314 
with the corresponding patterns on the vegetation map and the false colour photographs. The smaller lower 
area with the Aster. Spartina and Puccinellia vegetation was separated from the other part of the area but 
within this area these types were not very well allocated. It seems that Aster. Spartina and Puccinellia are 
not separable with Landsat and ihat these types will have to be combined to one class: vegetation on lower 
mtirsh with Aster. Spartina and Puccinellia. A major cause for this poor discrimination will be the lesser 
vegetation cover of this part of the area, besides the fact that the patches of the vegetation types are smaller 
here. Probably a lower vegetation cover is also responsable for the worse discrimination of vegetation types 
on the May and September images. 
The multitemporal classification with bands 3, 4 and 5 off all three images was not as good as the July 
classification and did not add any extra information. This worse result of the multi temporal classification 
can be attributed to the addition of ’noise’ by adding the May and October bands which are, for all 
vegetation types, less suitable for separation in comparison to the July bands. 
Agrostis ...... Phragmites Sdrpus — Elymus — Trichoglin 
A 
B 
r«fl*ctanc« (%) 
wave length (nm) 
6 May 1903 
reflectance (%) 
I 
rcflactanc* (%) 
r«fl*clWK* (%) 
Figure 5. Average reflectance curves (n= ca. 10) ± l standard deviation of vegetation types in May (left) and 
September (right); A: full spectrum, B; visual part of spectrum in detail