patrickbdevaney/mia9 · Datasets at Hugging Face

text stringlengths 0 6.44k
exposure within the urban region. One alternative approach,
recently advanced by Miami–Dade County, is to talk about a
“heat season” that could draw from the lessons of “hurricane
season” (Harris 2021). Thus, rather than focusing on specific
events, there would be an increased level of public and professional awareness and associated resources available in summer
months (Kotcher et al. 2021; Salas 2021), when the area is under persistent high heat exposure. This would be consistent
with our findings that levels of heat exposure that have been
deemed dangerous (such as the 1088F threshold) occurs in locations within the urbanized area of Miami–Dade County for a
significant part of the year (Fig. 4b) while avoiding the concerns
around “warning fatigue.”
Naturally, these results lead to the question of impacts. The
current NWS approach is tied to health impacts by examining
the rate of heat related emergency room admissions in central
Florida (Florida Department of Health 2011). However, this
approach misses the potential granularity of heat-health impacts since health data that have been used is most often at the
county scale (Anderson and Bell 2011; McElroy et al. 2020).
Even changing to a single county-specific heat index threshold,
instead of a statewide one, may miss varying exposures and
impacts, since we have shown considerable variations in
heat index within the county (Fig. 3). There are also disproportionate risks to different groups such as the elderly (Kenney
and Munce 2003; Semenza et al. 1999), pregnant women
(Zhang et al. 2017), people with preexisting conditions (Leon
and Bouchama 2011), outdoor workers (Uejio et al. 2018), and
members of low-income communities who often cannot afford
adequate air conditioning (Dahl et al. 2019). This suggests that
disaggregation of both health and heat exposure data may be
important to identifying impacts. Further, there are other heatrelated impacts such as productivity, that may not be easy to tie
to a single threshold given the varying exposures throughout
the county, since they will depend on the length of time of exposure and physical activity (Uejio et al. 2016; Masuda et al.
2019; Vanos et al. 2020; Oppermann et al. 2021). This may be
even more challenging in Miami where high temperature and
humidity, producing elevated heat indices, persist throughout a
significant part of the year, generally from May to October. These
chronic heat conditions can lead to impacts that may be difficult
to detect when judged by a correlation with extremes (Bolitho
and Miller 2017; Casanueva et al. 2019; Oppermann et al. 2021;
FIG. 3. Scatterplots of daily iButton vs MIA (Airport) (a) minimum temperature (Tmin), (b) maximum temperature (Tmax), and
(c) maximum heat index (HImax). For (a) and (b), red points represent 65 iButtons measuring temperature in locations determined
by participants to be sunny (open canopy); blue points represent 54
iButtons measuring temperature in locations determined by participants to be shaded (treed). For (c), red points represent 6 iButtons
measuring relative humidity in addition to temperature in sunny
locations; blue points represent 4 iButtons measuring relative humidity and temperature in shaded locations. The line of best fit based on
linear regression with 95% confidence (light shading) is calculated
for the sunny and shaded values and is shown in the iButton environment’s respective color. Each regression line’s R-squared value is
given in the legend. The gray line is a 1-to-1 reference line.
868 JOURNAL OF APPLIED METEOROLOGY AND CLIMATOLOGY VOLUME 62
Unauthenticated \| Downloaded 02/29/24 07:14 AM UTC
Strathearn et al. 2022) and may require new ways of relating
heat and health data (Hondula et al. 2021, 2022). These
questions clearly require a more complete analysis of where
and how different groups are being impacted by heat to design more locally relevant and appropriate policy actions
(Turek-Hankins et al. 2020), as recommended in earlier
Florida Department of Health (2011) reports. This kind of
analysis should also be paired with heat exposure data that
are closer to the experience of vulnerable groups.
The question of what dangerous levels of heat occur
within urban communities is a pressing one. Our data suggest that the heat index is higher and more persistent in many
parts of Miami–Dade County than is measured at the Miami
NWS station and can place many parts of the county well
above a single heat index threshold when accounting for local
conditions. Local conditions can introduce significant variability in heat exposure, which is important in attributing health
and other impacts that may be missed at the scale of a county
jurisdiction. Monitoring heat exposure at more locally relevant
scales, ideally in real time, is an important part of how cities can
plan proactively for current and future heat impacts (McCormick
2021), and to assess whether management interventions are
working as intended. We argue that the best way forward is a
multiplatform collection of sensors collecting at multiple spatial
and temporal scales for a more integrated and relevant reporting
of heat exposure, including stationary, mobile, and remote
sensing coupled with NWS monitoring data as has been
identified in previous heat studies (e.g., Shi et al. 2021).
There are also arguments to be made for including indoor
temperature measurements in order to understand heat exposure in the home, which can differ considerably from
outdoor temperatures (Uejio et al. 2016, 2018). Hyperlocal
data are an important part of this monitoring and have the
additional opportunity of directly connecting to community members who are affected by and should be part of
the discussion of mitigating heat exposure in urban
communities.
Acknowledgments. We acknowledge SLSC volunteer
Bertha Goldenberg, citizen scientists, Catalyst Miami, City
of Miami, Miami–Dade County, City of Miami Beach,
Kresge Foundation, Miami’s Urban Heat Research Group,
Jane Gilbert, Lynee Turek-Hankins, Kenny Broad, and
NOAA’s Climate and Equity Roundtables (https://www.
noaa.gov/regional-collaboration-network/noaas-climate-andequity-roundtables). We are grateful to Robert Molleda for
information he shared from the NWS Miami office. Partial
funding support for this work came from the University of
Miami Laboratory for Integrated Knowledge and the NOAA
Adaptation Science program.
FIG. 4. The number of days per year in which the heat index, calculated (a) using data at MIA
and (b) using modeled iButton data, reached or exceeded 1028, 1048, 1068, and 1088F (38.98, 408,
41.18 and 42.28C, respectively) from 1950 to 2020. The modeled iButton data for heat index were
based on the regression line from the iButton temperature scatterplot slopes.
JULY 2023 CLEMENT ET AL. 869
Unauthenticated \| Downloaded 02/29/24 07:14 AM UTC
Data availability statement. NWS data are publicly available
online (https://www.ncei.noaa.gov/cdo-web/). The hyperlocal
observations shown in this paper are available at our website
(https://amyclement.weebly.com/data.html).

exposure within the urban region. One alternative approach,

recently advanced by Miami–Dade County, is to talk about a

“heat season” that could draw from the lessons of “hurricane

season” (Harris 2021). Thus, rather than focusing on specific

events, there would be an increased level of public and professional awareness and associated resources available in summer

months (Kotcher et al. 2021; Salas 2021), when the area is under persistent high heat exposure. This would be consistent

with our findings that levels of heat exposure that have been

deemed dangerous (such as the 1088F threshold) occurs in locations within the urbanized area of Miami–Dade County for a

significant part of the year (Fig. 4b) while avoiding the concerns

around “warning fatigue.”

Naturally, these results lead to the question of impacts. The

current NWS approach is tied to health impacts by examining

the rate of heat related emergency room admissions in central

Florida (Florida Department of Health 2011). However, this

approach misses the potential granularity of heat-health impacts since health data that have been used is most often at the

county scale (Anderson and Bell 2011; McElroy et al. 2020).

Even changing to a single county-specific heat index threshold,

instead of a statewide one, may miss varying exposures and

impacts, since we have shown considerable variations in

heat index within the county (Fig. 3). There are also disproportionate risks to different groups such as the elderly (Kenney

and Munce 2003; Semenza et al. 1999), pregnant women

(Zhang et al. 2017), people with preexisting conditions (Leon

and Bouchama 2011), outdoor workers (Uejio et al. 2018), and

members of low-income communities who often cannot afford

adequate air conditioning (Dahl et al. 2019). This suggests that

disaggregation of both health and heat exposure data may be

important to identifying impacts. Further, there are other heatrelated impacts such as productivity, that may not be easy to tie

to a single threshold given the varying exposures throughout

the county, since they will depend on the length of time of exposure and physical activity (Uejio et al. 2016; Masuda et al.

2019; Vanos et al. 2020; Oppermann et al. 2021). This may be

even more challenging in Miami where high temperature and

humidity, producing elevated heat indices, persist throughout a

significant part of the year, generally from May to October. These

chronic heat conditions can lead to impacts that may be difficult

to detect when judged by a correlation with extremes (Bolitho

and Miller 2017; Casanueva et al. 2019; Oppermann et al. 2021;

FIG. 3. Scatterplots of daily iButton vs MIA (Airport) (a) minimum temperature (Tmin), (b) maximum temperature (Tmax), and

(c) maximum heat index (HImax). For (a) and (b), red points represent 65 iButtons measuring temperature in locations determined

by participants to be sunny (open canopy); blue points represent 54

iButtons measuring temperature in locations determined by participants to be shaded (treed). For (c), red points represent 6 iButtons

measuring relative humidity in addition to temperature in sunny

locations; blue points represent 4 iButtons measuring relative humidity and temperature in shaded locations. The line of best fit based on

linear regression with 95% confidence (light shading) is calculated

for the sunny and shaded values and is shown in the iButton environment’s respective color. Each regression line’s R-squared value is

given in the legend. The gray line is a 1-to-1 reference line.

868 JOURNAL OF APPLIED METEOROLOGY AND CLIMATOLOGY VOLUME 62

Unauthenticated | Downloaded 02/29/24 07:14 AM UTC

Strathearn et al. 2022) and may require new ways of relating

heat and health data (Hondula et al. 2021, 2022). These

questions clearly require a more complete analysis of where

and how different groups are being impacted by heat to design more locally relevant and appropriate policy actions

(Turek-Hankins et al. 2020), as recommended in earlier

Florida Department of Health (2011) reports. This kind of

analysis should also be paired with heat exposure data that

are closer to the experience of vulnerable groups.

The question of what dangerous levels of heat occur

within urban communities is a pressing one. Our data suggest that the heat index is higher and more persistent in many

parts of Miami–Dade County than is measured at the Miami

NWS station and can place many parts of the county well

above a single heat index threshold when accounting for local

conditions. Local conditions can introduce significant variability in heat exposure, which is important in attributing health

and other impacts that may be missed at the scale of a county

jurisdiction. Monitoring heat exposure at more locally relevant

scales, ideally in real time, is an important part of how cities can

plan proactively for current and future heat impacts (McCormick

2021), and to assess whether management interventions are

working as intended. We argue that the best way forward is a

multiplatform collection of sensors collecting at multiple spatial

and temporal scales for a more integrated and relevant reporting

of heat exposure, including stationary, mobile, and remote

sensing coupled with NWS monitoring data as has been

identified in previous heat studies (e.g., Shi et al. 2021).

There are also arguments to be made for including indoor

temperature measurements in order to understand heat exposure in the home, which can differ considerably from

outdoor temperatures (Uejio et al. 2016, 2018). Hyperlocal

data are an important part of this monitoring and have the

additional opportunity of directly connecting to community members who are affected by and should be part of

the discussion of mitigating heat exposure in urban

communities.

Acknowledgments. We acknowledge SLSC volunteer

Bertha Goldenberg, citizen scientists, Catalyst Miami, City

of Miami, Miami–Dade County, City of Miami Beach,

Kresge Foundation, Miami’s Urban Heat Research Group,

Jane Gilbert, Lynee Turek-Hankins, Kenny Broad, and

NOAA’s Climate and Equity Roundtables (https://www.

noaa.gov/regional-collaboration-network/noaas-climate-andequity-roundtables). We are grateful to Robert Molleda for

information he shared from the NWS Miami office. Partial

funding support for this work came from the University of

Miami Laboratory for Integrated Knowledge and the NOAA

Adaptation Science program.

FIG. 4. The number of days per year in which the heat index, calculated (a) using data at MIA

and (b) using modeled iButton data, reached or exceeded 1028, 1048, 1068, and 1088F (38.98, 408,

41.18 and 42.28C, respectively) from 1950 to 2020. The modeled iButton data for heat index were

based on the regression line from the iButton temperature scatterplot slopes.

JULY 2023 CLEMENT ET AL. 869

Unauthenticated | Downloaded 02/29/24 07:14 AM UTC

Data availability statement. NWS data are publicly available

online (https://www.ncei.noaa.gov/cdo-web/). The hyperlocal

observations shown in this paper are available at our website

(https://amyclement.weebly.com/data.html).