Stratified Landsat classification of north-central Idaho and adjacent Montana (p. 263-266)

Publication Type:

Conference Paper

Source:

Proceedings—land classification based on vegetation: applications for resource management; 1987 Nov 17–19, US Forest Service, Intermountain Research Station, Ogden, UT, Volume INT-252, Moscow, ID, p.pages 263-266 (1989)

Call Number:

A89BUT01IDUS

Other Number:

GTR-INT-257

URL:

https://doi.org/10.5962/bhl.title.100156

Keywords:

grizzly bear, Landsat classifications

Abstract:

In 1985 we began a stratified Landsat classification to classify and map vegetation on 3.5 million acres of land in north-central Idaho and adjacent Montana and provide a habitat data base to wildlife managers and researchers. Our objective is to classify and map vegetation in sufficient detail to serve as a data base for elk and grizzly bear habitat research. However, we realize that we are creating a highly detailed and generic classification that has many potential applications, wildlife being just one. The classification involves five steps: 1) topographic classification, 2) Landsat spectral classification, 3) collection of reference, or ground-truth, data, 4) stratification, and 5) incorporation of additional data layers. Defense Mapping Agency digital elevation data were used to create three topographic data planes: elevation, aspect, and slope. Landsat 5 MSS data were classified into 150 spectral classes using a guided clustering approach. Reference data have been collected for 1 ½ field seasons and, when completed, will be used to correlate the spectral classes with plant communities. We have been able to correlate some groups of spectral classes to general vegetation types with our existing reference data. The topographic and spectral data layers will be combined and used to split, or stratify, spectral classes that represent two or more vegetation types into distinct classes based on topographic modeling. Additional data planes can be used to stratify spectral classes based on features such as hydrology, snowchutes, geographic location, or any feature that can be mapped or modeled. Stratified Landsat classes that represent the same or similar plant communities will be pooled together to form working habitat classifications. Classes can be pooled in different combinations to produce different vegetation classifications for a wide variety of applications.

Notes:

ELECTRONIC FILE - Ecology

CSE citation:
Butterfield BR, Davis DL, Unsworth JW. 1989. Stratified Landsat classification of north-central Idaho and adjacent Montana. In: Ferguson DE, Morgan P, Johnson FD, compilers. Proceedings—land classification based on vegetation: applications for resource management; 1987 Nov 17–19; Moscow, ID. GTR INT-257. Ogden (UT): US Forest Service, Intermountain Research Station. p. 263–266. https://doi.org/10.5962/bhl.title.100156.

Conference was in Nov 1987; papers published in a General Technical Report in Feb. 1989.