Microscale segregation patterns in Gauteng and its implications for spatial transformation

South African urban form remains in dire need of spatial transformation. This need arises from the spatially ingrained nature of race-based segregation and inequality that was established under apartheid, and which endures to some extent in the present urban form.

Segregation—particularly the presence or absence of multiple race groups in newly developed areas—is one of the lenses through which we can view progress towards spatial transformation. This is because apartheid used race as a basis for separation and because socio-economic inequality in South Africa remains race-based. There have been changes, but their impact on spatial transformation is limited. At macro scales, cities have transformed to become more diverse in terms of race, but this masks the micro scale separations that remain or are reproduced.

Residential expansion—or development on “new” land—provides a blank canvas for creating truly post-apartheid settlements and are key components and opportunities for spatial transformation. In this paper, I analysed the impact of residential expansion on spatial transformation over the last 30 years by investigating how residential expansion is related to racial diversity in Gauteng, South Africa. I did this through a micro-scale perspective. The goal was to determine the extent to which residential growth reproduces or disrupts past patterns of racial diversity to understand the trajectory toward post-apartheid goals.

I deliberately focused on residential expansion in Gauteng for three reasons. First, past segregation analysis has not been directly viewed in relation to residential growth to understand how residential growth re-produces or disrupts segregation. Second, a focus on residential expansion at the level of 400 m-by-400 m hexagons highlight important micro-scale changes. The micro-scale analysis is unique and highlights patterns that are not evident at larger geographies, and change our perception of segregation patterns, causes and consequences. Third, the relevance of analysis in ‘former white Group Areas’ has been surpassed by substantial residential expansion.

The methods and data used for the analysis can be described in two parts:

1. First, the segregation component was completed by calculating the racial diversity of hexagons in Gauteng with an entropy index of segregation. Population estimates and the associated disaggregation by population group was obtained from the GeoTerraImage (GTI) proprietary hexagon dataset for the year 2020 and used in the calculations.
2. Second, the analyses involved tracking residential expansion since 1990. Land use data, which was derived from satellite imagery, was also obtained from GTI. Residential areas were defined as hexagons in which more than 30% of the total land use in the hexagon was classified as residential. Land use data from 1990, 2000, 2010, and 2020 was used to map residential expansion in Gauteng over time.

The results show that the average racial diversity of new residential development (0.279) is lower than the average racial diversity of residential development that occurred on the 1990 urban footprint (0.559). There are two main reasons for this. First, many of the older residential areas are former white-only suburbs that have desegregated substantially. Second, the total land area and the total estimated population associated to low diversity expansion (62% of the land and 80% of the population) is much higher than the land area and total estimated population associated high diversity expansion. Thus, in Gauteng as a whole, residential expansion is not related to a higher racial diversity.

It is clear that racial diversity is low in areas that were added next to townships such as Mamelodi, Atteridgeville, Olivenhoutbos, and Soweto. It is also clear that suburban expansion around south-eastern Tshwane, northern Johannesburg, and central Ekurhuleni, is associated with high racial diversity. The expansion of low diversity areas (townships and informal settlements) is spatially more significant and larger in terms of population than the expansion of other residential areas (upper and middle class) and contributes to lower overall diversity of urban expansion.

Fleurhof and Cosmo City are presented as two examples of public housing developments made important contributions to the goal of creating a spatially transformed and equal-opportunity city-region. Their impacts on racial diversity, socio-economic sorting and overall spatial transformation are positive, but limited in spatial extent. At a local scale, Fleurhof is associated with higher racial diversity while macro scale racial diversity is also increased between areas with otherwise distinct levels of racial diversity. In the case of Cosmo City, even though the area has low micro scale racial diversity, the location of the development (near the expanding middle class neighbourhoods of northern Johannesburg) means that this development increases macro scale racial diversity in a way that other public housing development which are located close to townships do not.

While the overall impact of residential expansion on spatial transformation is limited due to the dominance of residential expansion related to low racial diversity, and while there remains a distinct spatial separation between new residential areas, with low and high racial diversity, there are important instances of high racial diversity in residential areas that has developed since 1990.

The residential developments that have made some positive impact is limited to a handful of public housing developments that have contributed towards creating a more equal city-region.

Ultimately, urban planning has not altered the availability and distribution of affordable housing to contribute to increased levels of race and class diversity. These findings imply that urban planning has not maximised the potential benefit of new residential development for spatial transformation.

Key points

• The distribution of population groups in Gauteng represents a high degree of continuity with past patterns, but not without change over time.
• Calculations show that between 1990 and 2020, roughly 905 km2 of predominantly residential areas were added to the Gauteng residential footprint.
• In Gauteng as a whole, residential expansion is not related to higher racial diversity.
• The expansion of low diversity areas (townships and informal settlements) are spatially more significant and larger in terms of population than the expansion of other residential areas (upper and middle class), and contributes to lower overall diversity of urban expansion.
• Neighbourhoods with high or low racial diversity are separated from each other through socio-economic sorting.

Recommended citation

Hamann, C. (2025). Microscale segregation patterns in Gauteng and its implications for spatial transformation. Urban Forum, 37, 55-77. https://doi.org/10.1007/s12132-025-09547-4.

For more details, please contact Christian Hamann [christian.hamann@gcro.ac.za]