Data Models: RasterVector

1
Data Models Raster/Vector

• Raster - tells what occurs everywhere
• at each place in the area
• every cell has a value, even if its missing
• Vector - tells where everything occurs
• gives the location of every object

2
Raster data model

• uses a grid made up of cells in a rectangular
  matrix
• attribute is represented as a single value for
  that cell
• often used for data from
• remote sensing images
• scanned maps
• USGS elevation data
• generally best for continuous features
• elevation
• temperature
• soil type
• land use (perhaps)

3
Raster

• a cell has a resolution, given as the cell size
  in ground units
• e.g., 30m Landsat data has cells 30m by 30m on
  the ground

4
Types of raster analyses

• Display
• e.g., each value in the grid is assigned to a
  color
• can also display the data as a surface, e.g., for
  elevation, and make contour lines
• Local Operations
• a layer is created from one or more input layers
• value of the new cell, or pixel is determined
  by the value of same cell on input layers
  (neighboring/distant cells have no impact)
• Several options for doing this, e.g., z x y,
  z (xy)/2, z min(x,y), etc.

5
Types of raster analyses

• Local Neighborhood Operations
• value of a cell in the new layer is determined by
  the local neighborhood of the cell on the old
  layer
• e.g., slopes - for elevation layer, we can
  compute steepness by looking at the difference
  between a cells value and the value of its
  neighbors
• e.g., filtering - computes new cells value as
  the weighted average of surrounding cells, used
  to enhance detail or smooth layers to show
  general trends

6
Types of raster analyses

• Operations on extended neighbors
• distance - calculate distance of each cell from a
  cell or nearest of several cells
• buffer zones
• visible area/viewshed - for elevation layer,
  find the areas visible from a given point (used
  for site analyses of unsightly facilities,
  transmission facilities, fire towers, etc.)

7
Types of raster analyses

• Operations on zones (groups of cells/pixels)
• e.g., identifying zones - compare adjacent cells
  and identify those with same value
• e.g., areas of zones
• e.g., perimeter of zones
• Operations to describe contents of layers
• one layer mean, median, most common value...
• more than one layer - comparisons of layers
  pattern using statistics

8
Vector data

• Geographic features represented as
• points/dots/nodes, e.g., trees, poles, airports,
  cities, intersections, crash locations
• lines/arcs, e.g., streets, rivers, sewer lines,
  utility lines
• areas/polygons, e.g., land parcels, cities,
  counties, forested areas, soil types
• Lines and areas are built from sequences of
  ordered points
• Which to use (points, lines, areas) depends in
  part on scale, e.g., cities, airports

9
Vector data model

• location referenced by x,y coordinates which can
  be linked to form lines and polygons
• attributes referenced though unique ID numbers to
  tables
• used for data from
• TIGER (US Census geographic data)
• USGS DLGs for roads, streams, etc.
• US Census attribute data
• generally best for features with discrete
  boundaries
• property lines
• political boundaries
• transportation
• Census boundaries

10
Topology

• Topology - description of the relationships
  between geographic features, including adjacency,
  linkage, inclusion, and proximity
• Building topology - calculating and encoding
  these relationships between the points, lines,
  and areas

11
Raster vs. Vector - Which to use

• Depends on nature of data, type of analysis,
  available software

12
Raster

• simplicity of organization
• fast processing speed for buffers, overlays, etc.
  (cells stack up on each other)
• good for continuously changing data
• good for scanned or remotely sensed data
• good for some types of analyses, e.g., models of
  flow of water across a terrain, spread of fire,
  spread of air pollution, etc.
• easy to understand
• may sacrifice detail, especially for lines and
  points
• problem of mixed pixels
• must often include redundant or missing data

13
Vector

• good at representing point, line, and area
  features, e.g., roads, rivers, boundaries
• can represent features very accurately
• more efficient for typical geographic data
• only place points where needed
• not good at continuous coverages, e.g., elevation