First Advisor

Joseph Poracsky

Date of Publication

9-12-1997

Document Type

Thesis

Degree Name

Master of Science (M.S.) in Geography

Department

Geography

Subjects

Urban forestry -- Oregon -- Portland, Remote sensing

DOI

10.15760/etd.7276

Physical Description

1 online resource, (131p.)

Abstract

Digital pattern recognition methods were used to produce two maps of urban vegetation from LANDSAT Thematic Mapper for the City of Portland, Oregon. The image was acquired on July 7, 1991 and the spectral bands used were 2, 4, 5 and a ratio of 3 and 4. The two maps represent vegetation amount and vegetation type.

Due to the extreme heterogeneous nature of the urban environment, these maps were developed using techniques to reduce and manage the amount of spectral variation. This included purging non-vegetated cells from the multispectral image, and then stratifying the image into "spectral subdivisions" using tools for unsupervised classification. Each image subdivision was then classified using the standard unsupervised technique and the subdivisions reassembled.

The two resulting maps were then statistically assessed for classification accuracy. The map representing the amount of vegetation cover was found to have an overall accuracy of 80% and the map representing the type of vegetation was found to have an overall accuracy of 73%. Although previous studies have used similar imagery and techniques to map urban vegetation, none have reported accuracy figures, so direct comparison was not possible. However, these levels of accuracy are well within the range of accuracy reported by traditional remote sensing studies.

Uses of these maps include urban forest macro assessment, inventory, evaluation of ecosystem health and public education. Limitations of these maps are mostly a function to their spatial resolution. Stem counts, species detail and clearer identification of small amounts of vegetation are not directly obtainable.

Comments

If you are the rightful copyright holder of this dissertation or thesis and wish to have it removed from the Open Access Collection, please submit a request to pdxscholar@pdx.edu and include clear identification of the work, preferably with URL.

Persistent Identifier

https://archives.pdx.edu/ds/psu/32359

Share

COinS