Definition of entity: houses can be subdivided into rooms but not into houses ex: in Geomedia a feature (Smith Road segment 1) is an entity, the feature class (roads) is an entity set
  Entity Types ex: Smith Road may be a member of the entity "County Road" in Washington County's model and of "Local highways" in USGS Forest Service's "official" definition of "forest" is different than NRCS "official" definition of forest - maps do NOT line up. Entity attributes Special geographic (geometric) attributes

coordinates (x, y and z)

accuracy of coordinates

attributes reflecting "spatial" relationships such as connectivity

e.g. intersection A connects to streets B, C and D
valve 1 connects to pipe q
adjacency and others "topological" relationship
beside
within
near


In many "older" GIS (e.g. 1998) - "spatial" attributes were special. Kept separate from "normal" attributes
 
 

Special versus spatial attributes

Meyer argues that there are "fundamental" differences between spatial and non-spatial data

spatial data is

  • multi-dimensional (x, y and sometimes z)
  • autocorrelated (e.g. things close together are more related than things farther from each other)
    •  thus he argues:


    One focus of THIS course is that this is not correct - given recent software developments in databases!
     
     

    Nonetheless - first the old (1998) way!

    Bernhardsen uses word "object" to name the "carriers" of information in a data model.

    Unfortunate usage as object has many meanings object = feature = entity

    An object in B's terms is "Church Road" where this is composed of many different "segments." In GeoMedia Church Road would (probably) be many features (entities) in the feature class road.

    Its gets weird - is the object church road or is it the multiple segments and each of which have the attribute name = Church Road?
     
     

    B's real world objects are characterized by


    Objects have (in B's world) graphical representations (or components)

    Graphical representation are (according to B)


    B's graphical representation definitions

    point: zero dimensional object
    line: one dimensional object that is a straight line between two endpoints
    string: a sequence of line segments
    Area/polygon: a two dimensional object bounded by at least three one-dimensional line segments

    Some key points

    To be added to these should be geometric "graphical objects" - eg. spline, circles, ellipses etc. more on this later
     
     

    Entities versus fields

    A major area of consideration is whether a particular class of phenomena should be represented as an entity or a field

    field = cloud of data values - continuously varying

    e.g. elevation


    Contours may be viewed as one way to characterize a field as an entity

    Consider forest/grassland

    Entity or field?

     

    Planar enforcement, layers and entities

    In many GIS (e.g. Arc/Info) an entire map area must be divided into separate areal entities and no entity can overlay another

    Terms for the aggregated entities are coverage or layer

    Implications of this are substantial


    The "vector" model

    simple definition - set of x and y coordinates that define points, lines and polygons actually much more complicated

    polygon centroid - a link is a one dimensional object that is a direct connection between two nodes (question does link exist or is it a "relation") - a directed link is a link with direction specified

    string vs chain vs arc- a chain is a "directed set of non-intersecting line segments with a node at each end - arc is locus of points that forms a curve that is defined by a mathematical function (but not so in "Arc"/Info

    Area and area like features

    ring vs polygon - ring set line segments with closure (e.g. may be the "boundary" of an area

    simple polygon - area inscribed by an outer ring with no holes

    complex polygon - area inscribed by a outer ring and with one or more "holes"

    compound polygon (not one of B's) object defied by set of simple polygons that together comprise the object "e.g. the state of Hawaii

    If Entity Set is "states" then Hawaii would need to be defined as a complex polygon - but it is possible to define entity class "islands" and have seven entities with one attribute being Hawaii

    Older data model and area attributes

    In older data models (1998) it was common to have a special "point" called polygon centroid and the attributes of the area were associated with the centroid in the database

    The older data model isolated the "graphics" from the attributes

    Done because database access was difficult and slow

    Modern databases now are faster and access easier

    Current data model has geographic "parts" stored as just another "attribute"
     
     

    GeoRelational
     
     
    ID Number Owner Area Attribute n etc
    101 Jones 234 ac    
    102        

    AND - associated graphics file with stuff like

    101 2.3,2.3; 4.3, 4.7; ….

    102 2.3,5.6; x,y; x,y; …
     
     

    Object relational
     
    ID number Owner Area Geometry Attribute n
    101 Jones 234 ac 2.3,2.3;

    x,y;x,y;

    x,y;…

    		  

     

    This is the data model used by GeoMedia, Oracle 8i and the soon to be released Arc8
     
     

    Some further concepts

    spaghetti models versus topological data

    what's a topological relationship?

    next semester we will formally defined this but for now:

    topologically structure data is data in which connections and spatial relationships are described invariantly from any coordinates Definition of attribute states

    whenever possible it is useful to categorize possible attribute state in advance and (in some settings) create a "look up" table structure.

    Land use codes

    urban 1
    forest 2
    agri 3


    It is possible to only store the number

    Note that older systems (e.g. Arc/Info 7.x) required the formal creation of such structures. Modern systems (e.g. Access and others) have capability to create "domains" values for a field that can be selected form a "drop down" menu - works the same!
     
     

    Some database basics

    The spread sheet as a database
     
    Parcel ID Building type Owner Address  
    1 brick house smith 123 main  
    1 wood barn smith 123 main  
    2 wood frame house smith 123 main  
    3 stone house jones 345 water  

    Redundancy - potential for error etc.
     
     

    Relational database

    Decompose data into "relations" express each relation via a table

    Owner ID Owner name Owner address
    10 Jones 345 Water
    12 Smith 123 Main
    Parcel id Owner id
    1 12
    2 12
    3 10
    Parcel ID Building type
    1 wood barn
    1 brick house
    2 wood frame

     

    Entity relationships

    The ER Model

    A way to formalize relationships between entity classes
    Not all people accept idea but is a useful one
    Entities
     
    town ( name, mayor, state)
    building (size, age, construction type)
    parcel (number, size, date recorded, date tax paid)
    owner (last name, first name, age)


    Relations

    Each town contains one to many parcels

    Each parcel contains have zero to many buildings
    Each parcel can have one to many owners
    Each building must be on only one parcel

    General structure for data modeling

    The "real" physical reality

    Real world model (some level of mental abstraction) Data model - formal "logical/mathematical" structure (later) Database - actual computer representation (later)