ver
face
sat
It
have
ive
rban
put
(in
en-
"mi À
E uen M LEONE es me ER Bi ms eA SE
Variance
sut
0 Housing
Density
FIGURE 3: Schematic relationship between response variance over a
3x3 pixel neighbourhood and increasing housing density.
(b) Band TM 7 (2.08-2.25 um) should have particular advantages in roof
material discrimination. This band is particularly sensitive to hydroxyl
ions in rocks due to clay minerals which are essentially hydrous aluminium
silicates. These clays (particularly potter's clay and Kaolin) form the
major component of terra cotta roof tiles, which at present is the major
residential roofing material used in the Sydney region.
(c) Other roofing material, such as concrete tiles, is impregnated with iron
oxide, which has particular sensitivity in the 0.45-0.52 range (TM 1).
Many of the newer residential areas on the edge of Sydney are now being
roofed with cheaper concrete tiles (in contrast to older areas which are
predominantly clay tiles) and so this band may contain additional urban
information.
(d) The thermal infrared band TM 6 (10.4-12.5) will have a spatial resolution
of 120 metres which is generally too coarse for urban studies, however
broader scale differentiation of natural and man-made features will be
possible as man-made surfaces (particularly roads) are efficient absorbers
and therefore emitters in this spectral range.
COMMENTS AND CONCLUSIONS
The new generation of satellite systems wili be of considerable advantage to
urban studies primarily because of their increased spatial resolution which will
result in an improved detection ot urban elements, a reduction in the number of
surfaces per pixel and an increased potential for the use of textural data.
The spectral bands of Landsat D TM, because of their placement and number,
should allow greater discrimination of most urban surfaces.
Of the two systems considered, SPOT wili have the higher spatial resolution and
it would seem reasonable to suggest that ideally Landsat D TM data should be
used primarily to discriminate surface types while the SPOT panchromatic data
(at 10 metre resolution) be used to provide higher resolution cartographic and
contextual information. Individually, Landsat D TM and SPOT will have their own
advantages, one having better spectral characteristics, the other better spatial
characteristics.
At the research level remotely sensed data is increasingly being analysed with
existing auxiliary data, (such as topographic, socio-economic and tenure
boundaries), and incorporated into spatial information systems. The increased
spatial resolution of satellite remotely sensed data will aid this research and
should lead to operational systems having benefit in a number of areas of urban
study, for example;
$95
AR
e ONE MEE. N 08
RS