| Abstract | Hyperspectral imagery has the potential to become a useful tool for monitoring and extracting biophysical properties of vegetated areas. Exploitation of this potential relies on the ability to relate
at-canopy spectral reflectance to biophysical characteristics of vegetation and derive both sunlit and shaded component proportions and spectral profiles. Increased application of hyperspectral imagery to these areas is expected with the advent of
space borne hyperspectral sensors (such as EO-1 Hyperion and CHRIS-PROBA). Such imagery of vegetated scenes is influenced however by the well known bidirectional reflectance distribution (BRDF) effect. One method of determining the contribution of
shaded overstorey vegetation and background to observed spectral reflectance is to determine, by model inversion, the proportion of shaded surfaces viewed by the sensor, and the relative intensity of the radiative flux incident on these surfaces.
This can be achieved by modelling the overall reflectance as composed of mean sunlit and shaded reflectance components, combined with an analytical description of the shaded radiant flux. Assuming a land cover type with consistent mean foliage and
background reflectance, inversion of a semi-empirical model can be used to determine BRDF coefficients, which can then be applied to normalize the imagery to a specific view/sun geometry. If the modelled spectral coefficients directly relate to
canopy properties, then BRDF normalization can also provide information to help directly relate the canopy architectural and biophysical properties to the remotely sensed signal. One such model, FLAIR, has been successfully used to investigate canopy
characteristics from airborne and satellite spectral imagery. |