Many cities are experiencing more frequent extreme heat events during summer because of the global temperature rise and the urban heat island effects. Yet, how the rising temperature impacts urban residents differently is still not clear. Here we leveraged urban microclimate modeling to map the Universal Thermal Climate Index (UTCI), a more complete indicator of human heat stress at an unprecedentedly fine spatial resolution (1 m) for 14 major cities in the United States. We also examined the sensitivity and vulnerability to summer heat across different socioeconomic and racial/ethnic groups in these cities, finding that income level is most consistently associated with heat stress. Results show that a hypothetical 1 ºC increase in air temperature would have a substantial impact on human heat stress, with impacts that differ across cities. The results of this study can help us better evaluate the impact of extreme heat on urban residents.
Article
Sensitivity and vulnerability to summer heat extremes in major cities of the United States
https://doi.org/10.21203/rs.3.rs-3088700/v1
This work is licensed under a CC BY 4.0 License
Version 1
posted
You are reading this latest preprint version
Extreme heat events have become an increasingly severe threat to public health and urban sustainability due to global warming and rapid urbanization (Li and Wang et al., 2021). Every year, many people especially elderly people, are hospitalized or even die because of increasingly frequent heat waves. The 2003 European heat wave has killed more than 70,000 people over a few months (Robine et al. 2008). In the 2018 Northeast Asian heat wave, hundreds of people died from heat-related causes and thousands of people were hospitalized because of heat stroke. In 2021, a heat wave in Western North America had a death toll of over 1,400 people. The urban heat island effect is believed to exacerbate the mortality increase (Gabriel and Endlicher, 2011; Iungman et al., 2023). Extreme heat events have led to mortality increase in many cities all over the world and many cities expect extreme heat events more often with climate change (Alexander and Arblaster, 2009; Masselot et al., 2023). In addition, heat stress deteriorates the livability and walkability in cities (Lee et al., 2013; Jin et al., 2017) and impacts human physical activity levels.
Therefore, studying the impacts of extreme heat on urban residents is of great importance for building thermally comfortable and climate-resilient cities. Rising temperature would put more people out of the human climate niche and have huge human costs (Lenton, et al., 2023). On a local scale, not all neighborhoods and communities are equally impacted by extreme heat (Reid et al., 2009; Chakraborty et al. 2019; Hsu et al., 2021). Fine level quantitative information about where and which populations are vulnerable to heat is important to identify the most vulnerable neighborhoods and populations in order to minimize the negative impacts of heat stress on urban residents (Gronlund, 2014; Reid et al., 2012). The land surface temperature (LST) derived from remotely sensed thermal imageries is the most widely used metric to indicate the heat distribution at large scale because it is readily obtained at reasonably high resolution (< 100m) over large geographic areas. Hoffman et al. (2020) used LST to study the impacts of historical housing policy on the current heat exposure across 108 US urban areas. Results show that redlined neighborhoods are significantly warmer than non-redlined neighborhoods. Hsu et al (2021) studied environmental inequities across the United States using LST and found in most cities, people of color live in neighborhoods of higher heat intensity than non-Hispanic whites. Yin et al., (2023) found that lower-income people live in neighborhoods with higher LST in Los Angeles County, United States.
While the LST derived from satellite imagery provides an efficient way to quantify the spatial distribution of heat intensity across neighborhoods, LST cannot fully indicate human outdoor heat stress level, since it does not consider factors that impact human heat stress level such as, the shade, radiation, wind, and humidity (Budd, 2008; Lindberg, 2016; Li, 2021; Park et al., 2014; Di Napoli et al., 2018; Lau et al., 2015; Chakraborty et al. 2022). In addition, the satellite-view LST indicates the surface temperature of all satellite-visible surfaces, including building roofs and treetops, which are not the places where human activities take place.
In contrast to remote sensing-based LST, urban microclimate modeling provides an option to map hyperlocal urban heat exposure from a more human-centric perspective. Li (2021) investigated the heat exposure in Philadelphia using the 1m resolution mean radiant temperature (Tmrt) generated from urban microclimate modeling and found that the distribution of Tmrt is more significantly associated with income than with race/ethnicity. However, the microclimate modeling on high resolution urban geometrical model is very time-consuming, which is the major obstacle for large scale analyses. In 2021, Li and Wang (2021) proposed a GPU-accelerated algorithm to make the process much more efficient, which dramatically increases the applicability of the method for mapping the large-scale urban heat exposure. Here, we adopt the GPU-accelerated algorithm based on high-resolution 3D city models to calculate and map a more human-centric heat stress indicator, the Universal Thermal Climate Index (UTCI), in 14 major cities of different climate zones in the United States. The UTCI considers the human body’s energy balance, accounting for air temperature, shade, wind speed, and humidity. It has been widely used as an indicator of human thermal comfort. In order to investigate the sensitivity to higher air temperature in those cities, this study further simulated and mapped the UTCI in the scenarios of a local 1 ºC air temperature increase. This study also examined heat exposure levels across different socioeconomic and racial/ethnic groups in different cities.
Spatial distributions of UTCI
Figure 1 shows the spatial distributions of the universal thermal climate index (UTCI) with the spatial resolution of 1m in different cities in the current weather situation of the year 2017. Different cities have different levels of outdoor heat exposure level in the hottest month. Generally, in cities of Boston, Chicago, Seattle, San Francisco, Philadelphia, New York City, San Diego, the heat stress levels are moderate based on the simulation results that use current meteorological parameters inputs, since the maximum average UTCI in the hottest month is lower than 32 ºC, which is the threshold of strong heat stress. Cities of Baltimore, Washington, DC, Dallas, Atlanta, Los Angeles, Houston, and Miami have peak daily UTCI values larger than the strong stress threshold of 32 ºC. Please note that this is the averaged UTCI from 8 am to 5 pm every day in the hottest month, which smooths the highest UTCI values in one day. For Atlanta, Dallas, and Houston, the averaged UTCI for the whole cities are higher than 32 ºC, and some parts of the cities even experience UTCI higher than 38 ºC, which is considered as very strong heat stress level.
Fig. 2 shows the spatial distribution of the aggregated UTCI at the census tract level using the mean values by masking out the building roof pixels since the building roofs are generally not frequented by people.
Sensitivity analysis of heat stress level to a 1 ºC air temperature increase
Figure 3 shows the histograms of the UTCI values based on the current meteorological parameters input and in the scenario of 1 ºC air temperature increase in different cities at the pixel level. All cities have skewed distributions of UTCI to the high values, except for Atlanta. With 1 ºC air temperature increase, the histograms shift toward higher heat stress level for all cities. The shapes of the histograms of the UTCI in different cities also show that different cities may have different susceptibility levels to the potential extreme heat events. For cities with histograms more skewed to higher UTCI value, larger areas will be exposed to higher heat stress as the air temperature increases. This is especially for true Chicago, Seattle, San Diego, and Los Angeles. Although the UTCI values are not very high in those cities, most pixels are in the far-right side of the histograms. This means that as the air temperature increases, most of those cities would be exposed to strong heat stress. Atlanta is different from other cities in that most pixels are in the left side of the histogram. This is because of the large tree canopy coverage (more than 46%) in the city. However, in Atlanta most pixels have UTCI values higher than 32 ºC, which indicates the city is experiencing strong heat stress in the hottest month. Although the large tree canopy covers help to lower the heat exposure level in Atlanta, the general heat exposure level is still too high for human thermal comfort. Therefore, other measures need to be taken to mitigate outdoor heat exposure level in Atlanta.
Table 1 presents the numbers and percentages of population that are exposed to strong heat stress (UTCI larger than 32 ºC) in different cities in the current situation and the scenario of air temperature 1 degree higher. In the current situation, about 89.6% of residents in Washington, D.C, 79.3% of residents in Los Angeles, 100% residents in Dallas, 100% residents in Houston, 100% residents in Atlanta, 97.1% residents in Miami, 32.9% residents in Baltimore are exposed to strong heat stress. In the scenario of 1 ºC higher air temperature, almost all residents in those cities are exposed to strong heat stress. For Dallas, about 11.3% residents are experiencing very strong heat stress with UTCI higher than 38 ºC and for Houston, about 7.3% of residents are experiencing very strong heat stress (UTCI larger than 38 ºC). For cities of New York City, Seattle, Chicago, Boston, Philadelphia, San Francisco, San Diego, no census tracts have UTCI higher than 32 ºC in the current climate setting. However, in the scenario of 1 ºC higher air temperature, small population in Boston and San Diego starts to experience strong heat stress.
|
Cities |
Population exposed to strong heat stress (UTCI higher than 32 ºC) |
|||
|---|---|---|---|---|
|
Current weather setting |
Air temperature increases by 1 ºC |
|||
|
Population number |
Percentage |
Population number |
Percentage |
|
|
Boston |
0 |
0 |
1,020 |
0.2% |
|
Chicago |
0 |
0 |
0 |
0 |
|
Seattle |
0 |
0 |
0 |
0 |
|
San Francisco |
0 |
0 |
0 |
0 |
|
Philadelphia |
0 |
0 |
0 |
0 |
|
Washington, DC |
613,127 |
89.6% |
680,305 |
99.4% |
|
Baltimore |
202,008 |
32.9% |
585,324 |
95.2% |
|
New York City |
0 |
0 |
0 |
0 |
|
Dallas |
1,501,499 |
100% |
1,501,499 |
100% |
|
Atlanta |
479,854 |
100% |
479,854 |
100% |
|
San Diego |
0 |
0 |
10,817 |
0.7% |
|
Los Angeles |
3,259,731 |
79.3% |
3,259,731 |
100% |
|
Houston |
4,774,220 |
100% |
4,774,220 |
100% |
|
Miami |
439,885 |
97.1% |
453,211 |
100% |
Correlations between the UTCI and socioeconomic and racial/ethnic variables
Figure 4 shows the correlation coefficients of the UTCI and different socioeconomic and racial/ethnic variables in different cities. Generally, the per capita income has a significant and negative correlation with the UTCI in all cities except Atlanta, Seattle, Baltimore, and Boston. The proportion of non-Hispanic Whites has a significant and negative correlation with the UTCI in New York City, Los Angeles, Houston, Miami, Chicago, Philadelphia, San Francisco, and San Diego, while for other cities, there is no significant correlations. The proportion of Hispanics has significant and positive correlation with the UTCI in cities of New York City, Los Angeles, Houston, Miami, Chicago, Philadelphia, and San Diego. The proportion of African Americans generally has no significant correlation with the UTCI, except the significant and positive correlation in cities of New York City, Los Angeles, and San Diego. The proportion of Asian Americans has significant and positive correlation with the UTCI in cities of Dallas, Houston, Atlanta, and San Francisco, while there is no significant correlation in the other cities. The educational variables have relatively consistent correlation with the UTCI in different cities, as neighborhoods with more people of higher education have lower UTCI, and neighborhoods with higher proportion of people without high school degree tend to have higher UTCI. However, such correlations are not significant for all cities. The portion of people under 18 years of age has a significant and positive correlation with the UTCI for New York City, Los Angeles, Chicago, Philadelphia, and San Francisco, while in Atlanta, Seattle the correlation is negative. There is no significant correlation for the other cities. The proportion of people older than 65 has a consistently and significantly negative correlation with UTCI in cities of Washington, D.C, New York City, Los Angeles, Dallas, Houston, Chicago, Baltimore, Boston, Philadelphia, and San Diego, while there is no significant correlation for the other cities.
Regression analysis results
Figure 5 shows the regression analysis results of the UTCI and different socioeconomic and racial/ethnic variables in different cities. The per capita income has a significant and negative association with the UTCI in cities of Washington, D.C, New York City, Los Angeles, Dallas, Atlanta, Philadelphia, and San Diego. However, in Chicago, the per capita income has a significant and positive association with the UTCI. The per capita income has no significant association with the UTCI in Houston, Seattle, Miami, Baltimore, Boston, and San Francisco. However, after controlling the spatial autocorrelation in the residuals of the OLS, the per capita income has a consistently and significantly negative association with the UTCI in most cities, except cities of Seattle, Miami, Chicago, Baltimore, and Boston.
In the OLS model, the proportion of Hispanics has a mixed association with the UTCI in different cities. For example, in cities of New York City and Dallas, the proportion of Hispanics has a significant and negative association with the UTCI, while for the cities of Los Angeles, Houston, Miami, Chicago, Baltimore, Philadelphia, San Francisco, and San Diego, the proportion of Hispanics has a significant and positive association with the UTCI. There is no significant association between the UTCI and proportion of Hispanics in cities of Washington, D.C, Atlanta, Seattle, and Boston. After controlling the spatial autocorrelation, the proportion of Hispanics has a consistently and significantly positive associations with the UTCI in cities of Los Angeles, Houston, Atlanta, Miami, Chicago, Philadelphia, and San Diego, while for the other cities, there is no significant association.
The proportion of African Americans has a mixed association with the UTCI in different cities. In the OLS regression model, the proportion of African Americans has a significant and positive association with UTCI in Los Angeles, Houston, Miami, Chicago, and San Francisco, while the proportion of African Americans has a significantly negative association with UTCI in cities of Dallas, Atlanta, and Boston. There is no significant association between the proportion of African Americans and UTCI in other cities. After controlling the spatial autocorrelation, the proportion of African Americans has significant and positive association with UTCI in Los Angeles, Houston, and Chicago, but negative and significant association with UTCI in Boston. For all other cities, there is no significant association between the proportion of African Americans and UTCI.
The proportion of Asian Americans has a consistently and significantly positive association with the UTCI in cities of Los Angeles, Dallas, Houston, Seattle, Miami, Chicago, and San Francisco, while for the rest cities, there is no such significant association. After controlling the spatial autocorrelation, the proportion of Asian Americans still have a relatively consistent and significantly positive association with the UTCI in cities of New York City, Los Angeles, Dallas, Houston, and Chicago.
Extreme heat is an increasingly dangerous threat to human well-being in cities in the context of climate change. This study investigated the heat distributions in 14 major US cities at an unprecedented fine level (1 m) using urban microclimate modeling based on high-resolution urban geometrical models and meteorological data. We computed and mapped the spatial distributions of UTCI (Universal Thermal Climate Index), which is a physiologically relevant indicator of human heat stress with consideration of meteorological parameters that impact human body’s energy balance, for both the current climate setting and a scenario of 1 ºC higher air temperature. Fine level urban geometrical models were used to model the solar radiation flux in complex urban geometries. The UTCI maps generated in this study can better indicate the urban heat distribution across neighborhoods in different cities. In addition, this study examined the fine level outdoor heat stress by excluding rooftops to focus on areas that are accessible to the urban residents, those places of building roofs are neglected in order to better represent human actual heat stress. The scalable framework based on nationally available dataset presented in this study can be used for urban heat management for cities across the United States. The generated hyperlocal urban heat map can provide more insightful guidance for urban environment planning in cities in the context of warming climate.
We also conducted a sensitivity analysis of the heat stress change with an air temperature increase of 1 ºC to examine the heat exposure change in future higher air temperature scenarios. Sensitivity analysis results show that different cities show different sensitivities to rising air temperature and many cities are on the brink of strong heat stress level. With 1 ºC increase of air temperature, 0.2% residents in Boston and 0.7% residents San Diego start to experience strong heat stress (with UTCI larger than 32 ºC). The more skewed histogram of San Diego shows that more areas in San Diego will experience strong heat stress with air temperature increase. More than 95% of residents in Washington, DC, Baltimore, Dallas, Atlanta, Los Angeles, Houston, and Miami will experience strong heat stress in hot summer. For cities of Chicago, Seattle, San Francisco, Philadelphia, New York City, the heat stress level still does not reach a strong level with 1 ºC higher air temperature. However, because of the skewed distributions of UTCI histograms in Chicago, Seattle, San Francisco, and Philadelphia, with the increase of air temperature in future climate scenarios or during extreme weather events with high air temperature, a large portion of those cities will experience strong heat stress. This would make those cities more vulnerable to extreme heat events in a warming climate.
Statistical analyses results show that the per capita income has a relatively consistent and significantly negative association with the UTCI in most cities, which indicates that richer people tend to live in neighborhoods in relatively cooler neighborhoods. The proportion of Hispanics is significantly associated with stronger heat stress in cities of Los Angeles, Houston, Atlanta, Miami, Chicago, Philadelphia, and San Diego. Since Los Angeles, Houston, and Atlanta experience strong heat stress, this is more serious public health and wellbeing threat to Hispanics neighborhoods. There is no consistently significant association between the proportion of African Americans and the heat stress level across cities. The proportion of African Americans has significant association with higher heat stress levels only in cities of Los Angeles, Houston, and Chicago, while in Boston the association is negative and in other cities the association is not significant. Highly Asian American neighborhoods experience consistently stronger summer heat stress in cities of New York City, Los Angeles, Dallas, Houston, and Chicago. Since cities of Los Angeles, Dallas, and Houston are experiencing strong heat stress or even very strong heat stress level in summertime, this situation is even worse for the Asian American communities in those cities.
Although this study generated hyperlocal outdoor urban daytime heat maps for 14 US major cities based on fine level multispectral aerial images and LiDAR data with spatial resolution of 1m, there are still several limitations in this study. First, this study only computes the daytime and outdoor heat exposure in those cities and the nighttime heat exposure is not considered. However, the nighttime and indoor heat stress level also have meaningful impact on human wellbeing. Therefore, future studies should also focus on indoor and nighttime heat exposure. For modeling the indoor heat exposure level, more detailed information about the buildings should be further considered, such as the building types, floors, roof colors, windows, etc.
In addition, the ground covers and pavements were not considered in the modeling and the albedo of the ground was assigned as a universal value. However, in the heterogenous urban environment, different land use/covers and pavement materials have very different reflectance of the solar radiation. Therefore, in future studies, fine level ground covers should also be considered to investigate the impacts of the ground covers and pavement materials on human outdoor heat exposure. The wind speed, relative humidity, and air temperature were set as universal numbers for whole city based on the National Renewable Energy Laboratory data, which is estimated based on ground measurements. However, the wind speed, relative humidity, and air temperature also vary spatially across neighborhoods in on city. Future studies should also consider the spatial variations of the air temperature and wind speed, both of which are impacted by the complex urban geometry settings.
In addition, the census data used in this study only indicate the residence of urban residents, while residence place is not the place where all human activities happen, particularly during daytime. Future studies should also consider the human mobility pattern to better estimate human actual exposure to extreme heat. Future studies should also incorporate people’s travel behaviors ad daily mobility to better model the impact of outdoor heat exposure on urban residents.
Data preparation
This study selected 14 major cities across the United States in different climate zones. Figure 1 shows the locations of those selected cities. The datasets used in this study include the building footprint maps, LiDAR cloud point data, National Agriculture Imagery Program (NAIP) imageries, and metrological data from National Renewable Energy Laboratory (NREL). The building footprint map was collected from Microsoft building footprint (https://www.microsoft.com/en-us/maps/building-footprints), which was derived from high resolution remotely sensed imageries based on deep convolutional neural network with 99.3% precision and 93.5% pixel recall accuracy. The most recent high-resolution LiDAR data available in the form of pre-processed x, y, z points cloud files for all selected cities were collected from United States Geological Survey 3D Elevation Program (https://usgs.entwine.io/). In this study, the open-sourced tool PDAL was used to convert the LiDAR cloud points into digital elevation model (DEM) and digital surface model (DSM) automatically. The building height model and the generated DSM were combined to generate the building height model. The most recent NAIP aerial imageries with the spatial resolution around 1m have four bands, near infrared, red, green, and blue. The normalized difference vegetation index (NDVI) was calculated based on the multispectral NAIP imageries to extract the tree canopy cover maps using the thresholding method for each city. The tree canopy cover maps were then refined based on the generated DSM by removing those pixels lower than 2m, since the shade of trees lower than 2m is negligible. The generated tree canopy maps were then overlayed on the DSMs to generate the tree canopy height model for each city. Validation of the tree canopy cover was performed through comparisons with human-manually digitized tree canopy covers. Results show that the classification accuracy is higher than 90%.
The meteorological data of 2017 was collected from the National Renewable Energy Laboratory (NREL) (https://nsrdb.nrel.gov/). The meteorological data include the air temperature, global horizontal radiation, direct radiation, diffuse radiation, wind speed, relative humidity, etc. The current meteorological data is modeled using multi-channel measurements from geostationary satellites.
Universal thermal climate index (UTCI) mapping
Human heat stress is a comprehensive index influenced by the air temperature, solar radiation, humidity, and wind speed, along with physiological factors and possibly other environmental considerations. This study quantifies potential heat stress using the UTCI, which has units of °C, has been used widely to indicate the human heat stress level (Young et al., 2021). UTCI higher than 32°C is generally defined as strong heat stress level (Fig. 2c). The mean radiation temperature (Tmrt) is the most important input parameter for calculating the UTCI. The Tmrt is the net shortwave and longwave radiation that human body exposed from the surrounding environment and the Tmrt is the most significant meteorological input parameter for the human energy balance especially during clear and calm summer days (Mayer and Höppe 1987). Based on the Stefan-Boltzmann law, Tmrt can be calculated as,
where \({\epsilon }_{p}\) is the emissivity of the human body (standard value 0.97), σ is the Stefan-Boltzmann constant, and R denotes total radiation exposure as the sum of short and long wave radiation from above, below, and the four cardinal directions, and the R can be calculated as,
$$R={\xi }_{k}{\sum }_{1}^{6}{K}_{i}{F}_{i}+{\epsilon }_{p}{\sum }_{1}^{6}{L}_{i}{F}_{i}$$
2
where Ki is the shortwave radiation component from 6 directions (north, south, west, east, top and bottom), Li is the longwave radiation, Fi is the angular factor between a person and the surrounding environment, \({\xi }_{k}\) is the absorption coefficient for shortwave radiation (standard value 0.7). This study adopted the previously developed GPU-accelerated SOlar and LongWave Environmental Irradiance Geometry model (SOLWEIG) model (Li and Wang, 2021) to calculate the Tmrt based on the input urban 3D model and the meteorological data (Fig. 2). Based on the estimated Tmrt, this study rewrote the UTCI approximation algorithm based on the original Fortran UTCI algorithm (Bröde et al., 2014) with consideration of wind speed, humidity, and air temperature to calculate and map UTCI in all selected cities.
Sensitivity and vulnerability analysis to the extreme heat
The UTCI was calculated at the hourly level from 8am to 5pm everyday in the hottest month in each city in 2017. In San Francisco, Seattle, and San Diego, the hottest month is September, and the hottest month for other cities is July. In order to examine the sensitivity of heat stress level to higher air temperatures, this study modeled and mapped the spatial distributions of the UTCI in the scenario of 1 ºC higher air temperature in all cities. The generated hourly level UTCI maps were averaged to indicate the general pattern of the UTCI values distribution in both the current weather setting and 1 ºC higher air temperature scenario.
In order to examine the distributions of the heat stress levels among different socioeconomic and racial/ethnic groups, this study further compared the distributions of UTCI with socioeconomic and racial/ethnic statuses in different cities. To represent the socio-economic statuses of residents, we selected variables of per capita income, proportion of non-Hispanics Whites, proportion of African American, proportion of Hispanics, proportion of Asian Americans, proportion of people older than 65, proportion of people younger than 18, and proportion of people with bachelor or higher degrees, and proportion of people without high-school degrees based on previous studies (Li, 2021; Hsu et al., 2021; Landry et al., 2011). All the socio-economic variables were collected from American Community Survey 5-year data of 2015–2019. To make the heat maps comparable to the census data, this study aggregated the pixels level UTCI maps to census tract using the mean value for all cities and those building roof pixels were masked out.
The ordinary least regression models (OLS) were applied to investigate the associations between the social variables and UTCI values. In order to examine the heat stress level for residents of different racial/ethnic groups, therefore, only those minority variables were selected in the OLS regression models, which is defined as,
$${y}_{i}={X}_{i}\beta +{ϵ}_{i}$$
3
where \({y}_{i}\)is the aggregated UTCI for different census tracts, \({X}_{i}\) indicates the independent variables, \(\beta\) is the regression coefficient, \({ϵ}_{i}\) is the sigle error term. The global Moran’s I statistics were used to examine the spatial autocorrelation in the residuals of the OLS regression models. The spatial error regression (SER) models were then be applied when the residuals have significant spatial dependence in the OLS models for different cities. In the SER model, the single error term \({ϵ}_{i}\) in the OLS model can be split this error into random error \({u}_{i}\)and spatially structured error \(\lambda {w}_{i}{ϵ}_{i}\) (Anselin, et al., 2009),
$${y}_{i}={X}_{i}\beta +\lambda {w}_{i}{ϵ}_{i}+{u}_{i}$$
4
where \({w}_{i}\) is the weight matrix was calculated based on the Euclidean distances between centroids of census tracts, \(\lambda\) is the spatial error coefficient, \({u}_{i}\) is random error.
Data availability
UTCI maps are available for exploration on an interactive website, available at https://xiaojianggis.github.io/heatexpo/ and also for download upon request. Sociodemographic data were collected from the US Census Bureau 2017 5-year ACS via the API at https://api. census.gov/data/2017/acs/acs5/variables.html.
Code availability
Code to reproduce the figures is available upon reasonable request.
- Alexander, L. V., & Arblaster, J. M. (2009). Assessing trends in observed and modelled climate extremes over Australia in relation to future projections. International Journal of Climatology, 29(3), 417-435.
- Algeciras, J. A. R., Consuegra, L. G., & Matzarakis, A. (2016). Spatial-temporal study on the effects of urban street configurations on human thermal comfort in the world heritage city of Camagüey-Cuba. Building and Environment, 101, 85-101.
- Ali-Toudert, F., & Mayer, H. (2006). Numerical study on the effects of aspect ratio and orientation of an urban street canyon on outdoor thermal comfort in hot and dry climate. Building and environment, 41(2), 94-108.
- Aminipouri, M., Knudby, A. J., Krayenhoff, E. S., Zickfeld, K., & Middel, A. (2019). Modelling the impact of increased street tree cover on mean radiant temperature across Vancouver’s local climate zones. Urban Forestry & Urban Greening, 39, 9-17.
- Anselin, L., Syabri, I., & Kho, Y. (2009). GeoDa: an introduction to spatial data analysis. In Handbook of applied spatial analysis: Software tools, methods and applications (pp. 73-89). Berlin, Heidelberg: Springer Berlin Heidelberg.
- Armson, D., Rahman, M. A., & Ennos, A. R. (2013). A comparison of the shading effectiveness of five different street tree species in Manchester, UK. Arboriculture Urban Forestry, 39, 157-164.
- Borden, K. A., & Cutter, S. L. (2008). Spatial patterns of natural hazards mortality in the United States. International Journal of Health Geographics, 7(1), 1.
- Buyantuyev, A., & Wu, J. (2010). Urban heat islands and landscape heterogeneity: linking spatiotemporal variations in surface temperatures to land-cover and socioeconomic patterns. Landscape Ecology, 25(1), 17-33.
- Budd, G. M. (2008). Wet-bulb globe temperature (WBGT)—its history and its limitations. Journal of science and medicine in sport, 11(1), 20-32.
- Chakraborty, T., Hsu, A., Manya, D., & Sheriff, G. (2019). Disproportionately higher exposure to urban heat in lower-income neighborhoods: a multi-city perspective. Environmental Research Letters, 14(10), 105003.
- Chakraborty, T., Venter, Z. S., Qian, Y., & Lee, X. (2022). Lower urban humidity moderates outdoor heat stress. Agu Advances, 3(5), e2022AV000729.
- Di Napoli, C., Pappenberger, F., & Cloke, H. L. (2018). Assessing heat-related health risk in Europe via the Universal Thermal Climate Index (UTCI). International journal of biometeorology, 62, 1155-1165.
- Reid, C. E., O’neill, M. S., Gronlund, C. J., Brines, S. J., Brown, D. G., Diez-Roux, A. V., & Schwartz, J. (2009). Mapping community determinants of heat vulnerability. Environmental Health Perspectives, 117(11), 1730-1736.
- Fahmy, M., Sharples, S., & Yahiya, M. (2010). LAI based trees selection for mid latitude urban developments: A microclimatic study in Cairo, Egypt. Building and Environment, 45(2), 345-357.
- Lau, K. K. L., Lindberg, F., Rayner, D., & Thorsson, S. (2015). The effect of urban geometry on mean radiant temperature under future climate change: a study of three European cities. International journal of biometeorology, 59, 799-814.
- Lindberg, Fredrik, et al. "The impact of urban planning strategies on heat stress in a climate-change perspective." Sustainable Cities and Society 25 (2016): 1-12.
- Gabriel, K. M. A., & Endlicher, W. R. (2011). Urban and rural mortality rates during heat waves in Berlin and Brandenburg, Germany. Environmental Pollution, 159(8–9), 2044–2050.
- Gal, T., Lindberg, F., & Unger, J. (2009). Computing continuous sky view factors using 3D urban raster and vector databases: comparison and application to urban climate. Theoretical and Applied Climatology, 95(1-2), 111-123.
- Google, 2015. Google Maps API (accessed 2015) https://developers.google.com/maps/
- Giridharan, R., Ganesan, S., & Lau, S. S. Y. (2004). Daytime urban heat island effect in high-rise and high-density residential developments in Hong Kong. Energy and Buildings, 36(6), 525-534.
- Gronlund, C. J. (2014). Racial and socioeconomic disparities in heat-related health effects and their mechanisms: a review. Current epidemiology reports, 1(3), 165-173.
- Huang, K. T., Lin, T. P., & Lien, H. C. (2015). Investigating Thermal Comfort and User Behaviors in Outdoor Spaces: A Seasonal and Spatial Perspective. Advances in Meteorology, 2015.
- Hoehne, C. G., Hondula, D. M., Chester, M. V., Eisenman, D. P., Middel, A., Fraser, A. M., ... & Gerster, K. (2018). Heat exposure during outdoor activities in the US varies significantly by city, demography, and activity. Health & Place, 54, 1-10.
- Hoffman, J. S., Shandas, V., & Pendleton, N. (2020). The effects of historical housing policies on resident exposure to intra-urban heat: a study of 108 US urban areas. Climate, 8(1), 12.
- Hwang, R. L., Lin, T. P., & Matzarakis, A. (2011). Seasonal effects of urban street shading on long-term outdoor thermal comfort. Building and Environment, 46(4), 863-870.
- Iungman, T., Cirach, M., Marando, F., Barboza, E. P., Khomenko, S., Masselot, P., ... & Nieuwenhuijsen, M. (2023). Cooling cities through urban green infrastructure: a health impact assessment of European cities. The Lancet, 401(10376), 577-589.
- Johansson, E. (2006). Influence of urban geometry on outdoor thermal comfort in a hot dry climate: a study in Fez, Morocco. Building and Environment, 41(10), 1326-1338.
- Johnson, G. T., & Watson, I. D. (1984). The determination of view-factors in urban canyons. Journal of Climate and Applied Meteorology, 23(2), 329-335.
- Krayenhoff, E. Scott, Mohamed Moustaoui, Ashley M. Broadbent, Vishesh Gupta, and Matei Georgescu. "Diurnal interaction between urban expansion, climate change and adaptation in US cities." Nature Climate Change 8, no. 12 (2018): 1097.
- Klemm, W., Heusinkveld, B. G., Lenzholzer, S., & van Hove, B. (2015). Street greenery and its physical and psychological impact on thermal comfort. Landscape and Urban Planning, 138, 87-98.
- Lee, H., Holst, J., & Mayer, H. (2013). Modification of human-biometeorologically significant radiant flux densities by shading as local method to mitigate heat stress in summer within urban street canyons. Advances in Meteorology, 2013.
- Li, X., Zhang, C., Li, W., Kuzovkina, Y. A., & Weiner, D. (2015). Who lives in greener neighborhoods? The distribution of street greenery and its association with residents’ socioeconomic conditions in Hartford, Connecticut, USA. Urban Forestry & Urban Greening, 14(4), 751-759.
- Li, X., Zhang, C., Li, W., & Kuzovkina, Y. A. (2016). Environmental inequities in terms of different types of urban greenery in Hartford, Connecticut. Urban Forestry & Urban Greening.
- Li, X. (2021). Investigating the spatial distribution of resident’s outdoor heat exposure across neighborhoods of Philadelphia, Pennsylvania using urban microclimate modeling. Sustainable Cities and Society, 103066.
- Li, X., & Wang, G. (2021). GPU parallel computing for mapping urban outdoor heat exposure. Theoretical and Applied Climatology, 1-11.
- Lindberg F, Grimmond CSB, 2011: The influence of vegetation and building morphology on shadow patterns and mean radiant temperature in urban areas: model development and evaluation. Theoretical and Applied Climatology. 105(3), s. 311-323.
- Lindberg, F., Onomura, S. and Grimmond, C.S.B (2016) Influence of ground surface characteristics on the mean radiant temperature in urban areas. International Journal of Biometeorology. 60(9), 1439-1452.
- Lin, T. P., Tsai, K. T., Hwang, R. L., & Matzarakis, A. (2012). Quantification of the effect of thermal indices and sky view factor on park attendance. Landscape and Urban Planning, 107(2), 137-146.
- Lin, T. P., Matzarakis, A., & Hwang, R. L. (2010). Shading effect on long-term outdoor thermal comfort. Building and Environment, 45(1), 213-221.
- Lo, Y. E., Mitchell, D. M., Gasparrini, A., Vicedo-Cabrera, A. M., Ebi, K. L., Frumhoff, P. C., ... & Uhe, P. (2019). Increasing mitigation ambition to meet the Paris Agreement’s temperature goal avoids substantial heat-related mortality in US cities. Science Advances, 5(6), eaau4373.
- Masselot, P., Mistry, M., Vanoli, J., Schneider, R., Iungman, T., Garcia-Leon, D., ... & Aunan, K. (2023). Excess mortality attributed to heat and cold: a health impact assessment study in 854 cities in Europe. The Lancet Planetary Health, 7(4), e271-e281.
- Mitchell, B. C., & Chakraborty, J. (2015). Landscapes of thermal inequity: disproportionate exposure to urban heat in the three largest US cities. Environmental Research Letters, 10(11), 115005.
- Norton, B. A., Coutts, A. M., Livesley, S. J., Harris, R. J., Hunter, A. M., & Williams, N. S. (2015). Planning for cooler cities: A framework to prioritise green infrastructure to mitigate high temperatures in urban landscapes. Landscape and Urban Planning, 134, 127-138.
- Oke, T. R. (1981). Canyon geometry and the nocturnal urban heat island: comparison of scale model and field observations. Journal of Climatology, 1(3), 237-254.
- Onishi, A., Cao, X., Ito, T., Shi, F., & Imura, H. (2010). Evaluating the potential for urban heat-island mitigation by greening parking lots. Urban Forestry & Urban Greening, 9: 323-332.
- Park, S., Tuller, S. E., & Jo, M. (2014). Application of Universal Thermal Climate Index (UTCI) for microclimatic analysis in urban thermal environments. Landscape and Urban Planning, 125, 146-155.
- Raciti, S. M., Hutyra, L. R., & Newell, J. D. (2014). Mapping carbon storage in urban trees with multi-source remote sensing data: Relationships between biomass, land use, and demographics in Boston neighborhoods. Science of The Total Environment, 500, 72-83.
- Rahman, M. A., Smith, J. G., Stringer, P., & Ennos, A. R. (2011). Effect of rooting conditions on the growth and cooling ability of Pyrus calleryana. Urban Forestry and Urban Greening, 10(3), 185–192.
- Ratti, C., & Richens, P. (2004). Raster analysis of urban form. Environment and Planning B: Planning and Design, 31(2), 297-309.
- Ribeiro, A., Fernández-Quintanilla, C., Barroso, J., García-Alegre, M.C. (2005). Development of an image analysis system for estimation of weed. In: Proceedings of the 5th European Conference On Precision Agriculture (5ECPA), pp. 169–174.
- Shashua-Bar, L., & Hoffman, M. E. (2000). Vegetation as a climatic component in the design of an urban street: An empirical model for predicting the cooling effect of urban green areas with trees. Energy and Buildings, 31(3), 221-235.
- Stone, B., Hess, J. J., & Frumkin, H. (2010). Urban form and extreme heat events: are sprawling cities more vulnerable to climate change than compact cities? Environmental health Perspectives, 118(10), 1425-1428.
- Spangenberg, J., Shinzato, P., Johansson, E., & Duarte, D. (2008). Simulation of the influence of vegetation on microclimate and thermal comfort in the city of São Paulo. Revista da Sociedade Brasileira de Arborização Urbana, 3, 1-19.
- Steyn, D. G. (1980). The calculation of view factors from fisheye lens photographs, Atmosphere-Ocean 18: 254-258.
- Tan, Z., Lau, K. K. L., & Ng, E. (2016). Urban tree design approaches for mitigating daytime urban heat island effects in a high-density urban environment. Energy and Buildings, 114, 265-274.
- Theodosiou, T. G. (2003). Summer period analysis of the performance of a planted roof as a passive cooling technique. Energy and Buildings, 35(9), 909-917.
- Thorsson, S., Rayner, D., Lindberg, F., Monteiro, A., Katzschner, L., Lau, K. K. L., ... & Velho, S. (2017). Present and projected future mean radiant temperature for three European cities. International Journal of Biometeorology, 61(9), 1531-1543.
- Venter, Z. S., Chakraborty, T., & Lee, X. (2021). Crowdsourced air temperatures contrast satellite measures of the urban heat island and its mechanisms. Science Advances, 7(22), eabb9569.
- Yin, Y., He, L., Wennberg, P. O., & Frankenberg, C. (2023). Unequal exposure to heatwaves in Los Angeles: Impact of uneven green spaces. Science Advances, 9(17), eade8501.
- Watson, I. D., & Johnson, G. T. (1987). Graphical estimation of sky view‐factors in urban environments. Journal of Climatology, 7(2), 193-197.
- Xu, C., Kohler, T. A., Lenton, T. M., Svenning, J. C., & Scheffer, M. (2020). Future of the human climate niche. Proceedings of the National Academy of Sciences, 117(21), 11350-11355.
- Yang, J., Hu, L., & Wang, C. (2019). Population dynamics modify urban residents’ exposure to extreme temperatures across the United States. Science Advances, 5(12), eaay3452.
- Young, E., Kastner, P., Dogan, T., Chokhachian, A., Mokhtar, S., & Reinhart, C. (2021). Modeling outdoor thermal comfort along cycling routes at varying levels of physical accuracy to predict bike ridership in Cambridge, MA. Building and Environment, 108577.
There is NO Competing Interest.