Assessing change in woody vegetation cover in the Kruger National Park, South Africa, using spectral mixture analysis of a Landsat TM image time series

Abstract

Fire and elephants are the main influences on woody vegetation cover in the Kruger National Park (KNP). This study assessed the change in woody cover using geometrically registered Landsat TM images (1989, 1995 and 2007), their dry season period selected so as to exclude herbaceous vegetation. Using spectral mixture analysis (SMA), the images were classified in four woody fractions of cover classes (0.20–0.39, 0.40–0.59, 0.60–0.79 and 0.80–1.00). Field site-derived woody vegetation end member locations were used in the process. The southern-central section of KNP showed losses in woody vegetation of about 33% in the high fertility (basalt) areas compared to about 11% for the low fertility zone. Within classification errors introduced by the presence of cloud and herbaceous vegetation, a spatial indicator of susceptibility of the woody vegetation to change was developed for the 0.80–1.00 fraction, which indicated higher cover stability in the less frequently burnt areas of the park.