Remote sensing of structurally complex semi-natural vegetation— an example from heathland

Abstract

Preliminary results are presented from a study of heathland in the New Forest, southern England, using ground radiometry and airborne multispectral survey. Spectral data collected at ground level with a 1 m sampling cell size showed considerable internal variability for most of the vegetation types studied, and much overlap between classes in the multispectral feature space. The airborne scanner data (5 m nominal pixel size) averaged out the spatial variability in reflectance to some extent, and suggested that limited spectral discrimination between some of the vegetation types may be possible. However, effective remote sensing of structurally complex canopies such as shrubs, requires an understanding of the three-dimensional form of the spectral reflectance function and its relationship to the geometric arrangement of canopy components and to phenological changes in the canopy.