SEPTEMBER 2012 V A N O S E T A L . 1639

Human Energy Budget Modeling in Urban Parks in and Applications to Emergency Heat Stress Preparedness

JENNIFER K. VANOS,JON S. WARLAND, AND TERRY J. GILLESPIE School of Environmental Sciences, University of Guelph, Guelph, , Canada

GRAHAM A. SLATER AND ROBERT D. BROWN School of Environmental Design, University of Guelph, Guelph, Ontario, Canada

NATASHA A. KENNY Teaching Support Services, University of Guelph, Guelph, Ontario, Canada

(Manuscript received 29 November 2011, in final form 13 March 2012)

ABSTRACT

The current study tests applications of the Comfort Formula (COMFA) energy budget model by assessing the moderating effects of urban parks in contrast to streets, and it also looks at the influence of park types (‘‘open’’ or ‘‘treed’’). Exploration into energy budget modeling is based on empirical meteorological data collected in Toronto, Ontario, Canada, on fair-weather days plus the effects of a heat wave and climate change, at various metabolic activity levels. Park cooling temperature intensities ranged from 3.98 to 6.08C, yet human energy budgets were more closely correlated to incoming solar radiation than to air temperature. A strong linear dependence was found, with absorbed radiation (correlation coefficient squared r2 5 0.858) explaining the largest fraction of energy budget output. Hence, although the four parks that were examined are classified as urban green space, the distinctive treed areas showed a greater budget decrease than did open 2 park areas (225.5 W m 2). The greatest difference in budget decrease was found when modeling the highest 2 2 2 metabolic rate, giving 220 W m 2 for ‘‘whole park,’’ 232 W m 2 for treed sections, and 23Wm 2 in open park areas. These results are intuitive within energy budget modeling and indicate that blocking radiant energy is a vital aspect in lowering high budgets under the conditions tested. Strong empirical support was provided through successful prediction of emergency-response calls during a heat wave in Toronto (5–7 July 2010) and surrounding days. Calls were found to be significantly dependent on the energy budget estimations (r2 5 0.860). There is great potential for outdoor energy budget modeling as a meaningful guide to heat stress forecasting, future research, and application in bioclimatic urban design for improving thermal comfort.

1. Introduction 2007), fewer heat-related casualties, and a decrease in rates of obesity and cardiovascular disease as a result of Evidence of the benefits of a comfortable microcli- exercise (Galea and Vlahov 2005; Brotherhood 2008). mate within parks is extensive and has been correlated Many human health indicators are more strongly related with increased use of outdoor space (Mayer et al. 2008). to the amount of green space than to the degree of ur- Lower heat stress, coupled with increased frequency and banity (deVries et al. 2003), with vegetation found to be duration of exercise in parks, also results from improved significantly correlated with TC in urban areas such as thermal comfort (TC) (Watkins et al. 2007; Gaitani et al. Phoenix, Arizona (Harlan et al. 2006). 2007). Such benefits are transferable into higher quality Vital changes such as increases in vegetative shad- of life and functionality (Eliasson 2000; Watkins et al. ing, albedos of ground and building surfaces, presence of water, and alternating building heights can reduce

the air temperature Ta in an urban climate (Eliasson Corresponding author address: Jennifer K. Vanos, School of Environmental Sciences, University of Guelph, 50 Stone Rd. E., et al. 2007; Watkins et al. 2007). According to Lin et al. Guelph, ON N1G 2W1, Canada. (2010), shading is an important component of long- and E-mail: [email protected] short-term outdoor TC, specifically within urban canyons

DOI: 10.1175/JAMC-D-11-0245.1

Ó 2012 American Meteorological Society Unauthenticated | Downloaded 09/27/21 09:30 AM UTC 1640 JOURNAL OF APPLIED METEOROLOGY AND CLIMATOLOGY VOLUME 51

(Ali-Toudert and Mayer 2007). Hence, energy budgets moderating effect of urban parks on outdoor TC and and potential instances of heat stress are spatially vari- under various human physiological states. Various tan- able, with effects of population characteristics, vegetation gible empirical applications are presented and proposed density, socioeconomic factors, and coping resources through the use of energy budget modeling. This is com- confounding our understanding and prediction of heat pleted through accomplishment of the following: 1) as- stress (deVries et al. 2003; Harlan et al. 2006). sessment of the TC-moderating effects in urban parks According to a review completed by McGregor using an energy budget approach while evaluating the (2012), human TC models have illustrated greater de- impacts of gender, activity intensity, and conditioning velopmental advancements than their empirical coun- level on TS; 2) provision of an example of heat stress terparts, which is reflective of the complex physiological modeling that displays the impacts of current and fu- processes responding to a variety of human- and climate- ture extreme-heat events on those performing physical based factors. Very few studies in human biometeorology activity in an urban park; and 3) empirical evaluation of have been completed on exercising individuals in realistic energy budget modeling to extreme-heat occurrences, field settings using physiological heat balance models. demonstrating the use of energy budget modeling as

Thermal balance is influenced by metabolic activity Mact, a tool for predicting heat-related emergency-response micrometeorological conditions, clothing properties, and calls. psychological perceptions. Overall thermal sensation (TS), This paper is not intended to study the UHI or park often expressed as TC, is influenced by personal differ- cool intensity (PCI) effects in detail, as the spatial scope ences in mood, culture, social factors (Djongyang et al. and temporal scope of the data are limited. It is intended 2010), expectations, perceived control, time of exposure, to show that human comfort at any point in a built or fitness, aesthetics, season, weather, perception, and read- park environment can be quantified for urban design and iness to exercise (Thorsson et al. 2004; Vanos et al. 2012a). management using the COMFA energy budget model. Hence, studies have been completed to note overall adaptive processes: behavioral, physiological, and psy- chological (de Dear and Brager 1998; Fanger and Toftum 2. Methods 2002; Yao et al. 2009). Human adaptive strategies to a. Site selection and characteristics climate change—in particular, in urban areas—is a principal focus in current research (Smoyer et al. 2000; Toronto, Ontario, Canada has a population of ap- Nikolopoulou and Lykoudis 2006; Golden et al. 2008; proximately 4.28 million people in dense occupation, Matzarakis and Endler 2010), with an expected in- with a heterogeneous mix of modern core built-up areas crease in magnitude and frequency of heat waves as and compact housing in residential and suburban areas climate change advances (Gosling et al. 2007; O’Neill (Stewart and Oke 2009). The city experiences a humid and Ebi 2009). continental climate (Peel et al. 2007), having warm, hu- Heat is suggested to remain the most important extreme- mid summers with cold winters. Centered in the warmest weather-related killer in the United States (Sheridan et al. climatic zone in Canada, Toronto has experienced above- 2009). During intense heat waves, individuals may not average temperatures in six of the seven summers be- be able to escape the heat, with urban heat islands tween 2004 and 2011 (Environment Canada 2010), is (UHI) diminishing nighttime cooling and thus heat subjected to UHI effects (Mohsin and Gough 2010), and relief. Particular importance should be placed on pro- is predicted to have doubled heat-related mortality by tecting vulnerable individuals—the elderly (.65 yr; 2050 (Cheng et al. 2005). Lemmen et al. (2008) antici- Luber and McGeehin 2008), young, sick, unacclimatized, pate growing public-health risk to heat, reporting a high and unconditioned—who may have compromised ther- likelihood that Toronto will face extreme-heat events moregulation (Smoyer et al. 2000). This vulnerability is of heightened intensity, duration, and frequency. increased in poor populations that do not have adequate Table 1 lists the four urban-park study sites, as well resources (Curriero et al. 2002; Harlan et al. 2006), such as as the nearest corresponding weather station to each air conditioning (Luber and McGeehin 2008). Havenith park that is used for wind speed vectors. The four parks (1997) gives evidence that cardiovascular fitness is a are found within 1.5–2.8 km of the urban core, as dis- more important determinant of heat mortality than age, played in the GIS map of Toronto in Fig. 1. The parks however. did not contain ponds and were surrounded by similar The purpose of the current study is to test the appli- urban form of high-density housing in older neighbor- cation of the Comfort Formula (COMFA) model to hoods with narrow streets (7–7.25 m), and 2–3-story practical urban design issues using an existing dataset detached/semidetached houses set back 6–9 m (Slater from Slater (2010). This study will spatially assess the 2010). Daily population densities (people per square

Unauthenticated | Downloaded 09/27/21 09:30 AM UTC SEPTEMBER 2012 V A N O S E T A L . 1641

TABLE 1. Characteristics of transects, surrounding neighborhoods, and corresponding weather stations of the four urban parks near downtown Toronto.

Bordering height:width Transect Corresponding Park (area) coordinates ratio* length* (m) Park length (m) weather station** Withrow (8.5 ha) 438409300,798209490 0.32 1310 192 Leslieville Trinity Bellwoods (15.1 ha) 4383893520,798249510 0.38 922 646 Toronto Island Dovercourt (2.4 ha) 438399550,798269010 0.35 961 126 Davis (5.8 ha) 438399220,798209560 0.35 928 230 Davis

* Mean values. ** Nearest weather station at similar altitude to corresponding outdoor park. kilometer) range from 12 000 to 47 000 for all parks b. Microclimate data collection and modeling excluding Trinity Bellwoods, which has a density of 86 000 to 228 000 (Glazier et al. 2007, 119–129). These Four urban parks within Toronto, Ontario, were ranges are relatively wide because of daily migration of chosen on the basis of the ability to complete a relatively people to and from activities such as work, school, straight transect line through the park from the streets business, and recreation. All four parks have been pre- and hence not to alter the wind direction effects on the served and thus contain mainly deciduous and co- instruments significantly. Figure 2 displays an aerial niferous tree species that are typical to the area, which view of Dovercourt Park as an example transect taken include various mature species of ash (Fraxinus), oak through each park. (Quercus), maple (Acer), elm (Ulmus), and pine trees Instruments monitoring microclimatic parameters (Pinus). were securely fastened at a height of 1.5 m to the front of

FIG. 1. GIS map of Toronto, displaying the locations of the four chosen urban parks in black: (label A), (label B), Dovercourt Park (label C), and (label D). Urban green space is shown in gray; the open black rectangle represents the urban core. Source: Urban-parks map from City of Toronto (2011).

Unauthenticated | Downloaded 09/27/21 09:30 AM UTC 1642 JOURNAL OF APPLIED METEOROLOGY AND CLIMATOLOGY VOLUME 51

FIG. 2. A September 2009 aerial view of Dovercourt Park (43.678N, 79.438W, altitude 5 112 m), displaying an enlarged portion of the transect (white line), as was completed in the four urban-parks sites used in the current study (source: Ó 2012 Google Earth, v. 6.1.0.5).

21 a bicycle, which moved with a speed of ;9kmh along shortwave radiation Kin from the atmosphere, buildings, a street–park–street transect. Microclimate data were and trees was measured using a lightweight portable LI- collected along the transects in September of 2009 at 200 pyranometer (Li-Cor, Inc.). Data were collected times between 1300 and 1630 LT on sunny days (see at 0.8-s intervals using a 21X datalogger (Campbell Table 2). These data define the meteorological environ- Scientific, Inc.), with all instruments mounted on a bi- ment in which the human activity takes place, which is cycle (Manoeuvre model from Diamondback Bicycles required for input to the model. Co.), at 1.5-m height. A quick-release apparatus was

Air temperature Ta was measured using a fine-diameter used for consistent and timely mounting setup and for 2 (0.02 mm) copper–constantan (Type ‘‘T’’) thermocouple easy portability. Speed (m s 1) and distance (m) were (Omega Engineering, Inc.). Relative humidity RH was measured using two wired computer sensors (dZ4L measured using an HC-S3 temperature and RH sensor model from Filzer Co.). Transects consisted of two or probe (Rotronic Instrument Corp.), with the metal mesh four passes each, in which one pass of each pair was re- cap removed for improved response time. Total incoming versed for averaging of all data at an observation point.

TABLE 2. Summary of meteorological conditions recorded on-site and from corresponding weather stations expressed as a mean over 21 each sampling period (LT) in September 2009: Ta (8C), e (kPa), yw (m s ; compass-point wind direction is in parentheses), and t (sky transmissivity; Kenny et al. 2008). Sampling periods consisted of two or four transect passes. [Averaged wind data for all transect passes were obtained from the nearby weather station (10-m height)].

Park Date Sampling period No. of transect passes Ta e yw t (dir) Withrow 6 Sep 1555–1615 2 22.0 1.5 4.0 (E) 0.80 7 Sep 1435–1636 4 23.1 1.7 5.8 (E) 0.66 8 Sep 1330–1450 4 23.9 2.0 4.4 (E) 0.72 9 Sep 1440–1600 4 22.9 1.9 4.0 (E) 0.79 20 Sep 1540–1600 2 19.7 1.2 8.8 (E) 0.87 Dovercourt 4 Sep 1500–1520 2 23.9 1.3 9.3 (S) 0.91 15 Sep 1525–1615 2 24.0 1.0 6.7 (N) 0.78 Trinity Bellwoods 4 Sep 1545–1615 2 23.4 1.4 6.9 (S) 0.91 15 Sep 1430–1500 2 24.5 1.0 6.7 (N) 0.78 Dufferin Grove 15 Sep 1555–1610 2 23.8 1.0 6.7 (N) 0.78

Unauthenticated | Downloaded 09/27/21 09:30 AM UTC SEPTEMBER 2012 V A N O S E T A L . 1643

Moving averages of budget values were plotted, but 2012a). Most-recent improvements involve more ac- point-averaged meteorological data and budget compo- curate and efficient predictions of human physiological nents were used for analysis. Averaging-distance windows effects and behavior (Vanos et al. 2012b). These im- increased with activity speed, yielding walking, running, provements are adaptation of Icl with respect to Mact and cycling windows of 6, 25, and 140 m, respectively. and Ta, incorporation of individual conditioning level Intensities of air temperatures, budget outputs, and radi- (discussed below), and corrections for wind angle to ation levels between street and park were determined us- body movement (Vanos et al. 2012b). ing nonaveraged data of the maximum on the urban street The intrinsic Icl (in clo units, defined below), referring (streetmax) minus the minimum within the park (parkmin). to the whole body, was estimated from ambient Ta using Kenny et al. (2008) provide a method for reliable Eq. (2), shown below (Havenith et al. 2011). Bare and absorbed radiation Rabs estimates for a human on the clothed body fractions were accounted for as in Tikuisis basis of outdoor horizontal radiation fluxes that has et al. (2001) and Vanos et al. (2012a,b), as were accurate been successfully applied in subsequent studies (Vanos Icl values for each body segment, obtained from ISO et al. 2012a,b). For estimation of outgoing shortwave (2007). Equation (2) was originally designed for use with 22 radiation, Kin was multiplied by ground albedo agr. Al- metabolic rates of 175 W m . To account for clothing bedo agr was estimated with respect to the surfaces over adjustments when a user experiences a higher Mact (greater which the transect passed. Wide concrete/asphalt sur- endogenous heat load), the reference Icl value from Eq. faces and paths on streets and within parks were given an (2) was shifted according to the change in metabolic rate albedo of 0.12, grass surfaces were estimated as 0.25, and (Vanos et al. 2012b). This was done by computing on the locations at which both surfaces would influence short- basis of Mact an ‘‘equivalent Ta’’ value for use in Eq. (2). wave radiation reflected toward a human were esti- This equivalent temperature is higher than the actual Ta mated as a midrange value of 0.18. and therefore lowers the Icl value with increasing Mact at Wind velocity measurements and directions were any actual Ta. Equation (2) is given by obtained from the weather station that was nearest to the corresponding outdoor park. Wind speed at the rep- 5 : 2 : 2 : 2 Icl 1 372 0 018 66Ta 0 000 484 9Ta resentative human environment height of 1.5 m (Souch 2 : 3 and Souch 1993) was found by using a ratio of the loga- 0 000 009 333Ta , (2) rithmic wind profile as follows:

where Ta is in degrees Celsius and and Icl is in clo,   2 which is an arbitrary unit (1 clo 5 186.6 s m 1 5 z2 ln 2 8 21 21 y z 0.1555 m C W ). 2 5  2o, (1) y z Average Mact was quantified using a detailed energy 1 ln 1 expenditure estimation method by Strath et al. (2000) z 1o and is discussed further in Vanos et al. (2012b) as well

as within the current paper. The Mact levels for energy y y y where 2 and 1 are the wind speeds w and z2 and z1 are budget input are based on maximum volume of oxygen the heights z at 10 and 1.5 m, respectively. Also, z2o and intake levels VO2max for individuals who are unconditioned z1o are the roughness lengths of the given areas, esti- (lower VO2max ) and conditioned (higher VO2max ). The mated as 0.8 m for built-up street areas and 0.4 m when overall activity VO2 (VO2max ), and hence Mact, can be inside parks (Wieringa et al. 2001). determined through knowledge of the level at which a person is working (we assume a level that yields car- c. COMFA model diorespiratory benefits from exercise; Whaley et al.

To test an application of the COMFA model revision 2006), their VO2max , and their energy expenditure at rest 2 (Vanos et al. 2012b) to practical urban design issues, an (which is 1 metabolic equivalent, being 58.15 W m 2). existing dataset from Slater (2010) was employed. The Themetabolicmodelingwasbasedonanestimateof COMFA energy budget model requires meteorological the average age of a person who would be using the inputs (yw, Ta, vapor pressure e, and radiant tempera- park for the chosen activities; hence an age of 25 was ture RRT) and physiological inputs (Mact and clothing chosen. insulation Icl) to produce a human energy budget out- Three activities (walking, running, and cycling) at put in watts per meter squared. It has been extensively three conditioning levels (unconditioned, conditioned, tested and revised under various dynamic outdoor and elite) were employed in the current study. Values conditions for agreement with tested subjects (Brown for Mact and corresponding activity speeds ya are listed and Gillespie 1986; Kenny et al. 2009a,b; Vanos et al. in Table 3. The main controllers of Mact are gender and

Unauthenticated | Downloaded 09/27/21 09:30 AM UTC 1644 JOURNAL OF APPLIED METEOROLOGY AND CLIMATOLOGY VOLUME 51

y TABLE 3. List of activities and corresponding Mact and a that 2009. For scenarios 2 and 3, 28 and 38C were added, re- were used for energy budget prediction. Values for males are listed, spectively, to the scenario-1 Ta transect pattern. with values for equivalently aged females in parentheses. The RH and yw for the latter scenarios were assumed 22 21 Activity Mact (W m ) ya (m s ) to stay constant with average 5 July values, which were 21 Running 63% and 1.2 m s , respectively. It was assumed that Unconditioned 320 (262) 1.5 (1.4) enhanced evaporation would increase both moisture Conditioned 436 (326) 2.1 (1.7) content and temperature of the air, and therefore the Elite 756 (582) 3.5 (3.0) RH remained unchanged from values in July 2010. We Cycling altered only T for the future climate scenarios because Unconditioned 320 (262) 4.5 (4.0) a Conditioned 436 (326) 5.1 (4.5) other accompanying changes are more uncertain and Brisk walking this approach isolates the impact of increased temper- Avg 221 (221) 1.3 (1.3) ature by holding other variables equal. d. Heat stress emergency calls during heat wave

a person’s overall fitness level (VO2max ), which posi- The final objective of the current study involved tively relates to how hard one must work to gain car- evaluating the application of the COMFA energy bud- diorespiratory benefits. The model has been shown to get model as a tool for prediction of heat stress emer- be the most sensitive to gender input when modeling gency response. Therefore, we employ the use of elite runners, causing a budget differential between meteorological data from the 5–7 July 2010 Toronto 2 genders of 65 W m 2 (Vanos et al. 2012b); budget heat wave, as well as heat-related morbidity. Morbid- output and Mact are generally insensitive to changes in ity was quantified using daily dispatch-call data ob- age, however. tained from Toronto Emergency Medical Services (J. Kilch 2011, personal communication) for the dates HEAT-WAVE AND FUTURE-CLIMATE SCENARIOS of 28 June–15 July, which allowed heat stress analysis The methods outlined in this section will address before, during, and after the heat wave. Emergency a heat wave in Toronto from 5 to 7 July 2010, denoted medical services (EMS) ambulance calls are categorized scenario 1. In addition, future climate change pre- as a complaint of symptoms of heat, such as ‘‘light-headed’’ dictions of increased air temperature for the region are or ‘‘poor breathing,’’ rather than as environmental heat modeled, giving DTa of 128C for the year 2040 and stress (J. Kilch 2011, personal communication). There- 138C for 2070 (scenarios 2 and 3, respectively) fore, analysis was completed using the following heat- (CCCma 2011). The COMFA model was applied using related illness categories related to heat stress: breathing the three scenarios for Dovercourt Park for analysis of problems, headache, heat/cold exposure, stroke/heat 5 July in 2010, 2040, and 2070. Measured weather vari- stroke (cerebrovascular accident), unconscious/faint, ables of yw, wind direction yd, Ta, and RH on 5 July 2010 cardiac/respiratory arrest, and chest pain (Luber and were used from the Davis weather station (See Table 1). McGeehin 2008).

Values of Kin above the tree canopy were computed Energy budget modeling was completed for the du- by using a theoretical estimation approach for 5 July ration of the heat wave, as well on surrounding days. 22 2010 (where Kin 5 incoming beam 1 incoming diffuse This was accomplished using an Mact of 221 W m for radiation) (Kenny et al. 2008). The proportion of Kin sedentary–slow walking and clothing insulations that are transmitted through the tree canopies was calculated as based on Ta, as described in section 2c. Using the Davis the difference between estimated Kin (above the trees) weather station, Ta,RH,andyw were collected. These and measured Kin (1.5-m height) in September of 2009. data, along with theoretically estimated radiation data This percentage difference was subtracted from July from 1100 to 1800 eastern standard time, were averaged

2010 Kin estimations at the same time of day. Therefore, into 5-min aggregates and then into a single daytime the proportions of Kin reaching 1.5-m height along the energy budget average. The empirical temperature– transects are equivalent, yet the magnitudes differ with humidity index [‘‘humidex’’ (HX)] was also assessed for time of year. heat stress prediction [HX (8C) 5 Ta 1 5/9(e 2 10), In a similar way, the Ta patterns along the transect at where e is in hectopascals]. Daytime HX and energy 1.5 m also give equivalent variations, but differing mag- budget, as well as daily (24 h) maximum, minimum, nitudes, among the scenarios. In scenario 1, the difference and average air temperatures (Tmax, Tmin,andTavg), of average measured September 2009 Ta along the tran- were tested as predictive tools for heat-related EMS sect from mean heat wave Ta was calculated and added to calls during the heat wave using linear regression each measured Ta point along the transect in September analysis.

Unauthenticated | Downloaded 09/27/21 09:30 AM UTC SEPTEMBER 2012 V A N O S E T A L . 1645

Then, the prediction of total EMS calls was calculated TABLE 4. Descriptive statistics of energy budget values 22 using a multiple linear equation [Eq. (3), shown below], (W m ) 6 SD along transects. Select averages are shown for each developed specifically for the city of Toronto by Dolney study site, as well as results for distinctive experimental weather conditions and physiological states. Here, n indicates the number of and Sheridan (2006) on the basis of total EMS calls for data points. a 4-yr period (1999–2002). Therefore, this study imple- ments this equation during the heat wave to test the Total Park Park Park park (treed) (open) Street agreement and validity of this equation from the liter- ature, using the total EMS calls, rather than only heat- Dovercourt (n 5 927) 97 6 8 109 6 12 6 6 related calls, as utilized in the rest of the current study. Heat wave 206 12 217 4 S2 230 6 12 242 6 14 6 6 5 : 2 3 1 : S3 242 12 254 14 Calls 444 9 11 Day 3 1(AT1700), (3) Dufferin (n 5 891) 123 6 18 112 6 4 140 6 17 140 6 14 Unconditioned 65 6 17 54 6 4806 17 81 6 13 6 6 6 6 where Day 5 0 (weekday) or 1 (weekend) and AT1700 is Conditioned 99 18 88 4 116 17 116 13 the apparent temperature (temperature–humidity in- Elite 207 6 18 195 6 4 224 6 17 227 6 14 dex) at 1700 LT (8C) according to Steadman (1979). Be- Trinity Bellwoods Walk (n 5 1563) 29 6 17 14 6 14 41 6 10 38 6 13 cause this algorithm utilizes the apparent temperature Run (n 5 1554) 119 6 16 105 6 13 130 6 9 128 6 11 index, we use this index as opposed to HX for use in this Bike (n 5 1468) 41 6 14 31 6 10 49 6 12 54 6 8 equation only. The HX was used throughout the re- Withrow (n 5 1227) 107 6 18 114 6 19 mainder of the current study, rather than the AT used in the United States, because this is Canada’s current heat 2 stress warning index and is thus more meaningful in the ‘‘warm’’ (from 1151 to 1200 W m 2), and ‘‘hot’’ 2 current study than AT would be. (.1201 W m 2). The listed ranges are based on an ‘‘activity skewing effect,’’ in which there is a widening of 3. Analysis and results the comfort zone toward the warm end during exercise (Kenny et al. 2009b), as compared with sedentary ranges a. Application of COMFA energy budget model in (Brown and Gillespie 1986). At a high M , humans will urban-park transects act be more accepting of slight thermal discomfort (Vanos All averages are reported as means 6 standard de- et al. 2012b) and motivational effects during exercise viation (SD). Bivariate correlation analysis was reported (Roberts 2007; Brotherhood 2008). Metabolically active as a Pearson correlation coefficient r assessing the linear humans have been found to report lower TS than that dependence of one variable to another, using a level of predicted by an energy budget model, even when in heat statistical significance p of 0.05, unless otherwise stated. stress; thus, it is more important to follow guidelines Energy budget analyses under various physiological and predicted by the model than those reported by a human meteorological conditions were completed graphically to stay in a ‘‘thermally safe’’ budget range. Descriptive and statistically for each urban park as listed below results of human energy budgets for each study site, as (these were chosen randomly to display all possible well as correlation analysis and select intensities, are combinations of varying microclimates, exercise, cloth- listed in Tables 4 and 5, respectively. ing, gender, and vegetative cover on thermal energy 1) URBAN-PARK TRANSECTS budgets within separate parks): Figure 3 displays the spatial budget and R of a hu- 1) Trinity Bellwoods Park—4 and 15 September 2009, abs man for three activities (walking, running, and cycling) with an M of brisk walking, conditioned running, act within Trinity Park. Open areas lacking tree cover—such and conditioned cycling, as that indicated by the ‘‘B’’ boxed portions—resulted in 2) Withrow Park—6–9 and 20 September 2009, with an increased Rabs when compared with tree-covered portions M of conditioned running, 2 act within the park (average 5114 W m 2;maximum5 3) Dufferin Grove Park—15 September 2009, with an 2 134 W m 2), directly causing the energy budget to Mact of unconditioned, conditioned, and elite run- 2 increase (r 5 0.947). The Rabs intensity (streetmax 2 ning males and females, and 2 park ) was 53 W m 2. Increased y aided strong in- 4) Dovercourt Park—4 and 15 September 2009 and 5–7 min a creases in the convective heat loss C, giving C 5 72 6 3, July 2010, with climate change scenarios and an M 2 act 90 6 4, and 168 6 6Wm 2 for walking, running, and of conditioned running. cycling, respectively. Within-park wind speeds averaged 2 The budget results found in the current study fell within 0.70 m s 1 as a consequence of fair weather and high 22 21 the ranges of ‘‘neutral’’ (from 220 to 1150 W m ), surface roughness; hence, the ya of cycling (4.5 m s )

Unauthenticated | Downloaded 09/27/21 09:30 AM UTC 1646 JOURNAL OF APPLIED METEOROLOGY AND CLIMATOLOGY VOLUME 51

22 TABLE 5. Average budget [budget intensity (W m ) 5 was 3.98C and is dramatically displayed in this figure. 2 streetmax parkmin] and PCI [maximum park cooling intensity This was found to have a weak relationship with budget 8 ( C; Slater 2010) from high-frequency raw data], and Pearson cor- (r 5 0.504, with p , 0.01), as compared with budget relation analysis r (Pearson correlation coefficient with high- 5 , frequency budget outputs; an asterisk indicates significance at p , being strongly related to Kin (r 0.951, with p 0.01).

0.01 level) using raw values of Kin and Ta with energy budget outputs Hence, the human energy budget is controlled more so for the four selected parks (see Table 4 for n values) in Toronto. by receiving and absorbing radiative fluxes, which over-

rides the effect of Ta. Park Budget intensity PCId r(K ) r(Ta) in Figure 5 displays the effect of gender and individual Dovercourt 40 5.5 0.922* 0.369* conditioning levels on energy budgets in Dufferin Grove Dufferin 65 4.2 0.966* 0.574* Trinity Bellwoods 54 6.0 0.923* 0.514* Park. Weather conditions present on 15 September Withrow 84 3.9 0.951* 0.504* produced neutral budgets in and outside the park (65 2 and 81 W m 2, respectively) for unconditioned in- dividuals exercising at levels that would elicit a cardio- gave a greater reduction in energy budget than when vascular improvement. Elite athletes enter or become 2 running at 2.1 m s 1. Thus, cycling budgets fell below close to the warm budget zone both in and outside the 22 that of running even though Mact was equal. park (male 5 207 and 227 W m ;female5 140 and 22 Figure 4 demonstrates the variations in Ta with Rabs 160 W m , respectively), however, even in moderate and energy budget variations along a transect in With- weather conditions. The overall energy budget corre- row Park. A total of 16 transects on days with easterly lations with Kin and Ta were found to be statistically winds (6–9 and 20 September) were averaged for the significant, giving r 5 0.966 and 0.574, respectively. three variables. The path chosen through Withrow Park Street-to-park budget intensity increased with Mact, 22 had very little tree cover, shown by slight Rabs variations with elite athletes experiencing a 69 W m difference, and reductions and hence decreased reductions in bud- as compared with unconditioned athletes experiencing 2 get output. The budget follows a pattern strongly related 61 W m 2. The greatest difference in street-to-park 2 to the plotted Rabs (r 5 0.784) because of relatively budget decrease was found in elite athletes (highest 22 22 constant Mact, convection, evaporation, and emitted Mact), giving 220 W m for ‘‘whole park,’’ 232 W m 22 longwave radiation. The PCI effect of Ta (streetmax 2 for treed sections, and a minimal 23Wm in open- 21 parkmin) (Spronken-Smith and Oke 1998; Slater 2010) park areas. The average surface wind speed of 0.6 m s

22 FIG. 3. For Trinity Bellwoods Park, the modeled energy budget and Rabs (W m ) for brisk walking, conditioned running, and conditioned cycling on 4 Sep (south wind) and 15 Sep (north wind) 2009. Each panel consists of 6-, 25-, and 70-m moving averages for the three respective activities. Boxes labeled A and B indicate examples of visible tree cover and uncovered path areas within the park, respectively.

Unauthenticated | Downloaded 09/27/21 09:30 AM UTC SEPTEMBER 2012 V A N O S E T A L . 1647

FIG. 4. For Withrow Park, 5-m moving averages of mean air temperature (8C), energy budget, and absorbed radiation 2 (W m 2) for conditioned running, on test days with east winds (6–9 and 20 Sep).

2 (6.7 m s 1 at 10 m) from the north extended urban air an average daytime HX during the heat wave of 39.58C temperatures southward into the park for approximately (110.18C greater than surrounding six days). Even with 22 65 m and extended park cool air temperatures approxi- low metabolic rate modeling (Mact 5 221 W m ), the mately 50–80 m beyond the park. budgets were consistently in a ‘‘heat stress zone’’ 2 (.121 W m 2) (Brown and Gillespie 1986; Harlan et al. 2) HEAT-WAVE AND CLIMATE SCENARIOS 2006), and commonly fell above warm and hot levels of TS. Figure 6 displays original data from Dovercourt Park Figure 7 displays temporal changes of EMS calls, from three sources: 1) September 2009 data; 2) 5 July Tmax, Tavg, and energy budget. Dispatch calls follow 2010 heat-wave weather data; and 3) two climate change a close pattern with both budget and Tmax, showing scenarios applied to the 2010 heat wave. A decrease was significant linear dependence (r2 5 0.860 and 0.603, re- found in the budget values within the park relative to spectively). Weak dependence of EMS calls was found 2 those on the streets for all scenarios, with similar mean with daily Tmin (r 5 0.116). Peak values for all curves budget difference and maximum budget intensity found occurred on 5 July when discomfort was at its greatest. in the latter three scenarios (Label ‘‘Orig’’ 5 12 and The average HX was also found to be a significant pre- 2 2 40 W m 2, respectively, and 12 and 41 W m 2 for all dictor of EMS calls (r2 of 0.86); the HX lacks the vital three scenarios). Although these difference values are energy budget component of spatial variations in radi- equal for the heat-wave and climate change scenarios, ation, however, and hence Rabs. The current study did the absolute budget averages are significantly different not apply a spatial comparison with EMS calls and en- when paired together and compared with the original ergy budget modeling and hence did not exhibit the 2009 data (See Table 4). benefit of this. Such spatial analysis is possible given accurate microclimate measurements using a denser b. Emergency-response prediction using the COMFA network of simple weather stations throughout a city, energy budget model during a heat wave further addressed in the discussion section. DuringtheJuly2010heatwaveinToronto,heat- related dispatch calls within the city of Toronto in- creased by an average of 29% from the prior 3-day 4. Discussion average. During the heat wave, mean 24-h T and T max min a. Urban-park transects were 32.48 and 24.18C, which were 7.18 and 8.18C higher than the surrounding six days, respectively. The com- Overall results from testing the COMA energy budget monly used humidex in Canada was also assessed with model displayed the ability to adequately predict the

Unauthenticated | Downloaded 09/27/21 09:30 AM UTC 1648 JOURNAL OF APPLIED METEOROLOGY AND CLIMATOLOGY VOLUME 51

22 FIG. 5. For Dufferin Grove Park for 15 Sep 2009 (north wind), (a) modeled energy budget values (W m ) for unconditioned (Uc), conditioned (C), and elite running (E) for male (black lines) and female (gray lines) and (b) corresponding Ta (8C). All data are 25-m moving averages. Gray and black double arrows indicate urban air extension into park and park air extension into urban street, respectively. moderating effect of microclimate variations within ur- in open space or on the street, as well as strong budget ban parks on outdoor TC. In all four parks under the correlations with Kin and moderate to weak correlations tested climatic and metabolic conditions, decreases in Ta with Ta. Hence, a park-induced decrease in Ta is not occurred, yet this did not cause the human energy bud- enough to aid in heat stress reduction and to keep rec- get to decrease to the same extent as a reduced incoming reational athletes out of dangerous budget zones (as solar radiation. Solar radiation had the largest effect on compared with low Mact and/or sedentary individuals). energy budgets in all cases; therefore, blocking this ra- These results clearly demonstrate that addition of diation is the most effective way to lower high human deciduous trees with minimal summertime transmittance heat budgets. These results show that the model per- will reduce the radiant heat loads on humans (Koppe and forms in a realistic fashion and therefore can aid users Jendritzky 2005). According to Matzarakis and Endler such as researchers, planners, and urban designers in (2010), under shady conditions (low-transmitting trees), creating usable and effective urban green space by the mean radiant temperature Tmrt is assumed to be highlighting the sensitive variables when modeling hu- equal to Ta yet is much greater in open areas, as sup- man energy budgets. ported by our results. Conveying more information of

The relatively lower heat capacities and evaporative Tmrt and/or other TC measures gives a more realistic cooling of existing moist vegetation and soil in green measure for human health impacts of future climate space—in comparison with surrounding concrete and scenarios (Thorsson et al. 2011). asphalt—aided in the decrease of overall sensible heat- Differing levels of conditioning (fitness) and gender ing and Ta in the park yet was overcome by incoming displayed in Fig. 5 demonstrate how diverse the energy shortwave and longwave radiation when a human was budget can be as a result of an extra endogenous met- not modeled beneath a tree canopy. This was revealed abolic heat load. When a person has the fitness ability by the differences found when beneath a canopy versus and motivation to reach a higher Mact, their risk of heat

Unauthenticated | Downloaded 09/27/21 09:30 AM UTC SEPTEMBER 2012 V A N O S E T A L . 1649

22 FIG. 6. Budget outputs B (W m ) in Dovercourt Park from original September 2009 data (Orig) and three heat-

wave scenarios—1) 5 Jul 2010 heat-wave weather conditions and 2) 128C Ta (year 2040), and 3) 138C Ta (year 2070) above July 2010 heat-wave values, on the basis of Canadian coupled global climate models (CCCma 2011). Metabolic 2 activity rates are for a conditioned male running (436 W m 2). Heat stress ‘‘extreme caution’’ and ‘‘danger’’ levels 2 2 occur at B . 121 W m 2 and B . 200 W m 2, respectively (Harlan et al. 2006); because of the ‘‘activity skewing effect,’’ however, humans have been shown to respond as ‘‘warm’’ and ‘‘hot’’ at higher B levels of 150 and 2 250 W m 2, respectively (Kenny et al. 2009a). stress is much greater (Vanos et al. 2012b). Heat casu- and 0.78 for running/walking (Nielsen et al. 1988; alties in athletes have been reported in ambient tem- Nielsen 1990; Campbell and Norman 1998). This ac- peratures of less than 208C (Hughson et al. 1980); hence, counts for the differences in Rabs displayed between awareness of heat injury possibilities is essential. Met- upright and sitting activities. The comparisons of walk- abolic levels, commonly greater by a factor of 10–20 ing, running, and biking in Trinity Bellwoods Park and during physical exertion, are a greater contributor than of running at various speeds based on conditioning in

Ta to heat stress. Dufferin Grove Park display the impacts of ya on budget Trinity Bellwoods tested all three activities, where the balance. With the higher ya, the energy budget is sig- effective body areas Aeff modeled were 0.70 for cycling nificantly lowered, even with a constant Mact, because of

FIG. 7. Day-to-day variability of number of heat-related dispatch calls, daytime energy 22 22 budget (W m ) at sedentary/slow walking Mact (221 W m ), and 24-h maximum and average Ta.

Unauthenticated | Downloaded 09/27/21 09:30 AM UTC 1650 JOURNAL OF APPLIED METEOROLOGY AND CLIMATOLOGY VOLUME 51 increased convective and evaporative heat losses. Al- into zones of thermal heat stress (Vanos et al. 2012a). though a greater wind results in convective cooling Hence, the activity level must be considered when as- within the park, the greater wind also results in more sessing a person’s actual TC versus their subjective re- turbulent mixing and the diminishment of park-cooling sponses. Understanding the underlying microclimatic extension into the surrounding streets. Cooler air from mechanisms surrounding energy budget output values the park canopies extends outside the park in the di- allows researchers to meaningfully quantify the human rection of prevailing winds, with greater effect and ex- impact of urban design decisions during heat waves, tent on calm days, yet busy roadways can mix and which may become more frequent with climate change. diminish the cool air (Slater 2010). Lower average y for w c. EMS prediction during heat wave test days within Dufferin Grove and Withrow Parks gave cool air extensions of ;50 and 20 m, yet warm air Energy budget assessments are shown to be a strong from the street was also brought into each park ;80 and predictive tool for emergency response to heat morbidity.

20 m, respectively, by the prevailing wind. Although elevated Tmin has also been associated with high mortality (Greene and Kalkstein 1996; Sheridan and b. Heat-wave and future climate scenarios Kalkstein 2004), the current study did not find a pre-

Under extreme heat, the main interest is the length of dictive relationship with EMS calls and Tmin, with tem- time in which a person is under heat stress and their peratures that fell to an average of 24.18C each night. ability to achieve heat balance given the environmental The heightened number of EMS calls continued after constraints. The low cooling effect of Dovercourt Park the heat subsided (9–11 July). This lag effect can be due during the extreme-heat scenarios is problematic con- to a greater ability to withstand excessive heat in the sidering that this is when people are most susceptible to short term (Luber and McGeehin 2008) from delayed heat illness. This susceptibility, however, is dependent on physiological response and/or the perception of heat the combination of sensitivity to TS, along with the in- threat declining with time. tensity, length, and seasonal time (Koppe and Jendritzky In addition, the accuracy of the predictive equation 2005). Although preservation of ‘‘open’’ green space has [Eq. (3)] shows that investing in accurate and ad- been cited as a great benefit for cooling cities by Harlan vanced heat health warning systems (HHWS) for et al. (2006) and Luber and McGeehin (2008), abundant Toronto (Sheridan and Kalkstein 2004) is an essential tall greenery and trees is a much greater microclimatic step for emergency preparedness and protection of heat- controlling factor during the day and is essential during vulnerable populations and areas. To do so, however, heat waves (Harlan et al. 2006). This agrees with the further important variables must be considered, because current study, finding that budget cooling of the body heat stress situations are affected by time and space from corresponds more with vegetatively dense and covered such factors as local microclimates and population dy- areas, largely because of the presence of trees for PCI namics (deVries et al. 2003; Harlan et al. 2006). There is development (Spronken-Smith and Oke 1998). need in the literature for such a study, with further re- Figure 6 shows that the model is able to identify im- search applying accurate energy budget models. Ac- portant differences that are representative of reality counting for both energy budget and population density now and forecasts of the future. Although these results on an hourly time scale gives a finer temporal assessment. display humans entering into dangerous energy budget Although the HX was also a significant predictor of zones within scenarios 1–3, the heat stress remains EMS calls, it lacks the vital energy budget component lower within urban parks than in the streets. According of variations in radiation and wind speed, which vary to Harlan et al. (2006), ‘‘extreme caution’’ and ‘‘dan- the greatest spatially. Given the sensitivity of the energy 2 ger’’ levels of heat stress occur at budgets .121 W m 2 budget output to variations in radiation, investing in a 2 and .200 W m 2, respectively. The current modeling higher spatial density of urban weather stations will im- was completed on humans running at moderate levels prove spatial modeling of heat stress–prone zones, hence 22 (Mact 5 436 W m ) who will report a lower TS than guiding the deployment of EMS and coping resources. that actually predicted because of the activity-skewing From this, urban planners and designers can be aware of effect. This effect is displayed when an exercising subject areas of high heat stress, and the most oppressive hours of 2 responds as neutral with a budget level up to 150 W m 2 the day within that area, thereby knowing the most vital yet a sedentary individual would be warm at this level areas in which to implement corrective bioclimatic design. (Kenny et al. 2009a). This offsetting of thermal stress This can be demonstrated with GIS mapping, as in with psychological adjustments (broader sensation/ Dolney and Sheridan (2006), to visually recognize the comfort ranges during exercise; Havenith et al. 2002; location and frequency of heat stress situations and Kenny et al. 2009b) and acceptances can draw people other potential confounders of heat stress, such as

Unauthenticated | Downloaded 09/27/21 09:30 AM UTC SEPTEMBER 2012 V A N O S E T A L . 1651 socioeconomics. Hence, we can identify areas in need (2010) for quantitative analysis, the capabilities of of improved or corrective bioclimatic design, the most the COMFA model, as well as numerous application important one being incorporating trees to induce the methods, were displayed. On the basis of this analysis, effects of PCIs, as displayed throughout this study. the following findings were presented: 1) there is a lack Furthermore, populations within predefined at-risk areas of energy budget decrease without tree cover, 2) there is can benefit from the removal of structural constraints for a small extension of cool park air into nearby streets and improving environmental conditions, added green warm air into parks, 3) activity level, gender, and con- space, and additional coping resources—such as shade, ditioning are important influences on heat stress, and 4) water, and air conditioning (Harlan et al. 2006). Posi- there was successful prediction of EMS calls for heat tioning an emergency vehicle in areas found to experi- stress using the COMFA energy budget model, showing ence higher heat stress during heat waves and ensuring independent correlation with empirical data. proper staffing for a number of vehicles are both strong Significant budget decreases were not found with applications for lowering serious heat-illness cases a lack of tree cover. The presence and high density of tall (Dolney and Sheridan 2006). trees (.5 m) that have the ability to block solar radia- The International Panel on Climate Change tion is of greater importance than just vegetation density (Confalonieri et al. 2007) confidently classifies heat waves alone, because this may only refer to low-lying vegeta- as extreme events that even high-income countries are tion. To have a strong ‘‘sensible’’ effect on TC and TS, not well prepared to cope with, with adaptive capacity with extension into surrounding neighborhoods/streets, improvements needed everywhere. Public-health offi- parks must be larger, but, more important, they must cials and researchers cannot assume that all individuals contain adequate amounts of canopy cover from trees to will be adapted to heat during the summer months—in block out penetrating solar radiation and provide evap- particular, in conditions of heat plus exercise—given orative cooling. In doing so, human energy budgets can increasingly sedentary and indoor lifestyles. Further be decreased and individuals can be kept within the quantification of seasonal and long-term adaptation is ‘‘neutral’’ safe zone so that heat stress and morbidity do needed, as done by Matzarakis et al. (2011) for a 30-yr not occur. period in Vienna, Austria. They suggest implementation Implementing the use of energy budget models using of HHWS as a tool to reduce the impact of heat waves. a small number of readily available variables can allow In a highly populated and growing midlatitude city such for easy determination of TC, cooling potential, and as Toronto, the implementation of a location-specific UHI mitigation of designs by urban designers and land- HHWS is vital now and in the future. The 30-yr trends of scape architects (Slater 2010). This study has discussed the extreme-heat air masses of moist tropical and dry the threats of amplified urban growth and air tempera- tropical (MT and DT) in Toronto show a weak increase tures causing UHI effects related to poor human health on the basis of synoptic weather-type classification and morbidity. This becomes increasingly important with (Kalkstein et al. 1996; Sheridan 2002). Heat waves are future predictions of climate change causing intensified, linked to specific atmospheric conditions (Meehl and more frequent, and lengthier heat waves. Development Tebaldi 2004), with MT and DT stagnant and oppres- and ongoing sprawl call for further consideration of sively hot air masses associated with 9.8% and 9.4% in- spatial and temporal variations in population density and crease in daily deaths in Toronto, respectively (Sheridan heat stress levels. and Kalkstein 2004). This is a broader implication of Future TC studies for climate adaptation should in- heat-wave and UHI problems; it does, however, highlight corporate the change in temperature distribution sepa- the need for decision makers to focus on both climate rately from variability (Gosling et al. 2009) for quantifying change and UHI mitigation (through bioclimatic design) spatially diverse adaptive processes (Lin 2009; Yao et al. to lessen the likelihood of such air masses reaching ‘‘of- 2009) and comparative analysis of varied geographies fensive’’ levels (i.e., resulting in heat death; Sheridan and (Golden et al. 2008). Such studies can be coupled to Kalkstein 2004). sport-specific heat stress to identify areas and individuals most at risk of heat stress, along with remote sensing or GIS methods for effective public-health response (Luber 5. Conclusions and McGeehin 2008). In addition, the relatively small size The current study demonstrated finescale modeling of of the studied parks in the current study warrants further human thermal energy budgets while under various research to assess the geographical extent of budget metabolic activity rates, fitness levels, clothing, and cooling potentials with respect to size and tree cover. vegetative covers on heterogeneous street–park–street Additional applications of this model include quantifying transects. Through using an existing dataset from Slater crucial weather parameters for heat events, estimating

Unauthenticated | Downloaded 09/27/21 09:30 AM UTC 1652 JOURNAL OF APPLIED METEOROLOGY AND CLIMATOLOGY VOLUME 51 and mapping green space present and/or needed in built- Dolney, T. J., and S. C. Sheridan, 2006: The relationship between up areas, analyzing health trends of various urban re- extreme heat and ambulance response calls for the city of gions, and linking TS with ground-level air pollution. Toronto, Ontario, Canada. Environ. Res., 101, 94–103. Eliasson, I., 2000: The use of climate knowledge in urban planning. There is a breadth of knowledge and information re- Landscape Urban Plann., 48, 31–44. quired to address current and future TC situations, yet ——, I. Knez, U. Westerberg, S. Thorsson, and F. Lindberg, 2007: preventative measures are available for creation of com- Climate and behaviour in a Nordic city. Landscape Urban fortable places for people to use. Plann., 82, 72–84; Corrigendum, 83, 363. Environment Canada, cited 2010: Climate change science and re- search. [Available online at http://www.ec.gc.ca/sc-cs/Default. Acknowledgments. Funding for this project was gen- asp?lang5Enn556010B41-1.] erously provided by the Natural Sciences and Engi- Fanger, P. O., and J. Toftum, 2002: Extension of the PMV model to neering Research Council of Canada (NSERC). The non-air-conditioned buildings in warm climates. Energy authors thank Toronto Emergency Medical Services for Build., 34, 533–536. Gaitani, N., G. Mihalakakou, and M. Santamouris, 2007: On the providing emergency-dispatch data. The use of the data use of bioclimatic architecture principles in order to improve was allowed through a confidentiality agreement that thermal comfort conditions in outdoor spaces. Build. Envi- was reached between Toronto Emergency Medical Ser- ron., 42, 317–324. vices and the current study’s lead author. Galea, S., and D. Vlahov, 2005: Urban health: Evidence, challenges and directions. Annu. Rev. Public Health, 26, 341–365. Glazier, R. H., G. L. Gillian, L. Booth, P. Gozdyra, M. I. Creatore, and A. Tynan, Eds., 2007: Neighbourhood Environments and REFERENCES Resources for Healthy Living—A Focus on Diabetes in Tor- Ali-Toudert, F., and H. Mayer, 2007: Effects of asymmetry, gal- onto. Institute for Clinical Evaluative Sciences, 329 pp. leries, overhanging facades and vegetation on thermal comfort Golden, J. S., D. Hartz, A. Brazel, G. Luber, and P. Phelan, 2008: in urban street canyons. Sol. Energy, 81, 742–754. A biometeorology study of climate and heat-related mor- Brotherhood, J. R., 2008: Heat stress and strain in exercise and bidity in Phoenix from 2001 to 2006. Int. J. Biometeor., 52, sport. J. Sci. Med. Sport, 11, 6–19. 471–480. Brown, R. D., and T. J. Gillespie, 1986: Estimating outdoor thermal Gosling, S. N., G. R. McGregor, and A. Paldy, 2007: Climate comfort using a cylindrical radiation thermometer and an change and heat-related mortality in six cities. Part 1: Model energy budget model. Int. J. Biometeor., 30, 43–52. construction and validation. Int. J. Biometeor., 51, 525–540. Campbell, G. S., and J. M. Norman, 1998: An Introduction to En- ——, ——, and J. A. Lowe, 2009: Climate change and heat-related vironmental Biophysics. 2nd ed. Springer, 286 pp. mortality in six cities. Part 2: Climate model evaluation and CCCma, cited 2011: Climate modelling and analysis. Canadian projected impacts from changes in the mean and variability of Centre for Climate Modelling and Analysis. [Available online temperature with climate change. Int. J. Biometeor., 53, 31–51. at http://www.cccma.ec.gc.ca/diagnostics/cgcm3-t47/animation_ Greene, J. S., and L. S. Kalkstein, 1996: Quantitative analysis of sresa1b_st_pcan.shtml.] summer air masses in the eastern United States and an ap- Cheng, C. S., and Coauthors, 2005: Differential and combined impacts plication to human mortality. Climate Res., 7, 48–53. of winter and summer weather and air pollution due to global Harlan, S. L., A. J. Brazel, L. Prashad, W. L. Stefanov, and warming on human mortality in south-central Canada. Health L. Larsen, 2006: Neighborhood microclimates and vulnera- Canada Tech. Rep., 233 pp. [Available online at http://www. bility to heat stress. Soc. Sci. Med., 63, 2847–2863. toronto.ca/health/hphe/pdf/weather_air_pollution_impacts.pdf.] Havenith, G., 1997: Individual heat stress response. Ph.D. thesis, City of Toronto, cited 2011: Toronto open data [category 5 Catholic University Nijmegen, 189 pp. [Available online at http:// ‘‘Parks’’]. [Available online at http://www.toronto.ca/open.] dare.ubn.kun.nl/bitstream/2066/18586/1/18586_indihestr.pdf.] Confalonieri, U., B. Menne, R. Akhtar, K. L. Ebi, M. Hauengue, ——, I. Holmer, and K. C. Parsons, 2002: Personal factors in R. S. Kovats, B. Revich, and A. Woodward, 2007: Human thermal comfort assessment: Clothing properties and meta- health. Climate Change 2007: Impacts, Adaptation and Vul- bolic heat production. Energy Build., 34, 581–591. nerability, M. L. Parry et al., Eds., Cambridge University ——, and Coauthors, 2011: The UTCI-clothing model. Int. Press, 391–431. [Available online at http://www.ipcc.ch/ J. Biometeor., 56, 461–470. publications_and_data/ar4/wg2/en/ch8.html.] Hughson,R.L.,H.J.Green,M.E.Houston,J.A.Thomson, Curriero, F. C., K. S. Heiner, J. M. Samet, S. L. Zeger, L. Strug, and D. R. MacLean, and J. R. Sutton, 1980: Heat injuries in J. A. Patz, 2002: Temperature and mortality in 11 cities of the Canadian mass participation runs. Can. Med. Assoc. J., 122, eastern United States. Amer. J. Epidemiol., 155, 80–87. 1141–1142. de Dear, R. J., and G. S. Brager, 1998: Developing an adaptive ISO, 2007: ISO 9920: Ergonomics of the thermal environment: model of thermal comfort and preference. ASHRAE Trans., Estimation of thermal insulation and water vapour resistance 104, 145–167. of a clothing ensemble. International Organization for Stan- deVries,S.,R.A.Verheij,P.P.Groenewegen,andP.Spreeuwenberg, dardization, 104 pp. 2003: Natural environments—Healthy environments? An ex- Kalkstein, L. S., M. C. Nichols, C. D. Barthel, and J. S. Greene, ploratory analysis of the relationship between greenspace and 1996: A new spatial synoptic classification: Application to air- health. Environ. Plann., 35, 1717–1731. mass analysis. Int. J. Climatol., 16, 983–1004. Djongyang, N., R. Tchinda, and D. Njomo, 2010: Thermal com- Kenny, N. A., J. S. Warland, R. D. Brown, and T. J. Gillespie, 2008: fort: A review paper. Renew. Sustain. Energy Rev., 14, 2626– Estimating the radiation absorbed by a human. Int. J. Bio- 2640. meteor., 52, 491–503.

Unauthenticated | Downloaded 09/27/21 09:30 AM UTC SEPTEMBER 2012 V A N O S E T A L . 1653

——, ——, ——, and ——, 2009a: Part A: Assessing the perfor- ——, and L. S. Kalkstein, 2004: Progress in heat watch–warning mance of the COMFA outdoor thermal comfort model on system technology. Bull. Amer. Meteor. Soc., 85, 1931–1941. subjects performing physical activity. Int. J. Biometeor., 53, ——, A. Kalkstein, and L. S. Kalkstein, 2009: Trends in heat- 415–428. related mortality in the United States, 1975–2004. Nat. ——, ——, ——, and ——, 2009b: Part B: Revisions to the COMFA Hazards, 50, 145–160. outdoor thermal comfort model for application to subjects Slater, G., 2010: The cooling ability of urban parks. M.S. thesis, performing physical activity. Int. J. Biometeor., 53, 429–441. School of Environmental and Rural Design, University of Koppe, C., and G. Jendritzky, 2005: Inclusion of short-term adap- Guelph, 141 pp. tation to thermal stresses in a heat load warning procedure. Smoyer, K. E., D. G. C. Rainham, and J. N. Hewko, 2000: Heat- Meteor. Z., 14, 271–278. stress-related mortality in five cities in southern Ontario: Lemmen, D. S., F. J. Warren, J. Lacroix, and E. Bush, 2008: From 1980–1996. Int. J. Biometeor., 44, 190–197. impacts to adaptation: Canada in a changing climate 2007. Souch, C. A., and C. Souch, 1993: The effect of trees on summertime Natural Resources Canada Tech. Rep., 448 pp. [Available on- below canopy urban climates: A case study in Bloomingdale, line at http://www.nrcan.gc.ca/earth-sciences/climate-change/ Indiana. J. Arboricult., 19, 303–312. community-adaptation/assessments/132.] Spronken-Smith, R. A., and T. R. Oke, 1998: The thermal regime Lin, T., 2009: Thermal perception, adaptation and attendance in of urban parks in two cities with different summer climates. a public square in hot and humid regions. Build. Environ., 44, Int. J. Remote Sens., 19, 2085–2104. 2017–2026. Steadman, R. G., 1979: The assessment of sultriness. Part I: A ——, A. Matzarakis, and R. Hwang, 2010: Shading effect on long- temperature–humidity index based on human physiology and term outdoor thermal comfort. Build. Environ., 45, 213–221. clothing science. J. Appl. Meteor., 18, 861–873. Luber, G., and M. McGeehin, 2008: Climate change and extreme Stewart, I., and T. R. Oke, 2009: A new classification system for heat events. Amer. J. Prev. Med., 35, 429–435. urban climate sites. Bull. Amer. Meteor. Soc., 90, 922–923. Matzarakis, A., and C. Endler, 2010: Climate change and thermal Strath, S. J., A. M. Swartz, D. R. Basset, W. L. O’Brien, G. A. King, bioclimate in cities: Impacts and options for adaptation in and B. E. Ainsworth, 2000: Evaluation of heart rate as Freiburg, Germany. Int. J. Biometeor., 54, 479–483. a method for assessing moderate intensity physical activity. ——, S. Muthers, and E. Koch, 2011: Human biometeorological Med. Sci. Sports Exerc., 32, S465–S470. evaluation of heat-related mortality in Vienna. Theor. Appl. Thorsson, S., M. Lindqvist, and S. Lindqvist, 2004: Thermal bio- Climatol., 105, 1–10. climatic conditions and patterns of behaviour in an urban park Mayer, H., L. Katzschner, and M. Bruse, 2008: KLIMES—A joint in Goteborg, Sweden. Int. J. Biometeor., 48, 149–156. research project on human thermal comfort in cities. Tech. ——, F. Lindberg, J. Bjo¨ rklund, B. Holmer, and D. Rayner, 2011: Rep. 117, Meteorological Institute, Albert-Ludwigs-University, Potential changes in outdoor thermal comfort conditions in 101–117. Gothenburg, Sweden due to climate change: The influence of McGregor, G. R., 2012: Human biometeorology. Prog. Phys. urban geometry. Int. J. Climatol., 31, 324–335. Geogr., 36, 93–109. Tikuisis, P., P. Meunier, and C. Jubenville, 2001: Human body Meehl, G. A., and C. Tebaldi, 2004: More intense, more frequent, surface area: Measurement and prediction using three di- and longer lasting heat waves in the 21st century. Science, 305, mensional body scans. Eur. J. Appl. Physiol., 85, 264–271. 994–997. Vanos, J., J. Warland, T. Gillespie, and N. Kenny, 2012a: Thermal Mohsin, T., and W. Gough, 2010: Trend analysis of long-term comfort modelling of body temperature and psychological temperature time series in the greater Toronto area (GTA). variations of a human exercising in an outdoor environment. Theor. Appl. Climatol., 101, 311–327. Int. J. Biometeor., 56, 21–32. Nielsen, B., 1990: Solar heat load: Heat balance during exercise in ——, ——, ——, and ——, 2012b: Improved predictive ability of clothed subjects. Eur. J. Appl. Physiol., 60, 452–456. climate–human–behaviour interactions with modifications to ——, K. Kassow, and F. E. Aschengreen, 1988: Heat balance during the COMFA outdoor energy budget model. Int. J. Biometeor., exercise in the sun. Eur. J. Appl. Physiol., 58, 189–196. doi:10.1007/s00484-012-0522-1, in press. Nikolopoulou, M., and S. Lykoudis, 2006: Thermal comfort in Watkins, R., J. Palmer, and M. Kolokotromi, 2007: Increased outdoor urban spaces: Analysis across different European temperature and intensification of the urban heat island: Im- countries. Build. Environ., 41, 1455–1470. plications for human comfort and urban design. Built Envi- O’Neill, M. S., and K. L. Ebi, 2009: Temperature extremes and ron., 33, 85–96. health: Impacts of climate variability and change in the United Whaley, M. H., P. H. Brubaker, and R. M. Otto, Eds., 2006: States. J. Occup. Environ. Med., 51, 13–25. ACSM’s Guidelines for Exercise Testing and Prescription. 7th Peel, M. C., B. L. Finlayson, and T. A. McMahon, 2007: Updated ed. Lippincott Williams and Wilkins, 366 pp. world map of the Koppen–Geiger climate classification. Hy- Wieringa, J., A. G. Devenport, C. S. B. Grimmond, and T. R. Oke, drol. Earth Syst. Sci., 11, 1633–1644. 2001: New revision of Davenport roughness classification. Roberts, W. O., 2007: Heat and cold: What does the environment Proc. Third European and African Conf. on Wind Engineer- do to marathon injury? Sports Med., 37, 400–403. ing, Eindoven, Netherlands, IAWE, 8 pp. Sheridan, S. C., 2002: The redevelopment of a weather-type Yao, R., B. Li, and J. Liu, 2009: A theoretical adaptive model of classification scheme for North America. Int. J. Climatol., 22, thermal comfort—Adaptive Predicted Mean Vote (aPMV). 51–68. Build. Environ., 44, 2089–2096.

Unauthenticated | Downloaded 09/27/21 09:30 AM UTC