has
the
nt.
ind
not
rip
me
ing
ion
rne
the
ata
lel
of
ual
ted
rip
ere
ese
Erik de Min
Figure 2 illustrates this effect by showing an across-track profile of a small block of laser data. The character of the
height differences is strongly systematic, and can mostly be categorized as a vertical offset (z-shift) and an along-track
tilt (pitch) and across-track tilt (roll) in the height values. The amplitude of the offset and tilts can be up to several
decimeters. It is obvious that rms-values of the height values in overlapping areas with these height differences are much
larger than in areas where data is only available from a single strip.
In this paper, the various error sources during the acquisition of laserdata will be discussed and an approach to
determine and eliminate systematic height errors will be presented. These are, beside the mentioned offsets and tilts, also
cross strip parabolic effects, strip torsions and periodic effects in along-track direction.
eR i on
Figure 2: Across-track profile of shifted and tilted strips
2 SOURCES OF HEIGHT ERRORS
Laser altimetry is a rather new and complicated technique. The precision of the resulting height values depends on the
type of laser system used and on the measuring strategy. (Huising and Gomes Pereira, 1998) have given an overview of
errors in laser data measured by various systems, but a complete description of error sources is not yet available. The
errors can be categorized into four components which are illustrated in figure 3.
l. Error per point. Due to the measuring uncertainty of the laser scanner each point is affected with a random error.
This point noise is decreased by averaging several heights in order to compute mean height values for certain areas.
Error per GPS-observation. The GPS-observation interval usually is set to one second. Every GPS-observation is
affected with a random error as well. This error, however, is constant for all laser points measured during this
second. Usually, these points are lying in a strip-wide area of about 100 metres length (depending on flying speed).
3. Error per strip. GPS and INS sensors are needed to measure the position and orientation of the aircraft along the
flight path. This GPS/INS-system introduces a vertical offset for every strip as well as an along- and across-track tilt.
Sometimes even other systematic effects are caused by GPS/INS (see section 3.2).
4. Error per block. Terrestrial reference measurements (ground control ‘points’) are used to correct blocks of laser
measurements. Thus, errors in these reference measurements affect whole blocks of laser altimetry data.
ro
Error per strip GPS/INS
Error per GPS-observation /
L
A
Error
per
point
Error per block
Figure 3: Error components in laserdata
It is impossible to correct for error components 1 and 2. Error component 3 is the main cause of the occurence of
ribbels. Because this component is constant for large areas (in which a huge number of observations is available), it is
possible to subject it to correction procedures, which are described in the next section. In these procedures also
component 4 is corrected for.
International Archives of Photogrammetry and Remote Sensing. Vol. XXXIII, Part B3. Amsterdam 2000. 231