plain the poor correspondence for the dry and moist classes. This illustrates well the difficulties of
including subjective components in the field measurements in an otherwise objective and unbiased system.
FIGURE 2
Correspondence matrix for computer and field estimates of sub-surface waterflow.
Estimati
on in the
field:
Seldom
Shorter
Longer
Seldom
582
50
17
Estimation
by computer:
Shorter
82
57
19
Longer
38
52
50
FIGURE 3
Correspondence matrix for field and computer estimates of ground moisture.
Estimation in the field:
Dry Mesic Moist
Dry
Estimation
by computer: Mesic
Moist
5 165 3
25 523 29
1 162 30
DISCUSSION
It seems that site variables in the existing SCS system can be estimated digitally with good
reliability. It seems also that the subjective measurements in the field can give doubtful results. It
will probably always be troublesome to achieve, using the current technique, a field classification of
sufficient accuracy to evaluate alternative methods.
The continued work with digital site classification will not use the existing SCS as reference. This
decision has been made partly because of the need to gain good ground truth, and because of low-
precision in the regression equations due to the use of simplified parameters. The reference material
will instead be plot data with site index determined with the higher-precision and also objective method
based on measurement of the height and age of the trees.
REFERENCES
HAGNER, 0 1989. Computer aided forest mapping and estimation of stand characteristics using satellite
remote sensing. SUAS, Remote Sensing Laboratory, Umea.
HAGGLUND, B & LUNDMARK, J-E 1977. Estimation of growth potential through site variables, scots pine and
norway spruce in Sweden (in Swedish). SUAS, Inst for Vaxtekologi och marklara, report 28.
875