This study is about a non-invasive mapping technique of Welwitschia mirabilis in its habitat, the Namib-Naukluft Park, based on spatially high resolved satellite images. First, two satellite data sets from IKONOS and Quickbird satellite are geometrically corrected, including GPS ground control points and a digital elevation model. Second, Welwitschia individuals, bushes, sandy and rocky surroundings are successfully mapped with a supervised and object-oriented classification approach. GPS points, training samples and verification objects have been collected previously in a field survey. 49 of 55 (89%) Welwitschia individuals known in ground reference, were classified successfully in both data sets, despite the limited spectral and spatial resolution. Compared to a pixel-based classification of Welwitschia with IKONOS data also published on this conference, it was shown, that the object-oriented approach improves the mapping precision of Welwitschia species. The presented area wide mapping technique from space is effective and non-invasive, but limited by the spatial and spectral resolution of the input satellite data.