Woody vegetation cover in Namibian savannahs: a modelling approach based on remote sensing

Abstract

Savannahs are characterized by a co-existence of grasses and trees. Nowadays, these ecosystems are considerably affected by human impact, as they are used for various types of agriculture and support a major part of the population within the tropics. E.g., it is widely accepted that in many regions of Southern Africa, livestock farming with inadequate stocking rates has facilitated bush encroachment and strongly alters vegetation structure and composition in these environments. Similar observations are reported from national parks, especially if fences confine the natural mobility of animals. In this study, an approach is developed to model woody vegetation coverage from coarse resolution multitemporal NDVI data by making use of the different phenological characteristics of grasses and trees. A second order polynomial is used to describe NDVI curvature and to quantify phenological metrics. For model calibration and result validation, a reference map is initially generated based on extensive field work and medium resolution Landsat ETM data. The results suggest a close relationship between woody cover and dry season curvature and minimum, which are considered to be suitable phenological metrics for a modelling approach. Thus, spatial differences in woody vegetation cover were successfully measured on a regional scale by means of multitemporal NDVI data. The results clearly highlight the potentials of phenological parameters for mapping woody cover.