Development of Sampling Procedures Based Upon Satellite Derived Land Cover History for the NSF Digital African Cities Project

Abstract

This article discusses the sampling scheme employed by the Six Cities project to ensure that all areas of habitation have a chance of being selected, that we know what that chance is, and that we are able to critically evaluate the sampling strategy after it has been carried out. A weighting strategy that is slightly different from one used only to do research is therefore employed. The article describes a procedure for generating two kinds of random sample points for areas of change and of no change. Finally, a few simple rules for incorporating socioeconomic, demographic, and other relevant information into the sampling frame without introducing bias into the sample are discussed.

Key words: sampling strategies; random sampling; sampling bias; local knowledge; Six Cities project; remote sensing; urban areas in Africa