often present
as the choice
and filling in
ple available
t was started
acies for the
'esulting in a
1e automatic
combine the
mbination of
, with highly
| DTM based
atter DTM is
coincide with
r 2001
; available in
as a height
northern and
ern part. The
in several
)SMs derived
)00 that were
. For 1796 of
d by manual
. aerial b/w
ced for areas
:h are not too
in height was
ur lines on à
DEM for the
witched from
ally going up
omplemented
processing of
e imagery We
International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Vol XXXV, Part B4. Istanbul 2004
There is also a demand for a DTM which is to be used as a
reference plane in the future 3D geographical database. As the
accuracy of most of the topographical data is situated in the
0.5m to 2m range, a DTM with at least similar accuracy is
needed. This DTM should also be usable for the derivation of
contour lines for topographical map production.
The NGI performs regular surveys for the identification of
obstacles around airfields. The scope of these surveys has been
widened in the last year to almost the complete Belgian
territory. A reliable DTM is a prerequisite for this kind of
studies.
There is an increasing. demand for DEM products from
customers. The requirements are highly variable. They range
from small scale DSMs for draping satellite images over
detailed DTMs for geomorphological studies to line of sight
studies for the telecommunications sector. Priorities for most
customers seem to be: full coverage, reliability, ease to use and
a low price. On the one hand there is a general demand for an
off-the-shelf product, but on the other hand requirements vary
very considerably and frequently tailor-made solutions are
needed.
2.3. Production constraints
Available budget and personnel are limited. There is
considerable urgency, especially for the production of
orthophotos.
The regions in Belgium (Flanders, Wallonia and Brussels) are
also involved in production programs for DTMs. Although co-
operation is agreed upon, practical as well as legal and financial
issues still need to be settled. As a consequence we look
actually for a moderate cost intermediate solution. This situation
has an important influence on the data structure and working
philosophy.
2.4. Recovery and flexibility
A huge amount of information has already been collected in the
previous decades in the form of contour lines on a map scale of
1:10.000 that were stereoplotted and verified through extensive
field surveys (Vanommeslaeghe,2001). The contour lines are
still very useful as a source for DTM production. As the
recovery of the information (from analogue to digital form
through vectorization and identification) is a lot cheaper and
faster than starting over from scratch, it was decided to pursue
the recovery and upgrading of existing information as much as
possible.
The quality that should be achieved is an overall accuracy of
0.5m to 2m. If this quality level is not achieved or not yet
achievable, a quality should be achieved which is at least better
than the existing DTED level2.
Budgetary constraints make it necessary to use different types
of data sources and it is very likely that external data will need
to be incorporated rendering data-interoperability a very
important feature of the project. Likewise, a lot of different
derivatives are foreseen and possibly more will become
Un
O3
necessary. Because of time constraints, in certain areas it will
be necessary to use an intermediate suboptimal solution which
will have to be replaced with better data later on. In view of
this situation the project should work in a very flexible way. We
avoided making a very detailed conceptual model because it is
dangerous as it taxes the ability to adapt to new situations and
technical possibilities. You do however need an overall
conceptual framework as a tool to guide the actual work.
2.5.Working priorities
Whether a DTM qualifies as a “good DTM” depends on a
number of factors. First and foremost this question should be
rephrased into “Is the DTM making the things possible that you
want to do in your specific application ?” It is a question that
has a different answer for every different user. In our experience
we can corroborate the breakdown of the goodness of a DTM in
the following factors as given by Cory and McGill (1999):
completeness, reliability, consistency, uniformity, content,
accuracy . In our experience most customers’ questions pertain
to the first two factors (i.e. when these are not fulfilled there is
no interest and so other factors are not relevant).
Although every factor is theoretically important, due to
constraints it is not possible to focus on each factor at the same
time. Hence we will follow a phased production process. In a
first phase we tackle the completeness and reliability, relegating
the other factors to future phases. We strive to get a complete
coverage of Belgium with a quality at least better than the
existing DTED level 2. In a second phase we will perform a
systematical quality assessment, the objective of which is to
guide the following “upgrading” phase and to provide users
with useful metadata. In a third phase we need to upgrade the
accuracy where it is not within the 0.5-2m range. If necessary a
further upgrading to higher accuracies can be pursued.
After upgrading the whole data set, the next phase is that of
updating the data. Compared to other topographical data, this is
expected to be a smaller problem, if the accuracy range is kept
at 0.5 to 2m. Higher accuracies would of course cause a bigger
need for updating.
2.6. Data structure
A threefold structure is guiding the production. On a first level
we have the so called DEDS (Digital Elevation Data Set). These
are the original discrete data on terrain (or surface) height
whatever their origin. On a second level we have actual DTMs
(or DSMs) derived by interpolation of heights in between the
original discrete data. As there are a variety of interpolation
strategies and techniques and an equally wide variety in user
requirements, many different DTMs (and DSMs) should be (at
least conceptually) envisaged on this level. For the practical
reason of having only one easily manageable and off-the-shelf
product that is suitable for the majority of end users, one DTM
will be made as a general reference GIS layer and only locally a
second DEM will be made where the demands for different
applications cannot be met by a single DEM. The most obvious
application is the production of orthophotos where locally a
DEM should be provided including the surface of large
structures such as fixed bridges, etc... For easy distribution and
handling, it will be made available in tiles of 4 km’ like the
