Geotectonics, being one of the main geological disciplines, encounters conceptual difficulties that likely can be resolved by application of methods of knowledge engineering. However, a strategy of their application is needed. The role of ontologies in the knowledge-engineering process is to facilitate the construction of a domain model. This model can be either static, i.e., address only the observed geological structures and landforms, or dynamic, accounting for processes that operate in and below the earthcrust. Both types of model are required to overcome the conceptual problems of geotectonics, but while the former is more or less present in the literature, the latter represents complete terra incognita. Meanwhile, exactly the dynamic knowledge modeling is the most important for a field like geotectonics.
The term “tectonics” originates from a Greek word, “tekton”, which literally means
builder. Later this word acquired a wider meaning that included the whole process of
creation of something, including such connotations as techniques of construction,
properties of the material and principle of creation, or architecture (Laugier’s (1753),
Botticher (1852), Semper (1951) and Liu and Lim (2009)). In the Earth sciences, this
word is known at least since 1894, when it was said at the 6th International Geological
Congress, Switzerland, to describe the mammoth architecture of the Alps and Jura
Mountains
It is accepted in knowledge engineering that a model in particular domain is built in process of human-computer (or expert – knowledge engineer) interaction and thus largely reflects the way of thinking of the expert. Relying on multiple experts decreases the “human component” but still keeps the record of personal experiences. Here the emphasis is somewhat different; it is not intended to create a cognitive model, i.e. to simulate the cognitive process of expert. Instead, the challenge is to create a model that represents “bare knowledge” and, as such, offers as much bias-free results as possible. While the expert may consciously articulate some parts of his or her knowledge, he or she will not be aware of a significant part of this knowledge since it is hidden in his or her skills. This knowledge is not directly accessible, but has to be built up and structured during the knowledge-acquisition phase. Therefore, at some point (known to or felt by knowledge engineer) the acquisition of knowledge from particular experts (or texts) should be replaced by building a model, desirably as biasfree as possible. Certainly, any model is only an approximation of the reality. Apparently as well, the modeling process is infinite. However, in every knowledge engineering process the stages of knowledge acquisition and model construction should be, from one side, clearly divided, and from the other, tightly interrelated. In our opinion, the best connection between them could be ontology of considered domain of tectonic knowledge. Creation of ontology or relation of extracted knowledge to some pre-existing ontology should be the result of knowledge acquisition and starting point for knowledge modeling.
Ontology provides vocabulary of terms and relations to a model. The closer it is to the domain of interest, the better the model will be. For instance, if ontology perfectly suits the domain, then a domain model in some cases can be obtained just by filling the ontology classes with instances. However, this rarely happens, first, because the nature of ontology is to be generic, while domains of interest usually occur at intersection or as particular cases of such generic domains, and then, because only static model, assuming that modeled environment does not change, can be obtained right from ontology (see below). Also, ontology helps avoid mixture and overlap of meanings and figure out groundless meanings. For example, geologists often use ‘subsidence’ and ‘uplift’ to indicate crustal movements against sea level, however, ignoring the fact that the concept “sea level” is related to other concepts which indicate “exterior” phenomena (e.g., river flow discharge or precipitation from atmosphere) that may change simultaneously with crustal movements (i.e., there will be nodes in ontology denoting these exterior phenomena and nodes denoting blocks of the earth crust, and both types of nodes will be bound with the third type, indicating the periods of time, by similar relation, say, “change” or “vary within”).
One can evidently see a two-tier division in modeling of tectonics and related disciplines, (i) modeling of the morphologic features, or “anatomy”, of the lithosphere, its interior and surface, and (ii) modeling of the processes that govern the anatomy, i.e., the “physiology” of the lithosphere.
The case (i) implies static entities, like the shape and size of landforms studied by the morphological subdiscipline of geomorphology (when mathematically formalized, this subdiscipline is known as morphometry) or anatomy of geological structures (studied by structural geology that performs description of form, arrangement, representation and analysis of structures that are seen in rocks). It is noteworthy that the “anatomy” of the surface and that of the interior need not to be corresponding each other. Thus, hills may well correspond to synclines and vice versa. Sometimes, if the data are accurately presented, a detailed description may bring an illusion that the static part of scientific research gives full explanation to the phenomenon under study. Still it lacks understanding of the same phenomenon across time and under different parameters.
In case (ii), the processes (i.e., dynamic entities) that govern the anatomy are the
focus of study. The dynamic entities change in time and space under some external or
internal conditions. Nonetheless, unlike structural geology, this sub-discipline of
tectonics has no specific term, though may more or less pass under the term
geodynamics. Still, geodynamics is commonly meant to deal specifically with the forces and
processes of the interior of the Earth. Therefore, to avoid ambiguity, in this paper the
following terms are suggested, morphologic tectonics and dynamical tectonics. Such
division is natural for many sciences, e.g., anatomy and physiology (of plants, animals
and men), planetary science and cosmogony. “Static” (classification) branches are
clearly seen in history, while the main body of its knowledge is certainly “dynamic”.
In general, one may say, on one hand, that “static” subdisciplines address the
composition and structure of systems, and “dynamic”, the dynamics, function and evolution
of the same systems, in terms of Bogdanov (1926) later replicated by Von Bertalanfi
(1968). On the other hand, however, this is fully compliant with the division of
knowledge in knowledge engineering into static and dynamic suggested by
Pshenichny and Mouromtsev (2013) and earlier formulated classification of methods of
knowledge engineering into object-based and event-based, correspondingly
Geotectonics encounters conceptual difficulties from perceptional conflicts out of variant interpretation of same observation. The dilemma has to be resolved to bring forth unified scientific approach to earth system understanding. This paper considers the applicability and usefulness of knowledge engineering methods in the study of tectonics. For this, it explores the application of knowledge engineering (i) in morphological tectonics (structural geology) and (ii) in dynamical tectonics. Its main mission is to pave the way to future research in bringing a unified ontology which caters dynamic models in geotectonics as well as in other branches of geology.
Recent studies revealed a variety of perspectives to deal with object based-methods of
knowledge engineering in morphological tectonics. Zong et al. (2009) suggested class
hierarchy of geological structures (Fig.1). Similar hierarchies and ontologies exist in
other earth-scientific domains, on which the dynamical tectonics is based
Poole et al. (2008) suggest an approach that marries ontologies and Bayesian probabilistic computation as a possible solution. Here, the structure of probabilistic theories does not necessarily follow the structure of the ontology. For example, an ontology of lung cancer should specify what lung cancer is, but whether someone has lung cancer depends on many factors of the particular case and not just on other parts of ontologies (e.g., whether they have other cancers and their work history that includes when they worked in bars that allowed smoking). As another example, the probability that a room will be used for living depends not just on properties of that room, but on the properties of other rooms in an apartment. Similarly, in geological parlance, it is difficult to bring interpretation directly from the geological data. For instance, in geological mapping, geologist often tends to see what he wants to see, sometimes departing rather far from the facts – e.g., he “sees” faults which unlikely can be seen, finds stress deformations where an evidence of strain exists, traces rock block displacement in an opposite direction and so forth. The decisions made by geologist are often intuitional. It is observed that the instrumental data, geophysical and others, are being treated very broadly, often solely not to undermine the theory that the geoscientist ‘’believes’’ in. Now adding the probability distributions to the classes of ontology, which describes tectonic study, as proposed by Poole et al. (2008) may give a tool to show how probable is the suggested interpretation of given data. However, the result would not solve the remaining puzzle – the evaluation of the theory itself. An attempt to resolve the problem is addressed below, considering all special cases present in tectonics. 4
All existing theories in tectonics are genetic, that is, they not only involve description
of products (usually done within the realm of morphological tectonics) but also
involve the description and interpretation of processes. For example, the great mountain
arc of Himalayas is not described as a static feature; instead, in tectonics it is
considered as a product of ongoing phenomenon of uplift, run either by gravity mechanics (
principles of heat engine) or by quantum mechanics (principles of stress engine)
During last two centuries, the Earth science saw the rise and fall of many
hypotheses that intended to look into the dynamical tectonism. They all can be considered as
modern way of looking into the planet Earth and its evolution through the ‘absolute’
geological time (a concept introduced by Patterson (1953) and Houtermans (1953b).
Most of the hypotheses in dynamical tectonics that has been debated fall into one of
three classes assuming one of the following states of the Earth, the “contraction”
(Beaumont’s mountain formation model, the Dana - Hall model, the Suess model, the
Barrell model), “expansion” (the Egyed-Jordan model, the Vogel model, the Carey
model) and “steady state” (the Hayford-Bowie model, the Kreichgauer model, the
Wegner – du Toit model, the Vine-Mathews-Morgan-Wilson model). There are a few
more models like plume tectonics, surge tectonics, vortex tectonics, Belousov and
Kosygin concepts
Despite the acceptance or rejection of a model, each of them contains facts which
are evident – and each gives sufficient explanation only to a part of such facts. For
example, the contraction tectonic school easily interprets tilted strata and mammoth
relief features on the globe, but seldom looks at the jig-saw puzzle fit of continents
across oceans (the Atlantic case). The hypothesis does not give an apt account of the
“stripped pattern” of magnetic anomalies in north Atlantic ridge sector. Similarly, the
plate tectonics and the expansion tectonics logically reason the “stripped pattern” of
magnetic anomalies and the very existence of middle oceanic ridge structures, but
keep silent about the trans-oceanic submerged bridges (having continental
characteristics) connecting continents across oceans
In modern dynamical tectonics, however, only one hypothesis, the plate tectonics,
completely dominates. It was in beautiful accordance with the data, mainly
geophysical, at the time of its formulation, being the same time amazingly simple and
selfconsistent. However, many new facts have been reported. In order to fit them, both
the theory was modified and interpretations of facts were varied. This has made plate
tectonics an object of critique
While the compliance with facts should be likely addressed by the object-based methods as discussed in the previous section, the structure of the theory, as it describes the processes that are believed to operate in the Earth crust and mantle, may be a subject for event-based methods of knowledge engineering. Also, structure of other theories and their compliance with similar facts and with each other should be studied by the whole armory of concept- and event-based methods. These methods are truly new in dynamical tectonics a.
Earthquake Concept Map
The modern state of tectonics urges wide application of knowledge engineering approaches mainly to handle psychological and social aspects of this science (traditions, bias, herding and so forth) and to reduce their impact on scientific results. At present, however, these approaches are being dramatically underestimated and underused. Two dimensions of their application are straightforward in tectonics, (i) development of hierarchies and ontologies of geological knowledge used in tectonics and adding probability distributions to determine the probability of interpretation given the data and (ii) plotting the mechanisms suggested by each tectonic theory and corresponding scenarios of Earth evolution. The former corresponds to morphological tectonics, treats knowledge in a static way and can be performed by object-based methods of knowledge engineering. The latter relates to dynamical tectonics, models dynamic knowledge and needs event-based methods of knowledge engineering.
While the hierarchies and ontologies have been abundantly developed and their
integration with Bayesian computation based on assumed probability distribution has
been discussed
- Assess whether a theory is self-consistent and how well it covers the domain it
pretends to cover (i.e., how well it describes the tectonic processes that lead to the
observed results), identify gaps, uncertainties and ambiguities in a theory, as well as
its “protective belts”
- Estimate the relevance of contradictions between the theories (which may appear purely verbal)
- Enquire whether the mechanism proposed by a theory must be necessarily global or may operate locally in space or time (for instance, whether the plate tectonics may develop only where the asthenosphere is thick enough to enable the plate motion and whether it may wane and give way to other mechanisms otherwise). and - Look for compatibility of mechanisms from different theories. For example, spreading of the oceanic floor may appear the case not only a driving force of plate growth in plate tectonics but, without subduction, also a consequence of expansion of the Earth. 6
Conclusions 1. Modern geotectonics requires application of methods and approaches of knowledge engineering.
2. Static knowledge engineering techniques (hierarchies, ontologies and others) work well in structural geology or, broadly speaking, in morphological tectonics.
3. In dynamical tectonics the need for application of knowledge engineering methods is much greater; what is required in this domain is methods of modeling events, states, processes and scenarios, or engineering of dynamic knowledge.
4. These methods have been largely unused in the discussed domain, and up to now, even if used, are applied mainly not to compare theories and develop a selfconsistent tectonic body of knowledge but to show the advantages of one given theory.
Acknowledgement. The authors are deeply obliged to Cyril Pshenichny whose enthusiasm largely fueled up the creation of this paper; also, Lev Maslov and Paolo should be thanked for presenting their opinion regarding the work. Jishnu B.K (CUSAT, Cochin) and Santhosh P.R. (Tandem) for helping us with their expertise in graphics.