Application of Hyperspectral Data for Forest Stand Mapping

The objectives of this study are to evaluate the potential of Hyperspectral Mapper (HyMap) data and a spectral mixture model to characterize forest stands in a mixed coniferous and deciduous forest. A HyMap image was flown over a standard research site in Western Switzerland in summer 1998. The research forest-site can be characterized as heavily mixed forest. HyMap data were used to map the forest mixture-grade. The image was first evaluated qualitatively. There were no obvious noticeable geometric and radiometric errors. The HyMap image was geocoded using a parametric method (PARGE) based on a high-resolution digital elevation model and the simulation of the flight path. The image was also atmospherically corrected using ATCOR-4. Noisy wavebands were excluded. A linear unmixing method was used to model the spectral signature of each pixel. The image based endmember collection approach to derive the spectra for selected endmembers (pure coniferous, pure deciduous, clear cutting and shadow) was performed using high resolution airphotos. The fraction components derived from the unmixing model were compared with CIR-airphotos at the scale of 1:9000. The results show the capacity of hyperspectral HyMap data, and that linear unmixing models are very useful tools for automatically separating coniferous and deciduous stands and their mixture-grade, which is very important for forest management.