localization of poverty within the urban landscape via the proxy of building structures does have limitations. The general function of slums is to provide comparably cheap living spaces serving as possible access to the city, to its functions, and to its society (Saunders, 2010). This conceptual approach has been introduced by the term “Arrival City,” indicating these locations serve as a home for poor rural-urban migrants as well as for the existing urban poor.
Of course, this function is not only provided by slum shacks but also by other manifestations of the built landscape. In a thorough literature review, we documented morphologic categories from complex, organic, high dense slums to run-down old towns, to structured and planned environments even to such specific forms as rooftop slums or boat people (Fig. 3; Taubenböck, Kraff & Wurm, 2018). We found there is no homogeneous morphological unifying factor of Arrival Cities.
The blind spot – Reducing knowledge gaps in urban poverty
Yes, as stated above, even today in times of “big data,” we face a real lack of information on urban poverty. Earth Observation in interaction with other structured (e.g. census, in-situ surveys) and unstructured geodata (e.g. from social networks) allow for reducing these information gaps. The meaningful combination of new data sources and the use of recent methodological developments in the domain of artificial intelligence will allow further increasing current knowledge. However, we should be aware that every data set has its own limitations, and critical approaches are essential.
While it is argued that urbanization brings wealth to the people (Glaeser, 2011), it is often seen that urban poverty can be found in almost every city. In fact, all new satellites, data, and methods are useless if scholars around the globe do not know what and where they have to look for. Thus, it is time to think outside the box of one’s own discipline in order to gain new and deeper knowledge about this phenomenon. Let’s start now!
These findings are described in several journal articles conducted by Hannes Taubenböck, Nicolas J. Kraff & Michael Wurm.
- Glaeser E (2011): Cities, Productivity, and Quality of Life. Science 333 (6042), pp. 592–594.
- Kuffer M, Pfeffer K, Sliuzas R, Taubenböck H, Baud, I & van Maarseveen M (2018): Capturing the urban divide in nighttime light images from the International Space Station. IEEE Journal of Selected Topics in Applied Earth Observation and Remote Sensing, vol. 11(8), pp. 2578-2586.
- Saunders, D. (2010). Arrival cities. London, UK: Blessing.
- Tacoli C, McGranahan G, Satterthwaite D (2015): Urbanisation, Rural-Urban Migration and Urban Poverty; International Institute for Environment and Development: London, UK; ISBN 978-1-78431-137-7
- Taubenböck H & Kraff N (2014): The physical face of slums – A structural comparison of slums in Mumbai, India based on remotely sensed data. Journal of Housing & the Built Environment. vol. 29, issue 1, pp. 15-38.
- Taubenböck H & Wurm M (2015): Ich weiß, dass ich nichts weiß – Bevölkerungsschätzung in einer Megacity. SpringerSpektrum. S. 171-178.
- Taubenböck H, Staab J, Zhu XX, Geiß C, Dech S & Wurm M (2018): Are the poor digitally left behind? Analyzing urban divides using remote sensing and twitter data. ISPRS Internatl. Journal of Geo-Information. vol. 7(8), 304, pp. 1-18.
- Taubenböck H, Kraff N & Wurm M (2018): The morphology of the Arrival City – A global categorization based on literature surveys and remotely sensed data. Applied Geography, vol. 92, pp. 150-167.
- United Nations (2014). World urbanization prospects. New York.
- World Migration Report (2015). Migrants and cities. Geneva, Switzerland: International Organization for Migration.
- Wurm M & Taubenböck H (2018): Detecting social groups from space – Remote sensing-based mapping of morphological slums and assessment with income data. Remote Sensing Letters. vol. 9(1), pp. 41-50.