y receiver
nary air-
lings had
order to
ently fast
duce the
he initial
ed on the
interrup-
cause sig-
vn as Cy-
f a turn-
Nation of
y details
ffice here
the flight
d.
1ity solu-
reference
pted. Re-
ceeded to
and filter
affect all
| the help
Ware pro-
ing signal
hase am-
ns which
5 possible
' approx-
mbiguity
rer, linear
ral prob-
] tests on
atic GPS
der of 10
amongst
ift errors
pted as a
; may oc-
turns. It
GPS drift
iguity so-
her hand
corrected
it can be
e to avoid
with dur-
onsidera-
igulation.
ces. If we
t with the
here is no
s by sta-
amended,
ts, not to
attempt any stationary base-line determination before take-
off, i.e. to start flying and carry out the photo flight mission
in the usual way, without particular care about GPS conti-
nuity. The GPS recordings (at both receivers) need only be
switched on a few minutes before the mission area is reached.
In case of signal interruptions during flight turns (within a
strip no serious interruptions are expected) the phase am-
biguity solutions are redetermined in the post-processing by
using the C/A-code or P-code pseudo-range positioning, and
by considering a dynamic modelling of the aircraft move-
ment. The solution may leave some systematic errors for
the following stretch of the GPS trajectory. Such drift er-
rors are practically linear, according to available experience.
They may change after each major signal interruption. In
the extreme case each strip may have its own linear drift
error. It still needs to be investigated how large distances
(several hundred km) between stationary receiver and mis-
sion area and long flight missions (several hours) will affect
the GPS drift behaviour. Linear drift errors can be assessed
and compensated in combination with aerial triangulation,
by including additional parameters into the combined block
adjustment, as will be discussed in chapter 3.
2.3 The datum problem, ground control
GPS positioning refers generally to the earth-centered rect-
angular GPS coordinate system WGS 84. The same is true,
in principle, for relative positioning, although a transformed
coordinate system may be locally tied to the reference point
at the stationary receiver. It is to be stated clearly that aerial
triangulation with combined blockadjustment can be carried
out in that case, without any ground control point, provided
the GPS trajectory is not interrupted over the complete flight
mission. The result would refer to the original or a local
transformation of the GPS reference system WGS 84.
Normally, however, the results of aerial triangulation (and
of mapping) are wanted in a national horizontal and verti-
cal reference system. There are no absolute geodetic trans-
formation formulae available, at present, which would pre-
cisely enough link the WGS 84 to a national reference sys-
tem. Hence, the datum transformation must be provided in
the traditional photogrammetric way, i.e. by some ground
control points which would preferably be given in both coor-
dinate systems. The standard recommendation is, for aerial
triangulation, to use 4 XYZ-ground control points, located
at about the corners of a photo-block. They are sufficient
to provide the datum transformation (in GPS blocks control
points have no accuracy function any more), provided the
GPS trajectory is continuous. They are even capable of cor-
recting for overall drift errors. It is only the geoid reference
which is not completely solved by 4 ground control points
alone. If the geoid is known its undulations can be superim-
posed in addition, or additional vertical control points could
introduce the geoid indirectly. There have been discussions
about using only one control point. It allows the determi-
nation of the shift parameters of a datum transformation.
All other parameters must be derived from the known geo-
graphical position. The result can only be an approximate
solution which may, however, be sufficient for low accuracy
requirements.
693
Ground control points can be used in two ways in GPS sup-
ported aerial triangulation. If the GPS trajectory is continu-
ous the GPS-block may be adjusted without control or by us-
ing them as GPS control only. The subsequent datum trans-
formation (if necessary in 2 steps, or non linear because of
map projection and spherical vertical reference) would then
use the control point coordinates in the national system. It is
preferable, however, and in case of different drift errors nec-
essary, to include the control points directly in the combined
block-adjustment.
In the approach just described the linear datum transforma-
tion parameters can be treated as additional unknowns and
solved for in the combined blockadjustment. That solution
would include the correction for overall linear drift errors. In
fact, both effects cannot be separated completely from each
other. Drift correction is identical with a datum correction,
although it may include different parameters.
In section 2.2 it was discussed that there may arise indepen-
dent drift errors, even per strip. It means that there can be
a datum problem for sections of the GPS trajectory, even
per strip. To solve in that case for all unknown drift param-
eters in combined blockadjustment additional information is
required, in order to prevent singularities. The additional in-
Fig.1 Control scenarios for GPS blocks
/\ horizontal
e vertical
case c
LS AM
case a
LS A
case b