Geog 458: Map Sources and Errors

1
Geog 458 Map Sources and Errors

• January 9 2006
• Representing Geography

2
Outlines

• Importance of geographic representation
• How do human conceptualize surroundings?
• How is human conceputalization coded in computer
  databases?
• Reading Spatial Data Organization in FGDC
  Metadata Content Standard

3
1. Importance of geographic information

• What is geographic representation?
• Why is geographic representation important?

4
What is representation?

• Viewing the real world (or surroundings, reality)
  
• Forcing the real world into a manageable concept
• Human conceptualization of the world
• Putting the real world into a computer
• How would you represent the followings in a
  computer? Does the same data model fit all?
• Vegetation, soil, traffic volume, land use,
  tornado, clouds
• Data modeling (building data model)

5
Why is representation important?

• Forms the basis of metadata organization
• Identification, spatial data organization, entity
  and attribute
• Many operations depend on data model
• Routing
• Spatial interpolation
• Forms the basis for understanding data structure
  and data format used in a GIS
• Vector, point, line, node, area, topology
• Raster, pixel, tessellation, sampling, planar
  enforcement

6
2. How humans conceptualize surroundings?

• Object vs. field
• Dimensionality
• Space, time and attribute
• Scale

7
Do-It-Yourself

• Exploring different ways to represent spatial
  concepts
• Group A how land parcel is represented in a GIS
• Group B how elevation is represented in a GIS
• Group C how tornado is represented in a GIS
• Group D how traffic volume is represented in a
  GIS
• Each group should discuss the representation of
  each theme in terms of (1) object vs. field (2)
  dimensionality (3) how space, time and attribute
  are organized (4) appropriate scale of space and
  time

8
Land parcel Elevation

• We can observe discrete boundary of land parcel
  but we cant observe that of elevation
• Land parcel forms polygon in terms of geometry,
  but elevation does not conform to well-defined
  geometry
• Identity change over time is gradual in the
  corresponding time scale ? temporal element is
  largely ignored, the relation between space and
  attribute determines representation

9
Traffic volume Tornado

• Movement of things with some pronounced
  properties across geographic path
• Can have all dimensionalities by leaving out less
  important details
• Temporal element is important
• Finer temporal scale is required to describe the
  subject properly ? attribute is largely ignored,
  the relation between space and time determines
  representation

10
(1) Object vs. Field

• Discrete object
• Identifiable boundaries ? easier for manipulation
  like a tabletop object conceived easily by direct
  human experience
• Building, population, county boundary
• Continuous field
• Variation in a given spatial scale ? requires
  spatial sampling scheme (tessellation)
• Temperature, population density, tax rate per
  county
• Any limitations?
• Soil boundary, mental map of localities
• Planar enforcement (i.e. polygon doesnt overlap)

11
(2) Dimensionality

• By identifying dimensionality of object, you can
  place them in Euclidean geometry
• Zero-dimension (point)
• One-dimension (line)
• Two-dimension (area)
• Three-dimension (volume)
• Any limitations?
• Even though most of data format (e.g. shapefile)
  available in GIS is only allowed to have one of
  possible dimensions, reality is that some object
  (e.g. lake) can have multiple dimensions
  depending on scale and applications
• Multiple representation (i.e. representing
  geographic entities across multiple scales)

12
(3) Space, Time, and Attribute

• Geographic information has three components
  space, time and attribute
• Data model can be understood through measurement
  framework
• Measurement framework all of three components
  are not fully measured, but rather one of them is
  measured, controlled, and fixed respectively
  (Sinton 1978)

13
(4) Scale

• Scale determines the way in which phenomenon is
  described
• Astronomy, geography, human biology
• Universe time, geological time, geographic time
• Scale influences the way humans cognize
  surroundings
• Our experience of surroundings is quite different
  depending on scale e.g. tabletop object (direct
  experience) versus cities
• Scale ? human conceptualization ? data model ?
  GIS implementation
• Representing reality in some geometry (e.g. city
  as point) is a reasonable approximation only at a
  particular scale
• Scale determines accuracy and thus fitness of use
  of data to particular purposes

14
Is GIS a container of digital maps?

• Partially yes (mainly because we are used to it)
• DRG, DLG, DOQQ, Satellite image
• But pitfall of limiting your thought on
  geographic representation to this doesnt help us
  explore other potential geographic
  representations
• The way of describing geographic information is
  not necessarily limited to point, line, polygon,
  and surface, but rather relational or
  propositional (e.g. I live in Seattle, you should
  turn left at the intersection) ? first-order
  logic (deductive database)
• City does not have geometry and attributes, but
  also has different functions (as adminstrative
  unit, as economic unit, as ecological unit) ?
  object-oriented database
• Geography is not the same as geometry indeed!

15
3. How is human conceptualization coded in
computer databases?

• Putting discrete objects into a computer (vector
  GIS)
• Spatial primitives
• Topology
• Generalization algorithm
• Putting continuous fields into a computer (raster
  GIS)
• Gridded raster
• TIN
• Georelational model as a special case of vector
  topological model (ArcInfo coverage)

16
(1) Spatial primitives

• Three spatial primitives are used to represent
  discrete objects
• Point as a point (x, y)
• Line as a set of vertices
• Polygon as a set of vertices that form to close

17
(2) Topology

• What is topology?
• Non-metric properties of geographic objects that
  remain constant when the geographic space of
  objects is distorted (e.g. projection change,
  transformation)
• Why is topology important?
• From data perspectives, it validates geometric
  accuracy (e.g. network connectivity, line
  intersection, overlap, duplicate lines) ? link to
  logical consistency in data quality
• From analytics perspectives, it optimizes queries
  by storing spatial relation information in a
  table
• How are they stored in a table? (see next slide)

18

• Connectivity
• Arc 2 connects from 11 to 12
• Containment
• Polygon C is surrounded by arc 2,4, 9, and 6
• Contiguity
• Polygon B/C in the left/right of arc 6

19
(3) Generalization

• Douglas-Poiker algorithm
• Process for simplifying line by reducing the
  number of vertices in its representation
• E.g. Cape Cod in 11,000,000 map ? Cape Cod in
  125K
• How it works?
• For illustration, see Figure 3.17 at Longley et
  al (p. 82)
• What does it achieve?
• It attempts to preserve pronounced changes in
  angle within a given tolerance
• It reduces the size of data, and speeds up the
  process for display and further analysis

20
(4) Raster grid

• Variation in values are stored in each cell
• Many geographic data (e.g. satellite image, air
  photo) are derived from this data structure
• It takes up large space it leads to development
  of many different raster compression methods

21
(5) TIN

• What is TIN?
• Triangulated Irregular Network
• Represents a surface as contiguous
  non-overlapping triangular plane where three
  points have different z-values (See Figure 8.12
  in Longley et al 2005 (p. 189)
• Why is TIN popular?
• Allows for varying density in sampling points
• Less storage because it stores only critical
  points (cf. raster grid)

22
Georelational model

• Geometry of spatial object is stored in a file
  and attribute is stored in a table
• File is linked to attribute through a common
  identifier
• Separation between spatial and non-spatial
  attribute
• See Figure 8.10 (e.g. Arc/Info coverage)

23
4. Reading Spatial Data Organization in CSDGM
24
Do-It-Yourself
Air Quality Lake Monitoring Sites

• What is Indirect Spatial Reference? Take any
  example.
• What is Direct Spatial Reference Method?
• What is STDS point and vector type?
• What is Entity Point?
• What is the difference between point and node?
• Resources
• http//fgdc.gov/metadata/csdgm/03.html
• http//mcmcweb.er.usgs.gov/sdts/SDTS_standard_nov9
  7/part1b10.html

Metadata from Geospatial One-Stop