Using remote sensing for tree species discrimination in the narrow coastal forests of KwaZulu-Natal, South Africa

Abstract

Indigenous forests of South Africa occur in one of the smallest biomes covering less than 0.5% of the total surface area of the country. Regardless of the small and narrow extent of these forests, they offer a range of ecosystem services, such as fuel wood, construction material, food and medicine, especially for the poor. Mapping and monitoring these forests are crucial for their sustainability. Regular monitoring of vegetation composition and condition can be done through localised fieldwork, yet access may be difficult in mountainous terrain, or in swamp and mangrove forests, or where dangerous animals occur. Remote sensing technology offers a feasible alternative. The improved spatial and spectral resolution of space-borne sensors has enabled species mapping at canopy level, as well as assessing pigment and nutrients as indicators of vegetation health and condition. We studied six evergreen tree species found in the subtropical coastal, swamp and mangrove forests of the KwaZulu-Natal Province of South Africa over a period of four seasons: winter, spring, summer and autumn. We present the results and findings of our work, including: i) the ability to monitor pigments and nutrients over four seasons, using leaf-level data; ii) the ability to discriminate between these six species, using seasonal information at leaf level; iii) which bands were found to be important for these species over the four seasons; and iv) whether multispectral sensors, such as RapidEye, can be used to map these six species over four seasons. Our results indicate the importance of remote sensing in monitoring indigenous forests.