of the report by Dexter, Heislers and Sloan (1977). The fire damage
was classified into three burn classes in which the foliage of the
dominant vegetation was either not damaged (lightly burnt), scorched
(moderately burnt) or consumed (severely burnt). Only small isolated
areas were left unburnt and these could not be distinguished on the
photos from the slightly burnt class. More detailed planimetric base maps
of the general Mt Buffalo area at a scale of 1:50 000 were also made
available by the Forests Commission. Other support material by way of the
relevant 1:100 000 topographic map in the National Mapping series and a
vegetation map of the Mt Buffalo National Park (Rowe 1970) also existed.
The photointerpreted map of the fire area was to serve two purposes -
to aid in the analysis and to provide a datum against which the results of
Landsat analyses could be compared.
analysis Proceedure
Visual imagery
First generation visual images in each of the four M.S.S. bands of the
portion of the scene covering the fire area were prepared from the C.C.T.
on an Optronics Photowrite by the Division of Mineral Physics, CSIRO. In
addition to the image restoration process usually applied by this group
(Green 1977), the image was enhanced by contrast stretching using a linear
function to concentrate the available range of film densities on the
forested area at the expense of the cleared country. This imagery has been
examined only superficially at this stage.
Computer Analysis
The computer analysis system chosen for the study was the CSIRO-ORSER
multispectral data processing system described by O'Callaghan and Turner
(1977). This system is flexible in its capacity, general in approach
and relatively easy to use. It is based on the ORSER system (McMurtry
et al. 1974) developed by the Pennsylvania State University, Office for
Remote Sensing of Earth Resources. Implemented progressively from early
1977 on the CSIRO computing network centered in Canberra, the CSIRO-ORSER
system has been interfaced with a generalised image handling system, DISIMP
(O'Callaghan and Aujesky 1978) to facilitate the display of image data and
maps.
In order to minimise computing costs and permit concentration on the
area of the fire and the nearby unburnt area, a subimage of the Landsat
Scene was prepared. A standard colour composite image was used to locate
the relevant portion of the C.C.T. The subimage included appreciable areas
of unburnt vegetation types similar to those damaged by the fire so as to
provide data on signatures of unburnt forest for comparison with signatures
of the burnt types. This was also considered to: be important in examining
the effects of misclassification.
The digital image of the subset atea was preprocessed to remove the
six line striping effect by histogram equalisation and missing lines of
data were replaced by interpolation. The image was also corrected for
earth rotation and a new digital image produced. Being in ignorance of the
spectral characteristics of the burnt and unburnt types, recourse was made