DIGITAL ANALYSIS OF LANDSAT DATA
FOR SELECTIVE FOREST STAND TYPING
— AN INITIAL FEASIBILITY STUDY
Victor Zsilinszky
Associate Director
Ontario Centre for Remote Sensing
Ministry of Natural Resources
Toronto, Ontario, Canada
and
Dr. Simsek Pala
Head of Research and Training
Ontario Centre for Remote Sensing
ABSTRACT
To establish a methodology and system for supplementary forest
inventory from spaceborne data and to determine the parameters accept-
able to the inventory standards for the boreal forests of the Province
of Ontario, Canada, Landsat digital tapes were analysed applying the
supervised classification method,
Twenty-five 4-ha. test sites in the boreal forests were selected
so as to include the widest possible variety of species and combinations
of species, and of crown density. A 100% tally was done on these plots
of tree species, height, d.b.h., age, and crown density. These plots
were then used as training areas for computer analysis. The results
from one area containing three such plots are described.
The delineated boundaries of the extrapolated forest inventory
data derived from the tapes were analysed and compared with the bound-
aries as interpreted from panchromatic aerial photographs presently used
as the data base for forest inventory in Ontario.
; The correlations are analysed and problems and limitations that are
inherent in the methodology are discussed.