For
vere
lage
your
Map
GIS
area
for
for
ised
For
1atic
rior
tudy
mn)
onia
[ the
ture,
Jata.
ctral
case
lam-
Pedja area, this scheme was supplemented with mire cover classes
(all together 14 types) while anthropogenic classes were
presented only by some types. Differentation of land cover
categories depends on the homoge-neity of cover classes and their
size. An aggregation of these classes was made twice: into 9
categories for landscape moni-toring purposes on the landscape
level and into 4 categories for better visualisation of the changes
(transitions between 9 classes need 72 colors on map!).
Table 1. Classification scheme of Estonian land cover classes and aggregation possibilities.
maps
Features on change detection 9 classes for land cover monitoring 34 classes for land cover change detection (Aaviksoo,
1995b, modified)
WATER Water
and
floodplain grassland
Water
Sedges, reeds
NATURAL
COVER
Minerotrophic fen
Open fen
Shrub fen
Wooded fen
CLASS ES
Mixotrophic transitional bog
Swamp with birches and pines
Swamp with pines
Oligotrophic bog
Wooded bog
Complex bog
Open bog
Coniferous forest
S EMI-
Young spruce (incl. planted stands)
Old spruce
Young pine (incl. planted stands)
Old pine
Mixed (coniferous prevailing)
NATURAL
COVER Deciduous forest
CLASSES
Mixed (deciduous prevailing)
Birch
Alder
Aspen, ash, oak, maple
Thickets
Fresh clear-cuts
Thickets (natural re generation)
Grassland
Natural grasslands and hayfields
Cultural grasslands
ANTHRO-
POGENIC
COVER Arable land, mining of
CLASSES natural resources, settlement
Bare soil (dry)
Bare soil (fresh)
Bare soil (wet)
Summer crop (barley)
Summer crop (oats)
Winter crop
Vegetable fields
Peat mining
Abandoned peat mining
Sand and gravel pits
Urban and built up areas*
* urban cover class was cutted off the image, not classified.
Classification methods used. Investigation of the area was
started by identifying all spectral classes (automatic or
unsupervised classification, ISODATA). Using auxiliary material
we tried to label each of the proposed 52 classes. We succeeded to
identify 21 classes because many clusters belong to the same
class. For better results we carried out supervised classification
using all available material - classified cluster map and other
sources of information. Sometimes this classification type is
called hybrid (Figure 2).
Intemational Archives of Photogrammetry and Remote Sensing. Vol. XXXII, Part 7, Budapest, 1998 55
