Linear spectral unmixing of CASI data for habitat mapping and management in a coastal dune system

Linear spectral unmixing of airborne multispectral scanner (CASI = Compact Airborne Spectrographic Imager) imagery is performed using the constrained least squares method. The subpixel proportions of sand, vegetation and shadow/moisture are defined to decipher the status of habitats in Kenfig National Nature Reserve and dune ecosystem bordering Swansea Bay, south Wales, UK. Salix repens growth form, an indicator of vegetation density, is compared with the estimated fractions of vegetation and sand to demonstrate the validity of the mixture model