(4) 
SS 
NS 
Between 
agers, the 
CD pixels. 
(5) 
nd to the 
Equation 
he camera 
given by 
» antenna 
y antenna 
ra. 
has been 
; remarks 
he origins 
identical. 
e rotation 
ectors p* 
plies that 
ional and 
hanges in 
nal and 
uniform, 
nslations 
have not 
rientation 
d for the 
; in most 
(1) are 
time, the 
  
"m cei b 
third in post mission. The quantities dR à and dr? however, 
are determined by calibration, either before or during the 
mission; for details see Schwarz et al (1993). To define 
T by calibration, a minimum of three well determined 
ground control points is required. The scale factor s is 
changing with the flying altitude of the aircraft above 
ground. It can, therefore, either be approximated by 
assuming a constant flying altitude, calibrated by 
introducing a digital terrain model, or determined by 
measurement, using either stereo techniques or an auxiliary 
device such as a laser scanner. For precise georeferencing, 
the latter techniques are the most interesting to be 
investigated because they would provide all necessary 
measurements from the same airborne platform and thus 
avoid datum problems. 
The above equation can be used to evaluate the 
georeferencing requirements for photographic systems, 
scanning systems, CCD fram imagers, and radargrammetric 
systems. The overall accuracy will depend on the resolution 
of the remote sensing device and the accuracy with which the 
parameters in Equation (5) can be determined. The important 
parameters are the accuracy of the position and attitude 
determination on the one hand and the stability of the sensor 
configuration on the other. This will be further investigated 
in the next section. It should be noted, however, that for 
radargrammetric systems, velocity is an additional parameter 
which has to be determined with high accuracy. It is required 
for motion compensation which strictly speaking is not part 
of the georeferencing process but part of the remote sensing 
process and therefore has to be accomplished in real-time. 
Since the position and attitude sensors discussed in section 5 
will provide velocity as a by-product, it will be included in 
the following discussion. 
S. PERFORMANCE OF POSITION AND 
ATTITUDE SENSORS 
To achieve the required accuracies in position and attitude, 
two major systems are currently available, GPS and INS. GPS 
is primarily a positioning device, measuring distances to 
satellites whose positions are known. It can be used as an 
attitude sensor, however, by transforming vector changes in 
a fixed antenna configuration into attitude changes. INS has 
two independent sensor triples to measure accelerations and 
angular velocities from which linear velocity, position, and 
attitude can then be derived by integration. 
The two systems have very different error characteristics 
which are due to the type of measurements used. GPS 
accuracies are essentially uniform and time independent. 
Variations in accuracy are mainly due to satellite 
configuration and atmospheric conditions. The error 
spectrum for position is essentially flat and more or less 
stationary. INS accuracies are heavily affected by the fact 
that all measurements have to be integrated to obtain the 
required position and attitude parameters. Since the error 
spectrum is not flat but shows some low frequency spectral 
lines, position and attitude accuracies deteriorate in a 
systematic manner as a function of time. Thus, short term 
accuracy is excellent and equivalent or better than GPS 
accuracy, long term accuracy is not and needs updating to 
stay in the range required for precise georeferencing. 
From an operational point of view, the higher output rate of 
inertial systems (typically 50-100 Hz) is a major advantage 
because the exterior orientation of each image or each scan 
line can be determined without interpolation or prediction. 
Current GPS output rates (typically at 2 Hz, with emerging 
systems at 10 Hz) will not allow direct computation. 
5.1 GPS Performance 
Table 3 summarizes the positioning and attitude accuracies 
that are currently achievable using GPS. The single point 
positioning error budget is dominated by Selective 
Availability (SA), especially satellite clock dithering. The 
achievable RMS accuracy quoted in the table, i.e. 100 m 
horizontal and 150 m vertical, is therefore essentially 
independent of the type of receiver used. Although they are 
not shown in the table, velocity errors are also affected by 
SA. An RMS accuracy of about 0.5 m/s can be achieved in 
single point mode. 
  
Model 
Accuracy 
  
Pseudo range point 
ositioning* 
100 m horizontal 
150 m vertical 
  
  
10 km 0.5 - 3 m horizontal 
Smoothed pseudorange 0.8 - 4 m vertical 
differential positioning 
500 km 3 - 7 m horizontal 
4 - 8 m vertical 
10 km 3 - 20 cm horizontal 
Carrier phase differential 5 - 30 cm vertical 
positioning 
50 km 15 - 30 cm horizontal 
200 km (with precise orbits, same as 50 km) 
20 - 40 cm vertical 
  
Attitude determination 
  
  
] m separation 
5 m separation 
10 m separation 
10-30 arcminutes 
4 - 6 arcminutes 
2-3 arcminutes 
  
  
*Selective Availability on, PDOP < 3, 2DRMS (95%) (DOT/DOD, 1992) 
Table 3: GPS Positioning and Attitude Accuracies 
195