The general aim of the research is to assess the
suitability of using digital photogrammetry to survey and
model the coastal zone in order to assess both cliff
erosion and sediment movement (volume) along the
beach, both quickly and accurately. The conventional
method of surveying the beach by spirit levelling is
accurate, but can be an expensive, laborious and time
consuming exercise over large areas. Data capture by
photogrammetric techniques and the quantification of
sediment yield using digital terrain models, may offer an
improved method on those currently in use.
2. DIGITAL PHOTOGRAMMETRY AND THE
COASTAL ZONE
The coastal zone has a number of varied environments
which image very differently on aerial photographs. The
beach area, the cliff area and the zone inland from the
cliff are the three main areas of interest. One of the
greatest problems that is perceived by looking at the
imagery is that there can be considerable differences in
the appearance of features from frame to frame,
particularly along the shoreline on the beach.
Initial research using both the Zeiss P3 analytical plotter
and two digital photogrammetric workstations has shown
that the manual placing of the measuring mark in a
stereo model can be difficult due to the lack of texture
and the tonal changes in the images. Whereas the cliff
face offers a great deal of detail, some sandy areas of
the beach offer very little detail e.g. pebbles, and the
correct heighting of a point can prove to be difficult. This
is compounded by the waves being positioned differently
from one frame of photography to the next, making it
difficult in places, to create the stereo image.
One of the most powerful capabilities offered by digital
photogrammetry is automated digital elevation modelling
Early general experiences showed that automated digital
elevation modelling can produce a different quality of
result dependent on the image of the terrain. With these
very varied images of the coastal zone, mentioned
above, it is important to test the suitability of automated
elevation modelling before it is extensively used.
Therefore, research into the effects of the different
topography on the quality of results achievable is
important.
Automated digital elevation model (DEM) generation can
appear to be a simple operation but the underlying
algorithms are complex and contain a number of variable
parameters. These variable parameters have an impact
on the operation of the algorithms and ultimately on the
quality of the DEM. It is important to experiment to find
the effects of these variable parameters with particular
application to the coastal topography.
The study of the suitability of automated DEM generation
will be broadly divided into three aspects:
a) To investigate the effect of varying the parameters
used to control the automated DEM processes.
920
b) To compare the automated DEM with a DEM
observed on an analytical plotter, to determine its quality
against 'traditional' manual observations.
c) To undertake a comparison of the DEMs observed
photogrammetrically with those created from traditional
ground survey on site. The traditional survey is expected
to provide a significantly more accurate result.
Research is presently underway in a) and b) defined
above and the following sections show some of the
results obtained. Initially work is being undertaken using
selected single stereo pairs of photographs, later the
study will be more extensive with the availability of a strip
of photography.
3. TEST DATA
At present an investigation is taking place using 1:5000
scale (1993), full colour, vertical photography from along
the coast at Easington. Individual DEMs are being
created using the Intergraph ImageStation
(ImageStation) at the British Geological Survey (BGS)
and the ERDAS Imagine OrthoMAX system (ERDAS) at
the National Remote Sensing Centre Ltd (NRSC).
Results from the 1:5000 colour photography have been
compared with the DEM produced from an analytical
photogrammetric plotter (Table 1). This 'datum' DEM
has been produced by an experienced photogrammetrist
on a Zeiss P3 analytical plotter. It is being planned to
re-measure the datum to determine the precision of this
particular model. All subsequent automatic DEMs
created on both the ImageStation and the ERDAS
OrthoMAX workstations, are being compared to this
datum. Current research involves the extraction of the
DEM (X Y Z) files from all three workstations to enable a
comparison to be undertaken to determine height
differences.
It can normally be expected that for a given photo scale
of 1:5000, well defined ground control can be measured
on the analytical plotter to give a heighting accuracy of
+0.15m and points of detail to +0.2m. These are well
known heighting qualities for manual observation and
define what we might expect from automated DEM
generation. However, the ImageStation quotes three
different heighting accuracies for each of the three terrain
settings i.e. flat, hilly or mountainous. The heighting
accuracies quoted are, (h/10 000), (2h/10 0000) and
(3h/10 000) for flat, hilly and mountainous settings,
respectively, where ‘h’ is the flying height in metres.
Thus, for 1:5000 photography, ‘h’ is approximately 750m.
This gives heighting accuracies of +0.075m, +0.150m
and +0.225m for flat, hilly and mountainous settings,
respectively.
4. DEM COMPARISONS
The height differences between the analytical and
automatic DEMs are shown statistically in Tables 1 to 3-
They show some of the variable parameter settings that
International Archives of Photogrammetry and Remote Sensing. Vol. XXXI, Part B4. Vienna 1996
