Remote sensing of deciduous woodlands : a tool for ecological investigations

Deciduous woodlands are of great ecological importance within both a U.K. and World context, and there is an increasing need to monitor and classify these areas. Remote sensing has the potential to provide spatially comprehensive, and timely, information on deciduous woodlands, however this requires a deeper theoretical understanding of the controls upon the reflectance properties of these ecosystems. This research attempts to develop such an understanding through a series of controlled field experiments, and to use this knowledge to derive ecological information on deciduous woodlands from remotely sensed imagery. In order to overcome the problems of sampling the reflectance properties of woodland canopies a technique was developed in which a spectroradiometer was operated from a helicopter. These data were used in combination with ground-based measurements of biophysical properties, and it was found that variations in the degree of spatial dependence in reflectance could be attributed to variations in canopy structural complexity. Distinct temporal relationships were established between reflectance properties and phenological changes of deciduous woodland canopies, using a tower-mounted instrument array. The effects of such seasonal variations were given a spatial context by analysing multi-date imagery obtained with a Compact Airborne Spectrographic Imager. Data from this instrument were used to map spatial patterns within several deciduous woodlands, and these characteristics were quantified using a range of measures and indices within a GIS. These spatial statistics were then used to infer differences between various types of woodland in terms of canopy gap creation processes, regeneration mechanisms and species diversity. The potential of multi-date imagery to provide such information was established, and the major sources of variability in such data sets were determined using principal components analysis.