Extreme Heat Prevention Measures for Indoor and Outdoor Environments

Extreme heat events impact a wide range of health consequences (McMichael, Woodruff, and Hales 2006; Hondula, Balling, et al. 2015). The United States Center for Disease Control (CDC) defined the most at-risk groups to be: older adults, infants and children, people with chronic conditions and/or take medications that influence thermoregulation, low income, people with high heat exposure (athletes, outdoor workers). The definition of the vulnerable group expanded to include pregnant women, people with inadequate cooling systems, people who have or recovered from COVID-19 (Global Heat Health Information Network 2022). These vulnerable groups are exposed to extreme heat in both indoor and outdoor environments (Bock et al. 2021; K. L. Ebi et al. 2021; Eisenman et al. 2016). Therefore, it is important to establish extreme heat intervention measures for both environments. The primary goal of this dissertation is to integrate novel data streams to implement extreme heat intervention measures more efficiently. To achieve the goal, Chapter 2 evaluated a modified WBGT hindcast using the historical National Digital Forecast Database (hourly air temperature, dew point temperature, wind speed) and the European Centre for Medium-Range Weather Forecasts Reanalysis v5 (hourly solar radiation and wind speed). We verified the hindcasts with hourly WBGT estimated from ground-based weather observations. After controlling for geographic attributes and temporal trends, the average difference between the hindcast and in situ data varied from -0.64 °C to 1.46°C for different Köppen-Geiger climate regions reliable for decision making. However, the results showed statistically significant variances according to geographical features such as aspect, coastal proximity, land use, topographic position index, and Köppen-Geiger climate categories. The largest absolute difference was observed in the arid desert climates (1.46: 95% CI: 1.45,1.47), including some parts of Nevada, Arizona, Colorado, and New Mexico. This research investigates geographic factors associated with systematic WBGT differences and points toward ways future forecasts may be statistically adjusted to improve accuracy. Chapter 3 investigated the relationship between WBGT and shared city bicycle activity in New York City (NYC) and San Francisco (SF), US. Generalized Additive Models examined nonlinear relationships between WBGT and bicycle activity while controlling for rider demographics and temporal trends. Next, bootstrapping estimated the "peak point," when the relationship between the bike rentals and WBGT notably changed. The analysis also examined whether the heat warning messages affected cycling activities. We found that the number of rented bikes declined at different peak points in each city. The peak point was 27.4°C (95% CI, 25.0°C -29.9°C) in NYC and 23.2°C (95% CI 22.3-24.0°C) in SF. Somewhat paradoxically, bike rentals increased when heat warnings were issued in both cities. Chapter 4 leveraged property databases and identified Air conditioning (AC) ownership in Florida, US. This study examined census tracts with the highest/lowest air conditioning saturation applying Moran's I and Local Indicators of Spatial Association (LISA). Moran's I result showed significant spatial autocorrelation of AC ownership. Clusters of significantly high and low values (p < 0.05) of AC ownership were found across Florida. High-high clusters of AC ownership were found in the northern and eastern coastal Florida counties, with large metropolitan areas such as 39% in Duval County (Jacksonville), 20% in Sarasota County (Sarasota), 13% in Broward County (Fort Lauderdale), and 12% in Volusia County (Daytona Beach). Counties with clusters of no AC census tracts were located in the interior of Florida. We closely looked at AC prevalence clusters in Duval County, which contains the city of Jacksonville. High-value clusters of households without AC were found in the urban core, which has the most significant portion of poverty, poor housing conditions, and low education levels. Moreover, this study investigated the relationship between AC ownership and sociodemographic characteristics with a Spatial Durbin Model. We found a significant association between AC ownership and socioeconomic and urbanicity (urban and rural) variables. Among the socioeconomic variables: % Black/African American and % of historic Black/African Americans showed a significant positive relationship with the percentage of households without an AC. Our results can be used to identify neighborhoods that are in great need of AC and target heat prevention measures.