216 
It is NASA's plan to close down the STDN system when TDRSS becomes 
operational. Although post-flight analysis of tracking data can theo 
retically reduce errors, this is not done routinely. Thus even if 
spacecraft position were the only error, current tracking systems could 
at best meet planimetric accuracy requirements for 1:250,000-scale maps. 
However when the Global Positioning System comes on line there is a 
clear possibility of meeting requirements for 1:50 ,000-scale maps. 
Spacecraft and sensor attitude can be obtained by horizon sensors, 
inertial measuring units (IMU), stellar reference systems, or combi 
nations of these. There may also be a problem of spacecraft flexure 
("hot dogging") between the location of the attitude sensor and the 
terrain sensor. Attitude accuracies for current and proposed space 
craft are given in Table 3. 
Table 3 
SPACECRAFT ATTITUDE ACCURACY 
Spacecraft 
System 
roll 
pitch 
yaw 
Landsat-1, -2, -3 
horizon + gyro 
0.01° 
0.01° 
1.0° 
Landsat-4 
IMU + stellar 
0.01° 
0.01° 
0.01° 
Shuttle 
IMU 
0.5° 
0.5° 
0.5° 
star camera 
5" 
5" 
5" 
EO stellar 
2" 
2" 
2" 
Attitude errors (in radians) multiplied by altitude translate directly 
to positional errors. Their affect on topographic elevations is more 
complicated but easily calculated from standard formulas. 
LANDSAT-4 
Landsat-4 was launched July 16, 1982. The Thematic Mapper produces a 
30 m/pixel which is compatible with 1:500,000-scale image maps. The 
spacecraft carries a GPS transponder, but the full complement of GPS 
satellites will not be available until 1987. Consequently, except for 
limited experimental tests, Landsat-4 positional accuracy will be that 
available from STDN. The 0.01° attitude error results in approximately 
125 m error in ground point positions derived from the imagery. The 
resultant planimetric error would exceed that allowed for 1:500,000- 
scale maps. Landsat-4, like its predecessors, does not provide useful 
stereo so that topographic relief cannot be compiled. 
MAPSAT 
The U. S. Geological Survey, concerned with the cost and complexity of 
Landsat-4, funded a contractor study for an electro-optical system 
called MAPSAT with the objective of producing multispectral data and 
geometric accuracy sufficient for 1:50,000-scale mapping (Colvocoresses, 
1982). The spacecraft would be at 920 km altitude - the same as 
Landsat-1, -2, and -3. The sensors would be three multispectral linear 
arrays, one looking 23° forward, one vertical, and one 23° aft. This 
would provide a stereo system with B/H ratio of 1.0 to produce topo 
graphic elevation data. The optics would be capable of 10 m/pixel reso 
lution. The spacecraft would have the GPS for position and an electro- 
optical stellar system for attitude determination. 
According to the criteria developed in this paper, Mapsat would be able 
to meet the geometric requirements for mapping at 1:50,000 scale with 
a contour interval of 25 to 40 m. However the ground resolution would