Mapping - TR-01-40016

Figure 4. System of hierarchically classifying ecosystem components into integrated terrain units

(ITU) and further aggregating and simplifying ITUs into ecotypes.

tion of Landsat Imagery, geomorphology derived from

for their own individual purposes. Ground data can be

the ecosections map, and physiography from the digi-

reclassified and analyzed by regrouping characteristics

tal elevation model and ecodistricts map, we could in-

in Appendix A and applying the new organization to

tegrate several landscape components in a way that was

the plot database (Appendix B).

To classify ecosystems at smaller spatial scales, we

for mapping Fort Wainwright (Jorgenson et al. 1999).

used geomorphic and physiographic criteria (Table 2).

Ecosections were differentiated on the basis of their

Ecotypes

geomorphic patterns and processes, and we named them

Ecotypes were mapped by computer processing of

after geomorphic units. We classified ecodistricts and

Landsat TM imagery. The processing followed a series

ecosubdistricts on the basis of general physiographic

of sequential and iterative steps: acquiring data, cor-

characteristics that were related to associations of geo-

recting and classifying the image, using photo inter-

morphic units. Because each ecodistrict is unique, we

pretation techniques on it, developing conceptual mod-

named the areas on the basis of a general physiographic

els of ecological relationships, and developing a final

descriptor (e.g., lowland or highland) and a prominent

classification using a rule-based system that incorpo-

geographic feature (e.g., nearby creek or mountain).

rated ancillary data (Fig. 5).

Mapping

Data acquisition included compiling satellite imag-

ery, aerial photography, and other ancillary informa-

We mapped the ecosystems at three scales: ecotype

tion. The Landsat TM satellite image (Path 68, Row

(1:50,000), ecosection (1:100,000), and ecodistrict and

15), obtained on 10 August 1994, has a spatial resolu-

ecosubdistrict (1:250,000). The ecotype map was based

tion (pixel size) of 28.5 m. Aerial photography included

on the unsupervised classification of the spectral char-

complete coverage of 1979 and 1980 in color-infrared

acteristics of the Landsat Thematic Mapper (TM) im-

(1:63,000 scale) and 1996 in true color (1:24,000), and

agery and the ruled-based classification of spectral clus-

partial coverage of 1996 in true color (approximately

ters using ancillary data. The ecosection and ecodistrict

1:1000). The color-infrared photography was used for

maps were based on photo-interpretation. By incorpo-

mapping ecosections. We referred to the medium-scale

rating vegetation structure from the spectral classifica-

10