GIS - Lecture 4

8/20/2010

GIS Data Structures and Models

GE 517 – Geographic Information Systems
Engr. Ablao

Digital Geographic Data
Numerical representations that d ib real-world f
N i l i h describe l ld features
and phenomena
Must be encoded in digital form and organized as a
geographic database to be useful to GIS
This digital database creates our perception of the real
world

Geographic Information System 8/20/2010

1

8/20/2010

Conventional Data vs. Digital
Geographic Data
Conventional Data: Paper map
Static representation
Represents a general-purpose snapshot of the real world at a
given time only

Digital Geographic Data:
Dynamic representation
Allows a range of functions for storing processing analyzing
storing, processing, analyzing,
and visualizing spatial data
Has tools that allow users to interact with the data to meet their
objectives


FIGURE 1. Bringing real-world data into a GIS or a map requires
several abstractions and/or conversions of the original data. (Freely
adapted from Bernhardsen, 1992).

Physical Real world Map /
→ → Data model → Database →
reality model Report

Actual Entity Object Object Symbol, line,
phenomenon text

• Properties • Type • Type • Type
• Connection • Attributes • Attributes • Attributes
s • Relationship • Relationship • Relationship
• Geometry • Geometry
• Quality • Quality


2

8/20/2010

Figure 2

What is a Data Model?
The h
Th heart of any GIS
f
A way of representing data

A set of constructs for describing and representing
selected aspects of the real world in a computer
l ld computer.

Longley, Goodchild, Maguire & Rhind


3

8/20/2010

Geographic Data Models
REAL WORLD

Field-based model Object-based model

Raster data model Vector data model

Figure 3

Raster and Vector Data Models

Figure 4

4

8/20/2010

Raster Data Model
Treats geographic space as populated by one or more spatial
phenomena,
phenomena which vary continuously over space and having
no obvious boundaries.
Best used to represent geographic features that are continuous
over a large area (e.g. soil type, vegetation, etc.)
Uses an array of rectangular cells/pixels/grids to represent
real-world objects.
real world objects
Each cell is defined by a coordinate location and an attribute
that identifies the feature.
Cell’s linear dimensions defines the spatial resolution

Raster Data Model

Figure 5

5

8/20/2010

Raster Data Model

Figure 6

Raster Data Model

Figure 7

6

8/20/2010

A point is represented as a
single cell.

well

tree

rock

Figure 8

A line is represented as a string of cells.

river
road

canal

Figure 9

7

8/20/2010

An area is a contiguous block of cells.

grassland

barn

forest

Figure 10

Raster Data Model
The cell size determines the resolution of the data.
The grid cell has an extent.
When we map features onto the grid, there is
sometimes an imperfect fit.
Each grid cell can usually be “owned” by one feature,
that is, the one whose attribute it holds (mixed pixels).


8

8/20/2010

Figure 11

Figure 12


9

8/20/2010

Raster Data Compression
A single raster file usually contains several million grid
cells l d to a very l
ll leading large d f l
data file

Therefore, data compression is very important in the
digital representation of raster data!


Based on the realization that cells representing the
same ffeature type have identical values and that they
h d l l d h h
tend to be spatially clumped
Examples of raster data compression techniques are:
Run length encoding
Wavelet compression


10

8/20/2010

Run length encoding
g g
Adjacent cells along a row having the same value are
treated as a group known as a ‘run’
instead of storing the similar cell values individually,
the value is just stored once together with the
number of cells that comprise the run
b f ll th t i th


Run length encoding

Figure 13


11

8/20/2010

Wavelet Compression
p
More efficient compression
Can achieve a ratio between 15:1 to 20:1 for 8-bit
gray scale images and between 30:1 to 40:1 for 24-
bit color images


Raster File Formats
BMP (bitmaps) – used by graphics in Microsoft Windows applications; no
p
compression
DIB (Device Independent Bitmaps)
GIF (Graphical Interchange Format) – widely used for images to be used on the
World Wide Web
TIFF (Tagged Image File Format) – non-proprietary, system-independent
JPEG (Joint Photographic Experts Group) – primarily for storage of photographic
images and for the World Wide Web
GeoTIFF – extension of the TIFF format that contains georeferencing information
PNG (Portable Network Graphics) – patent-free; intended to replace the GIF format
PCX – supported by many image scanners
MrSID (Multi-resolution Seamless Image Database)
GRID – proprietary format used by ESRI in ArcInfo and ArcView GIS


12

8/20/2010

Quadtree Data Model
A hierarchical field-based model that uses grid cells
of variable sizes
f i bl i
Uses finer subdivisions in areas where finer detail
occurs
Geographic space is divided by the process of
recursive decomposition
More difficult compared to the simple raster model


Quadtree Data Model

Figure 14

13

8/20/2010

Digital Elevation Model

A raster-based
representation of
surface

Figure 15
Geographic Information System
Southwest Corner of the 8/20/2010
Morrison Quadrangle, Colorado

Remotely sensed
data are a special
type of raster data.

Figure 16

14

8/20/2010

Other forms of raster data

Scanned
Aerial maps
photos

Pictures

Converted data

Figure 17

Vector Data Model
Treats geographic space as populated by discrete objects, which
have identifiable b d i
h id tifi bl boundaries or spatial extent
ti l t t
Each object in the real-world is represented as either point, line,
or polygon features.
characterized by the use of sequential points or vertices to define
a linear segment, each vertex consisting of an X coordinate and
a Y coordinate.
Useful for representing discrete objects such as roads, buildings,
rivers, boundaries, etc.


15

8/20/2010

Vector Data Model

Vector Data Model Raster Data Model


Vector Data Model


16

8/20/2010

Vector data model
The vector is
composed of a
p
string of points,
each represented
by their exact
spatial
coordinates.

Points are represented by their
coordinates.

Vector data model

Lines are defined as a string of x- and y- coordinates.

17

8/20/2010

Vector data model

Areas are a string of x- and y-coordinates that terminate at the origin.


Types of Vector Data Model
1. Simple (spaghetti) data model
p (pg )
2. Topologic data model


18

8/20/2010

Simple (Spaghetti) Data Model
a one-for-one translation of the analog map
Features (lines and polygons) may overlap
No relationships between features/entities
Generally used where analysis is not important (eg.
Plotting)
each entity is a single, logical record in the computer,
coded as variable l th strings of x- and y- coordinate
d d i bl length t i f d di t
pairs
Examples are digitized maps and CAD data


g
Advantages:
Easy to create and store
Can be retrieved and rendered on screen
very quickly


19

8/20/2010

Disadvantages:
Spatial analysis (eg. Shortest path network analysis)
cannot be performed easily due to lack of any
connectivity relationships
Redundant since the boundary segment between two
polygons can be stored twice (once for each feature)
Inflexible since it is difficult to dissolve common
boundaries when joining polygons
Cannot be used in certain applications such as land
management because of overlapping features

Topological Data Model
spaghetti model but with topology!
Essentially just a simple d model structured using
E ll l data d l d
topologic rules
often referred to as an intelligent data structure
because spatial relationships between geographic
features are easily derived when using them
Most commonly used variants is the arc node data
arc-node
model
uses nodes (intersection of two lines; end point of
line, and any point on line) and arcs (line connecting
two nodes) System
Geographic Information 8/20/2010

20

8/20/2010

Topological Data Model Components
Point (vertex) – where a line originates or terminates
Node – where a link originates or terminates
Line – a segment between two points (vertices)
Link (or arc or chain) – a connection between two nodes
– may consist of several lines which are joined at points
(vertices)
– can only originate, terminate or be connected at nodes
Polygon (area) – composed of links
– adjacent polygons have only one link between them


21

8/20/2010

Elements of Topological Relationship
Connectivity
Information about linkages among spatial objects
I f ti b t li k ti l bj t
Keep track of which links are connected at a node
Adjacency
Information about neighbourhood among spatial objects
A link can determine the polygon to its left and its right
Containment
Information about inclusion of one spatial object within
another spatial object
What nodes and links and other polygons are within a
polygon

Elements of Topological Relationship


22

8/20/2010

Topological Map
Explains the concept of topological relationships
A map that contains explicit topological information on
top of the geometric information expressed in coordinates
(Masry and Lee, 1988)
All spatial entities are decomposed and represented in the
form of the three basic graphical elements (point, line,
polygon)


Concept of a Topological Map


23

8/20/2010



24

8/20/2010


Arc-node Data Model
Stores graphical elements rather than graphical
entities
Explicitly stores spatial relationships between the
graphical elements, and relationships between the
arcs and respective nodes
Stored t l i l l ti hi ll
St d topological relationships allow graphical
hi l
entities to be constructed using the basic graphical
elements

25

8/20/2010

Arc-node Data Model


Network Data Model
A special type of topologic model
p yp p g
Shows how lines connect with each other at
nodes
Applications:
Load analysis over an electric network
y
Vehicle routing over a street network
Pollution tracing over a river network

26

8/20/2010

Use of Topological Relationship
Areas in GIS wherein topological relationships are
applied are:
1. Data input and representation
2. Spatial search
3. Construction of complex spatial objects from
basic graphical elements
b h l l
4. Integrity checks in database creation


Data input and representation
Creating a topologic geographic database is
extremely time-consuming and error-prone
Possible to adopt the spaghetti digitization for
graphical data input and then later on let the
computer refine and structure them using
topological relationships


27

8/20/2010

Data input and representation


Spatial search
GIS uses topological relationships to perform spatial
searches efficiently
h ffi i tl
For example, the land parcels surrounding a particular
parcel can be located immediately using adjacency
information
Polygons that share a common boundary with the parcel of
interest are identified
Their IDs are used to retrieve the descriptive data from the
database


28

8/20/2010

Spatial search


Construction of complex spatial objects
Complex objects are represented as complex
polygons in a geographic d b
l h database
Two types of complex polygons are:
Those containing one or more holes (or
islands/enclaves) constructed from vector lines
using topological information
Those d
Th made up of two or more polygons th t are not
ft l that t
physically connected constructed using a common
identifier


29

8/20/2010

Construction of complex spatial objects


Integrity checks
Graphical data must be topologically cleaned
before using them in GIS applications
Must not contain any topological errors
During topology building, the computer detects
topological errors and marks them immediately
Operator can then decide what to do with the
hat ith
errors


30

8/20/2010

Integrity checks


Triangulated Irregular Network (TIN) Data
Model
Creates and represents surfaces in GIS
Surface i
S f is represented as contiguous, non-overlapping t i l
t d ti l i triangles
Size of triangles may be adjusted to reflect the relief of the surface being
modelled
Larger triangles for relatively flat terrain
Smaller triangles for rugged terrain
Manages information about the nodes that comprise each triangle and the
neighbors of each triangle
Applications:
Volumetric calculations for road design
Drainage studies
Landslide studies

31

8/20/2010

Triangulated Irregular Network (TIN) Data
Model


Advantages of TIN Data Model
Incorporates original sample points so the accuracy
of the model can be checked
Efficient way of storing surface representations
because the size of triangles can be varied according
to the terrain
Data t t
D t structure makes it easy t calculate elevation,
k to l l t l ti
slope, aspect, and line-of-sight between points


32

8/20/2010

Vector File Formats
GDF/DIME (Geographic Base File/Dual Independent Map Encoding) – originally used
for storing street maps
TIGER (Topologically Integrated Geographic Encoding and Referencing System) –
improvement for GDF/DIME
DLG (Digital Line Graphs) – USGS topographic maps
AutoCAD DXF (Data Exchange Format) – widely used as an export format in many GIS
IGDS (Intergraph Design System) – widely used in mapping
ArcInfo Coverage – stores vector graphical data using topological structure explicitly
defining spatial relationships
ArcInfo E00 – export format of ArcInfo
p
Shapefiles – format of ArcView GIS; defines geometry and attribute of geographically
referenced objects using the main file, index file and database table
CGM (Computer Graphics Metafile) – ISO standard for vector data format; widely used
in PC-based computer graphics applications
Page Description Languages (PDL’s)

Considerations in choosing between
Raster and Vector

source and type of data
analytical procedures to be used
intended use of data


33

8/20/2010

Raster vs. Vector
Raster Data Model
Advantages:
Simple data structure
Easy to conceptualize space representation
Easy data collection and processing
Easy and efficient overlaying
High spatial variability is efficiently represented
Easier and efficient for surface modelling, wherein individual
features are not that important
Allows easy integration of i
All i i f image d ( lli remotely-sensed,
data (satellite, l d
etc.)
Pixel values can be modified individually or in a group by using a
palette
Simple for own programming

Raster vs. Vector
Raster Data Model
Disadvantages:
Do not provide precise locational and area computation
information due to grid cells
Requires large storage capacity, based on size and number
of colors
No topological processing
Limited tt ib t data handling
Li it d attribute d t h dli
Inefficient projection transformations
Loss of information when using large cells
“Blocky” appearance when image is viewed in detail

34

8/20/2010

Raster vs. Vector
Vector Data Model
Advantages:
Far more efficient in storage than grids Compact data
structure
Provides precise locational information of points
Can represent point, line, and area features very accurately,
leading to more accurate measurements and map outputs
Topological models enable many types of analysis
Sophisticated attribute data handling
Efficient for network analysis
Efficient projection transformation
Work well with pen and light-plotting devices and tablet digitizers


Raster vs. Vector
Vector Data Model

Disadvantages:
Complex data structure
Difficult overlay operations
High spatial variability is inefficiently represented
Not compatible with RS imagery
Expensive data collection
Not good at continuous coverages or plotters that fill
areas

35

8/20/2010

Representation of a line using raster
and vector model

Raster representation Vector representation


Raster vs. Vector
Raster Vector
Data collection Rapid Slow
Data volume Large Small
Graphic treatment Average Good
Data structure Simple Complex
Area analysis Good Average
Network analysis Poor Good
Generalization Simple Complex


36

8/20/2010

Integrating raster and vector data


37

GIS - Lecture 4

More Related Content

What's hot (20)

Viewers also liked (13)

Similar to GIS - Lecture 4 (20)

More from sorbi (9)

Recently uploaded (20)

GIS - Lecture 4