2004
fined
| the
o be
data
ation
o be
ly in
Was
jonal
y on
‘any,
can
000,
.000
llert,
“the
ature
s the
d by
EST
[t is
lues.
1ges,
/ for
civil
ts of
, the
pean
.000
Map
obal
IRM
| but
1s at
Irces
exist
udes
'OSS-
sct.
lium
| the
thus
with
obal
mon
hese
jject
International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Vol XXXV, Part B4. Istanbul 2004
e be compliant with existing or emerging international
standards,
e be consistent with EuroGeographics specifications for
topographic datasets at medium and small scales
(EuroGeographics, 2003),
e serve the user requirements identified in the GiMoDig
project (Jakobsson, 2002),
e consider the needs of small display cartography
identified in the GiMoDig project (Nissen et al.,
2003),
e build on data that is available at the National Mapping
Agencies, and
e restrict the harmonisation operations from Local
Schema to Global Schema to those procedures that
can be performed on-the-fly, in real time.
For the classification into the FACC-code the object types of
the national topographic data. bases of Denmark, Finland,
Sweden and Germany are described together with their
attributes, their selection criteria and their data model.
The harmonisation operations used in the GiMoDig project do
not aim at creating a consistent cross-border topology, as it
seems to be unrealistic to make it fully automatically on-the-fly.
Therefore, the data in the Global Schema is not suitable for
cross-border spatial analysis like routing applications.
Edge matching is not considered in the GiMoDig project, so the
different geometry remains in the dataset and each national
dataset uses its geometry for clipping. In practice, the datasets
of the GiMoDig NMAs fit very well at the borders, and the
difference in geometry is only visible if the data is displayed at
a much bigger scale than the intended scale of 1:10.000.
4.3 Selection of relevant feature types
After the classification into the FACC-code it becomes evident
that from the approximately 470 feature types in the FACC, 180
have an equivalent in at least one of the national datasets. From
these 180 feature types a selection is proposed to be used in the
following steps of the GiMoDig project. As a criteria for
selection the examinations by Jakobsson, 2002 are raised. They
deal with user requirements on a harmonised topographic
dataset on mobile devices. Different concrete use cases such as
Tourist in a city’ or 'Hiking in national park’ are described
which have different requirements to the data content. Also
Nissen et al., 2003 are referring to that study and describe the
requirements for small display cartography, i.c. cartography for
small devices. This study is also considered in the Global
Schema. The initial idea for the selection of relevant feature
types was to start with the least common denominator, i.e. those
feature types that have equivalents in all national datasets or
can be derived from national data. However, the least common
denominator is small; actually less than 20 feature types have
an equivalent in each of the four national datasets. This set
lacks of some feature types that are very important to the
GiMoDig service. Therefore, we left the idea of the least
common denominator and kept those features types that are
supported by the majority (i.e. at least three) of the GiMoDig
countries (s. Table 1, (Illert, Afflerbach, 2003b).
FACC-Code FACC-Name
BA040 Water (except Inland)
FA000 Administrative Boundary
BH502 Watercourse(ERM)
BH080 Lake/Pond
BH095 Marsh/Swamp
AKI20 Park
ALOIS Building
CA010 Contour Line (Land)
EAO010 Cropland
ZD040 Named Location
AL020 Built-Up Area
AN010 Railway
AP030 Road
AP050 US-Trail/Footpath
GB005 Airport/Airfield
ECOIS Forest
EB010 Grassland
Table 1: Selected object types
In the very beginning of the GiMoDig project it was decided to
adapt the classification schema and semantic model of the
FACC. This should ensure the compliance with ERM and EGM
because both datasets are structured according to the FACC.
However, at a closer look it turns out that ERM does not always
follow the FACC. For example, ERM has introduced a new
feature type BH502 "Watercourse' which is not present in the
FACC but aggregates the FACC feature types BH020 'Canal',
BH030 'Ditch' and BH140 'River, Stream'. The ERM expert
group combined the FACC classes in one feature type because
they think the semantic difference between the 'Watercourse'
instances does not justify separate feature types. This change is
in line with the topographic datasets of many European NMAs.
In view of a uniform conceptual model for NMA data across
scales, the GiMoDig Global Schema follows the ERM
definition if conflicts occur between ERM and FACC.
4.4 Definition and Rules of the Global Schema
The first version of the Global Schema is created for the
selected feature types. For the definition of parameters for
collection criteria and geometry type, rules are set up to ensure
minimal conversion conflicts from the national core
topographic databases:
e Geometry type: in case the national specifications
define conflicting geometry types (point, line, area)
for a feature type, then that geometry type with the
majority among the national datasets is chosen.
* Collection criteria:
If the collection criteria in national specifications use
same rules but different parameters, then the least
common denominator is applied.
Example: feature type 'Park'
Germany: size of area 7 10.000 m? (1 ha)
Denmark: size of area 7 2.500 m? (0,25 ha)
Finland: size of area > 5.000 m? (0,5 ha)
Sweden: 900 n? (0,09 ha)
size of area
Global Schema: size of area 7 10.000 m? (1 ha)
If the collection criteria in national specifications use
different rules, then the Global Schema does not
impose collection criteria, allowing all national data to
be entered without harmonisation.
Example: BH502 'Watercourse', subdivision "natural
watercourse'