■ 
■ V 
Secondly, streams, lakes and open peatland were defined as Wet. Land adjacent to and within 1 meter in 
height from these areas was defined as Moist (for silt and fine morain, 2 meters). 
Thirdly, land was classified as Moist if the wetness index was high. The wetness index is defined by 
O'LOUGHLIN 1986 as: 
x,y 
MbL 
H 
qO 
— dA 
q 
where: w 
M 
b 
L 
T 
T 
q 
qO 
dA 
x.y 
wetness index in pixel x,y 
local slope [%] 
width of the studied slope, here 1 pixel [m] 
mean hillslope length [m] 
soil transmissivity [m^/s] (set relative to 
0.5-1 for morains and 5 for fine sand) 
reference soil transmissivity [m^/s] (set to 1) 
net flow (precipitation) [mm/year] in one pixel 
reference net flow [mm/year] (set to 1) 
ground water catchment area to pixel x,y [m^] 
A manual threshold was set on the wetness index to describe remaining Moist areas. 
Finally, all land still not classified was set to Mesic. The classification is diplayed in FIGURE 6. 
Site variables that were not included in the study 
All site variables could not be derived from the map material used. Of these, the most important is the 
ground vegetation type. This categorical variable is used to indicate the nutrition status at the site. 
Fifteen different classes of ground vegetation type are defined in the existing SCS. However, only about 
five occur in the study area. The vegetation type correlates to some extent with other variables, 
especially the sub-surface flow and ground moisture. Other, less important, variables not included in 
the study were exposure and presence/absence of ditching in peat lands. 
Estimation of site index for the entire study area 
Given the estimated site variables above for each pixel, the site index for Scots pine (FIGURE 7) and 
Norway spruce were estimated using regression functions from the existing SCS. 
EVALUATION 
Two estimated variables, the sub-surface flow and the ground moisture, were evaluated against the field 
sample plots. 
The classifications of sub-surface flow showed good correspondence (FIGURE 2). In 73% of the cases, the 
computer and field classifications were the same. In 6% of the cases the classifications were opposite 
(Seldom/Longer, or Longer/Seldom). These cases were rechecked in the topographic map and in the field. 
It was then concluded that all these plots were subject to misclassification in the field or to a 
difficult borderline judgement. In no case could the computer estimation be labelled invalid. 
The ground moisture showed a very poor correspondence between field and computer estimates (FIGURE 3). 
After rechecking a number of field plots, it was concluded that a big portion of the plots had been 
doubtfully classified in the field. The Mesic class had been clearly overrepresented, which might ex- 
874