Decomposition of mixed pixels of ASTER satellite data for mapping Chengal (Neobalanocarpus heimii sp) tree

Digital image classification of satellite remotely sensed data for forest mapping have been a great challenge in tropical rainforest areas due to high density of forest species in per unit area. This resulted in mixed pixels (mixels) in the captured image. Even with the advent ultrafine spatial resolution of current satellite images, the pure pixels of specific species within captured image is very rare. In automated digital image classification, mixels must be unmixed for an optimum output. This paper reports the results of study undertaken to spectrally unmixing of ASTER satellite data for detecting and mapping composition of Chengal trees (Neobalanocarpus heimi) within the mixels. Mixture tuned matched filtering (MTMF), a specialized type of mixture analysis, was used in this study. Results, indicated that decomposed mixels have high correlation with in-situ verification (r²=0.9, n=33, p<;0.005). It is therefore concluded that mixels of ASTER satellite data offers prominent application for mapping specific trees distribution and abundance map as exemplified with chengal trees.