Home Page

Initial attempts were made to acquire aerial photo data from Dr. Prestige Tatenda Makanga with 6cm by 6cm resolution. Due to confidentiality reasons, the data could not be released to us. An OpenStreetMap TIF file was instead used to attempt to count populations in Cape Town. Since OpenStreetMap doesn’t come with a projection system, when the image was imported into ArcGIS, additional steps were required to incorporate a projection system into the TIF file. The TIF file was georeferenced against the base map already present within ArcGIS. Twenty-five ground points were selected randomly to georeference the study area to the base map. The georeferenced TIF file was then clipped to the study area polygon given by Dr. Prestige Tatenda Makanga, which already identified areas that are informal settlements within Cape Town. Additionally, point data for all the houses within Cape Town was also given to provide us with reference data to evaluate our protocols.

The area sampling by manual count method was used instead of an object-based approach due to data quality. The first step is to start with random samples of thirty rectangular polygons with an area of 100m2 to 800m2 in size. The thirty random sample sites are shown in Figure 1. The literature revealed that the adequate number and size of square blocks are still in question (Brown et al. 2001). By statistical principle, the representativeness of the blocks are proportional to the increase of blocks and the increase of the block size. It should be considered that as the number of blocks or the size of the blocks increase there will be diminishing return in the accuracy of the results. Thirty blocks was an arbitrary number chosen base on factors such as manpower, time constraint, and quality of data. The counting of population within the blocks can be done by interviewing all heads within the household through on site surveying, but such a method cannot be done due to physical limitations (Brown et al. 2001). We have to resort to counting houses manually and using the average persons per household to estimate population.

The informal settlements within Cape Town also have a varying degree of density and could possibly be taken into account by “observation while walking around the camp” (Brown et al. 2001). Since on site investigation is impossible for us, the assumption has to be made that the average density of all the polygons will be sufficient to cancel out the variations in house density. The house count is entered manually into the attribute table in ArcGIS and the area of the block is calculated automatically in m2 with the geometry calculator. (Total area within blocks/total house count) is the average house density of the study area. The ratio is then extrapolated to the total area of the study area to get the estimated house count for the settlements. The estimated house count is used to compare to the point house data that is given to evaluate the effectiveness of the protocol. Finally, the estimated house count can be multiplied by the persons/house ratio found on Cape Town government website to find the estimated population for informal settlement in Cape Town.

The area sampling by manual count approach produced total area within the polygons of 13019.7m2 and the total house count within the polygons was 155 houses. The ratio of house/area is 0.12 house/m2 and the total area of study is was 3970808.4m2. By extrapolating the ratio to the entire Cape Town study area, the protocol gives an estimation of 47273 houses. The reference point data for the houses shows a count of 43260 in total, which means that the relative error for the area sampling method is approximately 10 percent. The ratio of dwellers per house is 3.17, which means that the population estimate of the informal settlements in the study area is 149855 people.

Brown, Vincent, Guy Jacquier, Denis Coulombier, Serge Balandine, François Belanger, and Dominique Legros. “Rapid Assessment of Population Size by Area Sampling in Disaster Situations.” Disasters 25, no. 2 (June 2001)