flight
| with
and 5
^S are
data
round
t the
its of
f the
ed as
ind 5.
VUE NA MN NAA ARA
at Em
of the
nation
———— ———
e —
A E ae
iz in a) is estimated poorer by about a factor of 4 than uxy,
due to the geometry of the three-line scanner (see chapter
2.1). The planimetric accuracy limit in b) increases by a
factor of 3 in comparison with that in a), because the
changed control information (125 XY GCP with o = 25m
instead of error-free GCP) defines the datum in planimetry
less accurately. The height accuracy limit in b), however, is
improved in comparison with that in a), because the
changed control information defines the datum in height
more accurately. The DTM namely supplies height control
information for all 11.905 object points.
Moreover, the resulting accuracy improves generally with
better precision of the position and attitude observations
and with an increasing DOI. This effect is more significant
using only 4 GCP than a lot of low accurate XY GCP and
a DTM. In case of a long distance (160 km) between the
orientation images observations for the exterior orientation
are dispensable.
From Figure 4 it is evident that the ground control
information derived from topographic maps is not sufficient
for attaining a high planimetric accuracy of about 1 to 3 m.
Therefore, accurate GCP are required.
Figure 5 shows, that the accuracy demands of the mission
(cz — 5 m) can be fulfilled with a short DOI (12 km) and a
standard deviation for the position and attitude data of 2 m
/ 10 mgrad, on condition that a DTM is available. Using
the same DOI (12 km) and the same standard deviation for
the position and attitude data (2m/10mgrad) the height
accuracy decreases to 12.2 m if there are only 4 XYZ GCP
instead of the DTM.
3.2.2 Non-standard flight configurations The
computations are carried out with a selected set of input
parameters only:
» DOL 12 km,
» standard deviations for position and attitude observations:
2 m/ 10 mgrad (relative accuracy),
» 4 error free XYZ GCP for each strip, located at the
corners of the 3-ray areas.
For these assumptions the aspired accuracy of 5 m
obviously can not be achieved with standard flight
configurations, as described above. However, the results are
considerably improved in case of non-standard flight
configurations.
Table 2 shows the rms values uy, and py; of the theoretical
standard deviations oy, o, and o;. The values are calculated
separately for 3-ray and 6-ray points. The number n of the
corresponding object points gives an idea of the area,
covered by the respective flight configuration (700 points =
1000 km?). For comparison the results, obtained by
standard flight configurations, are listed in the first line.
The simulations, assuming 30? cross inclination of the shuttle
results in an improved accuracy of the 3-ray points by a
factor of 1.7 in height and of 2.8 in planimetry, compared to
the normal flight attitude. Best results are obtained by the
simultaneous adjustment of two crossing strips. Within the
overlapping 6-ray area a constant accuracy level is achieved,
which is nearly independent of the intersection angle. This
result is of high practical importance. The weak 3-line
geometry, caused by the along track parallel perspective, is
completely overcome by the block adjustment of two (or
more) intersecting strips. According to the figures in
Table 2, small intersection. angles are more economic,
because they lead to larger 6-ray areas and better accuracy
in the 3-ray area. The acquisition of MOMS-02 imagery
from crossing orbits is planned. Their ground tracks will
mainly be located between 20? and 28.5? northern latitude,
which will cause small intersection angles.
4. CONCLUSION AND OUTLOOK
The aspired height accuracy of about 5 m can be achieved
either by means of DTM information, derived from existing
maps, or by the simultaneous adjustment of two (or more)
crossing strips within the overlapping area. The intersection
angle is of no importance.
The results are generally improved by more accurate
observations of the position and attitude parameters and by
an increasing distance between orientation images. High
planimetric accuracy is only achieved, if precise ground
control points, e. g. from GPS measurements, are available.
The weak 3-line geometry along track can essentially be
supported either by cross inclination of the shuttle or by the
simultaneous adjustment of two or more intersecting strips.
After the mission it will be possible to prove these findings
by the evaluation of practical data.
The three-line concept for the acquisition of digital stereo
imagery has been realized in conjunction with some other
important project: MEOSS, MOMS-PRIRODA and
HRSC/WAOSS.
MEOSS stands for Monocular (single lens) Electro-Optical
Stereo Scanner. It flew successfully onboard an aircraft in
the last years and will be payload on an Indian rocket,
scheduled for launch in autumn 1992.
number intersection Cross 3-ray area 6-ray area
of strips angle inclination n Mx Iz n Hxy Hz
1 - 0? 5951 13.6 122 - -
1 - 30? 5038 4.8 7.0 - -
2 S? 0? 1896 1.9 6.3 5307 1.6 4.9
2 30° 0? 9536 24 6.5 1803 1.5 4.7
2 55° 0° 11561 3.4 7.8 1097 1.6 4.8
Table 2: Number of object points n and rms values juxy and j4z for non-standard flight configurations, assuming 12km DOI,
2m/10mgrad std.dev. for position and attitude observations, 4 error-free GCP for each strip.
463
