Mapping recent built-up area changes in the city of Harare with high resolution satellite imagery

• Analysis of built-up area changes in Harare for 2004–2010 from satellite images.
• Combination of texture based feature extraction with grid based change analysis.
• The designed workflow provides approximate information on the location of population.
• Growth in Harare happens mainly in the urban fringes outside low-density areas.
• Results promote the use of remote sensing data for spatially refined information.

Spatial information about human settlements is key information in times of continuous urban growth and high rates of uncontrolled urbanisation in developing countries, but there is often little geographic information on where the population growth occurs. Built-up area is a good proxy for population that can be extracted from Earth observation data. At the example of the city of Harare in Zimbabwe, this research shows how remote sensing technologies can be used to obtain refined information on hot spots of built-up area change that allow a monitoring of population. This was achieved through an automated mapping of built-up area changes between 2004 and 2010 from multi-temporal high resolution satellite images. The designed workflow combines an automatic feature extraction method with an automatic, grid based change analysis. Built-up area was extracted from Spot images with accuracies between 78% and 84% and changes were computed as percentage change in a grid using an object-based hierarchical method. The derived built-up area and its changes were compared to population figures from the last two censuses at ward level. The comparison reveals a good match of the pattern of changes of both parameters; however, more detailed information was derived from the Earth observation data.