Warming Up the City: Mapping the hottest (literally) neighborhoods of Portland


The profound recognition that natural disasters are only disasters when people are affected brings with it a need to know who might be vulnerable to the next episodic event. Recent evidence suggests that the shift in the climate system will bring with it significant increases in the magnitude, frequency, and duration of climate-induced events including heat waves, droughts, floods, and fires. In fact, an emerging research area called attribution science (science that seeks to understand the mechanisms behind climate change) suggests that these shifts are underway and responsible for many of the natural disasters we see in the news every day.

While the recent results of the Paris climate talks (COP21) offer a glimmer of hope for addressing the imminent natural disasters that confront our cities, identifying those who are most vulnerable requires unprecedented collaboration among multiple sectors and across academic disciplines.

As elsewhere, many in the Portland region are asking: Who are most at risk from climate change events here? One aspect of that question focuses on the extent to which a population is exposed to a specific climate hazard—in other words, it’s a geographic question; another looks at those who may be most sensitive to and the least prepared for these hazards.

With support from the Institute for Sustainable Solutions at Portland State University and the Bullitt Foundation, a group of researchers and practitioners started a project in September 2014 to assess the impacts of a changing climate on neighborhoods in the city of Portland. Among the climateinduced natural disasters we examined was the urban heat island which is created when several landscape features, such as blacktop and buildings, absorb heat throughout the day making some neighborhoods much hotter than others.

Annually in the United States, heat waves kill more people than all other natural disasters combined; thus, an understanding of the factors that create local vulnerabilities is of paramount importance. To that end, in this edition of the Periodic Atlas we offer our analysis, which identifies those neighborhoods that bear a disproportionate level of heat stress during the hottest days of summer. We also identify areas that have the greatest social vulnerabilities because they have populations who may be most sensitive to excess heat and who may have the least capacity to cope.


During the summers of 2014 and 2015, a group of students, faculty, and other volunteers conducted a series of field campaigns to understand the differences in temperature throughout the region. They used GPS units and temperature gages, which were mounted on the passenger side of an automobile. The field teams drove for one hour (to minimize any changes in temperature) during three periods: morning (6:00 a.m.), afternoon (3:00 p.m.), and evening (7:00 p.m.). By statistically analyzing the data, we learned that areas within the city of Portland vary by upwards of 15 degrees Fahrenheit (figure 1).

But the story doesn’t end there. In fact, the afternoon is different than the morning (figure 2) and evening temperatures (figure 3). Figure 4 depicts the hottest locations in the city throughout the day.

previous arrow
next arrow
previous arrownext arrow

Isolating Vulnerable Populations

We looked to a group of expert practitioners and the literature to identify a set of 10 social factors that determine vulnerability to urban heat waves. Some of those factors include age, body mass index (BMI), the presence of air conditioning, and the extent of the tree canopy. These factors and the urban heat maps help us to identify potential areas of vulnerability (figures 5-8).

For example, a comparison of figures 4 and 5, indicates hot spots throughout the day just east of I205 and near I84 in areas with large populations of the elderly (figure 5). People in this age group often experience a disproportionate impact from urban heat; thus, this concentration of heat and the elderly suggests potential vulnerability. 

Those with pre-existing health conditions may also be more sensitive to excessive heat. While BMI is an imperfect measure, it is often used as a proxy for the existence of preexisting health conditions. Figure 6 shows large areas of the city where the average BMI is higher than normal and a few areas in North Portland and East Portland where it is much higher than normal. There is considerable overlap between these areas and locations where urban heat is concentrated. Figure 7 indicates the distribution of urban heat in the evening when temperatures tend to be warmer and most people are home. One mitigating factor is the availability of air conditioning. Figure 8 indicates the percentage of homes with central air conditioning.

Another important but sometimes overlooked mitigating factor is shading from vegetation, especially large canopy trees. As figure 9 illustrates, some neighborhoods in the city have a lot more large canopy trees than others. The lack of significant tree canopy is especially apparent on the east side of the Willamette where better canopy is present in neighborhood parks and some older, more affluent inner eastside neighborhoods such as Eastmoreland and Laurelhurst.

Potentially-deadly urban heatwaves are but one of many climate-induced stressors that will become increasingly important as we consider how the region changes in the future. Taking actions to reduce the impacts of such natural disasters will be in our collective interest because ultimately, they affect all of us. In the next phase of the project, which we expect to begin in early 2016, we will evaluate the extent to which specific built environmental interventions—such as expanding the tree canopy, changing the reflectivity of surfaces, and modifying building designs—will help to temper heat stress in specific neighborhoods.

In the meantime, we have developed an online tool (see www.climatecope.org) that houses all of our data and allows researchers, urban planners, public health practitioners, and others participating in urban climate discussions to understand how natural disasters will impact different places and people in the region. We welcome feedback about this and other climate related topics through the www.suprlab.org website.

Jackson Voelkel is a geospatial analysis consultant based in Portland, Oregon. His clients cover a wide range of clients and industries including university research labs, non-profits, private corporations, and all levels of government from city to federal. He works with an equally wide range of tools including advanced machine learning modeling, dynamic webmapping, spatial database design, custom 3D LiDAR analysis, and geospatial process automation. In addition to his consulting work, Jackson is an Adjunct Professor of Geography at Portland State University. He holds a B.S. in geography/GIS and Master’s of Urban Studies, has published his work multiple times in peer-reviewed journals, and has given lectures on his work around the world.

Voters’ Choice 11/14


Voters often ignore Midterm elections;  November 2014 was no exception. Nationwide, voter turnout, at 36.3 percent, was the lowest in 72 years, according to the United States Elections Project.

But in Oregon and Clark County, Washington, a governor’s race, a senate race, a several interesting ballot measures and a change in the Clark County charter upped the turnout to 51 percent for both Oregon statewide and for Clark County, Washington.

 While Metroscape readers prefer to view the region as integrated and interdependent, elections is one area where we’re challenged to take a regional view. Voters in each jurisdiction receive different ballots, consider different candidates, ponder different ballot measures, and even face different voting procedures depending on which side of the Columbia River they inhabit.

This version of the Atlas considers the results of the November 2014 elections for some key races and initiatives in Oregon, and separately considers some of the important issues faced by voters in Clark, County, Washington. The maps show precinct–level results; the circles within each precinct show the distribution of the vote within that precinct.

Oregon Statewide Elected Offices

Figure 2

Statewide, Senator Jeff Merkley secured 55.7 percent of the vote while Governor Kitzhaber captured 50.2 percent, according to the Oregon Secretary of State. This relatively small difference is visible in the metropolitan voting patterns shown in figures 1 and 2. As expected, areas voting most strongly Democratic are in the inner southeast and northeast Portland—neighborhoods that have become more reliably Democratic over the past decade. Areas farther east and west demonstrate a more conservative voting pattern.

This pattern is consistent with recent political science research that suggests that suburbs have become important battlegrounds in which elections are won and lost. With rural areas largely conservative and cities overwhelmingly liberal, the suburban voters have become the deciders in many elections.

In looking at Oregon’s recent senatorial and gubernatorial races, Senator Merkley had stronger support in  suburban counties than did Kitzhaber. And while Monica Wehby failed to attract a majority of voters in any of the metropolitan counties, Dennis Richardson’s bid for Governor did draw a majority in Yamhill and an even 50 percent in Clackamas and Columbia counties.

 Measure 88: Oregon Alternative Driver Licenses Referendum

Figure 3

Measure 88 would have offered residents an opportunity to obtain a driver’s card without requiring proof of legal residence in the United States. Statewide the measure failed with 66 percent voting “no.” Figure 3 indicates strong support in the liberal central precincts. Interestingly, there appears to be strong opposition in places such as Gresham, Cornelius, and Forest Grove that have relatively large Hispanic populations. Of course, many of them cannot vote.

Measure 90: Oregon Open Primary Initiative

Figure 4

Had it passed, Measure 90 would have created a top-two system of general election voting where all voters receive the same primary ballot that includes all candidates, regardless of political party. Candidates would have been identified on these ballots with their party registration and whether or not they have their party’s endorsement. The top two candidates, regardless of political party, would run in the general election.

Measure 90 lost with 68 percent of voters statewide voting “no.” The vote was strongly negative throughout the metropolitan area, demonstrating that overall many people questioned the need for this type of change in our voting system. Interestingly, the greatest support for the measure appears to be in Portland’s affluent West Hills.

Measure 91: Oregon Legalized Marijuana Initiative

Figure 5

Measure 91 legalizes recreational marijuana for people aged 21 and older. The measure allows adults to possess up to eight ounces of dried marijuana and up to four plants. It also authorizes in-state manufacturing, processing, and sale by and to adults. It gives the Oregon Liquor Control Commission the responsibility of regulating sales and it  provides for the taxation of producers. Proceeds are to be distributed to schools, mental health and drug treatment services, state police, local law environment, and the Oregon Health Authority.

Measure 91 was approved with 56 percent of the vote statewide and passed overwhelmingly in the central city precincts on both sides of the Willamette. Opposition was more significant in the outlying, more conservative precincts.

Measure 92, Oregon mandatory labeling of GMOs initiative

Figure 6

Had it passed, Measure 92 would have mandated labeling of foods that are produced with or contain genetically modified organisms (GMOs).

The measure was very narrowly defeated in the initial count by 802 votes, triggering an automatic recount. The recount showed the measure defeated by 837 votes—still an incredibly close contest.

The results don’t as closely follow the voting pattern typical of measures considered “liberal.”  This may be because some traditionally liberal voters could have been concerned about the potential impacts of pricing for lower income people.

Metro Charter Measure 26-160

Figure 7

This measure retains a provision in the Metro charter prohibiting it from requiring local governments to increase density in single-family neighborhoods.

The measure passed overwhelmingly even in the region’s most liberal precincts. The 94 percent “yes” vote seems to reflect voters’ general preference for local control regarding land use issues.

Clark County

Proposition No. 1 Home Rule Charter

Figure 8

The new Clark County charter replaces the 3-member Board of Commissioners with a 5-member Council.  It transfers supervisory authority over County employees, except employees of other elected officials, to an appointed County manager. The Council will be elected by district, except the Chair, who will be elected at large. The charter also adds provisions for initiative and referendum. The charter passed with 53 percent of the vote and had bipartisan support.

Clark County Commissioner District 3

Figure 9

Republican, Jeanne Stewart, was elected over Democrat, Craig Pridemore, with just over 50 percent of the vote—a difference of only 905 votes. Not surprisingly, Pridemore had strong support in the precincts within the City of Vancouver, while Stewart had stronger support in the eastern and northern, more rural, areas of Clark County.

Toll-Free Bridge Advisory Vote

Figure 10

This advisory vote asked voters whether the Clark County Commission should support and pursue a toll-free East County bridge across the Columbia River from SR14 at 192nd to Airport Way in Oregon. It passed with nearly 53 percent of the vote.

These results represent a four-point drop from the 2012 East County bridge advisory vote. The vote tended to be more strongly “no” in some Vancouver precincts and more strongly “yes” in some Washougal precincts, rural north Clark County, and in Battle Ground.

Initiative 594 Universal Background Checks for Gun Purchases

Figure 11

Measure 594 extends background checks, currently required for firearm sales by licensed dealers, to all firearm sales and transfers where at least one party in the sale is in Washington. The checks would be required at gun shows, for online transactions, and between unlicensed private individuals. The requirement includes gifts or loans but excludes gifts among family members and some other specific exceptions.

The measure passed statewide with 59 percent of the vote, and in Clark County with 58 percent of the vote. While it would be expected that voters in urbanized precincts would generally support the measure, the “yes” votes northwest of Vancouver, in Ridgefield and La Center were less predictable.

Sheila Martin is Director of the Institute of Portland Metropolitan Studies and the Population Research Center at Portland State University. She directs the Institute’s community-oriented research and service activities.

Meg Merrick, coordinated the Greater Portland Pulse regional indicator project as well as the Community Geography project (renamed Neighborhood Pulse) of the Institute of Portland Metropolitan Studies at Portland State University (PSU).  

Richard A. Clucas is professor of Political Science, and Executive Director of the Western Political Science Association.

Carolyn Long is an Associate Professor of the School of Politics, Philosophy and Public Affairs at WSU Vancouver.