Introduction

The concepts upon which human thermal comfort (TC) is based have become an increasingly prominent and debated topic in the literature. TC is defined as ‘that condition of mind which expresses satisfaction with the thermal environment’ (ASHRAE 1966; Fanger 1972; Parsons 2003; ISO7730 2005), and no preference to be warmer or cooler (Fanger 1972). National and international standards associated with human comfort have been developed, which include ASHRAE (1966) (American Society of Heating, Refrigerating and Air-Conditioning Engineers), ISO7730 (2005) (TC), ISO9920 (2007) (clothing), ISO7933 (1989) (hot environments) and ISO8996 (1990) (metabolic rate). Parsons (2003) followed up significant early work from Fanger (1970, 1972) and Kerslake (1972), extensively covering the effects of hot, cold and moderate environments on human health, comfort and performance in his book entitled Human Thermal Environments. Related comprehensive reviews have recently been completed on heat stress indices (Epstein and Moran 2006), heat stress and strain in exercise (Brotherhood 2008), clothing properties and metabolic heat production (Havenith et al. 2002), mechanisms of thermoregulation through sweating (Shibasaki et al. 2006), and urban climate research in the last two decades (Arnfield 2003). A universally applicable model is the Universal Thermal Climate Index (UTCI 2009). This index was recently completed in Europe by over 40 scientists in 20 different countries based on a variety of research (see, e.g., Richards and Havenith 2007; Fiala et al. 2007; Psikuta et al. 2008; Jendritzky et al. 2008; Mateeva et al. 2009).

Human biometeorology research highlights the increasing importance being placed on environment and human health, and their interrelationships. Knowledge of these is essential for addressing the problems associated with decreased use of outdoor spaces. In recent scientific literature, as well in the current review, a considerable amount of future research has been called for. The scope of this review centers on human physiology of exercising persons in warm-hot, dynamic urban environments, relating TC to physiological and psychological responses.

When an individual experiences thermal stress causing discomfort, it can negatively impact their health in many ways. Heat-related symptoms include, but are not limited to, fatigue, headaches, nausea, confusion, cramping, and reduced capacity to work or perform, which all are signs leading to heat stroke or exhaustion. Thermally comfortable outdoor spaces noticeably influence type and amount of human physical activity (Gaitani et al. 2007), and creation of comfortable spaces is needed to meet the demand of outdoor recreational users (Brown and Gillespie 1986). Many studies have specifically noted that urban dwellers show overall decreased health and functionality in everyday life (e.g., Johansson and Rohinton 2006; Eliasson 2000; Galea and Vlahov 2005; Harlan et al. 2006; Eliasson et al. 2007; Watkins et al. 2007), and also higher heat-related mortality (e.g., Smoyer et al. 2000; Basu and Samet 2002; Galea and Vlahov 2005; Hu et al. 2008; Revich and Shaposhnikov 2008; Golden et al. 2008; Gosling et al. 2009). Such studies have been completed in many urban areas throughout the world (e.g., Moscow, Sydney, Phoenix, Boston, Dallas, Budapest, London) and have found increased mortality during times of higher air temperatures, as well as high air pollution (Hu et al. 2008). However, heat mortality is still a debated topic, as displayed in a study completed by Sheridan and Dolney (2003) on rural, suburban and urban areas in Ohio, United States. They found no statistical difference (P < 0.05) between the three areas, thus showing that living in an urban area does not make one more susceptible to heat.

Vulnerable populations, such as the sick, young and elderly, have the highest mortality and risk of heat strokes during exercise in high temperatures (Smoyer et al. 2000; Basu and Samet 2002; Harlan et al. 2006). The future growth of cities will undoubtedly cause further problems associated with cardiovascular and respiratory problems, and death and illness from heat and air pollutants during the warmest times of the year. Research is needed to provide policy makers with evidence of health risks (Revich and Shaposhnikov 2008). Detrimental impacts can be further expanded to athletic performance, work and behaviour (Brotherhood 2008).

Health officials are partly blaming the poor construction of the built environment for a decrease in physical activity and rise in obesity (Handy 2004) leading to further health problems such as asthma, diabetes and cardiovascular disease, and economic strain on the health care system and economy. This is in agreement with Galea and Vlahov (2005), who compiled a detailed review on urban environment and health. Specific features of the built environment, such as street and trail connectivity, aesthetics, and scale of streets also impact physical activity (Handy 2004). According to Smoyer et al. (2000), facilitating cooler and more ventilated urban areas with wise land use is important to reduce the urban heat island effect. The studies cited above report the effects of heat on urban health, thus affirming the importance of bioclimatic urban design with comfort in mind.

There are large spatial and temporal variations of microclimatic conditions (Thorsson et al. 2007a); however, environments are often assessed only in terms of air temperature (T a) (Parsons 2003). This can lead to a lack of understanding of other important parameters. The six typical parameters, initially examined by Fanger (1970) in his book Thermal Comfort, include: (1) T a; (2) vapor pressure (e); (3) air velocity (v w); (4) solar and thermal radiation (R); (5) metabolic heat (M); and (6) clothing insulation (I cl). In terms of these parameters, researchers attempt to predict the thermal environment that will result in human comfort, and the effect of deviating from these comfortable conditions. Accurately modeling human comfort requires consideration of the physical environment and human physiology, as well as associated psychological responses (Hoppe 2002; Parsons 2003; Huang 2007; Kenny et al. 2009a). The shift towards focusing on energy balance during exercise and during varying outdoor conditions is vital in furthering the validity of models and improving outdoor recreational design.

This review will focus on how to correctly incorporate human exercise physiology into human comfort models. This is knowledge that urban planners, designers, and microclimatologists must apply in their modeling and designs. The overarching goal is to provide a solid foundation of pertinent and valid findings, which can then be used in further studies, as well as in bioclimatic design for improved urban health. The four main objectives of this review are:

  1. 1.

    To address the importance of metabolism and human physiology on the overall thermal sensation of an exercising subject; thus providing crucial knowledge required for microclimatologists and biometeorologists in human physiology and biophysical modeling.

  2. 2.

    To understand human physiology as it pertains to outdoor human comfort modeling.

  3. 3.

    To evaluate recent development of human psychological comfort models and the methodological issues involved in performing human comfort studies in outdoor environments.

  4. 4.

    To discuss the importance of developing outdoor urban environments conducive to physical activity and human thermal comfort.

Human physiology of exercise in outdoor environments

The heat balance of a human can be maintained within large limits; however, the conditions for thermal comfort are narrower and highly dependent on mean skin temperature \( \left( {{{\overline T }_{sk}}} \right) \) and sweat rate, which are dependent on the activity level (Fanger 1970). According to Hoppe (2002), the definitions and limits put forth by Fanger (1970) can be applied only under steady state conditions. Conditions outdoors are very dynamic and complex, with the major energy streams being convective heat loss (C), evaporative heat loss (E), conductive heat loss (K) radiative exchanges (R), and metabolic heat production within the body (M) (Brown and Gillespie 1986). The energy budget equation, developed from principles of heat transfer and human physiology, may be written as:

$$ 0 = M + R - C - K - E - \Delta S $$
(1)

where ∆S is the change in heat storage (W m−2), which is 0 at energy balance, >0 at energy surplus and <0 at energy deficit. This equation has been displayed and used in many different forms (e.g., Fanger 1972; Brown and Gillespie 1986; Butera 1998; Parsons 2003; Epstein and Moran 2006; Gaitani et al. 2007; Kenny et al. 2009a). Butera (1998) presents a comprehensive review of all components of the equation. When thermal imbalance occurs, it is generally assumed that thermal sensations of hot or cold are proportional to the magnitude of physiological responses of \( {\overline T_{sk}} \) and latent heat loss due to sweating (Yao et al. 2009).

All of the physical factors incorporated into Eq. 1 affect exchange of heat through the skin surface, which is the most important organ in human thermoregulation (Yao et al. 2007), and consequently plays a large role in the human thermal physiological state (Sakoi et al. 2007). Heat load from the sun warms the skin, increasing overall mean skin temperature \( \left( {{{\overline T }_{sk}}} \right) \) and localized skin temperature (T sk).

Metabolism

An accurate estimate of metabolic heat production is fundamental to the assessment of TC. Metabolic heat (M) increases with activity metabolism (M a) due to increased oxygen demands of the muscles; M also differs with gender, age and fitness level. Estimation of TC is compounded during exercise because the concept of TC changes when metabolic rate is high due to psychological mechanisms (Havenith et al. 2002). Thermoregulation is very complicated due to generation of heat related by muscular contractions (Shibasaki et al. 2006). Many studies have been completed in attempt to accurately predict M of an individual, whether exercising or stationary; however, procedures are often difficult and inaccurate. Havenith et al. (2002) tested three methods of M estimation, based on ISO8996 (1990): (1) description of the activity; (2) heart rate (HR) of the subject; and (3) direct or indirect calorimetry, in which the heat of chemical reactions within the body is measured (ACSM 2006). Direct and indirect calorimetry were deemed inappropriate for human TC studies, as they are impractical for workplace or field studies and also interfere with exercise. This study used measurements at low activity (<2 METs), where a MET is a metabolic equivalent, and 1 MET = 58.15 W m−2 is the resting rate. Therefore, Havenith et al. (2002) suggested increased research into both low and high metabolic activities; the latter due to the associated psychological mechanisms resulting in wider ranges of thermal acceptance. Havenith et al. (2002) also state that more precise measurements of M are needed for comfort assessments through classification by activity. Ainsworth et al. (2000) have classified specific physical activities and their corresponding MET, deemed the Compendium of Physical Activities. This publication has received widespread acceptance in the human health and exercise science fields (Ainsworth et al. 2000).

Use of HR, measured in beats per minute (bpm), is an acceptable guide for setting exercise intensity. HR is a measure of cardiovascular function because it has a relatively linear relationship with maximum capacity of the body to transport and use oxygen [VO2max (ml O2 kg−1 min−1); ACSM 2006], as well as with metabolic heat production when HR >120 bpm (ISO8996 1990). The applications and limitations of HR monitoring were reviewed by Achten and Asker (2003), who stated that estimating energy expenditure from HR is relatively inexpensive and easy to perform; however, complications arise with age, fitness level, weight, height and genetics. The relationship between HR and VO2max, or energy expenditure, is linear at low levels of intensity and becomes more non-linear as the intensity increases (Achten and Asker 2003). This non-linearity is due to physiological response mechanisms utilized by the body in an attempt to remain at a stable body temperature, in which the body attempts to lose heat by diffusion through skin. Thermal regulation is achieved mainly through alteration of bloodflow through vasodilation of the blood vessels when the body is warm (S >0) or vasoconstriction when the body is cold (S <0) (Butera 1998; Huizenga et al. 2001). If vasodilation is not effective enough to bring the body temperature to balance, then sweat production occurs as a mechanism to release heat from the body (Butera 1998).

Skin temperature and bloodflow

The ability to accurately measure mean skin temperature \( \left( {{{\overline T }_{sk}}} \right) \) of a human is a difficult process due to local temperature variations of the skin (T sk), and becomes more complex during exercise and sweating (Ramanathan 1964). TC is determined largely by \( {\overline T_{sk}} \) (Bulcao et al. 2000; Yao et al. 2007), and associated physiological responses, such as HR and sweat rate, are closely linked to core temperature (T c) (Bulcao et al. 2000). Actual thermal sensation (ATS) is the perception of heat or cold (what one feels) (Givoni et al. 2003). Yao et al. (2007) found that the conscious mind reaches conclusions about ATS directly from the temperature of the skin. One can sense a change in ATS through thermoreceptors located in the skin, with a high rate of temperature change (∆T) causing a greater effect on ATS. Kenshalo (1970) completed detailed research on the psychophysics relating physical stimulus and psychological sensation. He found that the firing of receptors depends on the rate of ∆T, and the threshold value for ATS or when ∆T is felt is dependent on the rate of change and the individual. The skin is relatively insensitive to slow rates of ∆T, and when in the range of 29°C to 37°C, physiological thermal adaptation occurs (Kenshalo 1970).

Physiological adaptation occurs when the body reaches a new physiological balance after being under stress or strain such as heat or cold. The adaptation and thresholds of humans will be individualized. Human comfort models generally have not considered individual variations (Huizenga et al. 2001), with essential psychological research still in its infancy (Nikolopoulou et al. 2001; Parsons 2003; Nikolopoulou and Steemers 2003). As noted by Havenith et al (2002), the meaning of ‘comfort’ changes while exercising, with reasons including: (1) an exercising individual is prepared and expectant of uncomfortable conditions (Parsons 2003); (2) different hormones are produced from the hypothalamus during exercise causing a sense of pleasure (Havenith et al. 2002); and (3) increased bloodflow will carry extra heat produced to the surface to increase convective loss (Huizenga et al. 2001).

The main function of \( {\overline T_{sk}} \) is to aid in regulation of bloodflow (Huizenga et al. 2001), and hence heat distribution throughout the body. The production of sweat can be linked closely to bloodflow distribution that leads to higher localized skin temperature (T sk) and overall \( {\overline T_{sk}} \) due to vasodilation (Fanger 1972; Huizenga et al. 2001). A well known multi-node human thermoregulation bloodflow model is the Stolwijk model (Munir et al. 2009), developed by Stolwijk and Hardy (1961). Studies by Huizenga et al. (2001), Tanabe et al. (2002), Salloum et al. (2007), Fiala et al. (2007) and Munir et al. (2009) have recently been completed using this model, which couples circulatory, thermoregulatory and energy balance systems.

Huizenga et al. (2001) assessed the bloodflow, sweating, metabolism and comfort of sedentary human subjects in complex thermal environments to improve the UC Berkeley Comfort Model, which is closely based on the Stolwijk model. This study improved the original 25-node model from six body segments to unlimited segmentation, and added five body layer segments (core, muscle, fat, skin tissue) plus clothing. The detailed model was found to be accurate in predicting bloodflow in a wide range of applications and environments. In a related study, Munir et al. (2009) found that in thermal-transient conditions and low activity, predictions of local T sk and \( {\overline T_{sk}} \) at low T a were accurate using the original 25-node Stolwijk model. When humans are not sedentary, evaporative cooling is the dominant process to maintain a stabilized T c (Parsons 2003), with \( {\overline T_{sk}} \) also falling (Fanger 1972). The calculated \( {\overline T_{sk}} \) in the Stolwijk model does not consider sweat accumulation on the skin, which produced higher experimental \( {\overline T_{sk}} \) values than those measured with latent heat loss; therefore, large errors would occur for exercise in hot environments (Munir et al. 2009). Huizenga et al. (2001) did not apply their model to exercising individuals; although this would increase complexity, it would also increase the validity of such a model. The Fiala model (Fiala et al. 1999), is a 340-node model, which takes into account all physiological and thermal aspects of the body, based largely on the Stolwijk model. The Fiala model extends the Stolwijk model to be valid under all metabolic rates, including very high activity. This, plus its ability to be applied to wide ranges of transient, physiological thermal conditions, make it a very strong model.

According to Zhang et al. (2001), the UC Berkeley comfort model must also consider individualized parameters such as height, weight, body density, body segment proportion and acclimatization, so is not applicable to a ‘typical’ person modeled by a manikin. A 65-node thermoregulation model was developed by Tanabe et al. (2002), with four tissue layers using thermal manikins and indoor simulations. They proposed the inclusion of computational fluid dynamics (CFD) with radiation and multi-node analysis, finding that the technique used is accurate in complex thermal environments. Murakami et al. (2000) also used CFD in conjunction with radiation on manikins to accurately predict actual human thermal sensation.

For studies completed on exercising individuals, in order to integrate central and local cues from the body, and predict impending fatigue, subjects can categorize their rate of perceived exertion (RPE) based on the 14-point Borg scale (6 = no intensity, 20 = highest possible intensity) (Maw et al. 1993; Parsons 2003; ACSM 2006). Maw et al. (1993) found that when subjects were asked to exercise at a ‘somewhat hard’ level of activity (60–69% of HRmax), they recorded a lower RPE when in cool temperatures (8°C) rather than in the heat (40°C). This can be due to the blood vessel vasodilation at higher ambient temperatures causing uncomfortable physiological responses such as irritation, hypotension (decreased/lowered blood pressure), and dyspnoea (difficult or laboured respiration) during exercise (Maw et al. 1993). An increase in the demand for oxygen during exercise increases cardiac output (HR) and thus bloodflow, which serves both thermoregulatory and metabolic demands (Maw et al. 1993). Normal bloodflow for a sedentary individual during thermal neutrality is 1.75 g s−1 m−2, and with each 1°C rise in T c above neutral (36.8°C), bloodflow increases by 56 g s−1 m−2 (Parsons 2003) due to vasodilation, ultimately leading to convective heat loss at the surface.

The thermoregulatory effects caused by hot and cold environments have been found to be unevenly distributed throughout the body due to mass, size and differing bloodflow resistance of the body part (Stolwijk and Hardy 1977; Sakoi et al 2007; Yao et al. 2007; Munir et al. 2009), as well as gender (Sakoi et al. 2007; Yao et al. 2007). The metabolic rate of a male is 11% higher than that of a female (Parsons 2003); therefore HR and local T sk of males is also higher (Yao et al. 2007). This was shown by Stolwijk and Hardy (1977) and Parsons (2003), who displayed how each segment (head, trunk, arms, hands, legs, feet) and their corresponding tissue compartments (core, muscle, fat, skin) have differing heat conductance (W m−2°C), bloodflow (L h−1) and heat production (W). For example, bloodflow in the core of the trunk reaches 210 L h−1, whereas it is 0.10 and 0.16 L h−1 in the core of the hands and feet, respectively, with heat production being the highest in the core of the trunk. In a study completed on sedentary subjects by Munir et al. (2009), the feet were found to have the largest standard deviation (SD) of T sk under thermally neutral conditions (SD = 1.4°C), followed by the abdomen (SD = 0.8°C). This agrees with Yao et al. (2007), who found foot T sk to be significantly lower than the rest of the body (P < 0.005), and forehead T sk the highest.

Due to this uneven temperature distribution, many different equations have been developed to estimate the \( {\overline T_{sk}} \) of a human. Yao et al. (2007) tested 12 different weighted, and two un-weighted, \( {\overline T_{sk}} \) measurement methods under steady-state conditions in a controlled chamber using copper-constantin thermocouples. The results of this study showed no significant difference between the Burton 3-point method (Burton 1934)—using the locations chest (weight = 0.50), forearm (0.14), anterior calf (0.36)—and other methods (P > 0.25), and should therefore be used preferentially for its simplicity.

The second simplest \( {\overline T_{sk}} \) weighting formula, tested by Mitchell and Wyndham (1969) was developed by Ramanathan in 1964, and is based on four points on the body: chest, thigh, calf and upper arm. From analysis of 112 measurements of three subjects over a 1-year period, Ramanathan developed the following formula for computing the \( {\overline T_{sk}} \) of a subject:

$$ {\overline T_{sk}} = 0.3{T_{chest}} + 0.3{T_{arm}} + 0.2{T_{leg}} + 0.2{T_{thigh}} $$
(2)

In Ramanathan’s study, correlations of the Hardy and Dubois 7-point formula (Hardy and Dubois 1938) were made with both the Burton 3-point formula and Ramanathan 4-point formula. Correlations of 0.94 and 0.98 respectively, were found, with “very high” significance, and a very low standard error found with the Ramanathan correlation (SE = 0.12). According to Ramanathan (1964), the basis for this formula was to consider important portions of the body that influence temperature changes of the skin, such as overactive and unclothed (exposed) muscles. Mitchell and Wyndham (1969) compared the agreement frequency of eight formulas, finding the Ramanathan 4-point method to be ideally suited for field studies on many subjects, due to its 98% rate of being within 1°C of actual \( {\overline T_{sk}} \). This formula was used by Sparks et al. (2005) in analysis of temperature on duathlon performances, and by Mora-Rodriguez et al. (2008) to measure thermoregulatory responses during constant and variable exercise.

Sparks et al. (2005) found that under higher ambient temperatures, duathletes experienced increased HR and cardiovascular strain, exhibited through high \( {\overline T_{sk}} \), sweat loss and T c. The heat stress from exercise has been shown to affect the level of performance of an athlete by decreasing their ability to perform adequately (Brotherhood 2008), and by behavioral modification at the expense of performance (Tucker et al. 2006). Significantly higher HR (P < 0.05) were also found during exercise in the heat by Tucker et al. (2006), who studied the regulation of intensity of elite cyclists in hot (35°C), normal (25°C) and cool (15°C) chamber environments (RH = 60%, vw = 10 km h−1). They found that when subjects were asked to exercise at a constant RPE of 18, or ‘very, very hard’, on the Borg scale, their work rate decreased linearly with duration in the hot chamber. This indicates that RPE does not always predict exercise intensity, and higher T a results in shorter exercise periods. Secondly, within the first 4 min of exercise at hot T a, the body showed a significant increase in \( {\overline T_{sk}} \) (P < 0.05) and heat storage (P < 0.05), which inhibited the athletes performance after >10 min of exercise.

Mora-Rodriguez et al. (2008) compared thermoregulatory responses between variable (50–90% of VO2max) and constant (60% of VO2max) exercise in a hot-dry environment to compare heat storage and heat dissipation. It was found that variable exercise increases thermoregulatory, cardiovascular and metabolic stresses in comparison with the same average metabolic heat production as constant, resulting in larger heat storage in variable despite similar heat production. They explain that variable exercise reduces conductance of heat through the skin, and consequently skin bloodflow, which would limit heat dissipation and increase storage. Secondly, increased heat storage was caused by inefficient sweating due to low windspeeds (v w = 2.5 m s−1), thus inefficient evaporative cooling would occur. This caused a reduction of the skin blood flow (11%) and decreased heat dissipation; however, \( {\overline T_{sk}} \) did not initially drop as expected when skin blood flow decreased due to the low evaporative cooling experienced. Evaporative cooling occurs during exercise in both cold and hot environments, in which the relative importance in hot environments is greater when T a > T sk, at which point convection adds to heat load (Brotherhood 2008). Under these conditions, evaporation and outgoing longwave radiation are the only heat loss terms at the skin surface.

Sweat production

The type of sweating responsible for thermoregulatory control in humans is eccrine sweating, produced from eccrine sweat glands located on every area of the body (Shibasaki et al. 2006). Shibasaki et al. (2006) thoroughly evaluated the basic mechanisms of eccrine sweat secretion as the primary controller of thermoregulation in humans during heat stress and exercise, and this paper should be referred to for more in-depth information on eccrine sweating.

There are two forms of perspiration arising from eccrine glands: sensible (E s) and insensible (E i) evaporation, where the latter is not subject to thermoregulatory control (Fanger 1970). This is because insensible perspiration is direct diffusion of vapor through the skin without droplets appearing on the skin surface, as would occur in sensible perspiration. Sweat production is dependent on the evaporative requirement (E req) for a balanced budget. This is determined by the net sum of heat arising from metabolism (M), radiation (R) and convection (C) \( \left( {{E_{req}} = M + C + R} \right) \), and must be equal achieve thermal balance (Brotherhood 2008) providing this sum is positive.

The regulation of this balance occurs through nerve signalling from receptors on the skin surface and within the body that send information to the hypothalamus (Wyss et al. 1974; Huizenga et al. 2001; Parsons 2003; Shibasaki et al 2006). The response of sweating is determined almost exclusively by the core temperature (T c) due to its direct impact on hypothalamus temperature (Wyss et al. 1974; Parsons 2003) and secondarily affected by mean skin temperature and local temperature of sweat glands (Shibasaki et al. 2006). Regulation by the hypothalamus will relay information to sweat glands on whether to open or close when the body or skin is too hot or cold, respectively (Huizenga et al. 2001). The concept of mean body temperature, which represents the sum of a weighted fraction of T c and \( {\overline T_{sk}} \), is frequently used to express sweating response (Shibasaki et al. 2006). These fractions are approximately 0.9 for T c and 0.1 for \( {\overline T_{sk}} \) (Gagge and Nishi 1977), thus displaying the relative importance of each temperature in sweating responses. Sweat rates and subsequent accumulation of vapor will vary due to individualized attributes, such as weight, height, shape, and acclimatization to the heat (Brotherhood 2008).

Acclimatization occurs due to an imbalance in the energy budget that results in a positive heat exchange towards the body. Through acclimatization, the relationship between internal temperature and sweat rate is shifted to a lower internal temperature, and the distribution of sweat may also change (Parsons 2003; Shibasaki et al. 2006), as shown in Fig. 1. Additional factors and individual characteristics may change the slope of the relationship shown. Undoubtedly, further studies are needed to understand the effects of non-thermal factors during exercise (e.g., psychology, acclimatization, expectancy), interactions between these factors on sweat response (Shibasaki et al. 2006) and further influences of sweat rate on human comfort.

Fig. 1
figure 1

The impact of heat acclimation or exercise training, and microgravity exposure or bed rest de-conditioning on the sweat rate versus internal body temperature relationship. Additional factors and individual characteristics may change the slope of the relationship. Source: Parsons (2003)

Full size image

Effects of clothing on skin temperature and thermal comfort

The impact of clothing on TC has a large effect on user satisfaction and hence consumer purchases (Huang 2007). The latest international standard for TC is ISO9920 (2007). The importance of this document can be attributed to its standardized approach for researchers worldwide to evaluate how a human is affected by the clothing they wear in a given environment. For an outline of how ISO clothing standards are produced, see Olesen and Parsons (2002). ISO9920 (2007) examines the effect of body movement and air penetration on the thermal insulation and water vapor resistance of clothing (r cvo) to evaluate thermal stress on an individual. It thoroughly outlines equations involving interactions of the physical environment with clothing ensembles and different elements of the body. For exercise specific research, the ISO document does not account for the absorption of water by clothing, or the buffering or tactile comfort of clothing (ISO9920 2007). It also does not deal with the separate insulation on different parts of the body relating to the discomfort caused by asymmetry of a clothing ensemble.

The type of clothing worn results in an added intrinsic insulation (I cl) in clothing units (clo). In this context it will be referred to as clothing resistance (r c), which acts in series with body tissue resistance (r t), and aerodynamic resistance (r a) (s m−1). Clothing adds to the resistance of convective heat and water exchanges, acting as an interface between the body and surrounding environment (Gaitani et al 2007). I cl values are typically from manikin static measurements, under ‘standard’ conditions: no air movement except for that in the test room (0.15–0.2 m s−1) (Havenith et al. 2002). According to Jones (2002), this is a large drawback since standard one-dimensional models are likely to be in error when applied to elevated air velocities. For this reason, Kenny et al. (2009b) proposed an equation using reduction factors for I cl and vapor resistances at various wind and activity speeds. From this, Eq. 3 was developed for r c:

$$ {r_c} = {r_{co}}\left( { - 0.37\left( {1 - ex{p^{\frac{{ - {v_a}}}{{0.72}}}}} \right) + 1} \right) $$
(3)

where r co is the static clothing resistance (r co = I cl × 186.6 s m−1, where 1 clo = 186.6 s m−1 = 0.1555 m2 °C−1 W−1) and v a is the movement of air over the body due to the activity being performed. It is essential to make use of a correction equation such as this in models for accurate comfort predictions due to the large effect of body movement and v a on I cl (Havenith et al. 2002).

The average insulation of body tissue from the core of the body to the skin surface is 0.16 clo during vasodilation, as compared to 0.64 clo for vasoconstriction (Monteith and Unsworth 1990). Nielsen (1990) measured the heat load imposed by the sun’s radiation on subjects cycling outdoors for 60 min wearing black (albedo = 0.28) or white (0.45) clothing. They found that black clothing resulted in a 3–4°C higher \( {\overline T_{sk}} \), a 10 bpm increase in HR, and 100 g h−1 increase in sweat rate. Nielsen (1990) commented that their initial study was completed outdoors, but weather conditions were so unstable that it was reproduced indoors under artificial sunshine. Hence, the dynamic microclimate changes could not be experienced by the subjects. Kenny et al. (2008) also found greater heat load on darker surfaces, finding that the mean change in radiation between albedo’s of 0.21 and 0.57 ranged from 115 to 157 W m−2, respectively. Integration of such findings—as well as accurate meteorological measurements—into existing models allows for improvement of human comfort prediction.

Modeling human comfort

Many human comfort models developed in the last half-century attempt to predict, at best approximation of reality, how a human feels in a given environment. This section will explore how human–environment interactions are integrated into thermal models in varying environments, and the role psychology plays in comfort. Meteorological measurements within the dynamic outdoor environment are required for human comfort modeling due to small- and large-scale changes within the environment, as well as ‘surface–plant–air’ urban interactions, as studied by Bruse and Fleer (1998). Such interactions are important; therefore, applying indoor research for prediction of outdoor human comfort should be used with caution due to the possibility of inaccurate applications of bioclimatic design. Comfort models range from simple, one-dimensional, steady-state simulations (e.g., Fanger model, Gagge two-node model) to complex, transient, finite element codes with thousands of nodes (Jones 2002). The limitations of each model must be understood by the user, and situations of valid or invalid accuracy must be known.

According to Spagnolo and de Dear (2003), the need for an outdoor TC model has led to the International Society of Biometeorology (IBC), and the World Meteorological Organization (WMO) to form a ‘working commission’ and integration of scientists to develop the UTCI. The UTCI is currently only available online (UTCI 2009); published literature involving components of its development include a clothing model by Richards and Havenith (2007), a physiological model by Fiala et al. (2007), and further components by Psikuta et al. (2008), Jendritzky et al. (2008), and Mateeva et al. (2009). Models currently in use in refereed literature for human comfort modeling include the COMFA model (Brown and Gillespie 1986), Givoni method (Givoni et al. 2003), the Rayman Model (Matzarakis 2000), the SOLWEIG model (Lindberg et al. 2008), ENVI-Met (Bruse 2004), and the Munich Energy Balance Model for Individuals (MEMI; Hoppe 1984); however, specific discussion of each model does not fall within the scope of this paper.

The MEMI and COMFA models are similar, both focusing on T c, \( {\overline T_{sk}} \) and clothing or surface temperatures (T cl, T s). Both differ from Gagge (1971) in that they calculate sweat rate; however, MEMI differs from COMFA in its estimation of metabolism and sweat rate from age and sex, as well as utilization of body surface area, bloodflow density from core to skin, density of blood, and specific heat capacity of blood, in its heat flux calculation from core to skin (Hoppe 1999). The Fiala-model (Fiala et al. 1999, 2007) also uses multi-layer and multi-node calculations, as discussed in the section “Skin temperature and bloodflow” above. Even though COMFA does not use such specific information, its ability to accurately predict ATS votes was shown by Kenny et al. (2009a, b) in exercising individuals under moderate rates of metabolism and sweat.

Microclimatic parameters in human comfort modeling

A model can only be as accurate as its inputs. Within the four meteorological input parameters being T a, RH, v w and R, TC is influenced specifically by R, or the mean radiant temperature (T mrt) (Matzarakis et al. 2007; Thorsson et al. 2007b; Kenny et al. 2008), with sunlight being the main component of environmental heat load (Brotherhood 2008). T mrt is defined as the uniform temperature of an imaginary enclosure in which the radiant heat transfer from the human body equals the radiant heat transfer in the actual non-uniform enclosure (ASHRAE 1966; Fanger 1972; Parsons 2003), or the sum of all short- and long-wave radiation fluxes required for the human energy balance equation (Matzarakis et al. 2007; Thorsson et al. 2007b; Lindberg et al. 2008). The radiant exchange imposed on a human subject has been said to be the main indicator of TC (Matzarakis et al. 1999, 2007; Parsons 2003; Hodder and Parsons 2007; Kenny et al. 2009a), supported by Gulyas et al. (2006) who found a strong relationship between heat load on the body, expressed as the physiological equivalent temperature (PET) (Hoppe 1999) and T mrt using the Rayman model. Hodder and Parsons (2007) studied the effects of direct solar radiation on comfort in sedentary people, mainly in vehicles. They calculated \( {\overline T_{sk}} \) using the Ramanathan T sk formula under four radiant conditions: 0, 200, 400 and 600 W m−2, with all other physical parameters held constant. ATS votes showed an increase in discomfort with radiation, and \( {\overline T_{sk}} \) values showed a significant difference between the control compared with each radiation (P ≤ 0.01), but no significant difference between 400 and 600 W m−2 (P = 0.12). Gulyas et al. (2006) found the Rayman model to be valuable for planners, decision makers and practitioners in new urban areas. When comparing street canyons with and without trees, they found the mean difference in T mrt and PET to be 7.0°C and 4.6°C higher without trees present.

According to Lin et al. (2010), shading is an important component of long-term outdoor TC, specifically within urban canyons (Ali-Toudert and Mayer 2007). Lin et al. (2010) completed research using 10 years of meteorological data in Taiwan—a hot, humid region where thermal index is high (Thorsson et al. 2007a). Also using the Rayman model, with PET as their primary TC index, Lin et al. (2010) found modeled T mrt and measured T mrt to be strongly correlated (R 2 = 0.85, P < 0.001) when not at a high T mrt. Through the use of an ENVI-Met three-dimensional microclimate model (Bruse 2004), Ali-Toudert and Mayer (2007) used PET patterns to map improved urban design aspects for shading, such as trees and overhangs, which showed improved climate quality in an urban street. Emmanuel et al. (2007) used the ENVI-Met model to study urban shading in Sri Lanka; however, shortcomings were found in the model, which include: (1) buildings have no thermal mass; (2) constant indoor temperatures and albedos of buildings; and (3) the thermal transmittance of walls and roofs is the same for all buildings. ENVI-Met also lacks the ability to simulate sea breezes within their model, which is an important component for urban ventilation (Emmanuel et al. 2007).

Further environmental parameters with a large capacity to impact heat load are vapor pressure (e) and v w, both highly associated with sweating. Including activity speeds (v a) in TC models can improve accuracy based on increased convective heat loss and local air movement, thus a relative velocity (v r) can be incorporated as a geometric mean of v a and v r, given by Holmer (1999) and used by Kenny et al. (2009b). During active sweating, the airflow over the skin and vapor pressure—rather than RH—determines the sweat rate (Brotherhood 2008). When sweating, the vapor pressure at the skin surface is assumed to be 100% RH; therefore, absolute humidity is used because it is a better estimator of evaporative demand of air above the skin caused by the vapor gradient above the skin surface. Metje et al. (2008) completed a study under varying weather conditions from August 2004 to February 2005. They used data sets from five European countries at pedestrian levels and determined that v w and T a have clear influences on human comfort (R 2 = 0.87 and R 2 = 0.88, respectively). However, Metje et al. (2008) found a weaker relationship between the apparent temperature index (linear combination of RH, v w and T a) and comfort values, (R 2 = 0.3125), confirming that meteorological values on their own cannot predict a person’s comfort, and that adaptation and subjective responses play a large role.

Modeling comfort during exercise in steady-state versus dynamic environments

Most well-validated heat budget models (e.g., Fanger 1970; Gagge 1971) have been developed from indoor laboratory studies (steady-state) that are non-complex (Huizenga et al. 2001; Hoppe 2002; Givoni et al 2003). These are not always appropriate for measuring heat stress and human comfort outdoors due to the inability to accurately recreate the ambient environment (Brotherhood 2008) and differing perceptions of indoor and outdoor environments (Hoppe 2002). Indoor experiments allow the body to achieve equilibrium with the environment in a short time (Huizenga et al. 2001; Hoppe 2002); however, when exercising outdoors for short periods of time, the body may not reach this equilibrium (Hoppe 2002). According to the ACSM (2006), it takes 10–20 min of transition from rest to exercise to reach a relatively constant metabolic activity rate (M a) considered adequate for moderate exercise (60–69% HRmax, where HRmax = 220−age) (Lamb et al. 1999), and T c may take an additional 10–20 min to stabilize (Maw et al. 1993; Sparks et al. 2005). For higher intensity exercise, time from rest to constant M a and stabilized T c will take longer if the goal is to maintain equilibrium at a higher intensity than moderate (ACSM 2006).

Outdoor TC studies can account for this range of metabolic output in their budget calculations, as well as a range of meteorological conditions. One such model is the COMFA mathematical model, which makes use of microclimatic, clothing and physiological data to predict the energy budget of a human based on Fanger’s 1970 model and Eq. 1. However, Kenny et al. (2009b) found the COMFA model did not account for the ‘lag’ in the rise in T c for the first 15 min of exercise, thus over-estimation of T c occurred until equilibrium was reached since T c is based on metabolic activity. There was also a corresponding lag (underestimation) in subjective discomfort, which can be explained by a slower change in brain and other internally sensed temperatures than that physically measured (Wyss et al. 1974). This may require future TC studies on human activity to directly measure T c, as well as mean skin temperature \( \left( {{{\overline T }_{sk}}} \right) \), rather than relying on equations based on metabolism, in order to account for this apparent lag.

TC is positively correlated with both local skin temperature from specific body parts (T sk) and mean skin temperature of the body \( \left( {{{\overline T }_{sk}}} \right) \), which have a large effect on ATS (Mairiaux et al. 1987; Yao et al. 2007); thus, adequate modeling of \( {\overline T_{sk}} \) is a very important aspect of TC prediction. Equation 4 displays how aerodynamic resistance (r a), tissue resistance (r t), and r c affect \( {\overline T_{sk}} \), as well as T c and T a.

$$ {\overline T_{c}} = \left( {\frac{{{T_{a}} - {T_c}}}{{{r_t} + {r_c} + {r_a}}}} \right)\left( {{r_a} + {r_c}} \right) + {T_a} $$
(4)

With use of this equation in the COMFA model, Kenny et al. (2009b) found COMFA to produce \( {\overline T_{sk}} \) values more consistent with Mairiaux et al. (1987), Maw et al (1993) and Sparks et al. (2005) than other equations; other current models also do not adequately address actual \( {\overline T_{sk}} \). The lack of accurate predictions in overall/local ATS is a weakness in modeling (Yao et al. 2007). This can be attributed to individual differences in magnitude of change in T sk and its distribution, as well as conflicting contributions from local T sk and/or internal T c. These issues are an integral part of understanding human TC and must be addressed in future research.

It is already known that, in uniform conditions, a strong linear relationship exists between ATS (what one feels) and TC (balanced energy budget) (Gagge 1971; Zhang and Zhao 2009). An indoor study that attempted to replicate outdoor dynamic conditions was completed by Zhang and Zhao (2009). They studied the relationship between predicted TC and ATS in a dynamic chamber environment, and found that ATS showed a significant relationship (R 2 = 0.91, P < 0.001) with predicted TC in steady-state uniform conditions. In steady-state non-uniform conditions, the change in space was found to be important in determining TC (R 2 = 0.89, P < 0.001); in dynamic environments, the change in time is the most important factor for TC. Researchers concluded that non-uniform environments should be studied because they affect thermoregulation and TC in all body segments.

The spatial and temporal environmental changes are a basis for designing with both comfort and energy conservation in mind. A three-dimensional model that fulfills such requirements is the ENVI–Met model (Bruse 2004), which considers further interactions within the urban boundary layer, such as flow around and between buildings, turbulence, vegetation vapor exchanges, and heat/vapor from ground surfaces and walls. These variables aid in the prediction of the microclimatic characteristics for improved accuracy on the microscale for such applications as effects of urban shading (Ali-Toudert and Mayer 2007).

Future research outdoors should continue, where higher work rates, airflows, and humidities drastically impact thermoregulation (Gavin 2003), and thus measurements of T c and T sk. However, personal preference and psychological perception of the type of outdoor design can also have an effect on whether a person has a pleasing or displeasing feeling of overall ATS.

Psychological aspects of thermal comfort modeling

The energy budget provides information on how variables ought to be combined and manipulated to create optimal TC; however, ATS is a physiological and psychological response. According to Jones (2002), the biggest limitation of comfort models is the accuracy with which comfort perceptions can be related to the simulated physiological variables. In addition to this, differing perceptions of what is ‘comfortable’ occur outdoors because people are more accepting of broader climatic conditions (Johansson and Rohinton 2006). Many non-physical factors, such as expectations of weather, affect comfort perceptions (Jones 2002; Hoppe 2002; Nikolopoulou and Steemers 2003; Metje et al. 2008; Knez et al. 2009). Factors that alter perception cause psychological adaptation, and include variables such as time of year, perceived control, time and length of exposure, naturalness, readiness for exercise, and experience (Nikolopoulou and Steemers 2003), as well as exercise frequency. Ahmed (2003) completed a study in the Tropics, finding that individuals who were walking at 3.0 METs reported being comfortable at 0.2 K higher than subjects at 2.0 METs, thus illustrating the broader range of acceptance. Models that make use only of meteorological and physiological parameters are missing critical components. How can psychological variables be measured or included in human comfort models?

Psychological evaluation of how one feels can be accomplished using the predicted mean vote (PMV) parameter developed by Fanger (1970), which ranges from −3 (cold) to +3 (hot), with 0 as neutral. This seven point ‘psychophysical scale’ makes it possible to predict ATS for any given combination of the six basic physical parameters through the use of the human energy budget equation (Fanger 1972). This subjective evaluation must be distinguished from how a person might like to feel and what the energy balance of the environment is. Additional indices for thermo-physiological assessment include physiological equivalent temperature (PET) (Mayer and Hoppe 1987; VDI 1998; Hoppe 1999), standard effective temperature (SET) (Gagge et al, 1986) and OUT_SET for outdoor use (Pickup and de Dear 2000; Spagnolo and de Dear 2003). Matzarakis et al. (2007) make use of the three model indices PMV, PET and SET as part of the Rayman radiation model, and Yao et al. (2007) as well as Thorsson et al. (2007b) made use of the PMV as an assessment tool. Lin et al. (2010) found the Rayman model to accurately predict PET on an hourly basis.

However, limitations to these indices still exist due to the inaccuracies associated with inputs, physical simulation, and psychological perceptions (Jones 2002; Yao et al. 2007), and they are not suitable to assess non-uniform environments (Zhang and Zhao 2009). PMV is also limited by the fact that it expresses warm and cold of the body as a whole (Olesen and Parsons 2002); however, as discussed previously, local discomfort is common, but persons engaging in physical activity are less thermally sensitive due to psychological and activity (metabolic) adaptations (Olesen and Parsons 2002) and have a broader comfort zone (Kenny et al. 2009a). The actual thermal sensation of an individual rated using thermal indices has a strong positive relationship with T a, as found by Metje et al. (2008), who used PMV as their psychological thermal index.

Yao et al. (2007) suggest using a thermal acceptability approach that is based on human physiology, focusing primarily on \( {\overline T_{sk}} \) since it is the most important parameter relating to TC, and removing the subjectiveness associated with psychological methods. Yao et al. (2009), and others (de Dear and Breager 1998; Fanger and Toftum 2002) renewed the notion of the adaptive PMV (aPMV) put forth by Fanger (1972), who discussed geographic location, gender, age, acclimatization, ethnicity, expectancy, and psychological changes associated with the comfort equation. Yao et al. (2009) used a “Black Box” adaptive model method in which stimuli act as inputs, with observed reaction being the system output. Through the use of thermal environment surveys, they found an adaptive coefficient for PMV for the relationship between ATS and predicted thermal sensation for both warm and cool conditions, as shown in Eqs. 5 and 6.

Warm conditions:

$$ aPMV = \frac{{PMV}}{{1 + 0.293 \times PMV}} $$
(5)

Cool conditions:

$$ aPMV = \frac{{PMV}}{{1 - 0.125 \times PMV}} $$
(6)

This experiment, however, was completed under indoor conditions and cannot be applied to the complex outdoor climate, and thus lacks critical information for outdoor use of the PMV as a conceptual assessment tool. Even so, it is important to note that some indoor studies (e.g., Paciuck 1990; Nielsen 1990; Brager and de Dear 1998; de Dear and Brager 2002; Hoppe 2002; Brager et al. 2004; Yao et al. 2007) may contain useful information for outdoor spaces, as there are similar phenomena that can be applied. Specific discussion of indoor modeling does not fall within the scope of this paper, and the examples listed should therefore be referred to for further understanding of this topic.

Humans have the physiological and behavioral ability to adapt more easily to cold conditions (Nikolopoulou and Lykoudis 2006), thus each climatic zone (see Peel et al. (2007)) should have distinct equations for each season for prediction of TC due to differing levels of exposure to hot and cold weather. Within workplace situations, individual behavior for control of indoor environment varies, based on personal factors and psychological mechanisms (Paciuck 1990); however, control of outdoor environments is not as easy.

Further psychological mechanisms and personal factors were addressed by Knez et al. (2009) and Nikolopoulou and Lykoudis (2006), who related conceptual models to the design of outdoor spaces. Employing a series of questions to 2,375 individuals in Gothenbburg (Göteborg), Sweden, Knez et al. (2009) found a significant relationship of outdoor activity with T a (P = 0.000) and clearness index (P = 0.001) and found that younger participants significantly estimated the current weather as calmest, independently of the weather (P = 0.027). Nikolopoulou and Lykoudis (2006) presented findings of psychological evaluation of the climate across many European countries from the RUROS dataset (Rediscovering the Urban Realm and Open Spaces), which explored 14 case studies, with over 10,000 questionnaires and corresponding microclimate data. They found that recent experience and expectations are responsible for over 10°C variation from neutral budget temperatures, closely following seasonal climatic temperatures. The comfort ranges for spring and autumn (17.6 and 13.5°C) were also found to be wider than for summer and winter (5.9 and 9.6°C), with summer having the smallest.

The subjective rating of ATS is a commonly used method where the subject rates their ATS on a 7-point scale (+3 = hot, 0 = neutral, −3 = cold) (Fanger 1972). ATS scales are commonly used in TC studies, whether sedentary or exercise-based. ATS votes have been found to shift from a normal distribution around ‘neutral’ (0) (Nikolopoulou and Lykoudis 2006) to a normal distribution around ‘slightly warm’ (+1) with moderate physical activity (Kenny et al. 2009a). Subjects are asked to rank their overall ATS, as well as RPE, at specific time intervals, and to avoid singular factors such as leg pain or shortness of breath (ACSM 2006) since overall feelings of comfort follow closely to parts of the body that are the most uncomfortable (Yao et al. 2007). It has been found that subjective responses and fatigue are affected by the surrounding microclimate (Maw et al. 1993; Sparks et al. 2005; Brotherhood 2008). Psychology and physiology are also altered at higher metabolic rates (Kenny et al. 2009b). Many researchers are in the process of studying such factors as the bias caused by expectancy, exercise, seasonal variation and climatic adaptation. However, it is clear that appropriate microclimatic design and planning can decrease negative aspects and increase positive exposures, resulting in increased use (Nikolopoulou and Steemers 2003). Nikolopoulou and Steemers (2003) suggest that careful planning and design can decrease negative aspects of the climate and increase the use of outdoor spaces, which requires the above noted information in order to successfully create outdoor urban areas.

In the Tropics, the negative aspects of poor urban planning are more pronounced, showing ‘progressive degradation’ of the physical environment (Ahmed 2003). Ahmed (2003) took the approach of linking indoor and outdoor TC models to facilitate design of free running buildings; thus, demand for TC indoors would result in improved application of bioclimatic design outdoors and less energy use. Linking indoor TC to outdoor TC may also be a critical factor in improving indoor conditions and improving human health for decreased mortality.

Bioclimatic urban design and human health

Bioclimatic urban design has become a critical tool in designing and describing climate-sensitive areas (Eliasson et al. 2007), with considerable focus on how to physically change the meteorological variables found in the human energy balance equation to result in a more balanced human energy budget. It is well known that air temperatures in densely populated urban areas are greater than in rural areas (Gaitani et al. 2007), differing by up to 12°C on calm clear nights due to daytime heat storage (Oke 2003) and displaying horizontal gradients of up to 5°C km−1 (Galea and Vlahov 2005). These are commonly referred to as ‘urban heat islands’ (UHI; Basu and Samet 2002; Arnfield 2003; Gaitani et al. 2007; Watkins et al. 2007). Arnfield (2003) completed a detailed review on urban climate research, citing 31 reviews, bibliographies and summaries specific to urban climatology that have been released since 1980, which display the large amount of conceptual advances made in the discipline.

With sprawling urbanization and increased apprehension concerning climate change, the importance of designing with climate in mind has lead to numerous studies involving bioclimatic urban design (e.g., Eliasson 2000; Nikolopoulou and Steemers 2003; Eliasson et al. 2007; Watkins et al. 2007; Knez et al. 2009). Proper urban design and planning is a mechanism of preventative medicine since physical inactivity is associated with a decline in human health, causing economic strain from rising health care costs.

It has been shown that an increase in thermal discomfort decreases the number of people present in a given area, self-motivation of individuals, and the amount and intensity of human activity or exercise (Givoni et al. 2003; Eliasson et al. 2007; Kenny et al. 2009a). Additionally, discomfort decreases work and athletic performance, and prompts poor behavior (Eliasson 2000; Galea and Vlahov 2005; Johansson and Rohinton 2006; Harlan et al. 2006; Eliasson et al. 2007; Watkins et al. 2007). Higher T a is associated with increased levels of air pollution and heat stress (Watkins et al. 2007), and ultimately a deterioration of social and outdoor activity (Johansson and Rohinton 2006).

Long periods of persistent high T a at the synoptic scale, relative to climate norms, are known as heatwaves, and have been found to increase heat-related mortality of vulnerable individuals (Smoyer et al. 2000; Basu and Samet 2002; Galea and Vlahov 2005; Watkins et al. 2007). The 2003 European heatwave resulted in 14,800 excess deaths within 3 weeks due to heat stress leading to high heat strain and ultimately death for vulnerable individuals (Watkins et al. 2007). The 1980 and 1988 heatwaves in the United States recorded 10,000 and 5,000–10,000 excess deaths, respectively (Smoyer et al. 2000). In many cases, vulnerable individuals are kept indoors due to sickness; thus, the above case studies show the large effect of heat waves on indoor TC, as well as outdoor. With the use of climate-sensitive designing practices, outdoor and indoor heat stress can be reduced, and heat-related mortality can be lessened.

Applications of bioclimatic design

Arnfield (2003) reviewed two decades of urban climate research due to a large expansion in the associated literature. A summary of notable conclusions and future research suggestions include: (1) furthering the development of valid methodology and modeling concerning the boundary layer within UHIs, water balances, and anthropogenic fluxes; (2) improving estimates of the effects of UHIs for practical applications, such as road climatology, energy conservation and weather forecasting; and (3) using the ‘explanatory power’ of bioclimatic design for future developments. Knowledge of the complex issues involved in outdoor TC is necessary to provide better guidance to those involved in urban planning and design (Thorsson et al. 2007a).

According to Harlan et al. (2006), land use patterns and surface cover are the two most important parameters that effect the spatial distribution of the UHI. Man-made changes to the surface, such as pavement and buildings, have high heat capacities and replace sources of surface moisture and vegetation (Pearlmutter et al. 2009). With respect to UHI intensification, Watkins et al. (2007) address the synergies between meteorological variables and their effect on TC and the UHI. Taking meteorological measurements at approximately average human height (1.5 m–1.8 m) (de Beer 2004) is important for accurately predicting TC of humans. Up to a certain height, energy heat exchange from the surface controls T a and is highly dependent on windspeed (Svensson 2004). The UHI effect creates a strong urgency to strive for improved urban climate conditions (Johansson and Rohinton 2006). Although planners are well aware of the importance of the microclimate on outdoor usage, it has been recognized that climate issues often have a low impact on the urban planning process (Eliasson 2000; Eliasson et al. 2007).

The vast range of knowledge required for urban microclimatic design was studied by Eliasson et al. (2007) in Sweden using scientists from the three disciplines of architecture, climatology and psychology. Their study combined three meteorological aspects—clearness index, T a, and v w—with behavioural aspects, and compared this to perceptions of current weather from volunteer surveys conducted in four study areas during summer and winter seasons. Results clearly showed that the three meteorological variables had significant impacts on the perceptions of weather, location, and feelings (P < 0.05), which explains emotional states and attendance at each study site, with T a explaining up to 63% of the variance of attendance (R 2 = 0.63). The stimulation of v w also allows for dispersion of low-level traffic pollutants that harm the respiratory systems of exercisers; hence, studies examining the relationship between air pollution and air temperatures have been warranted (Basu and Samet 2002).

Physiology, psychology, and bioclimatic design are the three main facets involved in modeling human TC. Surveying studies (e.g., Givoni et al. 2003; Johansson and Rohinton 2006; Gaitani et al. 2007; Eliasson et al. 2007; Metje et al. 2008; Knez et al. 2009) support the future importance and usage of climate-sensitive planning and site-specific design based on the particular microclimate, and the general importance of understanding variable perceptions of environments. Applications of human comfort models range from outdoor cafes and parks, to all sports field and golf courses, and can have significant economic benefits for businesses if utilized properly. Nakano and Tanabe (2004) studied semi-outdoor environments, such as atriums and cafes, and associated behavioral adaptation. They found occupants to be 2–3 times more tolerant to comfort ranges than the predicted dissatisfaction even though occupants were not fully outdoors.

Recently, the notion of the important effects of climate on tourism has been applied by Matzarakis and Rutz (2005) and Lin and Matzarakis (2008). This is of prime importance for tourism companies due to the short-term exposure of tourists to the local climate (Lin and Matzarakis 2008). Both these studies used the Rayman software for T mrt estimation. Lin and Matzarakis (2008) used different PET classes for Taiwan and Western/Middle Europe, while developing a ‘climate tourism information scheme’ to be used by tourists and tourism authorities. A more specific 10-day climate scheme was used, rather than monthly, in order to view variations within the two locations.

Conclusions

The objective of this paper was to provide an integrated view of the complex discipline of Biometeorology, and to present readers with a common ground of knowledge on a variety of topics. These range from human physiology and perceptive psychology, to application of climate-sensitive urban design and its effects on human health. Knowledge of human physiology, including skin temperature, sweat rate, and core temperature, are vitally important in understanding thermoregulation of the human body for application in human comfort models; however, urban planners and designers are commonly not well-acquainted with human comfort models and physiological knowledge. Urban planners must properly apply bioclimatic research and knowledge gained by human biometeorologists in order to develop sustainable and comfortable urban areas. Research in this field has shown large advancements in the past few decades, as illustrated by the recent developments mentioned within this review. Further research is required to fill gaps in human comfort research, such as skin and core temperature estimation, psychological factors, and clothing parameters (Kenny et al. 2009b).

The large role of perceptive psychology in human comfort is prominent in current research; however, further research is required in many aspects of psychological mechanisms, such as the effects of seasons, exercise, expectancy, acclimatization, and sport-specific heat stress (Jones 2002; Nikolopoulou and Steemers 2003; Knez et al. 2009). This is a relatively new area of research, where agreement on suitable psychological models has yet to be met. An adequate TC model should also incorporate the accurate meteorological variables affecting outdoor human comfort. The use of convenient and precise equipment by urban planners is essential for designing thermally comfortable city spaces. Brotherhood (2008) advocates further research into sport-specific heat stress and the related human physiology, which is critical for designing outdoor recreational facilities.

Many studies have failed to consider all above mentioned aspects simultaneously, and thus incorrectly predict human comfort. The use of adaptive models should be utilized in different regions of the world with respect to climates (Lin 2009; Yao et al. 2009). Indoor TC models are far more common in the literature than outdoor; however, they cannot be applied outdoors (Brotherhood 2008), and sedentary-based models cannot be used to model higher metabolic rates. For activity and sport-specific algorithms, more detailed and complex analysis is required in order to avoid heat stress (Budd 2008). The applicability of indoor models lies in the connection between outdoor and indoor comfort. Heat-related mortality statistics warrant increased use of climate-sensitive design in urban areas. This connection is vital in reducing the indoor effect of UHIs, and through simultaneous use of both outdoor and indoor comfort models, proper bioclimatic design can lead to improved indoor comfort as well.

There is no doubt that the interdisciplinary field of Human Biometeorology continues to be an increasingly prominent and fascinating topic in the literature, with a great deal of research needed in all aspects of the discipline. The possible benefits to society of such research has been successfully documented, with the health and overall well-being of urban dwellers of prime importance.