Studentized
difference
60
90
SSS
D
40
30
20
10
7
7
7
V4
7
ai A
v
uf
A
A
s
B Mean
e
I ERI
—10
t N1988—1990
p H1990—1993
i
s Mean
i H1990—1992
[__] Mean
H1992—1993
Unt C.thinn.
Unc. thinn.
Reg. cut
Prep. cut
Soil prep.
Clear cut
Figure 8. Studentized difference of treatment class means on different image pairs on TM channel 7.
4. CONCLUSIONS
Two problems in relation to preparing generic training data
for forest change detection were studied. First the effect of
timing of changes to spectral response of changes within
three year interval between Landsat TM images was
focused on. It was demonstrated that the timing of the
changes within this period did not affect the spectral
separability of the treatment classes under question.
Secondly, possibilities to compose generic training data for
supervised change classification was studied. It was
demonstrated that after regression calibration and
studentization, the image pairs covering the same
geographic location could be put to the same level. Range
scaling seems still necessary for making image pairs from
different areas radiometrically comparable after the
calibrations proposed. However, the data available was too
limited to make any final conclusions. It was estimated that
based on the calibration methods proposed, the
silvicultural treatments can be separated at stand level. In
addition, it can be expected that the forest damages at the
magnitude level of thinnings, for example, can be
separated. This means that the 20-30 % defoliation or wind
damage should be separable. However, the spectral changes
caused by normal growth after the canopy closure will not
be separable on stand level in short intervals in the Boreal
Forest conditions
REFERENCES:
Háme, T. 1991. Spectral interpretation of changes in
forest using satellite scanner images. Acta Forestalia
Fennica, 222, pp. 1-111.
Olsson, H. 1993. Regression functions for multitemporal
calibration of Thematic Mapper data over boreal forests.
Remote Sensing of Environment, 46, pp. 89-102.
Olsson, H. 1994. Monitoring of local reflectance changes
in Boreal Forests using satellite data. Report 7, Swedish
University of Agricultural Sciences, Remote Sensing
Laboratory, Umeá, Sweden.
Rousseeuw, P.J. and Leroy, A. M. 1987. Robust
regression & outlier detection. John Wiley & Sons. New
York, Chichester, Brisbane, Toronto, Singapore, pp. 1-329.
733
Singh, A. 1989. Review article - digital change detection
techniques using remotely-sensed data. International
Journal of Remote Sensing, 10(6), pp. 989-1003.
Varjo, J. 1996. Controlling continuously updated forest
data by satellite remote sensing. International Journal of
Remote Sensing, 17(1), pp. 43-67.
Varjo, J., and Folving, S. 1996. Regional monitoring of
forest changes using unsupervised methods; a case study
from Boreal Forest on mineral soils. Submitted to
Scandinavian Journal of Forest Science.
Weisberg, S. 1985. Applied linear regression. John Wiley
& Sons, New York, Chichester, Brisbane, Toronto, pp. 97-
119.
International Archives of Photogrammetry and Remote Sensing. Vol. XXXI, Part B7. Vienna 1996
