Built Environment

1
Built Environment Spatial Analysis

• Pavlos S. Kanaroglou
• Centre for Spatial Analysis
• McMaster University
• Pavlos_at_McMaster.ca

2
Objectives

• Provide an idea of the type of work we do in CSpA
• Enhance data collection efforts in PURE by
  providing a spatial context to the collected data

3
Spatial Analysis by Example

• Spatial Analysis Examines Processes and Patterns
  over the Surface of the Earth
• It Does So Through
• Spatial Statistics
• Location Theory
• It Uses Technology, such as
• Geographical Information Systems (GIS)
• Areal Photography (Air Photos)
• Remote Sensing (Satellite Images)

4
Spatial Statistics

• It is a set of statistical tools that allow to
  visualize, explore and model data that consist of
  observations over space
• Visualization uses techniques from cartography
  and is used to display data on maps
• Exploration and modeling correspond to
  descriptive and inferential statistics
• Use of ordinary, as opposed to spatial,
  statistics may lead to wrong inferences
• Here is an example

5
(No Transcript)
6
LISA MAP OF SMR for Myocardial Infarction
Morans 0.5724 using Rooks case for contiguity
matrix
7
Ordinary Least Squares (OLS) Model

• Y Xß e
• MYO.Trans 1.72 0.62(JAR.Trans) e

8
OLS Output
Dependent Variable MYO.TRANS Number of
Observations 188 Mean dependent var
4.57223 Number of Variables 2 S.D.
dependent var 0.2731 Degrees of
Freedom 186 R-squared
0.191731 F-statistic 44.1213 Adjusted
R-squared 0.187385 Prob(F-statistic) p
lt 0.001 Sum squared residual 11.33333 Log
likelihood -2.74332 Sigma-square
0.060932 Akaike info criterion 9.48664
S.E. of regression 0.246844 Schwarz
criterion 15.9595 Sigma-square ML
0.060283 S.E of regression ML 0.245527
-------------------------------------------------
--------------------------------------------------
--------------- Variable Coefficient
Std.Error t-Statistic Probability
-------------------------------------------------
--------------------------------------------------
--------------- CONSTANT
1.723711 0.4292173 4.01594 0.0000858
JAR.TRANS 0.620867 0.0934705 6.64239
0.0000000 ------------------------------------
--------------------------------------------------
---------------------------- MI/DF VALUE
PROB Moran's I (error)
0.448464 9.9815416 lt 0.0001
9

• Map of residuals from OLS Model

10
Simultaneous AutoRegressive (SAR) Model

• Y ?WY Xß e
• MYO.Trans 0.71(W)(MYO.Trans) 0.28(JAR.Trans)
  e

11
SAR Output
Dependent Variable MYO.TRANS Number of
Observations 188 Mean dependent var
4.572231 Number of Variables 2 S.D. dependent
var 0.273100 Degree of Freedom
186 Lag coeff. (Lambda) 0.708434
R-squared 0.539406 R-squared (BUSE) - Sq.
Correlation - Log
likelihood 37.856236 Sigma-square 0.034353
Akaike info criterion -71.7125 S.E of
regression 0.185345 Schwarz
criterion -65.239588 ----------------------
--------------------------------------------------
----------------------------------------
Variable Coefficient Std.Error z-value Probab
ility -------------------------------------------
--------------------------------------------------
------------------- CONSTANT 3.267428
0.3835552 8.518794 0.0000000
JAR.TRANS 0.284712 0.0825147
3.450446 0.0005598 rho 0.708434
0.0601201 11.78364 0.0000000 ---------
--------------------------------------------------
--------------------------------------------------
---
TEST DF
VALUE PROB Likelihood Ratio Test (rho)
1 81.19911 0.0000000
12

• Map Of Residuals from SAR Model

13
Location Theory

• Development of theories that explain the spatial
  arrangements of economic activities
• In a study area identify the best (optimal) spots
  for the location of facilities
• Here is an example where optimal location
  planning comes handy

14
Location Theory

• Study commissioned by Health Canada and funded by
  Health Canada and the Canadian Institutes of
  Health Research (CIHR)
• Objectives
• To derive optimal locations for Ogawa NO2
  monitors
• To derive exposure assessments using
  geostatistical and land use regression techniques
• To test associations between NO2 (marker for
  traffic pollution) and health outcomes
• We made use of an approach known as
  location-allocation

15
Southern Ontario Study Area
16
Ogawa Passive Monitors
17
Monitoring location allocations

• Based on two criteria
• Monitoring should be more intensive in areas of
• High pollution variability
• High density of vulnerable populations

18
(No Transcript)
19
(No Transcript)
20
Toronto Land Use Regression Pollution Surface
21
(No Transcript)
22
Transport and Land-Use

• Over the last few years a large portion of
  substantive research in CSpA has focused on
  transport and land-use studies in urban areas
• Development of urban simulation models, known as
  Integrated Transport and Land-Use Models (ITLUM)
• Such models can be used as decision support tools
  by urban planners
• Instrumental for land-use or transportation
  infrastructure development decisions

23
Transport and Land-Use

• For the development of ITLUMs it is important to
  understand how people go about their daily
  activities
• We are interested in understanding the behaviour
  of people in terms of the trips they undertake,
  for what purpose and by what means
• This behaviour is obviously conditional on who
  they are demographically, the built environment
  around them and the available infrastructure
• I will talk very briefly about one of the many
  GIS-based tools we have developed that helps us
  understand such behaviour

24
Activity Analysis Tools
25
Application

• The application is from Portland, Oregon, on the
  basis of data availability
• It is based on the 1994/95 Household Activity and
  Travel Behavior Survey
• It includes a detailed description of the
  activity/travel behavior for 4451 households and
  9471 individuals
• The total number of trips captured was 67891
  trips resulting in 122348 activities

26
Household Space-Time Trajectory
Female, Age 48, Self-Employed Full-time, Service
Sector Male, Age 52, Employed Full-time,
Managerial/Professional
27
Activity Field
UA Status Suburban UA Status Urban
Area 31.1 km2 Area 87.9
km2 Perimeter 26.3 km Perimeter 39.9
km In-home 6 In-home 14 Out-of-home
5 Out-of-home 18 Total 11
Total 32
28
Activity-Space Measures
29
(No Transcript)
30
Data Collection and PURE

• The type of studies I have presented are not
  possible without extensive spatial data
• Such data can be acquired from local governmental
  sources
• Because of the many countries involved in the
  study, data, if available, may be incompatible
• An alternative is to acquire data through areal
  photographs or satellite images
• Using such technology, one may create GIS
  coverages suitable for analysis

31
(No Transcript)
32
(No Transcript)
33
(No Transcript)
34
(No Transcript)
35
(No Transcript)
36
(No Transcript)
37
(No Transcript)
38
(No Transcript)
39
(No Transcript)
40
Data Collection and PURE

• Satellite images can also be used to obtain maps
  of the built environment
• There different types of satellites for different
  purposes
• We interested in Earth Resource Satellites
• Landsat 30m resolution
• SPOT 4m to 15m
• ICONOS (Space Imaging) 1m to 4m
• QUICKBIRD (Digital Globe) Less than 1m

41
Landsat
January 16th, 1973, March 12th, 1989 and
January 6th, 2003
The Mississippi Delta These images shows the
delta that has experienced significant changes in
just the past three decades. The most notable
change is the loss of marsh along the southern
edge of the delta on the left side of each image
42
Landsat
These images, from Landsats 2, 4, and 7, show the
progression of deforestation in Bolivia from 1975
to 2000. This area lies east of Santa Cruz de la
Sierra, Bolivia, in an area of tropical dry
forest. Since the mid-1980s, the resettlement of
people from the Altiplano (the Andean high
plains) and a large agricultural development
effort (the Tierras Baja project) has lead to
this areas deforestation.
43
SPOT Monitoring Urban Growth
44
SPOT Pipeline Leakage
45
(No Transcript)
46
(No Transcript)
47
Quickbird-Digital Globe
48
(No Transcript)
49
(No Transcript)
50
SPOT vs Digital Globe
51
LIDAR LIght Detection And Ranging

• Simple Principles
• Integration of INS GPS with laser
• Uses GPS base stations
• With post-flight processing, the laser range,
  scan angle, GPS data and INS data are combined to
  accurately determine the position of each LIDAR
  return or point

52
Urban Applications
All Points Data Colour Coded by Height
53
Urban Applications
54
Urban Applications
Building Extraction
55
Classification of Buildings, Vegetation, and
Ground
56
Classification of Lines, Towers, Buildings,
Vegetation, and Ground
57
Thank you