NADM Data Model Design Team (DMDT) Meeting Notes

October 29, 2002, Denver, Colorado

Attendees:

  • Éric Boisvert (GSC)
  • Boyan Brodaric (GSC)
  • Peter Davenport (GSC)
  • Jordan Hastings (UC-Santa Barbara)
  • Bruce Johnson (USGS)
  • Jim McDonald (Ohio GS)
  • Steve Richard (Arizona GS)
  • Peter Schweitzer (USGS)
  • Jerry Weisenfluh (Kentucky GS)

Guests

  • Martin Anctil (GSC)
  • Rob Wardwell (USGS)

  1. Confirmation of agenda
  2. Old business
  3. New business
  4. Action items: complete NORTON by DMT'03 (Pennsylvania)

    Persons Task Completion Date
    Boyan Brodaric Distribute NORTON 0.91 to NADMSC Nov. 4, 2002
    Boyan Brodaric Complete NORTON class hierarchy Dec. 1, 2002
    Steve Richard Complete NORTON cardinalities Jan. 1, 2003
    Bruce Johnson Complete scope notes and definitions Feb. 1, 2003
    DMDT Review NORTON Mar. 1, 2003
    Boyan Brodaric,
    Jerry Weisenfluh
    Document NORTON in PDF mid-Mar., 2003
    Peter Schweitzer Post NORTON draft to DMDT web site April 1, 2003
    NADMSC Review NORTON May 1, 2003
    DMDT Revise NORTON as per NADMSC review DMT'03
    Peter Schweitzer Post NORTON 1.0 to DMDT web site DMT'03
    DMDT Report on NORTON at DMT'03 DMT'03
    DMDT Submit NORTON to NADMSC for approval DMT'03

Details

NORTON-0.91-28Oct02 UNRESOLVED ISSUES

# Diagram Concept Issue
1. Geologic Portrayal Geologic Concept/
Geologic Class
disambiguate
2. Geologic Portrayal Extent add to diagram?
3. Geologic Unit Extent link to spatial object?
4. many Genesis rename to GeneticHistory?
5. Earth Material Void Description define, describe
6. Earth Material Fluid scope note: magma outside scope
7. Earth Material Fabric / Texture add xor dependency
8. Earth Material Form rename to Morphology?
9. Geologic Property Physical Property add to diagram
10. Geologic Property Physical Property add subtypes: Geotechnical Property, Petrophysical Property?
11. Geologic Structure Geometric Description delete; subsumed by Form/Morph.
12. Geologic Structure Geologic Structure scope note explaining level of detail of modeling; e.g. absence of describing the relation between parts of a compound structure
13. many many show explicit relations between?:
GeoUnit-GeoStructure
GeoUnit-EarthMaterial
14. Geologic Relation Geologic Relation how to (should we) model topology: relations between spatial objects?
15. Controlled Vocabulary Controlled Vocabulary scope note: re: multiple sources
16. many Properties survey from GEON document

Notes

Focus of this meeting is to describe to the DMDT the latest version of the conceptual model, called NORTON (NORTh american digital geological map ONtology). The comments below pertain to the most recent NORTON UML diagrams, which should be posted in PDF format alongside these notes.

The group of things that pertain to geologic maps and geologic information (loosely defined) is called GeologicConcept:

Higher structures:
Metadata
represents description of data elements (things in the database)
Graphic
an image
Occurrence
an instance of some concept.
SpatialObject
a description of geometry, referenced to the Earth in some fashion.

Everything can have
Name
arbitrary text string used for human identification of data instances.
TextDescription
arbitrary length text.

The diagram will be revised to (a) clean up names, (b) add cardinality, (c) add intended constraints that are not already explicit, and (d) add definitions, scope notes, and non-preferred terms.

Major concepts

EarthMaterial

EarthMaterials are the substance of the solid Earth (rocks, minerals, organic material, glass...), defined based on intrinsic properties independent of their disposition within the Earth.

An EarthMaterial can be composed of one or more of the following: CompoundMaterial, Mineral, OrganicMaterial, and/or Fluid (including glass)

CompoundMaterial can be consolidated (Rock), unconsolidated material, or Fossil [should be FossilMaterial], and is made of other EarthMaterials (including other CompoundMaterials). ConsolidationDegree is an attribute to help describe things that are intermediate.

When an EarthMaterial is made of several things (i.e. a Compound Material), the Constituents are described separately, including their role (e.g., phenocryst vs groundmass, cement, and so on), their proportion, and relationship to other constituents (EMCRelation, such as overgrowth).  The model includes a way to specify the role each constituent plays in the relationship.  This scheme could be used to document a modal analysis, for example.

CompoundMaterials, as well as their constituents can have Fabric and Texture.  Fabric is regarded as a type of GeologicStructure, focusing on the internal arrangement of the parts of the rock.  In contrast, Texture here refers to particle size and particle shape, which are described qualitatively, quantitatively, or both.  Sorting terms are part of the qualitative description of particle size.

There is a dependency ("XOR" relationship) between the Fabric of a Constituent and the Fabric of a CompoundMaterial.  Either Fabric or Texture may be described for a Constituent or for the entire CompoundMaterial, but the description for the Constituent must be consistent with the description for the whole. The logic of this consistency check is complex, and probably requires intervention by a geologist to enforce.

All EarthMaterials may have Color, ChemicalComposition, GeologicAge, Physical Properties (unspecified geotechnical or engineering properties) and EMGenesis (which is related to Genesis in GeologicProcess so that the history of the material's formation can be described).  EMGenesis includes the notions of input and output so that materials can be related to geological concepts such as "protolith".

It's important to note that on the model diagram, some of the boxes represent simple attributes, while others may represent more complex structures that are typically defined more fully in other parts of the diagram.

"Sediment" does not appear in the model as a separate concept; it is typically an Unconsolidated CompoundMaterial with a given history.

Some further discussion should focus on describing the void spaces in the rock, and possibly gases or fluids that occupy or move through the void space.  Some characteristics of the void space could be described in the context of an EMConstituent.

Magma, as an EarthMaterial, is regarded as outside the scope of the model at this time.

GeologicUnit

GeologicUnit represents an identifiable part of the Earth.

There are a number of subtypes of GeologicUnit.  Some were found in the North American Stratigraphic Code (NASC), and a few additional subtypes were defined for this model.

From NASC:

Chronostratigraphic

Polarity-Chronostratigraphic

Lithodemic

Lithostratigraphic

Allostratigraphic

Magnetostratigraphic

Pedostratigraphic

Additional unit types:

Pedodermic (Pedologic?)

Pedoderm is from the 4th edition of Jackson (1997), but I don't recall seeing the same term in earlier editions.  The def seems to cover the concept that we need, i.e., a mappable soil unit.

Lithotectonic

Geomorphologic

History of geologic units is described by GUGenesis, which may describe a single geologic unit or may relate one geologic unit to another.

Relationships among geologic units are described by sets of GeologicUnitParts [may be renamed GeologicUnitConstituent], with role and proportion of the parts, and relationships among the parts (Analogous to relationships used for EarthMaterials).  The role of the part in the relationship is also specified by GUPRelationRole.  A GeologicUnitPart may be a CompoundMaterial (e.g., a lithology) or another GeologicUnit (e.g. member of a formation), but the UML diagram implies that a GeologicUnitPart *has* a GeologicUnit).

Properties of geologic units are generally descriptive of the whole:

Rank (series, formation, member, bed)

WeatheringCharacter (description of rinds, varnishes, soils, ...)

OutcropCharacter (cliff-forming, slope-forming, ...)

GeologicAge

Extent (description of the distribution of the unit)

GeometricDescription (orientation, thickness, planar/linear, fold wavelength, ...)

(GeologicStructure should probably be used for this instead)

Form (lensate, tabular, conical, ...)

Color (including weathering color)

style='margin-left:0in;text-indent:0in'>MetamorphicGrade (including coal rank) 

Properties of GeologicUnitParts may include descriptions of the EarthMaterials that compose them.

Fossil

This problematic concept includes the subtypes cast, mold, trace fossil, and PreservedRemain.  Fossils are sometimes recognized by the form that is imprinted in an EarthMaterial, and sometimes recognized as the preserved remains of a life form. Fossils are typically thought of as objects (c.f. Bates & Jackson, 1987: “Said of any object…”), whereas EarthMaterial represents substance (see philosophical literature on object vs. substance). Geoscientists commonly describe fossils as constituents of rocks.

The current diagram distinguishes Fossils that are tangible CompoundMaterials (e.g., casts and preserved remains) from those that are GeologicStructures (e.g., molds and other trace fossils). These may need to be distinguished with different names (e.g. FossilMaterial, and FossilStructure) because of the differences in how they are used in rock description, and in how they are described themselves.

Fossil is explicitly related to both EarthMaterial and GeologicUnit.

The diagram allows that the proportion of fossils in a CompoundMaterial can be described explicitly, because the constituents of a CompoundMaterial may be other CompoundMaterials, and Fossil [i.e. FossilMaterial] is a kind of CompoundMaterial.   The mineral composition and structure of a Fossil [i.e. FossilMaterial] can also be described explicitly in the same way that the mineral composition and structure of a rock (or any other CompoundMaterial) can be described. 

GeologicStructure

GeologicStructure represents configurations of substance within the Earth.

Specific types of geologic structures defined include

Contact

3>PreLithificationStructure (Used for both sedimentary and igneous rocks.  Example: BeddingSurface)

Layering

Lineation

Foliation

Fault (with attribute Displacement)

Fold

Fracture (example Joint)

The same geologic thing can be seen as more than one of these types.

Properties of GeologicStructure

GeologicAge

Genesis (history)

Pervasiveness

GeometricDescription (orientation, thickness, planar/linear, fold wavelength, ...)

Extent

GSDefiningElement (what makes the structure what it is)

discontinuity

shapedsurface

orientedParticle

materialboundary

layer

GeologicStructures may be described as compound; structures can be made up of structures.  In the diagram, CompoundGeologicStructure consists of one or more GeologicStructures.  Fabric is a CompoundGeologicStructure.  [Should we add an attribute for the relationship between a CompoundGeologicStructure and its component GeologicStructure(s)?]

There was an extended discussion attempting to classify geologic structures in a hierarchical way (e.g., penetrative vs discrete; planar vs linear; rock forming vs rock modifying), but none of these satisfied our constraints of simplicity and logic.  We consequently resolved to use a relatively flat structure.

GeologicProcess

Processes include the following types:

HydrothermalAlteration

Crystallization

Deposition (Example: Sedimentation)

Erosion

Intrusion

Volcanism

Metamorphism

Diagenesis

Metasomatism

Deformation

(possibly others ...)

Each GeologicProcess has one or more GeneticEvent; the ordered set of GeneticEvents comprises the Genesis (history).  Each GeneticEvent has a GeologicAge and a GeologicEnvironment (which remains unspecified at this point but is construed broadly).  The history of EarthMaterials and GeologicUnits are types of Genesis.  GeneticEvents might be thought of as at least partially overlapping, but the diagram says that the set of GeneticEvents comprising a Genesis is ordered. We recognize that GeneticEvents are typically continuous rather than discrete (instantaneous) but suspect that this arrangement will suffice for most probable uses.  It may be inadequate for use in simulation.

GeologicRelation

This concept allows us to link any GeologicConcepts to any others. We recognize several specific types of relationships for illustrative purposes:

StratigraphicRelation

IntrusiveRelation

StructuralRelation

SpatialRelation

TemporalRelation

(probably others ...)

More than two GeologicConcepts can be described in the same GeologicRelation; each has a GeologicRole in the relationship. The GeologicRole would have a vocabulary including terms like "overlies", "is overlain by", "is younger", "is older", "intrudes", "is intruded by", and so forth.

Question raised by Eric: Does the model distinguish between things we know don't exist and things we don't know exist? [Steve: at this point it seems that this is not explicitly represented. An ExistentialConfidence attribute may be useful. The confidence in existence of some phenomena is subjective…]

Jordan: It seems like almost all of the information on the map could be described by this relationship mechanism. Steve: I think you’re right, but this would be a meta model; in making a schema for a particular model, we make statements about which relationships are important enough to represent explicitly.

Boyan:  We don't expect general relationships among spatial objects to be described explicitly with this mechanism, but to be encapsulated implicitly by the geometry inherent in the spatial objects themselves. But because some ways of transferring spatial objects don't include topological relationships, so at least some topological relationships may be described explicitly where that meaning is significant from a scientific perspective.

ControlledVocabulary

This provides a way to specify the terminology associated with concepts and the relationships among both the concepts and the terminology.  Many names can be associated with a concept; one is designated as preferred in order to facilitate communication and consistency in the database.

GeologicTimeScale and LithologicClassification are types of ControlledVocabulary, but there are many other kinds not specified here.

The ControlledVocabulary may describe a formal standard or any individual's [self-consistent] usage of terms; it is not restricted to formally-published vocabularies.

Note that the metadata associated with a concept would include the source of a given term.

Here GeologicRelation subsumes generic relationships such as broader, narrower, precedes, follows, that are found in formal thesauri and in ordered sets.

GeologicProperty

This general class exists to allow the properties of geological things to have all of the properties and behaviors inherent in GeologicConcept (metadata, for example??).

The properties appear elsewhere in the model with appropriate roles (implicit or explicit).

Boyan would like to map the properties defined by GEON to those we have defined here.  Bruce notes that our list is likely to be incomplete because we simply haven't fleshed out the model to the fullest level of detail.

GeologicPortrayal

The clearest motivation for this concept is the need to describe presentations [Steve: isn’t portrayal explicitly concerned with visual presentations? Why not use the term visualization?] of geologic concepts, either as information derived from the database or as information supplied to the database (so that, for example, authors of geologic maps used to populate the database can recognize that their work has been fairly represented in the database). Steve: another important reason to record individual visualizations of data from the database is that maps (visualizations, presentations…) are designed by a geologist to communicate a particular understanding of the geology, and the choice of what to symbolize, and how it is symbolized has a significant impact on how the map communicates. Different visualizations based on the same data might communicate different aspects of the geology.

Subtypes of GeologicPortrayal include

Correlation Chart

Cross-section

GeologicMap

GeologicGraphic

GeologicPortrayal may have one or more associated Graphics, which may be GeologicGraphics or CartoGraphics. GeologicGraphics are graphics in which the spatial arrangement of elements in the graphic has some correspondence with the spatial arrangement of features in the Earth (a mapping…, i.e. could be georeferenced). CartoGraphics are graphics in which the arrangement of the elements is not a mapping (in the technical sense) to spatial arrangement in the earth, but may follow some other logic related to spatial or temporal relationships. A Cross-section and a GeologicMap are geologicGraphics. Correlation charts, explanations of map units, ternary diagrams to show composition are all CartoGraphics.  (Some parenthetical definitions would be helpful here, but I can't remember what these mean).

Each GeologicMap has an associated MapDescription which is the conceptual set of spatial objects, their classification, and symbolization for a particular geologic horizon, scale and projection.  A Cross-section (which is considered a special kind of geologic map in a vertical plane) will generally also have a MapDescription.

MapDescription consists of a set of spatial objects, symbols, and the meanings of the symbols.  The Legend associates symbols with meanings (concepts). Some symbols may be unique to a particular occurrence, for example if a particular mine of special interest is given a unique symbol, different from other mines on the map.

The MapDescription has associated Extents, Scale, Projection, BaseMap, and Horizon. BaseMap refers to the actual graphical image that provides geographic and cultural reference for the geologic occurrences displayed in the visualization. Extent describes the boundary of the region displayed in a visualization. Scale and projection determine the mathematical relationship between locations in the visualization and locations in the Earth. Horizon refers to the surface on which features are mapped, commonly surface topography, but also may be a vertical surface (cross-section), a horizontal surface (mine level), an unconformity, a fault surface (fault cut-off map), etc.... Most lines on a geologic map represent the intersection of 3-D geologic surfaces with a map horizon. A BaseMap typically relates to a single map horizon, but geologic lines on the map may be related to more than one map horizon. For example, structure contours represent the intersection of a geologic surface with multiple horizontal map horizons at different elevations, all of which are probably different from the map horizon of the BaseMap on which they are displayed.

The Legend consists of the explained symbols both of GeologicClasses (which summarize many occurrences) and individual occurrences. SpatialObjects are associated only with Occurrences, not with GeologicClasses.

Each SpatialObject is associated with a projection, scale, and horizon (which refers to the surface on which the feature is mapped [possibly a volume in 3-dimensional models]).

Boyan is concerned that the things comprising the MapDescription need to be defined as types of something else.  There is an ongoing argument about the terminology associated with the box labeled "GeologicClass".

Steve—we need clear definitions of what Occurrence, GeologicClass, and GeologicConcept are supposed to mean.

There is a conceptual distinction between the horizon defined for mapping features and the plane (or volume) in which the features are portrayed graphically.

It is possible that MapDescription could be used to describe portrayals of the other relationships (for example stratigraphic) among concepts in, for example, correlation charts and other diagrams that are not in a strict sense maps.


This page is <http://geology.usgs.gov/dm/steering/teams/design/DMDT_20021029.html>
Maintained by Peter Schweitzer
Last updated 13-Dec-2002