Biomass estimation for land use management and fire management using Landsat-TM and -ETM+

Abstract

Grazing and fire are major factors influencing the savanna ecosystems of Southern Africa. In both grazing and conservation areas overgrazing is an important reason for degradation of vegetation and soil. Insufficient fire management can cause a change in the species composition and may influence the soil negatively. For adequate planning purposes the knowledge of available biomass is indispensable. High-resolution satellite systems can provide such knowledge on a large scale. Three study areas in Southern Africa contributed to this present survey. Gutu District is situated in Zimbabwe. In its Communal Lands a high population density leads to severe degradation of vegetation and soil. The South African test sites are located in Kruger National Park and Madikwe Game Reserve. Therefore a wide ecological range from highly degraded to slightly disturbed savanna ecosystems is included. Satellite images of both Landsat-5 (TM) and Landsat-7 (ETM+) were applied. After cross-calibration of the two different satellite systems, radiance and reflectance were derived from the raw data. Furthermore, digital numbers, radiance and reflectance were used to calculate different indices. These indices were then extracted and averaged for each test site and statistically analysed together with the different types of above-ground biomass. Significant correlations resulted from ratios of radiance and reflectance of bands 2 and 1 as well as bands 7 and 1 with grass biomass and total foliage biomass. In addition the Tasselled Cap Wetness Index is very useful to quantify grass biomass and the Brightness Index to calculate total foliage biomass. In contrary no significant correlation could be detected for woody biomass and total above-ground phytomass. The application of the regression models in the Gutu district indicated that grass biomass calculated from satellite data is within the range of the biomass measured in the field. Thus models for predicting grass and total foliage biomass are available on a regional scale.