Using High Resolution Imagery to Assess Local Urban Growth Patterns and its Relationship with Population Density in Accra, Ghana Open Access
Downloadable ContentDownload PDF
Accra, Ghana is a coastal, capital city in sub-Saharan Africa experiencing rapid urban growth with the most recent average annual growth rate of 3.3%. While the city is growing rapidly, it is characterized by very few high-rise buildings, and yet Accra has areas of remarkably high population density. As such, the objective of this research is to measure land cover change in Accra between 2002 and 2008 and to analyze the relationship of these results with population density data. Does population density have any influence on urban land cover changes happening within the Accra Metropolitan Area (AMA) and has a threshold in population density been reached anywhere within the area? The data used to measure the land cover change is high spatial resolution satellite data. Because these data are at a finer spatial resolution than 5 meters, they support an intra-urban analysis. However, these data also present several challenges. The main challenges are the wide spectral range of impervious surfaces and minor radiometric and geometric differences between the two time-periods of data from the QuickBird and Ikonos satellite sensors. This research made use of post-classification change detection and decision tree classification to overcome these challenges. Classification focused on identifying impervious surfaces and achieved 81% overall accuracy on the 2002 image and 83% for the 2008 image. The change detection was analyzed at two spatial scales held in common with the demographic data: the enumeration areas and neighborhoods. Urban land cover change was most pronounced in the northwest of the AMA, while the enumeration area scale revealed additional pockets of substantial changes in impervious surfaces. The relationship with population density was analyzed across space and time. Across space, a positive correlation between percent impervious surfaces and population density was found, especially at the neighborhood scale, even for this intra-urban analysis. Looking across time, a pattern was found that the most densely populated neighborhoods (greater than 31,000 people/km2 in year 2000) experienced little to no infill development; however the evidence was inconclusive that this level of population density marked a threshold that had been reached in this densely populated coastal urban area.