Extraction of forest attribute information using multisensor data fusion techniques: a case study for a test site on Vancouver Island, British Columbia

The use of multisensor and multitemporal remotely sensed data is important for the extraction of forest attribute information for Canada's 418 million km? of forests. Hyperspectral data can provide vegetation signatures for forest attributes and canopy chemistry. Hyperspatial data can be used for individual tree recognition. Data from P-band synthetic aperture radar (SAR) have yielded accurate timber volume information in experiments involving managed plantations. We present the results of a study undertaken to assess forest attribute determination over the Greater Victoria Watershed District (GVWD) test site on Vancouver Island, BC, Canada from the following airborne and satellite sensors: Multi-detector Electro-optical Imaging Scanner (MEIS), Airborne Visible Infrared Imaging Spectrometer (AVIRIS), AirSAR and LANDSAT-7. GIS information and field data from field spectrometers are used for validation and calibration of AVIRIS and LANDSAT-7 ETM+ data. Inventory information (e.g. stem density, biomass) for our plots is known as a result of both field sampling and data fusion of GIS and high spatial resolution MEIS data. Segmentation techniques are applied to identify spatially homogeneous objects for quantification of forest attributes.