Disproportionate Exposure to Urban Heat Islands

Introduction

By 2050 we will see an addition 2.5 billion people living in urban environments (United Nations, 2015).

Global temperatures are rising - current models predict increases of up to 5.7°C for the Pacific Northwest (Mote & Salathé Jr., 2010).

Worldwide economic damages from just 1°C rise in global temperatures could be as much as 448 billion USD (Tol, 2002).

Climate change is expected to reduce global food production by ~2.3% by the 2050s due to stress on agricultural systems (Calzadilla et al., 2013).

Urban Heat Islands (UHI)

• Date back to the early 1800's (Howard, 1820)
• Intra-urban temperatures
• High spatial variability

Satellite-based UHI Descriptions

Satellite-based UHI Descriptions

Cons:

• Poor Resolution (Sobrino, Oltra-Carrió, Sòria, Bianchi, & Paganini, 2012)
• Documented Inaccuracies (Song, Park, Song, & Park, 2014)

Vehicular Traverse-based UHI Descriptions

Dates back to before satellites.

Concept: many in situ measurements. Today, computers can model relationships.

Data

Health and Heat

• Hidden Killer, #1 natural hazard in post-industrial societies (Poumadère, Mays, Le Mer, & Blong, 2005)
• Intra-urban temperatures
• High spatial variability

Other studies have shown that marginalized populations tend to experience more exposure to temperature (Huang, Zhou, & Cadenasso, 2011).

Are people experiencing the same temperatures in Portland, Oregon?

Are marginalized and lower socioecnomic status populations experiencing higher temperatures in Portland, Oregon?

Background

Vulnerability: Conceptual Framework

Exposure

Direct contact to stressor by individual or population/subpopulation.

ie live/work in degraded environments

Sensitivity

Intrinsic: fixed characteristic

ie genetics, age

Acquired: adopted characteristic

ie tobacco/alcohol use, exercise

Adaptive Capacity

Ability to cope with - or understand the presence of - degraded environments.

ie income, race, individual sociality, social cohesion

Assessing Vulnerability

Why look for this area betwee adaptive capacity and exposure?

Morbidity, Mortality, and Disproportionate Exposure

Environmental Justice

Hypothesis

People with low adaptive capacity are experiencing hotter urban temperatures than their high adaptive capacity counterparts.

Measured through:
1. Socioeconomic status (SES)
2. Affordable Housing Placement

Methods: Data

Study Area

Portland, Oregon

Datasets

Urabn Heat Island Model

• Source: Sustaining Urban Places Research Lab
• Traverse-based
• Coverage: City of Portland
• Resolution: 1m
• Time: 7pm (chosen over 6am and 3pm)

Urabn Heat Island Model

Census Geometries

• Source: United States Census Bureau
• Block Groups
• American Community Survey
• Embeded sociodemographics

Census Geometries

Combining Census and Temperature Datasets

Zonal Statistics

• Calculates statistics for raster cells that fall within a single or multiple polygons
• Standard practice, even for UHI work (Huang, Zhou, & Cadenasso, 2011).

Combining Census and Temperature Datasets

Affordable Housing

• Source: Oregon Metro Regional Land Information System (RLIS)
• HUD data, Section 8, etc.

Affordable Housing

All Buildings

• Source: Oregon Metro Regional Land Information System (RLIS)
• Building-level vector data
• Intrinsic use data (residential, industrial, etc.)

City Boundaries

• Source: Oregon Metro Regional Land Information System (RLIS)
• Cities in the region
• Extracted the City of Portland

Methods: Analyses

Income

Income: Geographic Distribution

Income: Hypothesis

$H_{0}$: $\mu_{Low \space Income} - \mu_{High \space Income} = 0$

$H_{1}$: $\mu_{Low \space Income} - \mu_{High \space Income} > 0$

$\alpha$ = $0.05$

Where:
$\mu_{Low \space Income}$ = Mean temperature of low income block groups
$\mu_{High \space Income}$ = Mean temperature of high income block groups

Populations of Color

Populations of Color: Geographic Distribution

Populations of Color: Hypothesis

$H_{0}$: $\mu_{Non \space White} - \mu_{White} = 0$

$H_{1}$: $\mu_{Non \space White} - \mu_{White} > 0$

$\alpha$ = $0.05$

Where:
$\mu_{Non \space White}$ = Mean temperature of block groups with large population of color
$\mu_{White}$ = Mean temperature of block groups with small population of color

Education

Education: Geographic Distribution

Education: Hypothesis

$H_{0}$: $\mu_{Low \space Edu} - \mu_{High \space Edu} = 0$

$H_{1}$: $\mu_{Low \space Edu} - \mu_{High \space Edu} > 0$

$\alpha$ = $0.05$

Where:
$\mu_{Low \space Edu}$ = Mean temperature of block groups with larger population with less education
$\mu_{High \space Edu}$ = Mean temperature of block groups with small population with less education

Isolated Elderly

Isolated Elderly: Geographic Distribution

Isolated Elderly: Hypothesis

$H_{0}$: $\mu_{Isolated \space Eld.} - \mu_{Accompanied \space Eld.} = 0$

$H_{1}$: $\mu_{Isolated \space Eld.} - \mu_{Accompanied \space Eld.} > 0$

$\alpha$ = $0.05$

Where:
$\mu_{Isolated \space Eld.}$ = Mean temperature of block groups with large population of isolated older adults
$\mu_{Accompanied \space Eld.}$ = Mean temperature of block groups with small population of isolated elderly

Poor English Skills

Poor English Skills: Geographic Distribution

Poor English Skills: Hypothesis

$H_{0}$: $\mu_{Poor \space Eng.} - \mu_{Good \space Eng.} = 0$

$H_{1}$: $\mu_{Poor \space Eng.} - \mu_{Good \space Eng.} > 0$

$\alpha$ = $0.05$

Where:
$\mu_{Poor \space Eng.}$ = Mean temperature of block groups with large population with poor english skills.
$\mu_{Good \space Eng.}$ = Mean temperature of block groups with small population with poor english skills.

Building Level Analysis:
Affordable Housing

Affordable Housing: Hypothesis

$H_{0}$: $\mu_{AH} - \mu_{RH} = 0$

$H_{1}$: $\mu_{AH} - \mu_{RH} > 0$

$\alpha$ = $0.05$

Where:
$\mu_{AH}$ = Mean temperature within 100m of Affordable Housing
$\mu_{RH}$ = Mean temperature within 100m of Regular Housing

Affordable Housing: Workflow

Affordable Housing: Buffering

Results

Bifurcation of Continuous Variables

Normal Mixture Modelling (mclust in R)

Income

Income: Student's t-test Results

Result: Reject the Null Hypothesis

Race

Race: Student's t-test Results

Result: Reject the Null Hypothesis

Education

Education: Student's t-test Results

Result: Reject the Null Hypothesis

Isolated Elderly

Isolated Elderly: Student's t-test Results

Result: FAIL to Reject the Null Hypothesis

Poor English Skills

Poor English Skills: Student's t-test Results

Result: Reject the Null Hypothesis

Building Level Analysis:
Affordable Housing

Building Level Analysis: Student's t-test Results

Result: Reject the Null Hypothesis

Conclusions

Lower Adaptive Capacity... Higher Exposure!

(Except for Isolated Elderly)

Is this vulnerability?

Not quite: what about sensitivity?

Sensitivity

Such as:
• Age (old, young) - easy to get
• Obesity - hard to get (BMI from DMV... accurate?)
• Health Records - VERY hard to get (confidentiality)

Policy implications

• Fundamentals of environmental justice are being violated, however heat is often forgotten
• Targeting communities during heatwaves:
  • Cooling Centers
  • Movie Tickets
  • Door hangers: lack of heat education
• Mitigation through green infrastructure: urban trees

Urban Trees

• Major factor in heat
• Other benfits:
  • Reduction of PM2.5 (Nowak, Hirabayashi, Bodine, & Hoehn, 2013)
  • Reduction of PM10 (Bealey et al., 2007)
  • Reduction of NO2 (Rao, George, Rosenstiel, Shandas, & Dinno, 2014)
• Strongly associated with higher income in Portland, Oregon

Future Research:

• Change clustering
• Non-Linear Modeling
  • CART / MLR Hybrid (Hart & Sailor, 2008)

References