- The present day vegetation pattern of the Willamette Valley is a result of a long
past of human alteration of the landscape. Beginning with aboriginal burning of the
grasslands to present day land uses, the vegetation of the valley has been affected by
anthropogenic activity. Wet prairie is a vanishing habitat of the valley as both urban
and agricultural development continue to alter the landscape.
The W.L. Finley National Wildlife Refuge of the south-central Willamette Valley
contains within its boundaries one of the last remaining examples of open wet prairie.
In order to maintain the prairie community complex and associated biodiversity of the
Willamette Valley Floodplain Research Natural Area, a restoration project in which
controlled burning plays a significant role was initiated. As a part of the on-going
research a need was felt to develop a baseline data base identifying the spatial
distribution of principle features of the research area and to map the vegetationally
associated mount-intermound micro-topographic pattern prevalent throughout the
The purpose of this research project was to develop that baseline spatial data
base and to investigate the use of digital image processing and analysis as a tool for
inventorying micro-topographic features and associated open wet prairie vegetation
patterns. This paper is primarily a methodological study of applying geographic
techniques to an on-going biogeographical study
This study had three objectives: (1) defining the location and boundaries of
treatment spaces and sampling plots in a standard geographic coordinate system, (2)
delineation of spaces and plots onto a scanned image creating a spatial data base and (3) classification of vegetation and associated microtopography. The aim of this study is
to provide an evaluation of appropriate geographic information technology for the
planning and analysis of wetland restoration and management.
The objectives of this study were met with varying degrees of success. The use
of Trimble's Global Positioning System, IDRISI's low-cost image processing/
Geographic Information System software and a small scale, color aerial photograph
scanned to a single band, monochromatic digital image were utilized to create a
suitable visual and spatial data base.
Classification of the image into micro-topographic class was conducted using
three strategies. Two standard supervised classification algorithms were compared
using signatures generated from the original single band image. The influence of
including signatures created from an additional data band based on digital richness of
the original image was also examined.
Information provided by the single, wide spectral space image was lacking and
hampered classification of vegetationally associated micro-topographic patterns. The
addition of signatures based on a created Relative Richness band did not appreciably
effect classification accuracies. Regardless of classification strategy, insufficient
separability of class signatures led to overall classification accuracies of +1- 70%. A
high level of misclassification occurred within the intermound region resulting in overrepresentation
of the mound micro-topographic pattern.
The resulting thematic map was created using signatures generated from the
original single band image and maximum likelihood classification routine.
- File scanned at 300 ppi (Monochrome, 8-bit Grayscale, 24-bit Color) using ScandAll PRO 1.8.1 on a Fi-6670 in PDF format. CVista PdfCompressor 4.0 was used for pdf compression and textual OCR.
- description.provenance : Approved for entry into archive by Patricia Black(firstname.lastname@example.org) on 2012-07-12T19:40:45Z (GMT) No. of bitstreams: 1AndersonKarenL.pdf: 1994787 bytes, checksum: faa33aac3eb45a53993a7074c02613b0 (MD5)
- description.provenance : Submitted by Kaylee Patterson (email@example.com) on 2012-07-03T18:00:50ZNo. of bitstreams: 1AndersonKarenL.pdf: 1994787 bytes, checksum: faa33aac3eb45a53993a7074c02613b0 (MD5)
- description.provenance : Made available in DSpace on 2012-07-12T19:40:45Z (GMT). No. of bitstreams: 1AndersonKarenL.pdf: 1994787 bytes, checksum: faa33aac3eb45a53993a7074c02613b0 (MD5) Previous issue date: 1996-02