Exploring remotely sensed shadow in Amazonian regrowth forests

Bailey, Philip George (1997) Exploring remotely sensed shadow in Amazonian regrowth forests. University of Southampton, Doctoral Thesis.

Abstract

This thesis proposes the use of remotely sensed shadow to increase the accuracy with which tropical regrowth forests can be mapped.

There is uncertainty in the spatial extent of tropical regrowth forests, which are known to act as sinks for atmospheric carbon. The importance of these forests in the global carbon budget is therefore ambiguous, requiring the mapping of these forests, for which remote sensing is accepted as the only practical means of doing so. Current remote sensing (optical and microwave) successfully identifies the early stages of tropical forest regrowth, but relationships between biophysical variables and the remotely sensed data are non-existent or weak for the later stages. To identify the later stages, there is potential to exploit forest canopy structure, which continues to develop beyond the early regrowth stages. It was hypothesised that changes in canopy structure would affect canopy roughness, producing shadows within optical remote sensing imagery. Research focused on the estimation of canopy shadow, both on the ground and within remotely sensed imagery.

Biophysical measurements, collected during a field campaign to the Amazon region, were used to develop a geometric model of the canopy surface. Ray tracing was employed to estimate the shadow proportion by simulating solar illumination of the modelled canopy. This method was used to undestand the shadow proportion within Landsat Thematic Mapper (TM) imagery.

Spectral mixture modelling was used to estimate the sub-pixel shadow proportion. A new approach was developed for interpolating the endmember spectral response of shadow. This method was explored over time using a multi-temporal series of Landsat TM imagery and at a coarse spatial resolution using ERS-2 Along Track Scanning Radiometer (ATSR-2) imagery.

The results confirmed that the later stages of tropical forest regrowth can be identified accurately with shadow in remotely sensed imagery. Synergising this method with existing remote sensing capabilities, can extend the identifiable stages of regrowth beyond current capabilities. By these means, the spatial extent of these forests can be mapped, enabling the estimation of their carbon density and thereby reducing uncertainties in the global carbon budget.

This record has no associated files available for download.

Contributors

Download statistics