Geog 258: Maps and GIS
February 24, 2006 (Fri)
Geographic Information System
(GIS): Representation
Learning objectives
1. Understand how geographic phenomenon is measured
2. Understand how attribute is represented in GIS
3. Understand how space is represented in GIS
4. Understand how attribute & space are represented
in GIS
Measurement of geographic
phenomenon
Compare
information system with no spatial component to information system with spatial
component
How
would the databases look different? In other words, how differently would the
information be stored?
GIS
deals with georeferenced data and attributes attached
to spatial entities
Geographic
information has three main components
Space,
time, and attribute
Unlike
IS, database in GIS does not only consider attribute but also space and time
Spatial
component is central to GIS
Many
questions on GIS data model revolve around how space is combined with attribute
Can
you measure all of the three?
Example: population density per census tract in
In the map, what is measured among {space, time,
attribute}?
Time is fixed (e.g. source data: 2000 census)
Space is controlled (e.g. census tract)
The map does not measure the location of population,
but rather population count within census tract divided by areal
size of census tract is measured. Census tract (space) serves as the control in
which the attribute can be measured
Attribute is measured
Can
you measure all of the three?
Example: the location of cities in the
In the map, what is measured among {space, time,
attribute}?
Time is fixed
Space is measured
The location of cities is measured
Attribute is controlled
While the attribute (i.e. city) is controlled
Can
you measure all of the three?
elevation
land parcel
changes in city limits over the
last 50 years
sea level at the station
Measurement
of geographic phenomenon
► Three main components of
geographic information is space, time, and attribute
► All of them are not thoroughly
measured; but rather some of them should be sacrificed to allow for the full
measurement of one component
► One is measured, one is
controlled, and the other is fixed
► Control of other components
of a phenomenon permits the measurement of one component
The
case in which attribute plays role of control
1)
The precise location of geographic phenomenon (thing, event, entity, and so on)
can be measured e.g. land parcel, city location
2)
The line of same value (isoline) is derived in a way
that attribute is controlled and space is measured e.g. contour line
3)
The exhaustive boundary of geographic phenomenon is measured e.g. categorical
map
The
case in which space
plays role of control
1)
The phenomenon occurs continuously, thus it is infeasible to measure its values
at all locations, it is necessary to control space while attribute is measured
in the spatial unit e.g. DEM
2)
The attribute is measured in a predefined spatial unit (e.g. census) e.g. choropleth map
Representation of attribute
As a main component of
geographic information, (non-spatial) “attribute” has been stored and managed
in different ways, evolving from file system to database system
1. File system
Also
called flat file (e.g. ASCII file)
Still
common file format for distribution
Can
be converted to tabular format (use ms-access, or statistics package)
Comes
with data dictionary
2. Database system
Unlike
file system, a collection of “related” data
Unified
storage increases efficiency in data management
2-1. Relational database
·
Entities can be
viewed as tables
·
Row and column of
table represents entity and attribute, respectively
·
Building
relationship between entities are performed by common attributes
You can link attribute data to spatial data through a
common identifier
·
Primary key:
unique identifier
·
Foreign key: key
that allows for linking to other table
·
SQL: structured
query language for relational database, reflects standardization efforts
·
Normalization:
rules for reducing redundancy in table
·
Many commercial
GIS systems are built upon relational database
2-2. Object-oriented database
·
The world can be
seen as a collection of autonomous objects
·
Data and
procedure are not separated
·
Embodies
object-oriented concepts such as inheritance, polymorphism, and encapsulation
·
Handles abstract
data type (ADT)
·
Its idea has been
the engine for innovation in software design, database management and data
modeling and so on (e.g. UML: notation language for communicating system design
with object-oriented concepts)
·
Future GIS
system?
Representation of spatial
entities
There
are two common data models that represent geographic phenomenon
1. Raster model
Attributes
are stored in controlled spatial unit such as grid cell
Good
for representing continuous fields (e.g. elevation, temperature, soil type)
2. Vector model
Precise
location is stored while attribute is controlled
Good
for representing discrete objects (e.g. building, land parcel, lake
Vector
model can be classified depending on (1) whether topology exists (2) what kind
of spatial primitives are used (cartographic analogy: point, line, area)
What
is topology?
Simply
put, spatial relationships
2-1. Vector model by
topology
·
Spaghetti vector model: data without topology à mainly for display
·
Topological vector model: data in which topology is built à allows for complex
operations (e.g. network analysis, accurate spatial measurement)
2-2. Vector model by
spatial primitives
·
Point: 0-dimensional
·
Line: 1-dimensional
·
Area: 2-dimensional
How
are they stored in files (such as ASCII file)?
They
are scale-dependent
Q.
Identify data model
Topographic
surface of
Fatal motor vehicle accidents
& Road network & Lake Erie in
Orthophoto image and road network in SUNY-Buffalo Campus
Comparison between vector
and raster
|
|
Vector data |
Raster data |
|
Base
of reps. |
Coordinate |
Cell |
|
Good
to reps. |
Discrete
entities e.g.
school, lake, event-location, road |
Continuous
entities e.g. temperature, elevation. toxic level |
|
Example
data or product |
TIGER/Line,
USGS DLG |
Satellite
imagery, aerial photo, USGS DRG, DOQ, DEM |
|
Spatial
data format |
shapefiles, Arc/Info coverage, AutoCAD DXF file |
TIFF,
JPEG, MrSID, BMP, BIL |
|
Attributes |
Multiple
attributes are stored in a linked tabular data |
Usually
a single attribute (z-value) |
|
Note |
Topology |
Resolution
|
Translating
data between vector and raster model
Scanned maps are vectorized
when you have to create vector data from paper maps
Vector data are rasterized when you export vector maps into image files
Elevation (continuous surface) is represented by
vector model (e.g. contour line)
Advantage/disadvantage of
vector/raster model
·
Raster: many data
sources are already stored in this format (e.g. satellite image, orthophoto), equivalent to multidimensional array, requires
large space à compression technique (e.g. MrSID), precise
location is not measured, thus some measurement is not quite accurate
·
Vector: precise
location is measured (even though some approximation exists depending on
tolerance given to vertices: GBF/DIME vs. TIGER), storage space is saved, some
geographic phenomenon is not well represented in this data model (e.g. surface)
Representation of space and
attribute: GIS Architecture
1. Hybrid system
Spatial
data is stored as files separated from attribute stored in table
e.g. Arc/Info coverage: separation between Arc and Info (it’s called georelational model)
2. Integrated system
Spatial
database management system approach
Vector/raster
data model is stored as relational tables
3. Object-oriented system
GIS
data is modeled using OO concepts for natural representation of spatial
entities
e.g. Geodatabase supports (unlike previous data
model) rules and relationships
Lack
of consensus on how it can be implemented
Not
good for representing fields
