Stakeholder Consultation, Turin, November 3-4
Proceedings
The Turin consultation ran over a Friday afternoon and a full working Saturday.
On the first day there was a series of presentations on the current prototype
of the model. The second day consisted of discussions on a number
of aspects of transportation practice and appropriate modeling approaches.
Minor Issues
The general feeling about the prototype was that it was largely valid (with
some structural changes as indicated below) but that it needed to be elaborated
at lower levels.
Structural Changes
The prototype model presented 3 high-level objects:
-
Transport Network: unmovable assets such as pavement
-
Transport Assets: movable objects including vehicles, signs and construction
zones
-
Transport Processes: business processes
It was felt that the following high-level objects would be more appropriate:
-
Unetrans Network: physical network of unmovable assets with respect to
which objects can be positioned
-
Unetrans Assets: objects that are temporarily associated with the network,
e.g. signs, construction zones, inter-modal stations
-
Unetrans Vehicles
-
Unetrans Processes
Time and versioning
There is a need for transactional updates on an object. Each feature
requires a time associated with lifecycle events: design, construction,
in-service, in-database and out-of-service. Agencies may retire
an object when there is a major structural change, such as addition of
a lane. The object may have different versions, such as a
more accurate geometric representation, or even a less accurate
representation when an alignment is changed.
Linear reference points
The model accommodates point features, but in addition, calibration points
used for linear referencing (“reference points” or “landmarks”) need to
be explicitly defined.
Scale and accuracy
Users expressed needs at different mapping scales. At the high end,
users indicated that they work with 2-D geometry at ±0.2m, and develop
integrated applications where the geometry — not just linear measurement
— is critical. Other users indicated that geometry was desirable
but optional. Accuracy requirements for linear control are ±2m
to ±5m over a 1000m stretch, which translates to 0.2–0.5%.
Objects within objects
Bridges are an example of items that need to be treated as objects in their
own right. A bridge has properties such as height and clearance,
that need to be captured explicitly. A road section object may co-exist
as a bridge object.
Major Issues
An overriding objective of the model is to unify, but there were two issues
on which there were potential divisions.
Logical network or geometric network?
The model references the ArcGIS feature as the basic network object,
which integrates geometry and topology. It was proposed that the
model would be better built around a purely logical network (i.e.
a connectivity table with no associated coordinates), because a large community
of users have no geometric data. The position of the development
team is as follows:
-
Geometry and logical (topological) consistency are tied together in the
Arc feature-level definition. By referencing the Arc feature, the
model avoids having to define these objects and relationships at a low
level. There would have to be a strong argument for deleting this
reference and replacing it with entirely new thinking based on a logical
model.
-
The integrated geometry-topology of the Arc feature is not inferior to
the topology-only logical repesentation; it is in fact richer. Users
need not avail of this richness in the early stages of database development,
and may populate geometry fields with skeletal or even fictitious coordinates
(the London Underground schematic is an example), that evolve as more accurate
data become available. Versioning would allow ease of evolution.
-
We recognize that there is a significant constituency of users with poor
geometry or no geometry, and we will take appropriate measures to present
this aspect of the model and migration path in the accompanying book.
Transport modes
The properties of road nodes and links are so significantly different from
those of rail nodes and links (and similarly air and marine networks),
that it appears appropriate to build separate models based on different
transportation surfaces, with links between multi-modal transfer points.
This is analogous to the Water model's distinction between distribution
and collection (sewer) systems.
Technical Contacts
For further information please contact one of the UNETRANS technical team
reps (below).
ESRI
Jennifer Cadkin <jcadkin@esri.com>
Ernie Ott <eott@esri.com>
UCSB
Kevin Curtin <curtin@ncgia.ucsb.edu>
Mike Goodchild <good@ncgia.ucsb.edu>
Val Noronha <noronha@ncgia.ucsb.edu>