1. Introduction
Dual summertime environmental hazards facing many urban areas worldwide include excessive heat, compounded by the urban heat island (UHI) effect, and poor air quality. Often, however, solutions proposed to combat these issues may not consider the subsequent impacts related to the other. For example, one of the commonly proposed strategies for combatting ozone (O3) air pollution in urban regions worldwide is a reduction in on-road mobile sources by switching single-occupancy vehicle (SOV) travel to alternative transit (AT) methods, including walking, biking, and public transportation. In addition to reductions in air pollutant emissions, benefits of switching to AT include reducing noise (Nieuwenhuijsen 2020) and improving health for certain populations by increased physical activity (e.g., Maizlish et al. 2013; Olabarria et al. 2013; Woodcock et al. 2013; Buekers et al. 2015; Mueller et al. 2017; Pérez et al. 2017).
While AT usage may provide many benefits, AT modes may also expose users to other environmental hazards, including poor air quality and heat, that could lead to negative health outcomes. Research analyzing the exposure of walkers and bikers, including those walking to public transit stops, to heat in the San Francisco Bay area in California, has found that heat exposure varies both spatially in the region and with socioeconomic status (Karner et al. 2015). Numerous studies in different urban areas have shown reductions in AT usage during either heat waves or periods with high temperatures, including reductions in walking and biking (Ban et al. 2019), reduced usage of a bike-sharing network (Kim 2018), and decreased bus ridership (Ngo 2019). These issues are especially concerning given that O3 air pollution is typically a greater issue during the summertime. Zhang et al. (2017) found in a study of locations across the United States that extreme O3 days and high maximum daily temperatures were related, most likely due in part to factors such as meteorological conditions and sunlight amounts that favor both higher temperatures and increased O3. In urban areas, and especially those located in hot climates, combatting O3 pollution via increased AT usage may trade one problem for another by exposing AT users to greater amounts of heat. For those who have a vehicle available for travel, would they choose to use AT during periods of high temperatures? The answer to this question may depend on a variety of factors, including the planned travel time and the adequacy of AT infrastructure.
One prime location for analyzing these issues and potential solutions is the Phoenix metropolitan area in Maricopa County, Arizona. Maricopa County is the fourth most populous county in the United States and had the largest absolute growth in population of any county in the United States from 2010 to 2019 (U.S. Census Bureau 2020b). Maricopa County is located in the Sonoran Desert and experiences high summertime temperatures, with average daily maximum temperatures in June–August above 40°C (NWS 2021b). Furthermore, UHI impacts on heat in the urbanized portion of Maricopa County have been well studied in the region (e.g., Chow et al. 2012). Large portions of the county, including the urbanized areas, are designated nonattainment areas (NAA) for the Environmental Protection Agency (EPA) National Ambient Air Quality Standards (NAAQS) for O3 and particulate matter with diameter less than 10 μm (PM10) (EPA 2021). In addition to being a time of year with high temperatures, June–August is considered the peak O3 season for the Maricopa NAA (Maricopa County Air Quality Department 2019).
Studies in Maricopa County have examined aspects of heat impacts on public transit utilization. Fraser and Chester (2017a) analyzed the potential for heat exposure for transit riders by spatially determining the average times for transit riders to walk from their residences to transit stops and wait time at the transit stops. The authors found that on average the walk time was 6.2 min, and the average wait time was 11.1 min for public transit users. Dzyuban et al. (2021) examined the extent to which public transit stations provided thermal comfort for waiting riders during the summer in Phoenix and found that more than one-half of the riders surveyed were thermally uncomfortable. However, according to the Maricopa County Air Quality Department (MCAQD) 2017 Ozone Precursors Emissions Inventory, 51.2% of nitrogen oxides (NOx) emissions, which are precursors to O3, are due to on-road mobile sources and 55.1% of those on-road mobile source emissions of NOx are due to passenger cars and passenger trucks (MCAQD 2019). Therefore, a reduction in emissions of NOx via reducing SOV travel and increasing AT usage in the county presents a large opportunity for reducing O3 pollution in the region.
This study seeks to explore the following questions related to the viability of switching SOV to AT modes as a means for combatting O3 pollution in the study region: (i) what communities within Maricopa County have the most candidates for switching to AT to combat O3 pollution?; (ii) what does the relationship between regional weather conditions and air quality warnings imply for increased adoption of AT?; and (iii) what are the current and potential heat burdens for AT users, especially during O3 warnings? Overall, the goal of this work is to identify potential barriers to the viability of AT as a solution for combatting O3 pollution in a hot, desert region with a focus on travel associated with regular commuting for work. Although this research focuses specifically on the Phoenix metropolitan area, results should be applicable to other urbanized areas worldwide that face compounding impacts of extreme heat and poor air quality. These challenges will become even more important in coming years as heat waves have been increasing in frequency in urban areas across the world (Mishra et al. 2015).
2. Methods
a. Data descriptions
1) American Community Survey
Data about commuting methods and times were used from the American Community Survey (ACS) 5-yr estimates for 2015–19 (U.S. Census Bureau 2020a). Of interest for the present study were the following data tables from the ACS: “Means of Transportation to Work” (Table B08301) and “Means of Transportation to Work by Travel Time to Work” (Table B08134). In the latter table, estimates of commuters using each mode of commuting are provided within each of the following time categories (based on time from home to work): less than 10, 10–14, 15–19, 20–24, 25–29, 30–34, 35–44, 45–59, and 60 or more min. The ACS datasets include estimates and margins of error at 90% confidence levels for the included quantities (Fuller 2018).
2) Heat and air quality warnings
The Arizona Department of Environmental Quality issues advisories for days with forecast high concentrations of O3. Two types of warnings can be issued, health watches or high pollution advisories, depending on whether the O3 concentration is projected to approach or exceed the NAAQS level, respectively. The current NAAQS for O3 set in 2015 is 0.07 ppm averaged over an 8-h period (EPA 2022). Historical warnings for air quality in the region were obtained from the MCAQD website (Maricopa County Air Quality Department 2021).
As noted in section 1, the Phoenix metropolitan area is in nonattainment of two criteria pollutants, O3 and PM10. However, in the present study we have chosen to focus on solely O3. According to the historical warning file obtained from MCAQD, there were 251 warnings for O3 in 2017–20 as compared with only 28 warning days for PM10. In addition, 97.6% of these O3 warning days occurred during the months of April–September, as compared with only 71.4% of the PM10 warning days.
Historical excessive heat warnings for the region were obtained from the National Weather Service (NWS) Phoenix Heat Safety web page (NWS 2021a). These warnings are issued based on the level of risk determined by the NWS “HeatRisk” product, which is calculated using a variety of location-specific factors, including daily maximum temperature, overnight minimum temperature, time of year, and deviation from normal temperatures (NWS 2018).
3) Temperature data
Exposure to heat relevant to human health is a function of multiple environmental conditions, driven by factors such as air temperature, radiation (e.g., shade vs full sun), humidity, wind, etc. While mean radiant temperature (Middel and Krayenhoff 2019) can capture a variety of these exposure pathways in a single descriptive variable, these data are not available consistently across a larger spatial scale with high temporal resolution. Therefore, in the present study, we have used two proxies for determining exposure to heat that are available with greater spatial-temporal coverage: air temperature and land surface temperature (LST). While air temperature is used in determination of excessive heat warnings and has relatively high temporal coverage, LST from satellite-borne instrumentation yields higher spatial coverage. While neither metric perfectly captures the exposure of residents to heat, both provide unique strengths that are utilized in the present study. Air temperature data were obtained from MesoWest (University of Utah 2021) for Phoenix Sky Harbor Airport, which is centrally located in the study region.
LST data from the satellite-borne Ecosystem Spaceborne Thermal Radiometer Experiment on Space Station (ECOSTRESS) were used to characterize variations in heat across the metropolitan Phoenix region (NASA 2021). Data from ECOSTRESS have previously been used to examine variations in heat in the Los Angeles, California, metropolitan region (Hulley et al. 2019). Additional details related to the spatial and temporal coverage of the ECOSTRESS instrument can also be found in Hulley et al. (2019).
To determine the datasets used in this analysis, first, all days with both a heat warning and O3 warning in 2019–20 were identified (41 days). Next, all ECOSTRESS geolocation data granules from the dual warning days that included central Phoenix in the spatial domain of the bounding latitude/longitude coordinates for the granules were identified (31 granules on 22 unique days). From this initial list, only granules that fully covered the study region were kept for further analysis (15 granules from 15 separate days). Next, the ECOSTRESS data for LST and the cloud mask corresponding to the geolocation files were obtained. One of the dates did not have these corresponding data, resulting in 14 separate days of LST data. These data were filtered using the quality control (QC) flags and cloud mask provided to remove points obscured by clouds or with otherwise indicated lower quality (Hulley and Freepartner 2019). Visual inspection of the QC indicators resulted in the removal of 8 days of LST files (Fig. S1 in the online supplemental material), and 6 days were kept because of high-quality data coverage across entire the study region (Fig. S2 in the online supplemental material), although no file had high data quality for every single point in the study region. Each of these remaining 6 days had measurement times at different hours of the day. The highest-quality LST data for each of these 6 days were gridded to approximately 70-m resolution using procedures obtained from the NASA ECOSTRESS website (NASA 2022).
4) Alternative transportation infrastructure
To determine the extent to which AT use and relationships with heat may be impacted by the availability of AT infrastructure, spatial analysis of existing public transportation stops and bikeways were included in this analysis.
The regional public transit authority, Valley Metro, provides data about bus stops, light rail stations, and ridership via their Valley Metro GeoCenter website (Valley Metro 2021). For this work, the term “Valley Metro stops” includes both light rail stations and bus stops.
Regional bikeways data were obtained from the Maricopa Association of Governments data portal (Howard 2021). The types of bikeways included in this dataset are paved shoulder, bike lane, bike route, multiuse path–paved, multiuse path–unpaved, and recreational trail. For this analysis, all types of bikeways were included except for recreational trails, which are mainly found in locations such as regional parks that would not be considered transit thoroughfares.
b. Alternative transit utilization rate and candidate index
Metrics were developed to describe the areas with the best chances for increasing AT use based on both the current utilization of AT in each ZIP (U.S. postal) code and the number of commuters deemed the best candidates for switching to AT. The AT commuters were defined as those who walked, biked, or rode public transit to work. The commuters identified as the best candidates for switching to AT were defined as those with the shortest commutes based on the ACS data (<10 min) who drove alone to work via car, truck, or van. These commuters, subsequently referred to as SOV commuters with commute times of less than 10 min (SOV < 10), were combined with the number of current AT users to find a total number of prime candidates for AT commuting. Given that some fraction of these prime candidates already commute via AT, an AT utilization rate was defined as the ratio of current AT commuters to those who are prime candidates for AT use.
One goal of this study was to identify regions with the best potential for having commuters switch from SOV to AT. We used three assumptions to identify these regions. First, we assumed that regions with higher absolute numbers of SOV < 10 commuters presented a greater opportunity for reducing SOV commuting that regions with lower numbers of SOV < 10 commuters. While the commuting metrics are based on time and not distance, we believe it is reasonable to assume that shorter commuting times indicate shorter distances traveled for the same mode of commuting (e.g., SOV < 10 vs SOV > 10 min). Therefore, regions with more SOV < 10 commuters include higher numbers of shorter trips that would be “easier and more convenient” (Nieuwenhuijsen 2020) to switch from SOV transit as opposed to longer trips. Second, we assumed that regions with a higher AT utilization rate, defined as the ratio of current AT commuters to prime candidates for AT commuting, presented greater opportunities for increasing AT commuting. This second assumption may at first appear counterintuitive due to the idea that regions with higher AT utilization rates may have reached a saturation in the number of AT commuters. However, because the AT utilization rate is based on the prime candidates for AT commuting and not the total number of commuters in the region, a higher AT utilization rate may indicate increased availability of AT infrastructure. To illustrate this point, consider two hypothetical regions each with 1000 SOV < 10 commuters and varying AT utilization rates. In the first region, the AT utilization rate is 0.25, which means that there are approximately 333 AT commuters. In the second region, the AT utilization rate is 0.75, indicating 3000 AT commuters. Given the vastly different numbers of AT commuters in each region, we assume that AT commuting is a more readily available option in the region with the higher utilization rate, even with similar numbers of SOV < 10 commuters in both regions. Finally, the third assumption was that the number of SOV < 10 would more strongly influence the potential for switching from SOV to AT than the AT utilization rate.
To meet the above goal of capturing regions with higher potential for switching SOV to AT commuting, an AT candidate index was developed based on the two variables in the stated assumptions: the AT utilization rate and total number of SOV < 10. The index was created as a relative measure for different ZIP codes in the region by dividing the two variables into thirds (i.e., terciles) based on the 33rd percentile and 67th percentile for each. This process resulted in a 3 × 3 matrix that was numbered 1–9 to indicate areas of increasing potential for switching to AT from SOV based on the stated assumptions. Therefore, an AT candidate index of 1 corresponded to ZIP codes with values in the lowest third regionally for both AT utilization rate and SOV < 10, while ZIP codes with AT candidate index of 9 had values for these two variables in the highest third regionally.
c. ZIP code selection
For the portions of the analysis based on ZIP code, the following process was used to select ZIP codes in the urbanized area. Note that a map showing this process can be found in Fig. S3 of the online supplemental material. The initial ZIP codes selected were those in Maricopa County that contained at least a portion of one of the following urban areas defined by the census bureau: Phoenix–Mesa, Arizona Urbanized Area; Avondale–Goodyear, Arizona Urbanized Area; and Buckeye, Arizona Urban Cluster. These three census bureau urban areas in Maricopa County contain 97.1% of the total county population. While the Avondale–Goodyear, Arizona Urbanized Area and Buckeye, Arizona Urban Cluster fall entirely within Maricopa County, a small portion of the Phoenix–Mesa, Arizona Urbanized Area (3.9% of the urban area population and 5.4% of the land area) is located in Pinal County. From the initial list of ZIP codes overlapping with at least one of these urban areas, ZIP codes were excluded if less than 99.5% of the total ZIP code population lived in the urban area. Furthermore, ZIP codes were also excluded when the margin of error for either the number of AT commuters or SOV < 10 commuters exceeded two-thirds of the estimated value for either of these quantities. This process resulted in a total of 86 ZIP codes in the urbanized areas that were used for this analysis.
For subsequent maps and analyses related to density, the land area of the urban portion of the ZIP code was used. For 65 of the 86 ZIP codes analyzed, the entire area of the ZIP code was within the urban area. For the remaining 21 ZIP codes, between 0.5% and 86.1% of the total land area of the ZIP code was excluded because it was not defined by the census bureau as being in the urbanized portion of the ZIP code. However, as stated previously, for all ZIP codes used in the final analysis, at least 99.5% of the total ZIP code population resided in the urban portion of the ZIP code.
d. AT commuting times and exposure
For each ZIP code, the total amount of time for potential exposure to heat by AT commuters was calculated by summing the times for walking, biking, and walking/waiting at public transportation stops for commuters in the ZIP code. For walkers and bikers, the potential exposure time was determined by multiplying the midpoint of each time category from the ACS and the estimated number of commuters for that category. For each biker or walker in the 60-or-more-minutes category, the time used was 60 min. since an upper bound for this category is undefined. However, we estimate that the fraction of AT commuters falling into this category is only 1.4% for walkers and 11.9% for bikers. The sum of times for each walking category was then used as the total potential exposure time for walking commuters in the ZIP code. For the bikers, the time-delineated data provided by ACS is grouped with commuters who used a taxi or motorcycle. However, the ACS does provide the total number of commuters taking each of these three modes; therefore, the total time in each ZIP code calculated using the above method was subsequently multiplied by the relative fraction of bicycle commuters as compared with those who take a taxi or ride a motorcycle.
Finally, the level of exposure for public transportation users is more difficult to quantify since these commuters may be exposed to heat while walking to their stop and waiting, but not necessarily while riding a bus or light rail. We used the average values of walking to and waiting at the station for Valley Metro produced by Fraser and Chester (2017a): average walk time of 6.2 min and average wait time of 11.1 min, resulting in an average exposure time of 17.3 min per public transit commuter. However, since this value is greater than the amount of time some commuters use for their entire trip as documented in the ACS times, we did not include any estimates of walk and wait times for public transportation commuters with less than a 20-min commute. These public transit commuters with less than a 20-min commute only account for 13.6% of public transit commuters for the ZIP codes examined. In addition, the exposure time used for public transit commuters did not include the walk from the last station to their final destination due to lack of data.
3. Results and discussion
a. Current state of commuting in Maricopa County
An estimated 76.0% of workers aged 16 years and over in Maricopa County (total number of workers = 2 040 912 ± 6489) drove alone to work via car, truck, or van as shown in Fig. 1a. The commute times of the SOV commuters are also shown in Fig. 1b. Approximately 8.9% of SOV commuters had commute times in the shortest time category provided by ACS, less than 10 min (N = 138 745 ± 3088). Given these short SOV commute times, this group of commuters (SOV < 10) could be good candidates for changing to AT, since these commuters can be assumed to live with a reasonable distance of their destination. Furthermore, as stated by de Nazelle et al. (2010), short automobile trips “contribute disproportionately to emissions.” In addition, as shown in Fig. 1c, the number of SOV < 10 commuters exceeds the current number of AT commuters. While not explored further in this analysis, the number of SOV commuters with commute times of 10–14 min is greater than the number of SOV < 10 commuters and may also represent a suitable group of candidates for switching to AT.
Current status of commuting in Maricopa County, including (a) methods of commuting for all workers in the county, (b) commute times for SOV commuters in the county, and (c) description of prime candidates for AT commuting.
Citation: Weather, Climate, and Society 14, 3; 10.1175/WCAS-D-21-0158.1
The overall AT utilization rate for the county, as defined in the methods section (section 2) as the ratio of current AT commuters to prime candidates for AT commuting, is 0.39. However, this rate varies widely among the 86 ZIP codes examined, from 0.10 to 0.77, since various factors, including the availability of AT infrastructure influence commuting patterns. Therefore, Fig. 2 shows the spatial distribution of the AT candidate index of the ZIP codes examined across Maricopa County. As described in the methods section (section 2), this AT candidate index was developed to find the regions with the greatest chance for switching to AT and is based on AT utilization rate and total number of SOV < 10 in each ZIP code. As shown in Fig. 2, these two variables were divided into terciles for the region and AT candidate indices were assigned based on these delineations. Each AT candidate index contains 7–13 individual ZIP codes.
(a) The categories for the AT candidate index are defined by terciles for both the number of SOV < 10 commuters and AT utilization rate by ZIP code. (b) Regional map showing ZIP codes by AT candidate index, with major roadways displayed as gray lines. The approximate locations of a few important areas are shown, including downtown Phoenix, Phoenix Sky Harbor International Airport, and the ASU Tempe campus.
Citation: Weather, Climate, and Society 14, 3; 10.1175/WCAS-D-21-0158.1
Overall, a few trends emerge as to the locations most favorable to increase the rate of AT utilization. While ZIP codes in the central city area and to the west of downtown have high rates of AT utilization, these locations have fewer SOV < 10 commuters. However, the area to the east of downtown Phoenix and near the Arizona State University (ASU) Tempe campus exhibits a cluster of ZIP codes with the highest candidate index (9). Regions with a high number of SOV < 10 commuters but lower AT utilization rates are located farther from the central city area and tend to reside more in the outlying portions of the metropolitan Phoenix region.
As the number of AT commuters and AT utilization rate could be dependent on the available infrastructure, such as the number of public transportation stops or dedicated bikeways in a ZIP code, Fig. 3 shows the relationships between the density of public transit commuters, bicycle commuters, public transportation infrastructure, and bikeways infrastructure. In addition, the distribution of AT infrastructure in the ZIP codes for each AT candidate index is shown. While there is a fairly strong correlation between the density of public transit commuters in each ZIP code and the density of Valley Metro stops in each ZIP code (Fig. 3a), there is a very weak relationship between the density of bicycle commuters in each ZIP code and the density of bikeways in each ZIP code (Fig. 3b). For reference, Fig. S4 in the online supplemental material also shows the number and density of workers, Valley Metro stops, and bikeways per ZIP code, since infrastructure could have been developed based on population. As shown in Fig. S4, some of the most centrally located ZIP codes, such as those surrounding downtown Phoenix and Sky Harbor International Airport, contain low numbers and density of workers who reside in these ZIP codes. However, the density of Valley Metro stops is greatest in the central city and decreases while moving farther outward, probably in part due to the number of destinations for workers located in the central city area. In contrast, the distribution of designated bikeways is much more varied, with the highest lengths of bikeways and concentrations located in the southeastern portion of the region.
Density of AT infrastructure, including public transit stops and bikeways in each ZIP code. (a),(b) Scatterplots show the density of commuters for each AT mode vs the infrastructure density of the respective mode. (c),(d) The distribution of AT infrastructure for the ZIP codes in each AT candidate index. In (c) and (d), the whiskers show the minimum and maximum points, plus signs indicate outliers, and inner boxes show the 25th, 50th (median), and 75th percentiles for the data.
Citation: Weather, Climate, and Society 14, 3; 10.1175/WCAS-D-21-0158.1
For AT candidate indices within the same range for the number of SOV < 10 commuters, there tends to be an increase in the density of Valley Metro stops for the AT candidate indices with higher rates of AT utilization (Fig. 3c). For example, ZIP codes with an AT candidate index of 3 tend to have a higher density of Valley Metro stops than those ZIP codes with AT candidate indices of 1 and 2. However, there are a few notable exceptions, namely the minimum values for ZIP codes in AT opportunities indices of 3 and 7, which indicate that at least one ZIP code within each of these indices had zero Valley Metro stops. This may indicate that a higher number of walking and/or biking commuters were present in these ZIP codes or that nearby stops outside of the ZIP code boundaries are readily accessible.
For the density of bikeways (Fig. 3d), the opposite pattern seems to hold (i.e., higher bikeway density in regions with lower rates of AT utilization) except in the case of the AT candidate indices with the highest number of SOV < 10 commuters (7–9). For the other categories, a higher density of bikeways is often found in the ZIP codes that have lower AT utilization rates. As the estimated number of public transportation commuters in the region is far greater than the number of bicycle commuters (40 675 public transit commuters vs 15 862 bicycle commuters), it is expected that public transportation infrastructure as opposed to bikeways would be more strongly related to the AT candidate index. However, because the density of bicycle infrastructure and density of bicycle commuters in each ZIP code appear to be practically independent of one another, future research may be required to identify reasons for this apparent discrepancy.
b. Weather conditions during air quality warnings
High O3 pollution is generally considered to be a summertime issue given the need for solar radiation and elevated temperatures for O3 accumulation. In the Sonoran Desert, where Maricopa County is located, the summer is characterized by high temperatures and dry conditions prior to the onset of the southwestern monsoon during which peak temperatures decrease slightly and humidity increases. Therefore, conditions on days with projected high O3 in the region could also present peak temperatures that may result in less AT usage (i.e., extremely hot sunny days).
Table 1 summarizes various characteristics for days with O3 warnings in Maricopa County by month for 2017–20. For the purposes of Table 1, both high pollution advisories and health watches are included in the category “O3 warning.” The typical daily range in air temperature and air temperatures during “peak” commuting times as measured at Sky Harbor International Airport in Phoenix for days with O3 warnings are provided, along with the number of days with O3 warnings and the concurrence of these days with excessive heat warnings issued by the NWS.
Number of warnings and weather conditions of days with O3 warnings divided by month for 2017–20. The last column shows the total number of warnings and the average conditions of days with warnings for all months in 2017–20. For air temperatures, averages (± standard deviation) are shown.
As shown in Table 1, the vast majority of O3 warnings occur during the summer months. Furthermore, over a quarter of all days with O3 warnings coincide with days where an excessive heat warning was issued. Since high temperatures are a consistent feature during the summertime in the region, even on days without O3 warnings, the region is accustomed to advanced planning and adapting strategies for dealing with the heat. However, Maricopa County still accounts for a disproportionate number of the total heat-associated mortalities in the United States (Iverson et al. 2020). In addition, 5876 Arizona residents visited emergency departments for heat-related illness in Maricopa County from 2017 to 2020 (Arizona Department of Health Services 2022). Therefore, on days with both high temperatures and O3 warnings, AT usage as opposed to SOV travel may lead to greater heat exposure and subsequent negative health outcomes. As an estimated 87 806 commuters currently utilize AT in Maricopa County, including 31 269 workers who commute by walking, providing adequate shading and reducing the urban heat island effect will be critical efforts moving forward to protect these commuters from negative health impacts from extreme heat and will further be necessary for expanding the number of SOV < 10 commuters who choose AT. However, as previous studies have shown that heat exposure and heat-health outcomes are not distributed evenly across Maricopa County (e.g., Harlan et al. 2006, 2013; Hondula et al. 2015), potential heat burdens during AT usage may likewise show disparities across the region.
c. Commuting patterns and relationships with heat
To investigate the spatial distribution of heat as it relates to potential barriers to using and adopting AT, LST data were examined on six days with O3 warnings across the metropolitan Phoenix area. Whereas Figs. S5 and S6 in the online supplemental material show the pixel level data, ZIP code averages, and differences among average LSTs for the ZIP codes, Fig. 4 focuses on heat data from the morning of 6 May 2020. This dataset was chosen for this specific analysis because of its collection during a typical morning commute time [0815 local time (LT)]. Figure 4a shows the difference between the average LST for each ZIP code in the region and the ZIP code with the lowest average LST at 0815 LT 6 May 2020; these differences are further categorized by AT candidate index in Fig. 4b. While there is variability in the average LSTs for each ZIP code within each AT candidate index category, in general the AT candidate indices with higher rates of AT utilization had ZIP codes with hotter average LSTs than other AT candidate indices with comparable numbers of SOV < 10 commuters (i.e., 3 was hotter than 1 and 2, 6 was hotter than 4 and 5, and 9 was hotter than 7 and 8).
The average LST for 0815 LT 6 May 2020 was found for the urbanized portion of each ZIP code. (a) The difference between the average LST for each ZIP code and the average LST for the ZIP code with the coolest value in the region, shown as the “degrees above minimum.” (b) The degrees above minimum shown separated by AT candidate index. (c) The current average time for potential exposure during AT commuting. (d) The current degree-hours of AT exposure above minimum LST (DHAM). Also shown are the potential new (e) hours of AT exposure and (f) DHAM based on the scenario presented in the main text. Note that for (c) and (d), one ZIP code had a much higher number of hours and DHAM than all others; therefore, to increase figure readability, this ZIP code is shown in bright blue, whereas the other ZIP codes adhere to the given color scale.
Citation: Weather, Climate, and Society 14, 3; 10.1175/WCAS-D-21-0158.1
Using the method for calculating potential exposure times during AT commutes as presented in the methods section (section 2), we estimate that there are currently over 26 000 h of one-way (i.e., from home to work) potential exposure times by AT commuters each day. As shown in Figs. 4a and 4c, both heat and potential exposure times vary spatially across the region; therefore, these two metrics were combined to produce the DHAM for each ZIP code shown in Fig. 4d. The DHAM results show that certain regions, including areas surrounding downtown Phoenix, account for higher amounts of DHAM than the outer suburban regions (Fig. 4d).
Since we hypothesize that SOV commuters in regions with higher AT utilization rates may be more willing to switch to AT, due in part to both infrastructure and prevalence of AT usage, we mapped a scenario in which a variable percentage of SOV < 10 commuters switched to AT methods. If 5% of SOV < 10 commuters switched to AT in ZIP codes with the lowest AT utilization rates (i.e., AT candidate indices of 1, 4, and 7), 10% in regions with the middle AT utilization rates (i.e., AT candidate indices of 2, 5, and 8), and 15% in regions with the highest AT utilization rates (i.e., AT candidate indices of 3, 6, and 9), an estimated 10 678 SOV < 10 commutes one-way per day would be switched to AT means. If we assume that each of these trips switched to public transportation and we use the mean walking and waiting times for public transportation in the region from Fraser and Chester (2017a), an estimate of 3079 h of potential exposure time for one-way commutes would be added. Once again, this increased exposure time and related DHAM would not be distributed evenly across the region as shown in Figs. 4e and 4f.
To further investigate the spatial distribution of heat as it relates to AT infrastructure, LST data were examined for Valley Metro stops (bus stops and light rail stations) and bikeways across the region. Each Valley Metro stop that resides in one of the ZIP codes used in this analysis was assigned to the nearest LST pixel; note that this process sometimes resulted in multiple bus stops assigned to the same pixel. Then, for each LST file, all pixels that contained at least one Valley Metro stop (N = 6682) were used to construct quintiles and each pixel was subsequently assigned to one of those quintiles (i.e., bottom 20%, 20%–40%, 40%–60%, 60%–80%, or top 20%). This process was separately repeated for all pixels that were in one of the ZIP codes examined and overlapped with at least a portion of a bikeway (N = 60 968). The cut points for these quintiles for each file are shown in Table S1 in the online supplemental material. The difference between the maximum and minimum LST for each pixel containing at least one Valley Metro stop ranged from 8.9° to 15.9°C for the six different days for which LST was available, while the range was 11.5°–45.2°C for pixels containing bikeways. However, the difference between the cut point for the bottom 20% and top 20% only ranged from 1.6° to 3.4°C for Valley Metro pixels and from 2.1° to 4.0°C for bikeways pixels; therefore, the majority of LST values tended to fall within a narrow temperature range for each day. While previous studies in the region have shown variations in LST may be based on factors such as vegetation, land surface types, and measurement scales (e.g., Myint et al. 2015; Jenerette et al. 2016), the fairly narrow range of values for each day in the present study may be due to the location of Valley Metro stops and certain bikeways along roadways and therefore impervious surfaces that tend to be hotter.
As shown in Fig. 5, the distribution of hotter Valley Metro stops and bikeways was not even across all AT candidate indices. For example, in ZIP codes with AT candidate indices of 3 and 6 there were a higher percentage of Valley Metro stops in the top 20% of the hottest stops overall (30.2% and 26.3%, respectively) and a much lower percentage of Valley Metro stops in the coolest 20% (11.5% and 15.0%, respectively). In contrast, ZIP codes with AT candidate indices of 4 and 7 were associated with a much higher percentage of Valley Metro stops in the bottom 20% of all the LSTs (36.2% and 30.1%, respectively). The results for bikeways were similar, with bikeways in ZIP codes with AT candidate indices of 3 and 6 having an increased percentage of the hottest bikeways-containing pixels (39.5% and 31.3%, respectively). This means that two of the indices with the highest current rates of AT utilization have a disproportionate number of hotter stops and bikeways, while two of those with the lowest rates of AT utilization have a higher percentage of the coolest Valley Metro stops. These results, combined with the ZIP code level analysis of heat and commuting in Fig. 4, indicate that additional work is needed to ensure that current AT commuters are adequately protected from heat exposure and could indicate potential barriers for SOV commuters switching to AT commuting.
Distribution of LST pixels containing (left) Valley Metro stops and (right) bikeways based on AT candidate index and LST category, which is divided into quintiles for each of the six days with LST data. The cut points for each category for each day are provided in Table S1 in the online supplemental material. The numbers shown indicate the total number of pixels falling into each category over all six days. Note that excluding the “Total” and “% of pixels” columns, the colors in each row indicate the fraction in the LST category for each AT candidate index and therefore sum to a total of 1 for each row. The “% of pixels” column describes the percentage of pixels within each AT candidate index in the region that contain either a Valley Metro stop or a portion of a bikeway.
Citation: Weather, Climate, and Society 14, 3; 10.1175/WCAS-D-21-0158.1
d. Study limitations
There are a few important limitations to note with the current study. Data from ACS were used at the ZIP code level for the AT utilization analysis due to greater margins of error for the same dataset at smaller scales (i.e., census tracts). In addition, reasons for commuting via certain methods are not included in the dataset. However, other studies have shown that factors such as trip chaining (e.g., Jianchuan 2013), poverty levels (Burford et al. 2021), and the microscale built environment (Lanza et al. 2020) may influence mode of transportation.
Related to heat exposure, the level of physical activity for the different travel modes (i.e., riding a bicycle vs waiting at a bus stop) may exert some influence on outcomes. Furthermore, LST may not be fully indicative of heat exposure and levels of heat exposure sufficient for negative health outcomes. Additional environmental factors can impact personal heat exposure including air temperature and radiation (Kuras et al. 2017). However, research in the metropolitan Phoenix area has indicated a relationship between daytime LST and residents’ perceptions of and experiences with heat illness (Jenerette et al. 2016).
This study also did not closely examine one subset of AT commuters: those who commute by walking. However, we believe that the bikeways data may present a good proxy for walking routes as multipurpose pathways may be utilized by walkers and roadways with developed bikeways likely represent routes suitable for commuting by walking as well. As is the case with bicycle commuters, there is some uncertainty as to whether walkers and bikers utilize designated infrastructure for these modes (i.e., sidewalks or designated bikeways) or instead walk/ride alongside roadways.
This work did not address work-from-home/telecommuting policies, which can reduce SOV travel. This method may be appealing for workers who wish to avoid long commutes. However, work-from-home/telecommuting is not a feasible option for all job types. In addition, reports emerged in the region during the stay-at-home directive due to the COVID-19 pandemic that some new telecommuters struggled with higher electricity bills because of the need to air condition their homes during the hottest parts of the summer when they typically would be at work (Bausch et al. 2021).
4. Conclusions
This study examined the potential barrier that urban heat poses to the adoption of AT, including walking, biking, and riding public transportation, as a means for combatting O3 pollution in an urban desert region. While reduction in SOV travel and a subsequent increase in AT is a proven approach to both reducing emissions of air pollutants and increasing physical activity for typically sedentary populations, this strategy during periods of high temperatures may lead to negative health outcomes by exposing additional commuters to excessive heat.
First, this study examined areas of the study region with the most candidates for increasing AT usage. In Maricopa County, SOV commuting is by far the most widely utilized commuting method. However, over 138 000 SOV commuters in the region have commute times of less than 10 min, making them ideal candidates for changing to AT. By using the number of SOV < 10 commuters and the current number of AT commuters, an AT candidate index was developed to determine regions of the area that have the best candidates for changing from commuting by SOV to AT. Analysis shows that regions with a higher density of public transit commuters and a higher AT utilization rate tended to have a higher density of public transit stops as well. However, this pattern did not hold the same with regard to the density of bicycle commuters and the regions with a higher density of bikeways.
Second, the conditions present during O3 warnings for the region were examined. The vast majority of O3 warnings from 2017 to 2020 occurred during the hottest months of the year. Furthermore, over a quarter of days with an O3 warning also had an issued excessive heat warning. In addition, the average daily maximum air temperature on days with O3 warnings in the months of June–September 2017–20 (185 days) exceeded 41°C.
Last, LST data were examined to understand the heat burdens of current and potential AT commuters. Analysis of LST data by AT candidate index showed that ZIP codes and AT infrastructure (public transit stops and bikeways) in some of the regions with higher AT utilization rates tended to be hotter than those with lower AT utilization rates.
The results of this study suggest that caution should be used when advising AT as a means to combat O3 pollution (and therefore improve health outcomes from reduced exposure to O3). The association between high O3 pollution and high temperatures could result in alternative negative health outcomes due to exposure to heat for individuals who switch transit modes to AT. Therefore, a dual approach that considers both heat and air quality should be pursued for developing solutions to mitigate both hazards. Some potential strategies that could be used to mitigate exposure to extreme heat and therefore protect current and potential AT users include increasing frequency of public transportation services to reduce wait times at stops and increasing the network of public transportation to reduce walking times to stops. A case study from Los Angeles also found that reallocating existing public transportation resources to reduce wait times along certain routes could protect riders who are more vulnerable to heat (Fraser and Chester 2017b). In addition to reducing exposures to extreme heat, steps such as increasing frequency may also entice additional commuters to use public transportation (e.g., Redman et al. 2013), although many factors influence decisions to switch from SOV to public transit (Kang et al. 2020). Furthermore, providing shading through trees or engineered structures reduces daytime pedestrian heat exposure (Middel and Krayenhoff 2019) and Lanza et al. (2021) found significant differences in heat between shaded park areas and unshaded playgrounds. Implementing these design considerations at public transit stops and along walkways and bikeways may therefore protect both current and future AT users. While a study in Texas found that the availability of shading at bus stops had little impact on ridership numbers, the authors note that increasing shading at public transportation shelters will still be critical for protecting public transit users from heat exposure (Lanza and Durand 2021).
Although this work is focused on the metropolitan Phoenix area, because extreme heat waves and a warming climate bolstered by the urban heat island effect impact cities across the world, the results presented here could have wide-ranging implications to other regions as they work to improve urban air quality. In particular, urban regions may need to consider implementing strategies for improving thermal comfort and reducing heat exposure for AT users in conjunction with increasing AT usage to combat O3 air pollution. Extreme heat is a pressing issue in the metropolitan Phoenix area, but other urban regions may need additional research or strategies that focus on different applicable extreme weather events, such as extreme cold temperatures.
Acknowledgments.
This work was funded as part of the Healthy Urban Environments (HUE) initiative by the Maricopa County Industrial Development Authority (MCIDA), Award AWD00033817. Author Braun acknowledges support from the Mistletoe Research Fellowship.
Data availability statement.
All data used in this study are publicly available. Additional details and websites can be found in the references section for the following datasets: ACS (U.S. Census Bureau 2020a), MESOWEST (University of Utah 2021), ECOSTRESS LST (NASA 2021), air quality warnings (Maricopa County Air Quality Department 2021), heat warnings (NWS 2021a), bikeways (Howard 2021), and public transportation stops (Valley Metro 2021).
REFERENCES
Arizona Department of Health Services, 2022: Heat-related illness emergency department & inpatient admissions in Arizona by year (2016–2020). Accessed 11 March 2022, https://azdhs.gov/preparedness/epidemiology-disease-control/extreme-weather/index.php#news-publications.
Ban, J., W. Shi, L. Cui, X. Liu, C. Jiang, L. Han, R. Wang, and T. Li, 2019: Health-risk perception and its mediating effect on protective behavioral adaptation to heat waves. Environ. Res., 172, 27–33, https://doi.org/10.1016/j.envres.2019.01.006.
Bausch, C., and Coauthors, 2021: Are state and local policies addressing community concerns about extreme heat? A case study of American Indian and Latino Communities in Phoenix, Arizona. Arizona State University Morrison Institute for Public Policy Doc., 64 pp., https://morrisoninstitute.asu.edu/sites/default/files/are_state_and_local_policies_addressing_community_concerns_about_extreme_heat.pdf.
Buekers, J., E. Dons, B. Elen, and L. Int Panis, 2015: Health impact model for modal shift from car use to cycling or walking in Flanders: Application to two bicycle highways. J. Transp. Health, 2, 549–562, https://doi.org/10.1016/j.jth.2015.08.003.
Burford, K., L. A. Ganzar, K. Lanza, H. W. Kohl, and D. M. Hoelscher, 2021: School-level economic disparities in police-reported crimes and active commuting to school. Int. J. Environ. Res. Public Health, 18, 10885, https://doi.org/10.3390/ijerph182010885.
Chow, W. T. L., D. Brennan, and A. J. Brazel, 2012: Urban heat island research in Phoenix, Arizona. Bull. Amer. Meteor. Soc., 93, 517–530, https://doi.org/10.1175/BAMS-D-11-00011.1.
de Nazelle, A., B. J. Morton, M. Jerrett, and D. Crawford-Brown, 2010: Short trips: An opportunity for reducing mobile-source emissions? Transp. Res. Part D Transp. Environ., 15, 451–457, https://doi.org/10.1016/j.trd.2010.04.012.
Dzyuban, Y., D. M. Hondula, P. J. Coseo, and C. L. Redman, 2021: Public transit infrastructure and heat perceptions in hot and dry climates. Int. J. Biometeor., 66, 345–356, https://doi.org/10.1007/s00484-021-02074-4.
EPA, 2021: Nonattainment areas for criteria pollutants (Green Book). Accessed 8 October 2021, https://www.epa.gov/green-book.
EPA, 2022: NAAQS table. Accessed 21 February 2022, https://www.epa.gov/criteria-air-pollutants/naaqs-table.
Fraser, A. M., and M. V. Chester, 2017a: Transit system design and vulnerability of riders to heat. J. Transp. Health, 4, 216–225, https://doi.org/10.1016/j.jth.2016.07.005.
Fraser, A. M., and M. V. Chester, 2017b: Transit planning and climate change: Reducing rider’s vulnerability to heat. Int. Conf. on Sustainable Infrastructure, New York, NY, ASCE, 456–464.
Fuller, S., 2018: Using American Community Survey (ACS) estimates and margins of error. U.S. Census Bureau, accessed 17 June 2021, https://www.census.gov/programs-surveys/acs/guidance/training-presentations/acs-moe.html.
Harlan, S. L., A. J. Brazel, L. Prashad, W. L. Stefanov, and L. Larsen, 2006: Neighborhood microclimates and vulnerability to heat stress. Soc. Sci. Med., 63, 2847–2863, https://doi.org/10.1016/j.socscimed.2006.07.030.
Harlan, S. L., J. H. Declet-Barreto, W. L. Stefanov, and D. B. Petitti, 2013: Neighborhood effects on heat deaths: Social and environmental predictors of vulnerability in Maricopa county, Arizona. Environ. Health Perspect., 121, 197–204, https://doi.org/10.1289/ehp.1104625.
Hondula, D. M., R. E. Davis, M. V. Saha, C. R. Wegner, and L. M. Veazey, 2015: Geographic dimensions of heat-related mortality in seven U.S. cities. Environ. Res., 138, 439–452, https://doi.org/10.1016/j.envres.2015.02.033.
Hondula, D. M., and Coauthors, 2021: Novel metrics for relating personal heat exposure to social risk factors and outdoor ambient temperature. Environ. Int., 146, 106271, https://doi.org/10.1016/j.envint.2020.106271.
Howard, J., 2021: Bikeways in Maricopa and Pinal Counties. Maricopa Association of Governments, accessed 22 June 2021, https://geodata-azmag.opendata.arcgis.com/datasets/AZMAG::bikeways-in-maricopa-and-pinal-counties-arizona-2019/explore?location=33.469543%2C-112.025533%2C8.96.
Hulley, G., and R. Freepartner, 2019: ECOsystem Spaceborne Thermal Radiometer Experiment on Space Station (ECOSTRESS) Mission: Level 2 product user guide. California Institute of Technology Jet Propulsion Laboratory Doc. JPL D-103137, 19 pp., https://ecostress.jpl.nasa.gov/downloads/userguides/2_ECOSTRESS_L2_UserGuide_06182019.pdf.
Hulley, G., S. Shivers, E. Wetherley, and R. Cudd, 2019: New ECOSTRESS and MODIS land surface temperature data reveal fine-scale heat vulnerability in cities: A case study for Los Angeles County, California. Remote Sens., 11, 6–8, https://doi.org/10.3390/rs11182136.
Iverson, S. A., A. Gettel, C. P. Bezold, K. Goodin, B. McKinney, R. Sunenshine, and V. Berisha, 2020: Heat-associated mortality in a hot climate: Maricopa County, Arizona, 2006–2016. Public Health Rep., 135, 631–639, https://doi.org/10.1177/0033354920938006.
Jenerette, G. D., and Coauthors, 2016: Micro-scale urban surface temperatures are related to land-cover features and residential heat related health impacts in Phoenix, AZ USA. Landscape Ecol., 31, 745–760, https://doi.org/10.1007/s10980-015-0284-3.
Jianchuan, X., 2013: Research on the interdependencies between commute mode choice and trip chaining behavior. Int. Conf. on Management Science and Engineering, Harbin, China, IEEE 2042–2046, https://doi.org/10.1109/ICMSE.2013.6586545.
Kang, A. S., K. Jayaraman, K. L. Soh, and W. P. Wong, 2020: Tackling single-occupancy vehicles to reduce carbon emissions: Actionable model of drivers’ implementation intention to try public buses. J. Clean. Prod., 260, 121111, https://doi.org/10.1016/j.jclepro.2020.121111.
Karner, A., D. M. Hondula, and J. K. Vanos, 2015: Heat exposure during non-motorized travel: Implications for transportation policy under climate change. J. Transp. Health, 2, 451–459, https://doi.org/10.1016/j.jth.2015.10.001.
Kim, K., 2018: Investigation on the effects of weather and calendar events on bike-sharing according to the trip patterns of bike rentals of stations. J. Transp. Geogr., 66, 309–320, https://doi.org/10.1016/j.jtrangeo.2018.01.001.
Kuras, E. R., and Coauthors, 2017: Opportunities and challenges for personal heat exposure research. Environ. Health Perspect., 125, 085001, https://doi.org/10.1289/EHP556.
Lanza, K., and C. P. Durand, 2021: Heat-moderating effects of bus stop shelters and tree shade on public transport ridership. Int. J. Environ. Res. Public Health, 18, 463, https://doi.org/10.3390/ijerph18020463.
Lanza, K., and Coauthors, 2020: Transit environments for physical activity: Relationship between micro-scale built environment features surrounding light rail stations and ridership in Houston, Texas. J. Transp. Health, 19, 100924, https://doi.org/10.1016/j.jth.2020.100924.
Lanza, K., M. Alcazar, D. M. Hoelscher, and H. W. Kohl, 2021: Effects of trees, gardens, and nature trails on heat index and child health: Design and methods of the Green Schoolyards Project. BMC Public Health, 21, 98, https://doi.org/10.1186/s12889-020-10128-2.
Maizlish, N., J. Woodcock, S. Co, B. Ostro, A. Fanai, and D. Fairley, 2013: Health cobenefits and transportation-related reductions in greenhouse gas emissions in the San Francisco Bay Area. Amer. J. Public Health, 103, 703–709, https://doi.org/10.2105/AJPH.2012.300939.
Maricopa County Air Quality Department, 2019: 2017 periodic emissions inventory for ozone precursors. Maricopa County Doc., 177 pp., https://www.maricopa.gov/DocumentCenter/View/52917/2017-Periodic-Emission-Inventory-Ozone-PDF.
Maricopa County Air Quality Department, 2021: Historical-advisories-and-exceedances-2013-2021_202106141326403187. Maricopa County, accessed 15 June 2021, https://www.maricopa.gov/1643/Air-Quality-Status-and-Monitoring.
Middel, A., and E. S. Krayenhoff, 2019: Micrometeorological determinants of pedestrian thermal exposure during record-breaking heat in Tempe, Arizona: Introducing the MaRTy observational platform. Sci. Total Environ., 687, 137–151, https://doi.org/10.1016/j.scitotenv.2019.06.085.
Mishra, V., A. R. Ganguly, B. Nijssen, and D. P. Lettenmaier, 2015: Changes in observed climate extremes in global urban areas. Environ. Res. Lett., 10, 024005, https://doi.org/10.1088/1748-9326/10/2/024005.
Mueller, N., and Coauthors, 2017: Urban and transport planning related exposures and mortality: A health impact assessment for cities. Environ. Health Perspect., 125, 89–96, https://doi.org/10.1289/EHP220.
Myint, S. W., and Coauthors, 2015: Does the spatial arrangement of urban landscape matter? examples of urban warming and cooling in Phoenix and Las Vegas. Ecosyst. Health Sustainability, 1 (4), 1–15, https://doi.org/10.1890/EHS14-0028.1.
NASA, 2021: ECOSTRESS land surface temperature. NASA EOSDIS EarthData Search, accessed 21 July 2021, https://search.earthdata.nasa.gov.
NASA, 2022: ECOSTRESS frequently asked questions. California Institute of Technology Jet Propulsion Laboratory, accessed 11 March 2022, https://ecostress.jpl.nasa.gov/faq.
Ngo, N. S., 2019: Urban bus ridership, income, and extreme weather events. Transp. Res., 77D, 464–475, https://doi.org/10.1016/j.trd.2019.03.009.
Nieuwenhuijsen, M. J., 2020: Urban and transport planning pathways to carbon neutral, liveable and healthy cities; A review of the current evidence. Environ. Int., 140, 105661, https://doi.org/10.1016/j.envint.2020.105661.
NWS, 2018: Western region heat services. NWS Western Region Supple. 01-2018, 11 pp., https://www.nws.noaa.gov/directives/sym/pd01005015w012018curr.pdf.
NWS, 2021a: Historical heat warning dates. NWS Phoenix, accessed 15 June 2021, https://www.weather.gov/psr//Heat.
NWS, 2021b: Monthly climate normals Phoenix Area, AZ. NOAA, accessed 8 October 2021, https://www.weather.gov/wrh/climate?wfo=psr.
Olabarria, M., K. Pérez, E. Santamariña-Rubio, A. M. Novoa, and F. Racioppi, 2013: Health impact of motorised trips that could be replaced by walking. Eur. J. Public Health, 23, 217–222, https://doi.org/10.1093/eurpub/cks015.
Pérez, K., M. Olabarria, D. Rojas-Rueda, E. Santamariña-Rubio, C. Borrell, and M. Nieuwenhuijsen, 2017: The health and economic benefits of active transport policies in Barcelona. J. Transp. Health, 4, 316–324, https://doi.org/10.1016/j.jth.2017.01.001.
Redman, L., M. Friman, T. Gärling, and T. Hartig, 2013: Quality attributes of public transport that attract car users: A research review. Transp. Policy, 25, 119–127, https://doi.org/10.1016/j.tranpol.2012.11.005.
University of Utah, 2021: MesoWest. Accessed 17 June 2021, https://mesowest.utah.edu/.
U.S. Census Bureau, 2020a: 2015–2019 ACS 5-year estimates. Accessed 5 February 2021, https://www.census.gov/programs-surveys/acs/technical-documentation/table-and-geography-changes/2019/5-year.html.
U.S. Census Bureau, 2020b: Most of the counties with the largest population gains since 2010 are in Texas. Accessed 8 October 2021, https://www.census.gov/newsroom/press-releases/2020/pop-estimates-county-metro.html.
Valley Metro, 2021: Valley_Metro_Bus_Stops_View and VALLEYMETRO_LightRailStations. Valley Metro GeoCenter, accessed 22 June 2021, https://geocenter-valleymetro.opendata.arcgis.com/.
Woodcock, J., M. Givoni, and A. S. Morgan, 2013: Health impact modelling of active travel visions for England and Wales using an Integrated Transport and Health Impact Modelling Tool (ITHIM). PLOS ONE, 8, e51462, https://doi.org/10.1371/journal.pone.0051462.
Zhang, H., Y. Wang, T. W. Park, and Y. Deng, 2017: Quantifying the relationship between extreme air pollution events and extreme weather events. Atmos. Res., 188, 64–79, https://doi.org/10.1016/j.atmosres.2016.11.010.
