No description available.

No description available.

No description available.

No description available.

No description available.

No description available.

No description available.

No description available.

Accurate quantitative biophysical estimates derived from remotely sensed data are required for effective forest monitoring and to enhance the understanding of biospheric processes at regional and global scales. The aim of this thesis was to examine relationships between forest biophysical properties and polarimetric synthetic aperture radar (SAR) backscatter. Multiwaveband polarimetric SAR data were acquired by the NASA/JPL. AIRSAR over coniferous plantations in Thetford Forest, East Anglia and the Tywi forest, central Wales. Thetford Forest was located on relatively flat terrain and the Tywi forest was an upland plantation with variable topography. A simple classification of the Thetford Forest SAR data showed that species and age classes could be distinguished to a high level of separability when using data from all wavebands. Backscatter, represented by mean amplitude, increased with compartment age until 20 years, with the strongest relationships evident using long wavelength P- and L-band data. The strength of relationships, especially in C-band, were improved when variables representing changes in scattering behaviour (e.g. polarization ratios) were used. Backscatter from the Tywi forest was found to be highly dependent on viewing geometry; a function of radar look angle and topography. Relationships between mean amplitude and estimates of forest biophysical properties related to above-ground biomass (e.g. stem biomass) were significant only for stands situated at high incidence angles and with little topographic variability. Mean percentage canopy closure was significantly related to polarization ratios and coefficients of variation, but not to single wavebank/polarization mean amplitude. It was found that canopy gaps formed by windthrow could be identified and that the remotely sensed response was related to geometric properties of the gaps (e.g. area, perimeter). In conclusion, the research highlighted the potential for deriving quantitative estimates of a range of forest biophysical properties when multidimensional SAR data is available.

Species loss in fragmented tropical forest is the greatest single threat to biodiversity. Fragmentation involves changes in the size, shape and spatial distribution of existing forest. Understanding these effects and comparing their impact with the influence of other environmental variables is of utmost importance in prioritising forest blocks for conservation. The spatial geometry of forest blocks in Ghana, West Africa, was examined with respect to tree species number and composition. Tree species number proved to be positively related to forest block <I>area</I>, according to laws of probability and ecological determinism. In addition, tree species composition was dictated by forest block size, with larger reserves recording significantly more rate species than smaller ones. Additional species encountered in reserves of increasing size followed a predictable pattern, which allowed the identification of species vulnerable to habitat loss. An effect of forest block <I>shape</I> on the structure and composition of tree species was visible using coarse spatial resolution satellite sensor imagery. Fragmentation of a previously continuous canopy led to distinct forest edge effects. This was verified by ground data, which showed an increase in the proportions of common pioneer species along reserve peripheries. Significant trends were only detected across small areas: indicating that shape is not an important consideration in the design of larger forest reserves. The effect of forest block <I>proximity</I> on tree species number was examined using regression analysis, but any rôle of proximity was dominated by the influence of area. The influence of <I>climate</I> and <I>fire activity</I> was strongly expressed throughout the examination of spatial geometry in the form of residual variation about the regression line. Results of multivariate analysis indicated that the management of size and disturbance should be considered in preference to shape and proximity if higher levels of diversity and bioquality are to be maintained.