Raster vs Vector Data models and structures

Question 1

What is a data model?

Answer 1

Determining how to represent real world objects and phenomena
within a database requires data modeling
A data model is a set of constructs for describing and representing
selected aspects of the real world in a computer
Representation of geographic features in digital format requires:
◦ abstracting geographic objects
◦ converting all data relationships seen on a map into a series of numbers

Question 2

What is a Field Model (raster)?

Answer 2

◦ Models data represented as continuous spatial trends (elevation,
temperature)
◦ Often associated with data collected by remote sensing platforms
◦ Requires
◦ A spatial framework to specify the boundaries of the tessellations (or grid). Latitude/Longitude, UTM,
other irregular boundaries
◦ An attribute domain
◦ Field operations – local, focal, zonal

Raster
◦ Area is covered by grid with (usually square) equal-sized cells
◦ Origin point in the upper left of the grid
◦ Location of each cell calculated from origin of grid
◦ Attributes are recorded by assigning each cell a single value (binary, integer,
floating point) based on the attributes in the cell, such as crop type.
◦ Multiple attributes can be associated with a cell
◦ Analysis done through mathematical operations on cells (Local, Focal,
Zonal)
Simple data structure:
◦ directly store each layer as a single table (basically, each is analogous to a “spreadsheet”

Question 3

Why and how is raster data stored? (compression)

Answer 3

raster format has very large storage requirements so data compression techniques are sometimes
used

Run length encoding
◦ code sequences of cells with same attribute value

Chain codes
◦ define region boundary in terms of its origin and a sequence of unit vectors in the cardinal directions (east=0,
north=1, west =2, south=3

Block codes
◦ 2-dimensional run length encoding
◦ use square blocks to tile the area to be represented
◦ store the origin (centre or bottom left) and radius of each square

Quadtree
◦ recursive subdivision into quadrants

Wavelet
◦ Uses a wavelet based compression algorithm (JPEG 2000)

Question 4

What is a Object Model (vector)?

Answer 4

Object Model (Vector):
◦ Often representing networks (road, utility, drainage) or human defined parcels.
◦ Defined by spatial and non-spatial attributes

Geographic features represented either as:
◦ points: single coordinate pairs, and attributes (trees, poles, spot elevations)
◦ lines: a series of ordered coordinate pairs (streams, streets, sewers, gas lines)
◦ areas (polygons): one or more line segments that close to form a polygon (land parcels, soils, counties)

Coordinates may have 2 (x,y), 3 (x,y,z), or 4 (x,y,z, time) dimensions
Geographic entities are encoded as Features. Features of one geometric type (point, line,
polygon) are stored together in a feature class
1

Question 5

What are issues with Object (vector) data models?

Answer 5

Issues with this model
◦ The focus was on the production of drawings and maps, not the
spatial analysis of these objects. The spatial relationships between
objects (Topology) is not recorded in the database. Spatial Analysis
is more difficult, and requires more complex algorithms.
◦ redundancy, especially for polygons
◦ sliver and gap polygons
◦ difficult to detect errors
◦ problems with nested polygons e.g. area
calculation
◦ It is difficult to tag individual objects with attributes
◦ Coordinates may be local, not geographic

Question 6

Raster vs Vector Data

Answer 6

Raster Data, Vector Data
◦ simple data structure ◦ points, lines and polygons with topological relationships
◦ uses rows and columns of grid cells of same
size

◦ origin at upper left ◦ origin at lower left

◦ stores real x,y co-ordinates of the origin and
calculates others as required ◦ stores real x,y co-ordinates for all features

◦ resolution depends on cell size ◦ resolution depend on compilation method and scale of source data

◦ each cell has single attribute ◦ each feature has unique identifier linking it to descriptive attributes

◦ attribute may be identifier which links to
separate attribute table
◦ generally large storage requirements but
can use data compression techniques ◦ generally more compact than raster data

◦ implicit topology ◦ explicit topology

◦ overlays easy ◦ overlays difficult

◦ surface oriented spatial analysis easy ◦ object oriented spatial analysis easy