uropean 
n three 
uropean 
over; 
(E Land 
an and 
level. 
roviding 
ation of 
uropean 
on, land 
centrate 
rent and 
essment 
n of the 
on the 
lopment 
for the 
product 
lure and 
ment of 
the use 
graphic 
of the 
abase of 
archived 
and use 
of the 
1998); 
mission 
S. 
PPING 
ovide an 
ing the 
visual 
mapping 
rea to be 
nts (100 
nd cover 
  
  
EUROPEAN ENVIRONMENT AGENCY 
  
  
PHARE PROGRAMME 
  
  
Executive Director « 
> Task Manager 
  
Programme Manager - 
v 
Project Co-ordinator 
  
  
  
  
  
  
  
  
  
ETC ETC 
Partner 1 Partner 2 
Project Manager < > EEA Phare Expert 
^ A 
i extended EUROPEAN TOPIC CENTRE | i 
ETC Leader PTL Leader 
a M Min ip 
PTL PTL 
Partner 1 Partner 2 
  
  
  
  
  
  
  
  
  
[ Task team 1 | | Task team 2 1 [ Task team n ] 
EEA MEMBER COUNTRY PHARE PARTNER COUNTRY 
NFP/NRC NFP/NRC 
  
  
  
  
Figure 1: Institutional connections of the CORINE Land Cover project (see text for details) 
Raw satellite images first have to be pre-processed and enhanced to 
yield a geometrically correct document (satellite image map) in the 
desired map projection. Digital film writers are used as output device 
to produce an optimal hardcopy for visual photointerpretation. 
Recently large sized, high-resolution computer printers are becoming 
also a selectable alternative for output production. In the course of 
interpretation, the photointerpreter draws polygons on transparent 
overlay, fixed on the top of satellite image hardcopy. The 
photointerpreter should consult simultaneously all available ancillary 
data (topographic maps, aerial photographs, vegetation maps, land 
cover and land use statistics, etc.). The polygons are coded according 
to one of the items (a three-digit code) of the nomenclature. Although 
display and evaluation of satellite imagery on screen of image 
processing system is highly recommended to compensate for 
drawbacks of using hardcopy images, it depends on availability of 
appropriate facilities. 
  
satellite image processing: 
rectification, enhancement, hardcopy 
  
  
  
  
photointerpretation, 
field checking 
  
  
  
  
  
digitisation of interpreted map sheets 
  
  
  
  
  
  
  
  
  
  
  
verification 
| merging adjacent map sheets | 
| CORINE Land Cover database | 
  
Figure 2: Flowchart of CORINE Land Cover mapping 
Table 2: 
Basic characteristics of the CORINE Land Cover database 
  
e A Area coverage: 3,5 million km’, 23 countries (1998) 
fully completed 
e Method: computer assisted satellite photointerpretation 
e Satellite images: LANDSAT, SPOT, etc. 
e Ancillary data: topographic maps, vegetation maps, 
land use and land cover statistics, etc 
e Working scale: 1:100 000 
e. Minimum mapping unit: 25 hectares 
e Minimum linear feature width: 100 m 
e Nomenclature: hierarchically structured in 3 levels: 5 
classes for level-1, 15 classes for level-2 and 44 classes 
for level 3 
e Cartographic projection: Lambert azimuthal equal area 
e X Various projections in country databases 
e Implementation: national teams 
e Supervision: Land Cover Technical Unit 
e Overall classification reliability: better than 85% 
e Geometric accuracy compared to topographic map: 100 
meters RMS or better 
e Mean cost: 5 ECU/km? 
  
  
  
In the latest time however, with the rapid spread of lower cost 
hardware and software systems, computer assisted 
photointerpretation (CAPI) technology is more and more applied and 
provides better thematic and positional accuracy. 
Landsat Thematic Mapper (TM) data have been the most widely 
used satellite imagery in the course of the project, because it provides 
good area coverage, sufficient thematic and geometric details for a 
reasonable cost. Recently, new high- resolution satellites with a 
International Archives of Photogrammetry and Remote Sensing. Vol. XXXII, Part 7, Budapest, 1998 635