Abstract

1
Ontology of Experimental and Simulated Fault
Gouge Hassan A. Babaie1 and Jafar Hadizadeh2 1
Department of Geosciences, Georgia State
University, Atlanta, GA 30302-4105 Email
(hbabaie_at_gsu.edu) 2 Department of Geography and
Geosciences, University of Louisville,
Louisville, KY 40292
Geoinformatics 2006 May 10-12, Reston,
Virginia Art Hallway and Main Lobby
Abstract We have designed ontologies related to
an investigation of changes in gouge porosity in
small-displacement faults via 2D computer
simulations and experimental deformation. These
ontologies, built based on the state-of-the-art
domain models, will facilitate efficient online
interchange and reuse of the knowledge resulting
from research of physical process and materials
involved in the deformation. The groundwork
consists of ontologies for brittle rock
deformation that help the understanding of
deformation along natural and artificial brittle
fault zones. These ontologies are formal,
controlled vocabularies reflecting domain
theories, and are developed for shared use within
and across domains. The vocabularies include
terms that represent real world individual
entities such as a particular fault, mylonite, or
gouge zone, and formal relations (e.g., is-a,
part-of, participate-in, location-of, agent-of)
between these individuals. Instances (i.e.,
individuals) of these material (e.g., particle)
and immaterial (e.g., pore) entities have all
their spatial parts present at all times of their
existence (i.e., occur as wholes), and occupy
same or different spaces at different times.
These entities, sampled by geologists at specific
time instants and locations, may change their
properties (e.g., a fault changing length a
gouge becoming weakened) and qualities (e.g., a
gouge changing density or porosity, or becoming
hot) without losing their identity (e.g., San
Andreas Fault remaining the same fault over a
period of time). Some entities evolve into new
entities (with new identity) over time through
transformation, for example, particular granite
(e.g., a sample of the Westerly granite) becoming
a mylonite, or a 2D surface becoming a fold.
Process entities such as deformation, cataclasis,
and slip localization lack spatial parts, and
involve material entities as agents (e.g., water
is agent during hydrolytic weakening and
hydraulic fracturing) and participants (e.g.,
rocks participate in fracturing). While material
entities do not have any temporal parts, the
process entities have no spatial parts. Instead,
the latter entities have temporal parts that
unfold over time in phases. The spatial location
of material entities (e.g., fault, rock, mineral)
that participate in particular processes (e.g.,
faulting, cataclasis) define the location where
these processes occur. Processes (e.g.,
cataclastic flow) may destroy or create material
entities (e.g., foliated cataclasites) over
different times, and under different conditions.
Our ontologies are designed with a modular
architecture, allowing integration, and a more
realistic depiction, of the interaction of
individual entities that are involved in
connection with a computer simulation study of
porosity of various regions of experimental and
natural gouge. The ontologies include the
following classes and their object properties
(given in parentheses) fault, gouge (porosity),
region, particle (shape, number, contact, sliding
surface area), particle size (distribution,
fractal), packing (density), and gouge texture.
The process classes include shear displacement,
shear localization, comminution, weakening,
friction, cataclastic flow, and fracture. We
present the vocabulary (i.e., classes and
relations) of each of these non-overlapping
ontologies from processual and material
perspectives.
We observe and measure an endurant entity as a
mereological whole, at the present instant of
time, in the field or laboratory. In the
following we describe two orthogonal
perspectives, i.e., SNAP and SPAN (Smith and
Grenon, 2004) that are required to completely
capture both static and dynamic components of
reality (Figs. 1 2). The SNAP perspective
(Grenon, 2003 Grenon and Smith, 2003) provides
snapshots of endurants at specific time
indexes, vs. the SPAN perspective which provides
a video-like, continuous series of views of
perdurants over a time interval. The SPAN
perspective provides a measure of change over
time.
Dynamic natural and experimental gouge
ontologies We have applied the SNAP and SPAN
to represent the knowledge about natural,
experimental, and simulated brittle deformation.
The UML package diagram (Fig. 4), shows three
sets of packages covering the endurants and
perdurants (processes and activities) in this
these domains. Each package translates into a
single ontology or database, and contains several
classes that depict the SNAP or SPAN entities
which exist or occur in the domain of deformation
in a natural, experimental, or simulated shear
zone. Three types of relations exists
between the classes (1) intra-ontological
relations, which obtain between classes within
the same package, (2) inter-ontological
relations, that obtain between classes in
different ontologies (packages) of the same
category (SNAP or SPAN), and (3) trans-category
relations that relate SPAN classes to the
participating SNAP classes. The
inter-ontological and trans-ontological relations
lead to dependency relations among the relata of
these relations. This means that the dependent
package includes or imports the required classes
from other packages. The dependencies are
depicted in the package diagram by the dashed
arrows that point from the dependent package
(ontology) to other ontologies. This modular
structure is an efficient way to maintain and
curate the ontologies. This means that
structural geologists who collect their data
through field work, rock deformation experiments
in the lab, or simulation using a computer, can
develop their own ontologies and databases, and
then share their data. The modular architecture
will facilitate merging and integration of the
ontologies developed by autonomous groups of
researchers working on related aspects of the
same problems.
Entities in the Earth realm can be divided into
two disjoint (i.e., non-overlapping) groups (1)
endurants (continuants) and perdurants
(occurrents). In a shear zone, endurants include
objects such as mylonite, gouge, fault, and
fracture. These endurant entities (depicted by
the SNAP entities in Fig. 1) occupy same or
different spatial regions at different times, and
acquire different properties by participating in
processes such as deformation, weathering, and
metamorphism. These entities are said to endure
over time by maintaining their identity despite
the changes that are realized through processes.
For example, the San Andreas Fault remains the
same fault even though it goes through continuous
spatial and qualitative change. Geologists
capture information about the endurants at
different times and places while these entities
are going through change.
One- to three-dimensional SNAP entities
(endurants) are divided into three categories
(Fig. 1) (1) Independent SNAP (substantial)
entities, (2) SNAP dependent entities, and (3)
spatial regions. The independent entities
include (i) substances, such as fault, breccia,
molecule, water, and sag pond (ii) fiat parts
(iii) boundaries, such as the boundaries of a
shear zone, and grain boundary (iv) aggregates
of substances, such as the sum of all components
of a strike slip fault (e.g., all its segments,
bends, and steps) and (vi) sites, such as
cavities, pores, holes, and other empty spaces in
rocks that could be occupied by substances (e.g.,
water, air). The SNAP dependent entities depend
on other endurants for their existence, and
include (i) qualities such as density, rigidity,
bulk modulus, shape, size, and temperature (ii)
function, such as function of water is to enable
diffusion of ions and (iii) roles, such as the
role of water in hydrolytic weakening or
hydraulic fracturing.
Figure 1. Top-level SNAP entities
Figure 2. Top-level SPAN entities
The perduring entities (depicted by the SPAN
entities in Fig. 2) include processes (e.g.,
cataclasis, frictional sliding, creep) that
unfold over time intervals that start and end
with instantaneous, or relatively short-duration,
events (e.g., slip event, seismic event). These
entities involve one or more endurant entities
(e.g., rock, grain) that display different
properties (qualities) under different ambient
conditions (e.g., pressure, temperature) (Fig. 3)
SPAN entities are divided into three categories
(Fig. 2) (1) processual entities, (2) temporal
regions, and (3) spatiotemporal regions. The
processual entities include (i) processes such as
brecciation, shear localization, and slip (ii)
fiat parts, such as the stick or slip part of
a stick-slip cycle (iii) aggregates, such as
slip which can be an aggregate of frictional
sliding, creep, and other processes (iv)
settings, which are spatiotemporal counterparts
of SNAP sites, such as Hollywood during the 2001
West Hollywood earthquake and (v) instantaneous
temporal boundaries (events), such as the first
foreshock and last aftershock detected for the
1994 Northridge earthquake.
Fig. 6A
Fig. 6B
 
Fig. 5B
Fig. 5A
Figure 3. Class hierarchy in the Quality Package
Fig. 6C
Figure 4. UML Class Diagram
Figure 6. Class diagrams of few of the endurants
in the domain Discontinuities (6A), Geomaterial
(6B), and Region (6C).
Components of a Dynamic Ontology Ontologies
are designed to formally capture, and explicitly
specify, domain theories and knowledge about the
world. Traditionally, ontologies in the Earth
sciences have focused solely on the static part
of reality (e.g., fault, mineral, mylonite), by
only formalizing the endurants, and completely
ignoring the perdurants (e.g., processes). This
approach misses the dynamic component of reality
(e.g., faulting, fracturing, mylonitization), and
cannot be useful if we intend to represent our
knowledge in a useful form. Endurant
entities such as fault and cataclasite only have
spatial parts for example, a fault has segments,
bends, and steps as parts. Endurants have
properties, both relational (fault displaces rock
sequence) and qualitative (a faults width,
thickness, temperature at specific points,
extent). An endurant occupies (i.e., is located
in) a spatial region as a mereological whole
(i.e., its entirety), which may change over time
as the entity changes, e.g., as a fault
propagates, thereby changing its length. This is
in contrast with perdurants (e.g., sliding) that
have temporal parts (phases or stages) but no
spatial part. Only temporal parts of perdurants
occur at specific time slices, i.e., they can
only be detected as a part, but not as a whole,
at specific times. The whole can only be
observed over a time interval. For example,
slip along a fault may start with an initial
phase of hardening (within a finite time
interval), followed by dilatation, cataclastic
flow or mylonitization (depending on depth), or
by frictional sliding. Only one of these
processes occurs at specific regions at specific
time instants (i.e., the present). Processes are
said to occur within spatio-temporal regions,
defined by both time intervals and space. This
is in contrast with the endurants which occupy
spatial regions (i.e., at x, y, z coordinates) at
the same or different times (t).
Fig. 5D
Fig. 5C
Fig. 8. Process importing SPAN event.
Fig. 7. Endurants using SNAP ontology
Introduction Shear zones are tabular,
spatial regions along which complex deformation
and other related processes lead to the
localization of deformation. These dynamic
processes, which are caused by plate motions,
involve a variety of one- to three-dimensional
spatial entities, at all scales microscopic to
lithospheric. In general, processes bring about
qualitative and spatial change to the spatial
entities (e.g., rock, mineral) over time
intervals that start with initial and terminal,
instantaneous temporal boundaries (i.e.,
events). Processes, for example, change the
spatial location, distribution, and orientation
of some objects through translation, rotation and
strain. They also lead to the creation of new
entities, e.g., new mineral, mylonite, fracture,
or fault may form during deformation. Processes
may also lead to the destruction of entities, for
example, bedding, fossil, and an original texture
or structure may be annihilated during
deformation. In this presentation, we present an
approach based on a perspective that combines
both the endurants and perdurants components of
reality in shear zones, and discuss a method to
design dynamic ontologies for shear zones.
Figure 5E
Figure 5F
Fig. 9A
Fig. 9B
Figure 5. Types of Activity (5A), Studies (5B),
Modeling (5C), Simulation (5D), Experiment (5E),
and Sample (5F) in the domain.
Figure 9. Fracturing (9A) and Brittle_Deformation
(9B) processes importing the SPAN ontologies, and
relating to endurants.