Detecting settlement expansion using hyper-temporal SAR time-series

Abstract

The detection of new informal settlements in South Africa using time-series data derived from coarse resolution satellite imagery has recently been an active area of research. Most of the previous methods presented using hyper-temporal satellite data in this particular study area relied on optical data as input. In this paper, the feasibility of using hyper-temporal SAR data for the detection of new informal settlements was investigated. Using numerous ENVISAT ASAR images during the period 2005/01 to 2011/01 , it was shown that new settlements could effectively be detected using an autocorrelation change detection approach which was previously only tested on optical data and subsequently adapted to use hyper-temporal SAR data as input. Preliminary results indicate change detection accuracies in the order of 85% at a false alarm rate of less than 1%.