Remote sensing and spatial-analytic techniques for monitoring landscape structure in disturbed and restored coastal environments

The purpose of this work is to design and evaluate a framework that integrates remote sensing and spatial analytic techniques for providing information on the spatial characteristics of vegetation structures. Vegetation structures are horizontal and vertical arrangements of plant species and biomass that control levels of ground cover. Information on the spatial characteristics of vegetation structures can be used to define landscape structure. Environmental monitoring and management activities from local to global scales require this information to assess ecosystem structure, dynamics and disturbance impacts. This is especially the case in coastal areas at local scales, up to 100 km{dollar}sp2{dollar}, where disturbed landscapes and restoration activities are common. Remotely sensed data and spatial analytic techniques are capable of providing information on vegetation structures at local scales. Two problems limit the application of these techniques: (1) identifying suitable spectral, radiometric, spatial and temporal data resolutions and (2) defining analytic techniques to provide appropriate information for specific monitoring objectives and environments. Both of these "scaling" problems result from not utilizing prior knowledge on the forms and processes controlling an environment's spatial structure, to select and interpret data. Therefore, this dissertation research addresses the problem of determining suitable dimensions of remotely sensed data for deriving information on vegetation structures. A framework integrating exiting concepts in physical geography, hierarchy theory and remote sensing was developed to monitor the spatial characteristics of vegetation structures in disturbed and restored coastal environments at local scales.; The first stages identify the type and scales of information required on the spatial characteristics of vegetation structures for disturbance and restoration monitoring. The spatial, spectral, radiometric and temporal resolution(s) of remotely sensed data are specified by combining the scale and type of information required on vegetation structures in a scene model and checking this with exploratory analysis of existing remotely sensed data. Scene model dimensions provide a basis to evaluate potential data sets and select suitable analytic techniques to produce the required information on vegetation structure. Three case studies requiring information on vegetation structures for disturbance and restoration monitoring in a coastal environment were conducted for evaluating the framework's utility, sources of uncertainty and error, and overall strengths and weaknesses. The framework's principal strength was its function as a general construct for providing synthetic assessment of the utility and limitation of different types of remotely sensed data and analysis techniques. The case studies demonstrated the framework's capability to address previous limitations of applying remotely sensed data to examine terrestrial landscape-ecosystem-vegetation structure by: (1) selecting spatial, spectral, radiometric and temporal data resolutions and analytic techniques suitable for the target environment; and (2) providing information able to address a particular management objective or research question.