756 
eastern Finland indicated some signitleant-level 
correlations. Index values calculated from the MSS 
channel 4 and 5 values indicated information of lake 
depths, Secchi disc depths, water colour, iron and 
turbidity. Channels 6 and 7 added some information 
of such production factors as nitrogen, phosphorus 
and chlorophyll a. Because of the low level of 
nutrients only a qualitative approach was possible 
within these lakes (Raitala et al., 1984 a,b). 
Figure 1. Landsat MSS derivative (channel 4/channel 
5) vs Secchi disc transparency (SDT) values (upper) 
and channel 6 vs total nitrogen (lower) for lakes of 
the Kuusamo area (Raitala et al. 1984b). 
3. VEGETATION MAPPING 
According to differences in the reflected spectrum 
it is possible to distinguish different life forms 
and vegetation units from one another and likewise 
evaluate the share of turbidity and production areas 
within distinct water areas. Helophytic vegetation 
reflects near-infrared radiation more than 
nymphaeids and much more than bryophytes, elodeids 
and isoetids. The density of the vegetation unit is 
also one of the main factors which must be taken 
into account. 
Several aquatic areas in different parts of 
Finland were classified using both supervised and 
unsupervised classification procedures. Once again 
attention must be paid to the critical value of 
field work when preparing and testing the classifi 
cation. Because of the relatively rough pixel 
ground resolution (0.5 ha) the classification does 
not provide a neat correspondence with a unicon- 
ceptual parametric map but makes it possible to 
inspect different surface complexes and environ 
mental gradients. 
The spectral reflectance of aquatic areas depends 
on water depth (Hammack 1977, Arkimaa and Raitala 
1984), bottom type (Raitala et al. 1984a), aquatic 
plants (Raitala et al. 1984 a,c, 1985), water 
quality and Secchi disc depth (Lindell 1980, Arkimaa 
and Raitala 1981, Raitala et al. 1984b). The 
effects of these variables may be substantially 
interwoven and intermixed. The most effective vari 
ables are most easily mapped by means of satellite 
classification, while the fainter ones can be 
identified only less distinctly. Computer-aided 
multispectral classification seems to provide an im 
portant tool for obtaining qualitatively different 
classes from within aquatic areas, but quantitative 
means of evaluation should also be developed further 
before this technique can be used in routine 
studies. 
MJOk. 
" 'W 
c 
srasfegh 
Figure 2. Classification of the coastal waters of the 
Gulf of Bothnia off Oulu. C = coastal open deep sea, 
D = open shallow sea, E = dilution and submersed veg 
etation, F = bottom and vegetation effects, G “ dis 
charge dilution and vegetation, H = discharge from a 
small river (Arkimaa and Raitala 1984). 
In the mi 
to obtain 
from withi 
supervised 
categories 
which slij 
categories 
quite reas 
classes an 
depend on 
present an 
ideal repr 
the classe 
forming a 
ference fi< 
tions of tl 
a small n 
forming an 
category m 
variations 
Landsat 
aquatic ar 
conjunctioi 
familiar i 
water area; 
date infor; 
aquatic er 
manner. 
<4 : 
Figure 3. I 
of the Bay 
dicates dil 
(Raitala el 
Eleocharis 
IV = Scirpi 
palustris, 
lustris, E! 
bottom, Vi: 
cularis on 
Limosella, 
4. MULTIT 
Because al 
reflected 
time-relat 
aquatic re 
of enviroi 
littoral p 
rent plac 
fluetuatin 
the northe 
upheaval o