Fuel type mapping from hyperspectral MIVIS andmultispectral Landsat TM data in the fragmentedecosystems of southern Italy

摘要

This paper is focused on the hyper- and multi-spectral characterization offuel type at different spatial scales in the fragmented ecosystems of the Pollino NationalPark. To this aim, multisensor remote sensing acquired from hyperspectral MultispectralInfrared and Visible Imaging Spectrometer (MIVIS) and ASTER data were analysed forthe test area. Fieldwork fuel type recognition, performed at the same time as remotesensing data acquisitions, were used to assess the results obtained for the considered testareas. Results from the classifications showed very satisfactory accuracy levels that werehigher than 90% and 81% respectively for MIVIS. and ASTER data.