f models
umerical
etry and
| depth it
relation.
'veloped,
and the
yography
thod on
erational
features,
ater sand
"he most
oves the
arable to
racy for
y reduce
actor of
s as an
onsuming
earch are
e system
3 million
arch and
od of 15
tem will
s from 2
tion of 1
stment of
3.3 Laser scanning for height measurements
In the Netherlands and Belgium the coastal zone managers are
considering the introduction of laser technology to derive height
information for the beach and coastal dune area. This new
technique will provide the necessary data sets for coastal zone
management, aiming both at the reduction of costs and the
improvement of the products. Laser scanning will provide an
elevation model of the sandy coast. From the resulting digital
elevation model (DEM) profiles of the beach and foredunes will
be extracted on a yearly basis, replacing the costly and time
consuming determination of profiles through photogrammetric
means. At the same time the laser scanning data, covering an area
in total, provides an opportunity to assess more accurately the
volumes of sand having been eroded from beach and foredunes.
For the inner dune area the elevation data will allow changes to be
monitored, for example with respect to erosion and accretion of
sand and to study morphological processes. However, in many
parts the construction of an accurate model of the ground surface
elevation is hampered by dense vegetation cover. The main
objective is therefore to provide a general overview of the area, of
benefit to the manager. To this end the laser DEM will be
combined with imagery providing thematic information, such as
land cover. The imagery will be obtained from airborne digital
video. Main consideration for the selection of video, among
alternatives like aerial photography and optical line scanners, is
the relative by low cost. However, the savings realised with
respect to the imagery might be annulled by the increase of cost
for the further processing of the video data (e.g. geometric
correction and mosaicing) and should be further investigated.
Another disadvantage is the relative by low resolution of the
video imagery in comparison to aerial photography. The user will
decide in this respect whether video imagery satisfies his
demands.
The laser DEM and video data will be placed in a GIS en-
vironment with specific tools to allow the user to integrate depth
and height measurements of the coast, extract cross sections of the
coastal zone, assess the quality of elevation models derived and to
generate bird's eye views of areas one is interested in.
In 1995 a number of experiments were performed to assess the
accuracy and reliability of the laser technique for DEM
acquisition as well as to gain insight in its operational aspects.
Large project areas were appointed in several parts in the
Netherlands along the coast and estuaries, the most important
being the group of Frisian Islands in the north. The total area
involved several hundreds of square kilometres. The quality
demands were specified as following:
- point density 1 per 16 m^;
- amaximum systematic error of 5 cm;
- a precision of 15 cm (standard deviation) for non-covered or
sparsely vegetated areas;
- a precision of 20 cm (standard deviation) for areas covered
with dense vegetation;
In every project area, small patches were chosen where control
measurements were carried out, both for checking on point
density as well as assessing the geometric quality by simple
statistical means. Furthermore, laser derived coastal profiles were
compared with the photogrammetric derived coastal profiles. In
figure 3 the results of such a comparison are presented, made
between 132 photogrammetrically surveyed coastal profiles of the
island of Ameland, and the laser profiles. The latter are generated
by interpolation from a regular grid of laser points to the
planimetric co-ordinates of the profile points. For each profile
differences are then calculated between the interpolated laser
heights and the corresponding photogrammetric heights. From
these differences, a systematic offset and a RMS error can be
calculated for each of the 132 profiles. In figure 3, the X-axis
gives the position of a profile along the coast, in kilometres from
the most western point of Ameland. On the Y-axis, the systematic
offset and the RMS error are given in meters above Mean Sea
Level. The RMS error is generally about 15 cm in magnitude. The
systematic offset starts low in the west (between kilometre 1 and
3), then suddenly increases to 15-20 cm.
The results turned out to be slightly disappointing, as in general
the quality demands were not met. Systematic offsets larger than
10 cm occurred, together with RMS errors of more than 20 cm in
magnitude. For assessing sand volumes at the beach, these are
unacceptable figures. Also, many parts of the delivered DEMs
showed the presence of remaining large outliers, that had not been
filtered out, due to a too inflexible filtering method applied and
the complexity of the terrain topography, which the laser scanner
could not handle properly.
: o4 | A | M
3-2 AMI mi a
Pob A hah Oa ANA NT VN
EZ RANA hl aA VE
Rccine A V. yu P
q J Ï T 5 10 13 > 20 "e
Distance along coast line (km)
Figure 3: Differences between photogrammetrically surveyed coastal profiles and laser profiles
15
International Archives of Photogrammetry and Remote Sensing. Vol. XXXI, Part B6. Vienna 1996
