UVicSpace | Institutional Repository
UVicSpace is the University of Victoria’s open access scholarship and learning repository. It preserves and provides access to the digital scholarly works of UVic faculty, students, staff, and partners. Items in UVicSpace are organized into collections, each belonging to a community.
For more information about depositing items, see the Submission Guidelines.
Recent Submissions
Application of the Feature Mixing Model (FMM) for vegetation land-cover detection in remote sensing
(2000) Law, Derek T.; Niemann, K. O.
The objective of this research was to develop and implement the Feature Mixing Model (FMM), and apply it to detect vegetation land-cover in remote sensing. A practical method to execute the model was created within the WiT image analysis program. The FMM was used as an image classifier for a Landsat TM subscene. The results were then compared against the classification results from both the industry 'standard ' spectral unmixing model: the unmix classifier, and the frequently used Maximum Likelihood Classifier (MLC). The comparison showed that the MLC was a more accurate image classifier than the two spectral unmixing models. However, the two spectral unmixing models have the potential to be more effective image classifiers with further research and development. In addition, a secondary study comparing the Normalized Difference Vegetation Index (NDVI) versus an extension of the FMM - the Modified Vegetation Index (ModVI), was performed to determine if one was less affected by radiometric sensing conditions. The comparison suggested that ModVI compensated for factors that affect that spectral signature of vegetation slightly better than NDVI.
An automatic system for hyperspectral remote sensing endmember unmixing
(2003) Han, Tian; Goodenough, D.
Subpixel information on the components within a pixel is usually desired for remote sensing imagery produced by the medium resolution (30m) sensors. This information includes the number of endmernbers in the image scene, endrnernber spectra, and the endmember fractions for each of the image pixels. With its property of high spectral and relatively low spatial resolution, hyperspectral imagery provides us a perfect data source to explore subpixel information with linear spectral unmixing techniques. In this thesis, linear spectral unmixing algorithms are classified according to whether one is considering the unmixing constraints or requiring all endmembers as a priori. Two algorithms are analyzed and extended for endmember extraction and unmixing. The endmember extraction algorithm is developed based on the unconstrained least squares regression, while the unmixing algorithms are derived from the constrained energy minimization and constrained least squares regression, respectively. These algorithms are implemented in the IDL language and tested with both simulated and real hyperspectral data. The computed result shows a very good agreement with the truth data. The accuracy is within 2% for simulated data and reasonable for the real data.
An Exploration of conifer canopy anisotropy: 3D modeling versus airborne imaging spectrometer and multispectral scanner imagery
(1999) Burnett, Charles Nils; Niemann, K. O.
Canopy biophysical data measured in the field was compared to modeled simulations of bidirectional reflectance (BR) and to airborne hyper-spectral and multi-spectral scanner imagery. The simulations were produced using a 3D computer graphics implementation of the geometric-optical model of Li and Strahler (1984). Two independent experiments were undertaken. In the first experiment, a solar position change of7° in zenith and 68° in azimuth produced changes in reflectance from Douglas-fir (Pseudotsuga menzeseii) forest stands on Vancouver Island. A VIRIS-measured change and 3D-modeled reflectance change were compared for 5 conifer stands ranging in age from 20 to 150 years. Sites 8, 9 and 10 contained mature (>30m in height) Douglas fir stands, while site 7 and 6 were populated by 30 year old and IO year old trees respectively. Change in values recorded by the spectrometer varied with wavelength. In the 700-800nm range, a grouping pattern of sites 8, 9, and IO separate from sites 7 and 6 was discernible, suggesting even with only a 7°/68° change in solar position, canopy shape may have a measurable effect on reflectance. Topography was incorporated into a second iteration of the first experiment, resulting in change estimates closer to the measured reflectance. Finn relationships between change and biophysical parameters were neither found nor expected with so few BR measurements and with the use of endmember reflectance values that were extracted from the literature rather than in situ.
In the second experiment, semi-hemispheric bidirectional reflectance functions were calculated at two spectral wavelengths for two stands (aged 40 and 150 years) using the 3D graphical modeling method described above, but with modifications to the input variables. These functions were compared against each other and then compared to canopy measurements from multi-spectral airborne imagery.
Ray-traced renderings demonstrate very accurate calculation of the areal extent of each of the four Li and Strahler geometric-optical model (GOM) components; however, the model is too simplistic for accurate BR modeling. For example, the differences in the BR change due to wavelength suggest the importance of introducing spectral parameters to the GOM. If the spectral information is incorporated into a 30 graphics/ray-tracing approach to modeling, the method could provide a powerful tool for the estimation of BR corrections for remotely sensed imagery acquired over topographically complex areas. This implementation of a simple version of the Li and Strahler model has demonstrated some of the strengths and weaknesses of using models in forest canopy research.
Toward the development of an exploratory spatial data analysis application for reporting sensitivity to the modifiable areal unit problem: a conceptualization
(1996) Setton, Eleanor M.; Keller, C. Peter
The modifiable areal unit problem (MAUP) has been the concern of geographers for many decades. As of this date, there is no generally applicable solution to MAUP. It has been suggested that in the absence of such a solution, the effect of MAUP on any particular analysis be reported explicitly. This research conceptualizes an exploratory spatial data analysis (ESDA) application for reporting MAUP sensitivity. The characteristics of MAUP and ESDA are identified through literature review, and these characteristics are used to formulate a conceptual framework for developing the ESDA application. The conceptual framework is extended by proposing measures for sensitivity and through the demonstration of the concepts using two existing software applications, Arc/Info and S+.
This research finds that there are three distinct levels for reporting MAUP sensitivity, each with distinct concepts of sensitivity and distinct measures. Reports can be generated for private or public use. It appears that the public report could prove to be lengthy and place an undue burden on analysts. It is suggested that interested parties should have access to the same ESDA application used to generate the report. Some limitations were imposed on this research by the software applications used for demonstration purposes. The applications do not share a common data model, data base, or user interface; therefore, dynamic linking between views in either application is not possible. Given that this configuration of software appears to be the only commercially available means of using both Arc/Info and a statistical program in tandem, it is concluded that the widespread use of ESDA and hence any application for reporting MAUP sensitivity based on the use of such software is not expected in the immediate future.
The rationalization of geographic information management practices with fair information practices
(2002) Michell-Viret, Christopher St. Elmo; Keller, C. Peter
This thesis undertakes to rationalize Geographic Information Management practices with Fair Information Practices. The thesis describes alternate ways in which to balance geographic information management technologies and geographic information management requirements of public and private organizations with the right to privacy. The dissertation addresses three primary issues. Firstly, the dissertation provides an analysis of the effectiveness and efficiency of the application of information management practices on geographic information management and analysis. Secondly, the thesis includes the identification of innovative geographic information management practices that permit the most diverse and intensive forms of geographic information analysis without compromising fair information management practices. Thirdly, the dissertation addresses the development of potential geographic information management remedies where geographic information is underutilized or where geographic information analysis practices contravene fair information management practices.