Community Resilience to Climate Change: Theory, Research and Practice

155 3.3.1. Exposure to Urban Heat The white and elderly populations have a negative relationship with UHI (−1.515 and −2.114, respectively), which is statistically significant. This means that census blocks with a higher number of white residents and older adults as a percentage of the total census block population are more likely to have lower temperatures during an urban heat event. For example, for every 10% increase in the white population, temperatures are lower by 0.1515 °C on average during a heat event. By contrast, a larger black/African American population, along with NHPI, Hispanics, and young children all have a positive linear relationship with UHI. For example, the coefficient for NHPI is 3.471, meaning that for every 10% increase in NHPI, of the total census block population, temperatures are higher by 0.3471 °C on average. The share of AIAN and Asians do not have a statistically significant relationship. This analysis is based solely on demographic characteristics as they relate to temperature and does not explicitly account for the presence of buildings, trees, or other factors which influence urban heat. However, it may be inferred that those groups experiencing the highest temperatures are located in areas which lack heat-ameliorating infrastructure, or possess built urban features that exacerbate heat [21,22]. 3.3.2. Central Air Conditioning (CAC) Units per Area Only the white population has a significant positive relationship with CAC per area (km2). On the other hand, only young children have a negative linear relationship with CAC, which is statistically significant. For every 10% increase in white population, of the total census block population, CAC units are likely to be higher by 9.6 units per square km, and for every 10% increase in young children, CAC units are likely to be less by 30.5 units on average. The share of black/African American residents, AIAN, Asians, NHPI, Hispanics, and the elderly do not have a statistically significant coefficient. 3.3.3. Public Cooling Centers When assuming the average walking speed, only the black/African American population has a positive relationship with accessibility to public heat refuges. On the other hand, Asians and the elderly have a negative linear relationship with public heat refuges in the city. For every 10% increase in black/African American population of the total census block population, the residents have more access by 0.22 public heat refuges on average, and for every 10% increase in Asians, the residents have less access by 0.20 public heat refuges. Other tested socio-demographic characteristics do not have statistically significant relationships with the accessibility to public heat refuges. 4. DISCUSSION This study examined socio-demographic factors in relation to the distribution of urban heat in an attempt to better understand vulnerability based on (1) disproportionate heat exposure among socio-demographic groups; and (2) disproportionate access to refuge (either public cooling facilities or residential central air conditioning), resulting in heightened or lowered adaptive capacity. Overall, results indicate that populations with low adaptive capacity characteristics also experience high exposure, and that access to refuge is significantly influenced by socio-demographic status. A series of Student’s t-tests were performed to test the hypotheses that the difference between “high” and “low” adaptive capacity groups were significantly greater than 0. The results of the study, with the exception of the variable isolated elderly, allow rejection of the null hypothesis, and indicate that populations with characteristics of low adaptive capacity do experience higher temperatures than those with high adaptive capacity within the study area. Additionally, the analysis showed significantly higher temperatures in the area directly surrounding affordable housing when compared to a random sample of non-affordable (i.e., regular) housing from similar block groups. We focused on isolated elderly specifically because they have historically been disproportionately impacted by heat waves in other parts of the U.S. [39,40], though similar patterns are not statistically significant in the City of Portland. In fact, the observed non- significance of the isolated elderly (percent of the population 65+ years old and who live alone) could be related to the spatial nature of the census block group geographies. A test for spatial autocorrelation conducted using Moran’s I [56] showed that, while census block groups with a high percent of isolated elderly have statistically significant clustering (z-score = 2.921, where 0 is random; p-value = 0.0035), they are far more random in spatial distribution than the variables for extreme poverty (z-score = 6.411; p-value ≈ 0), high racial diversity (z-score = 15.475; p-value ≈ 0), poor English skills (z-score = 15.673; p-value ≈ 0), and low education (z-score = 17.787; p-value ≈ 0). This notable difference in spatial autocorrelation shows that block groups with high levels of isolated elderly populations are more randomly distributed than the other socio-demographic variables, thus increasing the chances that they will have a more randomized exposure to extreme heat and a less significant Student’s t-test result. The accessibility analysis revealed that walking speeds, as they relate to the distribution of cooling centers, greatly affect the percentage of areas in the city having access to heat refuge. At the slower walking speed (1.4 km per hour), only 3.4% of residents have access; at the average speed (3.5 km per hour), the percentage increases to 16.9%; and at the fast speed (5.6 km per hour), it increases to 32.7%. This finding reveals that even in the best case scenario (fast speed), less than one third of the population can access a public heat refuge. This may be especially meaningful for individuals with mobility challenges, such as those using wheelchairs, those with pre-existing