®
Linked geodata, spatial ontologies, and making
place for the Web
Josh Lieberman
Harvard Center for Geographic Analysis
Dec 10, 2014
Credits
• Many participants in SOCoP workshops http://socop.org
• Many participants in the W3C-OGC Linked Geodata Workshop
http://www.w3.org/2014/03/lgd/report
An Ontology Design Pattern for Surface Water Features
Gaurav Sinha1, David Mark2, Dave Kolas3, Dalia Varanka4, Boleslo E. Romero5, ChenChieh Feng6, Lynn E. Usery4, Joshua Liebermann7, Alexandre Sorokine8
1Department
of Geography, Ohio University, Athens, OH, USA
sinhag@ohio.edu
2Department of Geography, University at Buffalo, Buffalo, NY, USA
dmark@buffalo.edu
3Raytheon BBN Technologies, Columbia, MD, USA
dkolas@bbn.com
4U.S. Geological Survey, Rolla, MO, USA
varanka, usery@usgs.gov
5 Department of Geography, University of California, Santa Barbara, CA, USA bo_romero@umail.ucsb.edu
6 Department of Geography, National University of Singapore, Singapore, Singapore geofcc@nus.edu.sg
7Tumbling Walls LLC, 85 High Street, Newton, MA 02464
jlieberman@tumblingwalls.com
8 Oak Ridge National Laboratory, Computational Sciences and Engineering Division, Oak Ridge, TN, USA
sorokina@ornl.gov
Copyright © 2014 Open
Geospatial Consortium
Connections
• Geodata – data describing features
• Ontologies – logical / graphical representation of theory of
reality
• Linking – relationships accessible over networks
• Linked Geodata – feature data where relationships in and
between features can be distributed
• Spatial Ontologies – logical graphical representations
including theories of space +/- geography
Semantics, Ontology, Networks
• Semantics – “meaning” or conceptual role
• Ontology – “conceptualization of a theory”
• RDF / Owl – network (subject – predicate – object) representation
of an ontology
• Network models are one general representation of logical
relational semantics
• Networks generalize most data model patterns (relational,
hierarchical, etc.)
• Network knowledge model relationships have multiple uses
–
–
–
–
–
–
Similarity / inheritance “Is-a”
Mereotopology “Part-of”, “Connected-to”
Axiomatic / logical “Disjoint”
Physical “mass-conserving”
Spatial “downhill-from”
Social “friend-of”
Features vs
Links
Mutual
interest
Franz Kibbe:
• how should we encode geometry?
• how and where should we implement topological functions?
• additional metadata is required for spatial datasets – how do we do
that?
• where is the software support for spatial datatypes and functions?
• geometries expressed as WKT literals are large objects — the Linked
data world is used to handling simple literals;
• how do we help developers handle (or avoid) the steep learning curve
to work with Linked Data?
Application: the National Map
Nationally extensive data layers pertaining to USGS Topos
–
–
–
–
–
–
–
–
Elevation (NED),
Orthoimagery,
Hydrography (NHD, WBD),
Geographic Names (GNIS),
Boundaries,
Transportation,
Structures,
Land Cover,
National Hydrography Dataset
• NHD: all streams and lakes at scales of 1:24000, 1:100000
in a network
• WBD: defines the areal extents of surface water drainage
to a point
• NHD+ connects each NHD reach to a catchment
Hydro Cycle
Ontology:
What exactly is the “lake”?
• Is a lake a large body of water?
• SDTS: Lake: “Any standing body of inland water”
• Or is it a (natural) basin that can hold a large amount of
water?
• Or is it a large area covered by water?
• If the water goes away, is the lake dry, or does the lake
cease to exist?
• If the water rises, what is lake and what is flood?
• As water flows through it, where does the lake begin and
end?
What exactly is the “river”?
• Is a river a large stream of water?
• SDTS: Watercourse: “A way or course through which
water may or does flow”
• If the water goes away, is the river dry, or does the river
cease to exist?
• If the water rises, what is river and what is flood?
• Where does surface flow stop and river begin?
• Where does river stop / start and lake start / stop
• How many streams is a braided stream?
®
Hayes on Lakes
Hayes, P. (1985b) Naive Physics I: Ontology of
Liquids. in: J. Hobbs and R. Moore (Eds.), Formal
Theories of the Commonsense World. Norwood,
NJ: Ablex, pp. 71-108
Consider a Lake
• “Consider now a lake.
• This is a contained-space defined by
geographical constraints. Lake Leman,
for example, is the space contained
between the Jura Mountains, Lausanne,
the Dent d'Oche, Thonon, and the
Rochers de Naye, below the 400 meter
contour (more or less).
• Its container is the surface of the earth
under it, i.e. the lake bed.”
• Hayes, P. (1985b) Naive Physics I: Ontology of Liquids. in: J.
Hobbs and R. Moore (Eds.), Formal Theories of the
Commonsense World. Norwood, NJ: Ablex, pp. 71-108
Consider a Lake, cont’d
• “I think the only way to describe lakes, rivers
and ponds in the present framework is to say
that they are contained-spaces which are full
of water:
–that is, the space ends at the surface of the
water.
• To be in the lake is then, reasonably, to be
immersed in water, while to be on the lake is
to be immediately above the water and
supported by the lake (cf. on the table), which
seems reasonable.
• Thus a lake is full by definition.”
Water and Voids
Water and Voids
Water and Voids
Origin Stories…
• Reech, M., 1858, Proprieté générale
des surfaces fermées, Journal de
l'Ecole Polytechnique
• Cayley, Arthur, 1859, On contour lines
and slope lines. Philosophical
Magazine Series 4, Volume 18, Issue
120
• Maxwell, James Clerk, 1870. On hills
and dales. Philosophical Magazine
Series 4, Volume 40, Issue 269
Surface Network
Source: Rana and Morley, 2002, Figure 2.1, p. 21
Contour Enclosure Tree
Source: Rana and Morley, 2
002, Figure 2.1, p. 21
Surface Network
Unfortunately, Nature is more c
Elevation: Contours
Aogashima aerial view
Contours
• Contour value
Contour line (has value, is a closed line, encloses a
continuous area)
• Contour region (region with a value greater/lower than
a particular contour value)
• Nesting of regions (spatial containment)
Relationship between neighboring contour lines
(“next” contour value) Local minima/maxima
• Implication of scale in discernment of terrain
• Many interpretations of surface features from contour
patterns
Water and Voids
Hahmann & Broderic
Wet vs Dry
• We divided into two models:
– “Dry” model is terrain features that *may* contain water
– “Wet” model is where water is currently flowing and standing
• Initial models are temporal snapshots
• Ignored erosional causation, groundwater leakance,
etc.
• Do not necessarily correspond directly to named or
topographic features
Dry Model
Junctions
Nodes
Channels
Depressions
Dry Model Classes
Wet Model
Influences
Exfluences
Confluences
StreamSegments
WaterBodies
Wet Model Classes
Fluence
Influence
Exfluence
Stream
Segment
Confluence
Water Body
Surface Water Patterns
• Fig. 1. Surface Water pattern’s Dry and Wet module
classes (brown/blue) and properties (grey).
Surface Water Features
• Fig. 2. An illustration of how surface water features can be
described as instances of classes defined in the Dry (left)
and Wet (right) modules of the Surface Water pattern.
Coming into Use…
• Varranka and Usery
Spatial Relationships
• Varranka and Usery
NHD Ontology
OGC 11-039r2
HY_Features
• Focus on
hydrography
“features” and
connection to
observations
Placement of
HY_Features in
the existing
OGC Feature
landscape
Atkinson &
Dornblut, OGC
DP 11-039r2
Copyright © 2014 Open Geospatial Consortium
O&M Ontology
From Simon Cox, CSIRO
Copyright © 2011 Open
Geospatial Consortium
The Work
• Justification
– Feature layers in the National Map (TNM) are a fundamental context for
spatiotemporal data collection and analysis.
– Little in the way of formal relationships has been created between layers to aid
computation
– Formal ontologies for hydrographic (water) features, containing topographies,
and water place names will facilitate automated discovery and processing of
related water quality and other observational data.
• Plan
– Formulate Ontology Design Patterns (3) for NHD, NED, WBD, and GNIS data
building on recent “wet-dry” work with water bodies and containing landscape
– Develop target application designs to exercise the ontologies together with
observational data
– Validate ontology candidates by translating selected TNM data into semantic
form and performing representative GeoSPARQL queries against them
– Compile and translate water data into compatible form and representation (2)
– Develop and deploy prototype applications for discovery and aggregation of
water data using TNM ontologies as a framework
– Refine ontologies and applications in response to open usage and feedback
Copyright © 2014 Open Geospatial Consortium
The Benefits
• Improvement on status quo
–
–
–
–
Explicit relationships between TNM feature layers
Validation in discovery and analysis Web applications
Incorporation of water data and analysis
Incorporation of time and community dependent names.
• Benefits
– Improve the capability to utilize TNM data in new and innovative
applications
– Increase TNM’s value as a framework for collection and analysis of
related data.
– Demonstrate value of agile semantic technology for discovery and
analysis of heterogeneous data in a geographic framework.
Copyright © 2014 Open Geospatial Consortium
Where Next
•
•
•
•
Graphical National Map research
W3C – OGC linked geodata standards
OGC Testbed 11 linked NHD – GNIS services
Harmonization
–
–
–
–
–
–
Physical world
Geospatial features
National Map data
Ontologies
Linked Data on the Web
Applications
W3C – OGC working group
• GeoSPARQL
• GeoRSS
• Semantic Sensor Network