Rokos, Demetrius
5.2 Exemplar of an ELCM layer produced. Conclusions.
A layer of the ELCM under implementation, showing buildings detected in a forest environment (category 4 of
potential law infringement) of the municipalities of Old and New Penteli, Attica, Greece, at scale 1:5000, is presented in
figure 6 and a map detail in figure 7. On this map layer (layer: Art. 71 Law 998/1979, Art. 45 par.] Law 998/1979), the
following information, appropriate and useful for judicial and information procedures, is shown:
e conditioned areas, ie. the forests and forest areas, resulting from the relevant maps produced by the Hellenic
Ministry of Agriculture. On these maps, land-use changes in forests and forest areas in the 1937 — 1978 period, in
the municipalities of Old and New Penteli, are shown at scale 1:5000. After being compiled with the information
included in the Maps of the Approved City Plan of Attica valid since 1985, this information was used to produce the
conditioned areas of the proposed ELCM.
e potential breach areas, i.e. buildings, detected using image processing methods and techniques and specifically,
classification based on texture analysis applied on the KVR1000 image captured in 1992 (Rokos and Karathanassi,
2000). This image was the most recent available satellite image of the area, which meets the image processing
spatial requirements. Consequently, the produced ELCM layer refers to 1992 data.
e other background information, i.e. road network, forest road network, drainage network, contour lines, municipality
boundaries, and annotations, which was selected from the respective topographical map (scale 1:5000) and overlaid
on the ELCM layer.
The spatial accuracy of the ELCM layer is 3m. Detected buildings in forests and forest areas present an omission error
of 8.39% and a commission error of 4.90%. The overall accuracy of the Image Processing products is 91.61%.
The methodology developed for the production of the ELCM was proved to be successful, despite the fact that the only
available very high resolution remotely sensed image (KVR) for the area was not of optimum quality. Consequently, the
certain prospect of continuous improvement of the spatial, spectral and temporal resolution of remotely sensed images
will result, with equal certainty, in the improvement of ELCM quality and the dramatic reduction in the probability of
confusion of land-use classes and/or objects, e.g. forests roads and/or rock outcrops confused for buildings.
ENVIRONMENTAL LEGISLATION COMPLIANCE MAP
1992 HOUSING DETECTION MAP IN FOREST ENVIRONMENT (Art. 71 & 45, Law 998/1979)
NaTU SS Lad.
Frag, Madang Technica University al Arens
NEXT Lasanry cif emat Sensing
Europes Sommisssn
ENVRONMENT 4-2 T97-4£37
Caumäinned ferestareas (Art. Z & 3, Um» 083/1879)
INN Nanferest areas hut. 2 33, Lus 32/1570)
HUE npproved ty plan, 1932
+ Patenlalbreach arean (buildings), 1982
Sa. KEN MD
Xe ME Sard nnd See
duae UEXD, Coop EU
tera ea
CM -a25438 24
ind E Ponisi, Attica, Greecs
Figure 6. Exemplar: Housing Detection Map in Forest Environment
International Archives of Photogrammetry and Remote Sensing. Vol. XXXIII, Part B7. Amsterdam 2000. 1275