The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences. Vol. XXXVII. Part Bl. Beijing 2008 
106 
Figure 1. Sun-shadowed view of the test site, Gold Coast City, 
Queensland, Australia. 
in Figure 1. The geomorphology of about half of the site is 
composed of low, flat alluvia; the remainder is formed by 
eroded hills and valleys of volcanic origin. 
0.4 
03 
>> 
o 
c 
CD 
3 
cr 
a> 
LL. 
0.1 
0 
0 50 100 150 200 
Elevation (m) 
Figure 2. Histogram of elevations of the test site. 
The highest point is about 377 m a.m.s.l. The mean and median 
elevations are about 29 m and 11 m, respectively. Figure 2 
shows a histogram of the terrain’s elevation, which resembles 
an exponential distribution. The degree of the terrain’s 
roughness can be seen in Figure 3. The mean slope for AOI is 
about 3°, which is classified as flat terrain. 
The AOI land cover can be divided into four land cover classes, 
e.g., a) agricultural land (predominantly under sugar cane and 
pastoral areas); b) trees and shrubs usually forming dry 
rainforest and open eucalyptus forests of grey gum (.Eucalyptus 
punctata) open-crowned tree, blue gum (Eucalyptus 
tereticornis), and stringybark and tallowwood (Eucalyptus 
microcorys) trees. Mean tree height is about 20 m, but it can 
reach 45 m in instances; c) high-density housing estates with 
very low to none tree cover; and d) water bodies in the form of 
small dams or lakes, and artificial channels connected to rivers. 
Land cover of about 10%, 57%, 19%, and 14% constituted 
classes a, b, c, and d, respectively. 
0.5 
Max = 31° 
Mean = 3° 
q ^ Median = 2° 
STD = 3.7° 
c 
CD 
3 
0.1 
Q 
0 5 10 15 20 25 30 35 
Slope (deg) 
Figure 3. Histogram of the terrain’s slopes of the test site. 
2.2 Data 
SRTM Data 
The SRTM.C, so-called finished data, version 2 (cell 
S28E153.hgt.zip) were downloaded from the JPS/NASA site: 
ftp://e0srp01 u.ecs.nasa.gov/srtm/version2/SRTM3/Australia/. 
This is the recommended source of the SRTM.C because its 
three arc-second downsampling has been achieved using the so- 
called averaging procedure (Becek, 2007). A 15 arc-minute 
SRTM.X one arc-second cell (starting at 153.25 E, 28.0 S - 
bottom left comer) has been purchased from the German 
Aerospace Center (DLR). All pixels falling within water bodies 
were removed. Every pixel from the SRTM.X data set was 
associated with the so-called height error map value (HEM), a 
pixel-based accuracy measure provided as a part of the X-band 
data package. The HEM value is statistically determined from a 
neighbourhood of pixels mainly considering the phase and 
baseline stability. Thus, it is a relative measure of the precision. 
The HEM varies in a range from 0 to 255. The C-band 
elevations are referred to the sea level means. The X-band 
ellipsoidal elevation was converted to mean sea level using the 
AusGeoid98 model http://www.ga.gov.au/geodesy/ausgeoid/. 
The C-band pixels were also associated with an accuracy 
measure derived from the slope of the terrain. In the following 
section, a procedure facilitating that is described. 
DTM Data 
As reference terrain elevation data, a set of spot elevations was 
used. The accuracy of the photogrammetically/lidar derived 
elevations is better than ± 0.3 m (1 a). The mean density of the 
spot elevations was about 48 points /ha. 
Land Cover Data 
No of points = 46,980 
Max = 377m 
Min = Om 
Mean = 29.43m 
Median = 11.08m