Social Applications of GIS : Raster Applications

1
Social Applications of GIS Raster Applications
Rachel Ambagtsheer
2
Introduction Outline

• GIS as a Problem Solving Tool
• Raster GIS - Overview
• Some Applications
• Conclusion

3
GIS as a Problem Solving Tool

• The problem solving process
• Identify the problem
• Set aims/objectives
• Inputs and processes (data and analysis)
• Outputs
• Conclusions and review

4
Raster GIS Overview

• Spatial data organised in either raster or vector
  data structure
• Raster structure layers are composed of an
  array or grid of cells
• Each cell represents a part of the real world
• A value is assigned to each cell
• The raster model is well suited to the analysis
  of continuous data but can handle discrete and
  linear features

5
Raster Procedures

• Some of the raster operations procedures used
  by applications in this presentation
• Distance operations
• Buffers
• Recoding of cells
• Interpolation
• Overlay (esp. in modeling) statistical
  analysis
• Identify visible area
• Identification of zones etc

6
Applications Overview

• Range from simple to complex problems
• Cover a range of topics and fields BUT
• Universal approach
• Applications overview

7
Raster Social Applications of GIS

• Modeling
• Network Analysis
• Site Selection
• Visualisation and Animation
• Hot Spot Analysis
• Probability surfaces

8
Raster Social Applications of GIS

• Identify the site/s with the highest potential
  for a particular purpose

9
Raster Social Applications of GIS

• Archaeological Modeling
• Waste Site Identification
• Retail Site Selection Targeting
• Criminal Profiling
• Recreation Planning Applications

10
Archaeological Modeling
11
Archaeological Predictive Modeling

• Research goal
• To understand the long-term (2,000 years)
  interaction of various cultures and the physical
  environment in the Arroux River Valley region of
  Burgundy, France

Project led by Scott Madry, Informatics
International Uni. Of North Carolina
12
Archaeological Predictive Modeling

• Why GIS?
• The ability of GIS to integrate data from a
  variety of sources and to model a variety of
  scenarios were considered essential functions for
  the purposes of the project
• Inputs
• Elevation
• Aspect and Slope
• SPOT images
• Land Use/ Land Cover
• Celtic Hill Forts

• Geology
• Faults
• Hydrology
• Ancient and Modern Roads
• Survey transects

13
Archaeological Predictive Modeling

• Processes Deriving Aspect and Slope
• A number of processes were run on the data sets
• Initially, a DEM (Digital Elevation Model) was
  created by manually digitising contours from a
  paper map and then interpolating these into a 20
  meter raster array
• Aspect and slope were then derived from the DEM

elevation
slope
aspect
14
Archaeological Predictive Modeling

• Processes Deriving Distance Information
• Relatively simple process in which basic data
  layers were buffered at certain distances
• This process creates new data input layers,
  shown below

distance to ancient roads
distance to water bodies
distance to hill forts
15
Archaeological Predictive Modeling

• Line of Sight Analysis
• Determines what is visible from any given
  location in this case the 4 corners of each
  Celtic hill fort in the area
• The 4 line of sight maps for each hill fort were
  combined to give a map of complete
  intervisibility from each hill fort
• These were then combined to give the total area
  within sight of the hill forts old roads
  correspond

16
Archaeological Predictive Modeling

• Site Location Modeling
• Model where archaeological sites might be based
  given environmental cultural data in the GIS
• Statistical analyses were run on existing sites
  in the defined study area combined with various
  layers in the GIS to look for patterns
• The region of analysis was then expanded further
  and new maps showing areas with the highest
  probability of site locations were created
• Found that areas with high probability of new
  sites corresponded with areas threatened by the
  gravel mining industry
• Currently using aerial photography and site
  surveys to investigate areas of high potential

17
Archaeological Predictive Modeling
Results of the predictive model
For more information visit www.informatics.org/fr
ance/gis.html
18
Site reference http//www.cast.uark.edu/kkvamme/
mnmodel/mnmodel.htm
Where P primary dataset R reclassification G
gradient operation D distance operation B
intersection C cover operation (drape known
sites over result)
19
Predicted probability of pa sites (fortified
Maori camp or village) Leathwick 2000
Site reference http//www.nzarchaeology.org/elecp
ublications/predictive.htm
20
Waste Site Identification
21
Model inputs include Geology, Population
Density, Conservation Areas, Coastal Regions,
Local Road Access, Local Rail Access Overall
Accessibility
22
(No Transcript)
23
End result Model based on the factors and
weightings assigned by the user displayed
immediately Users can submit their model,
personal details and comments online
Site reference www.ccg.leeds.ac.uk/mce/mce-urls.h
tm
24
Retail Site Selection Targeting
25
Site reference www.dsslink.com/app1098.htm
26
Site reference http//www.directionsmag.com/mapga
llery/
27
(No Transcript)
28
Criminal Profiling
29
Jeopardy surface of insurance agency robberies,
Vancouver
Site reference http//www.directionsmag.com/mapga
llery/
30
Recreation Planning Applications
31
Recreation Planning Applications

• Research goal
• To develop a GIS model to identify suitable sites
  for the placement of a nature park in Fairfax
  County, Virginia

Project conducted by Peter LaPlaca of TASC (The
Analytic Sciences Corporation) Virginia
Source Fairfax County Park Authority
32
Recreation Planning Applications

• Context
• Fairfax County is experiencing rapid population
  growth due to its location near Washington DC.
  Open space for recreational activities will
  become increasingly in demand as the population
  increases. A method of modelling potential
  suitable sites for nature parks was thus
  required.
• Once again, the integrative and modelling
  capabilities of the raster GIS model were
  considered of primary importance to the
  achievement of the research goal.

33
Recreation Planning Applications

• Inputs
• A survey of nature park users and other
  stakeholders was conducted in order to identify
  important factors in the siting of the park. 13
  factors were identified as important inputs to
  the system.
• Many of the input datasets were vector-based but
  converted to raster format for the analysis.
• The preparation and manipulation of the input
  datasets illustrate some of the common functions
  undertaken in a raster GIS analysis.

34
Recreation Planning Applications

• Input Railroads
• Due to noise safety, the park should not be
  within one mile of a railroad. The grid was
  reclassified with a NO DATA value for areas
  within 1 mile. The rest was reclassified a 10
  (maximum value).
• Input Solid Waste
• Park should be at least 2 miles from a solid
  waste site. The grid was reclassified with a NO
  DATA value for areas within 2 miles. The rest
  was reclassified as a 10.

35
Recreation Planning Applications

• Input Existing Parks
• Most people agreed that they did not want a new
  park too close to an existing one. The parks were
  reclassified to NO DATA, along with a half mile
  area around them. The remaining cells were
  allocated a value of 10.
• Input Police
• Grid cells were classified according to their
  distance from a police station, with 10 1 mile
  from a station, 9 2 miles from a station a
  value of 6 given to the remaining cells

36
Recreation Planning Applications

• Input Historic Sites (Protected)
• There were 4 protected sites in the research
  area. It was desired that new park sites be
  located at some distance from these. These were
  buffered according to significance. Areas within
  1-2 miles of each site were given a NO DATA
  value, with the remaining area coded a 10.
• Input Historic Sites (Established)
• It was considered important to site a new park
  near an established historic site. Areas 1 and 2
  miles from these sites were classified 10 9,
  with the remaining cells coded at 8.

37
Recreation Planning Applications

• Input Industry
• All industrial areas were classified as NO
  DATA. Everything else was reclassified to 10.
• Input Street Centre Line
• The street grid was buffered out to 500 feet to
  account for housing. These cells were coded NO
  DATA. The remaining cells were coded to 10.
• Input Airports Government Facilities
• Airports, Government Facilities and the Prison
  (!!) were coded NO DATA. Quote One does not
  want to build parks here. Everything else got
  coded a 10.

38
Recreation Planning Applications

• Input Public Facilities
• Distance from all public facilities (schools,
  shopping centres etc) calculated. Reclassified
  NO DATA out to 300 feet, everything else coded
  a 10.
• Input Universities
• Areas 1 mile from university sites were coded
  10, 5 miles were coded 9 and all other cells were
  coded 7.
• Input Slope
• A slope grid was calculated from a DEM and
  reclassified thus Slope 1 9, gt1-2 10, 3 7,
  gt3-8 4, and gt8 1

39
Recreation Planning Applications

• Input Land Use/Land Cover
• Residential 7
• Industry 1
• Transport Communications 1
• Other Urban 1
• Croplands and Pasture 8
• Other Agricultural Land 7
• Deciduous Forest 10
• Evergreen Forest 10
• Mixed Forest 10

• Streams Canals 8
• Lakes 8
• Reservoirs 8
• Forested Wetland 8
• Non-forested wetland 6
• Strip Mines 1
• Transitional Areas 1

40
Recreation Planning Applications

• Final Park Site Cost Grid
• After the individual grids were created for each
  of the 13 layers, a final grid was created by
  summing the 13 values for each cell. The final
  score was divided by 13 in order to get a final
  value scaled from 0 to 10 in terms of
  suitability.
• There are many possible variations on this
  process, both for this and similar applications.
  Improvements at all stages of the modeling
  process could be implemented

For more information, check out the ESRI 1997
User Conference Proceedings, ESRI web site
41
Recreation Planning Applications
Source LaPlaca 1997
42
Conclusion

• Comments
• Raster applications show a wide range of
  functionality but there is still a lot of
  potential for development
• There are clearly going to be situations when it
  is more appropriate to use a vector based model
• For some applications, use of both data models
  is the best solution developments still under
  way on this
• Once again, it is a case of clearly defining the
  aims of the project, assembling the inputs and
  choosing the most suitable methodology