Application of hyperspectral techniques to multispectral data: spectral mixture analysis (SMA) in mapping of emergent macrophytes in a water-hyacinth-infested area

Water hyacinth (Eicchornia crassipes (Mart.) Solms) is an invasive aquatic macrophyte that has infested the lake Victoria, East Africa, since the late 1980s. It has been associated with major negative economic and ecological impact of this important water resource in East Africa. Remote sensing technology has significant potential in mapping this fast growing floating weed, in a mostly inaccessible area for field measurements. Our study site is the Winam Gulf, on the Kenyan part of the Lake, which has had the highest reported infestation in recent years. The paper describes a study to evaluate the ability of ETM+ multispectral imagery in mapping water hyacinth and associated macrophytes in the hyacinth infested Winam Gulf. By applying hyperspectral techniques on multispectral data, a spectral mixture analysis was undertaken using image-derived endmembers. The study was also an evaluation of an alternative way of acquiring emergent macrophytic endmembers in cases where limitations like lack of hyperspectral data, spectrometric measurements and spectral libraries exist. The results demonstrate that whereas it is possible to discriminate and map the different spectral constituents, a spectral library of the endmembers under investigation would be required for positive identification, especially for macrophytes that are closely related spectrally, fast growing, have varying concentrations (density) spatially, and are non-static in nature.