MRT.pdf

Seediscussions,stats,andauthorprofilesforthispublicationat:http://www.researchgate.net/publication/235777384

Characteristicsofthemeanradianttemperatureinhighlatitudecities-implicationsforsensitiveclimateplanningapplications

ARTICLEinINTERNATIONALJOURNALOFBIOMETEOROLOGY·FEBRUARY2013

ImpactFactor:2.1·DOI:10.1007/s00484-013-0638-y·Source:PubMed

CITATIONS

6

DOWNLOADS

39

VIEWS

171

4AUTHORS,INCLUDING:

FredrikLindberg

UniversityofGothenburg

49PUBLICATIONS611CITATIONS

SEEPROFILE

BjörnHolmer

UniversityofGothenburg

40PUBLICATIONS491CITATIONS

SEEPROFILE

SofiaThorsson

UniversityofGothenburg

43PUBLICATIONS714CITATIONS

SEEPROFILE

Availablefrom:SofiaThorsson

Retrievedon:15July2015

ORIGINAL PAPER

Characteristics of the mean radiant temperature in high latitudecities—implications for sensitive climate planning applications

Fredrik Lindberg & Björn Holmer & Sofia Thorsson &

David Rayner

Received: 20 December 2012 /Revised: 26 January 2013 /Accepted: 27 January 2013# ISB 2013

Abstract Knowledge of how the mean radiant temperature(Tmrt) is affected by factors such as location, climate andurban setting contributes to the practice of climate sensitiveplanning. This paper examines how Tmrt varies within anurban setting and how it is influenced by cloudiness. Inaddition, variations of Tmrt in three high latitude cities areinvestigated in order to analyse the impact of geographicalcontext and climate conditions. Results showed large spatialvariations between sunlit and shaded areas during clearweather conditions, with the highest values of Tmrt close tosunlit walls and the lowest values in the areas shaded bybuildings and vegetation. As cloudiness increases, the spa-tial pattern is altered and the differences are reduced. Thehighest Tmrt under cloudy conditions is instead found inopen areas where the proportion of shortwave diffuse radi-ation from the sky vault is high. A regional comparisonbetween three Swedish coastal cities showed that Tmrt dur-ing summer is similar regardless of latitudinal location. Onthe other hand, large differences in Tmrt during winter werefound. Shadows, both from buildings and vegetation are themost effective measure to reduce extreme values of Tmrt.However, extensive areas of shadow are usually not desiredwithin outdoor urban environments at high latitude cities.One solution is to create diverse outdoor urban spaces interms of shadow and also ventilation. This would provideindividuals with access to a choice of thermal environmentswhich they can use to assist their thermal regulation, basedon personal needs and desires.

Keywords SOLWEIG . Urban geometry . Mean radianttemperature . Sweden . Göteborg . Luleå . Stockholm

Introduction

Mean radiant temperature (Tmrt) is one of the most importantmeteorological parameters governing human energy balanceand thermal comfort outdoors, especially during clear andcalm summer days (Mayer and Höppe 1987). Tmrt is definedas the ‘uniform temperature of an imaginary enclosure inwhich the radiant heat transfer from the human body equalsthe radiant heat transfer in the actual non-uniform enclosure’(ASHRAE 2001) and its spatial variations during the dayare chiefly influenced by shadow patterns generated byobstructing objects such as trees, buildings and generaltopography (Lindberg and Grimmond 2011b). Other factorsthat affect Tmrt are the thermal and radiative properties ofsurrounding surface materials (albedo, emissivity, thermaladmittance, etc). Compared to other parameters influencingthe thermal comfort, such as air temperature (Ta) and airhumidity, Tmrt shows large spatial variations over shortdistances. Mayer et al. (2008) demonstrated this by compar-ing Ta and Tmrt in one sunlit and one shaded locationadjacent to each other in the city of Freiburg, Germanyduring warm and clear summer weather. The differences inTmrt could be as much as 37 °C whereas Ta showed adifference of only 1–2 °C between the two locations.Cloudiness affects the spatial pattern of the radiative fluxesfrom solid surfaces (ground, buildings and vegetation, etc.)as well as from the sky vault, which in turn affect Tmrt. Thereduction of direct shortwave radiation with increasedcloudiness reduces the differences between sunlit and shad-ed areas. The proportion of diffuse radiation from the skyvault increases with increased cloudiness. The differencebetween the surface temperatures of sunlit and shaded sur-faces also reduces as cloudiness increases, which in turnaffects the long wave radiation fluxes (Lindberg et al. 2008;Lindberg and Grimmond 2011a; Konarska et al. 2012).Another important determinant of the outdoor human

F. Lindberg (*) :B. Holmer : S. Thorsson :D. RaynerGöteborg Urban Climate Group, Earth Science Centre, Universityof Gothenburg, Box 460, 405 30 Göteborg, Swedene-mail: [email protected]

Int J BiometeorolDOI 10.1007/s00484-013-0638-y

thermal comfort is the convective cooling of wind, whichalso tends to have large variations within short distanceswithin the urban environment (Mochida and Lun 2008).

The most accurate method to estimate Tmrt is to measurethe short-wave and long-wave radiation fluxes (upward,downward and from the four cardinal points) along withinformation on body posture (sitting, standing, etc.) andabsorption coefficients of shortwave and longwave radiationof the human body (Höppe 1992). The method requires theuse of pyranometers and pyrgeometers, arranged in the sixdirections, which makes it costly and difficult to implementin extensive measuring campaigns (e.g. Thorsson et al.2007). A simpler but less accurate way of measuring the Tmrtoutdoors is to use a globe thermometer along with air temper-ature and wind speed measurements (e.g. Nikolopoulou et al.1999). Observations are in general stationary and cannot pro-vide information on the spatial patterns of Tmrt. As a conse-quence, it is more common to make use of a model in order toobtain estimations of Tmrt (Bruse and Fleer 1998; Lindberg etal. 2008; Matzarakis et al. 2009).

Predictions of human induced climate change suggestincreases in near surface air temperatures anywhere between0.5 and 6.5 °C over the next 100 years (IPCC 2007). It isalso predicted that heat wave episodes will become morefrequent, more intense and longer (Meehl and Tebaldi 2004).It is further suggested that climate change could magnify theurban heat island effect in some locations (McCarthy et al.2010). This calls for an improved understanding of how thevariations of microclimates within an urban setting influencepeople’s health and well-being. Of particular importance arehuman thermal comfort issues. During extreme heat waves,like that which occurred in central Europe in 2003, heat stresscan have profound effects on people’s health and well-being,with substantial economic consequences (Pascal et al. 2006).By taking climate issues into consideration in the urban plan-ning process it is possible to improve outdoor thermal comfortconditions, i.e. reducing heat and cold stress as well asprolonging the periods of comfortable conditions, thus alsothe health and well-being of the urban citizens. It is suggestedthat Tmrt is a better measure to analyze the impact of weatheron people’s health compared to air temperature or apparenttemperature (Thorsson et al. 2012). By improving knowledgeabout how Tmrt varies within the urban environment duringdifferent weather conditions, and about the influence of urbanmorphology and vegetation on the radiative properties andthus Tmrt, it is possible to more accurately identify risk areasand take appropriate measures to reduce heat stress in theseareas.

In this paper we explore how the spatial pattern of Tmrt inan urban setting is influenced by cloudiness and how Tmrtvaries across different seasons in high latitude cities. Theresults are discussed in terms of their implications for urbanplanning and design. Three Swedish cities at different

latitudes were selected for this case study (Göteborg,Stockholm and Luleå). Tmrt was simulated using the SOlarand LongWave Environmental Irradiance Geometry(SOLWEIG) model (Lindberg and Grimmond 2011a;Lindberg et al. 2008).

Methods

Study area for spatial analysis

To explore the spatial variations of Tmrt for a standingperson in the urban environment, a 3-D representation ofbuildings and vegetation is acquired. In this paper, a ‘realworld’ study area was selected covering a part of the citycentre of Göteborg, located on the Swedish west coast(57.70° N, 11.94° E). The city centre has a classicalEuropean design, characterized by a compact mid-rise struc-ture with little vegetation. The study area selected covers thearea south of the main train station as shown in Fig. 1. In thesouth area of the domain one of the major parks in Göteborg(Trädgårdsföreningen) is situated. The Digital SurfaceModel (DSM) in Fig. 1 is derived from an extensivegeodatabase maintained by the Building and PlanningOffice in Göteborg including both building and vegetationgeometries. The DSM is derived according to Lindberg(2005) and has a resolution of 1 m pixel and an extent of637 by 400 pixels.

Spatial modelling

This study made use of the SOLWEIG model, which is ableto simulate spatial (2-D) variations of 3-D radiation fluxesand Tmrt as well as shadow patterns in complex urbansettings. As from Version 2, SOLWEIG can also includevegetation in the form of trees and bushes in the modellingprocedure. The model has been shown to accurately esti-mate the radiation fluxes for a number of different urbansettings and weather conditions as well as in different re-gional contexts (Lindberg and Grimmond 2011a). The mod-el requires the meteorological parameters air temperature,relative humidity and solar radiation (global and diffusecomponents) together with spatial data in the form of aDSM (see the section Study area for spatial analysis) and ageographical location. Tmrt is calculated for a (rotationallysymmetric) standing or walking person where the angularfactors are set to 0.22 for radiation fluxes from the fourcardinal points (east, west, north and south) and 0.06 forradiation fluxes from above and below. Absorption coeffi-cients for shortwave and longwave radiation are 0.7 and0.97, respectively. Albedo and emissivity for buildings andvegetation is set to 0.20 and 0.95, respectively. The trans-missivity of shortwave and longwave radiation through

Int J Biometeorol

vegetation is set to 5 % and 0 %, respectively, according toLindberg and Grimmond (2011a). Wind is not considered inthe current version of the model. Sunlit surface temperaturesof terrestrial surfaces such as building walls and ground areestimated using an empirical relationship for the differencebetween air temperature and surface temperature based onclearness and day of year (Lindberg et al. 2008). The pro-portion of sunlit walls seen at a specific location within themodel domain is derived using the concept of a cylindricalwedge based on the sky view factor (SVF) and sun altitude(see Fig. 3 in Lindberg et al. 2008). Since all surfaces areconsidered to have the same properties in the current versionof the model, errors could be introduced compared to a ‘realworld’ situation. For a detailed description of the SOLWEIGmodel see Lindberg et al. (2008) and Lindberg andGrimmond (2011a).

In order to examine how the spatial patterns of Tmrtchange during different weather situations represented bycloudiness (clearness), a sensitivity test was performed. Abase weather, representing a clear and warm summer day,the 6th of June 1997, was chosen. On this day, the globalsolar radiation (G) peaks at 821 Wm−2 during middaywhereas the diffuse component (D) peaks at 104 Wm−2.The direct radiation beam (I) peaks 2 h earlier at 894 Wm−2.Air temperature is relatively high, peaking in the afternoonat about 23.5 °C. The meteorological data were collected atan hourly resolution from a nearby weather station (500 m

east of the study area) run by the Swedish Meteorologicaland Hydrological Institute (SMHI).

For our sensitivity test, global radiation was reduced by 80,60 and 40 % of its original values, keeping Ta and RHconstant. Thereafter, the diffuse component was estimatedusing the statistical model developed by Reindl et al. (1990).Direct shortwave radiation on a surface perpendicular to theSun is then estimated:

I ¼ G� Dð Þ= sin η ð1Þ

where η is the Sun’s elevation angle above the horizon.Equation 1 is sensitive to very low values of η (sunrise andsunset). The current version of SOLWEIG (v2) corrects fornegative values of I. As for unreasonably high values of I, nocorrection is made and output needs to be evaluated andremoved if unreasonable values are identified. Figure 2 showthe increase and decrease in D (Fig. 2b) and I (Fig. 2c),respectively, when reducing G in 20 % intervals (Fig. 2a). AsG is reduced, D increases. However, D peaks when G isreduced by 60 % and not at 40 %. This is due to the fact thatG is reduced to a level where D can no longer be increased. Idecreases with decreased G and becomes almost 0 Wm−2

whenG is reduced to 40 % of its original value. One can arguethat Ta and RH should change when the shortwave radiationcomponents change but in order to make a solid sensitivityanalysis Ta and RH are kept constant. Furthermore, a general

Fig. 1 Model domain covering the area around the central station in Göteborg (Sweden). Spatial resolution is 1 m. Green circles shows locationand extent of vegetation units. White lines are base map features such as pavement edges, etc.

Int J Biometeorol

decrease of G might not have been appropriate if a short timeinterval was applied due to convective cumulus clouds.However, since 1 h averages are used, the effect of highlyvarying shortwave radiation is reduced.

Quantitative analysis and regional comparison

A 25-year meteorological dataset between 1983 and 2007from the same SMHI station in Göteborg as mentioned inthe section Spatial modelling is used to examine the behav-iour of Tmrt from a climatological and geographical perspec-tive. The data has an hourly resolution and incorporates allthe meteorological parameters necessary to executeSOLWEIG. For the regional comparison, two additionalSMHI datasets from Luleå (65.54°N, 22.11°E) andStockholm (59.35°N, 18.06°E) were used. The same periodas for the station in Göteborg was used. Both Luleå andStockholm are located on the east coast of Sweden.

For the regional comparison, a subversion of SOLWEIGcalled SOLWEIG1D (Lindberg 2012) was used to calculateradiation fluxes and Tmrt for a generic sunlit location withinthe urban environment. Unlike the full SOLWEIG simula-tions discussed above, where SVF and shadow patterns aredetermined for each pixel in a DSM, SOLWEIG1D has asingle, fixed, user-specified SVF and the location is as-sumed to be sunlit during the daytime hours. The latterwould not would be the case in a real world situation, wheresurrounding objects would block the sun at specific times ofthe day and year when SVF<1. Otherwise, the same settingsare used as presented in the section Spatial modelling.

Results and discussion

In this section, maps and statistics from the sensitivityanalysis (section Spatial modelling) are presented. A

regional comparison from a Swedish point of view is alsopresented.

The influence of cloudiness on the spatial patterns of Tmrt

Figure 3a shows the spatial variation of Tmrt at 3 PM on aclear and warm summer day, i.e. the 6th of June 1997.During a clear sky situation Tmrt reaches its highest valuesclose to sunlit building walls and in sunlit dense environ-ments. At open, sunlit locations the Tmrt show somewhatlower values compared to close to sunlit walls. The spatialpattern in sunlit positions has two main explanations; first,the longwave radiation fluxes originated from the warmbuilding walls and second, the cooler sky is partly blockedby the same buildings. Furthermore, reflection of shortwaveradiation is greater at locations closer to sunlit buildingwalls. At the open locations, less building walls and moreof the cool sky is visible. The lowest values of Tmrt arefound in shadowed positions. Figure 4 illustrates the spatialpatterns of some of the short- and longwave radiation com-ponents for the same point in time as is shown in Fig. 3a.Spatial variation of shortwave radiation is foremost depen-dent on the shadow patterns generated by buildings andvegetation. The direct component is dominating (Imax=633 Wm−2) and only affected by shadow patterns (Fig. 4a)whereas the diffuse (Fig. 4b) and reflected (Fig. 4c) compo-nents are controlled by SVF and of much less importance(Dmax=92 Wm−2 and Reflectedmax=38 Wm−2). Outgoinglongwave radiation (Fig. 4d) is controlled by the groundview factor which is a measure of the amount of sunlit areasthat is ‘seen’ by a fictitious down-looking pyrgeometer atthe height of 1.1 m (for details, see Lindberg and Grimmond2011a). The incoming longwave radiation is controlled bySVF (Fig. 4e) whereas the longwave fluxes from the fourcardinal points (Fig. 4f) are a combination of the outgoingand incoming fluxes (Lindberg and Grimmond 2011a).

Fig. 2 The relationship between the three shortwave radiationcomponents at four different levels of global radiation (G). Directbeam radiation and diffuse irradiation for G (80 %), G (60 %)

and G (40 %) are calculated using the Reindl et al. (1990)approach. a Global shortwave radiation (G). b Diffuse radiation(D). c Direct beam radiation (I)

Int J Biometeorol

Apart from shadow patterns, the SVF is a good measurethat explains much of these spatial patterns (Fig. 5). Forexample, SVF decreases as sunlit building walls are

approached, which results in increased values of Tmrt closerto the same walls. As from version 2, SOLWEIG can com-pute continuous maps of “complete” SVFs incorporating

Fig. 3 Spatial variations of Tmrt (standing man) (°C) during clear andwarm weather in the city centre of Göteborg, Sweden at 1500 h LST on6 June 1997 (a). B through D show spatial patterns when globalradiation is reduced to: 80 % (b), 60 % (c), and 40 % (d) of its original

value. E through G show normalized differences between G (100 %)(a) and G (80 %) (e), G (60 %) (f), and G (40 %) (g). The diffuse anddirect beam components are recalculated according to Reindl et al.(1990), respectively

Int J Biometeorol

buildings, ground topography and vegetation units whenavailable (Lindberg and Grimmond 2010; Lindberg andGrimmond 2011a). Visual examinations of the SVF imagein Fig. 5 shows the large reduction of SVF underneathvegetation where a very large proportion of the hemisphereis blocked by the vegetation canopies. Even though the SVFis very low underneath the canopies, the Tmrt is less than inareas close to sunlit building walls. This is because thesurface temperatures are considered to be much higher forbuildings compared to vegetation in the model.

Reducing the global radiation and recalculating the dif-fuse and direct component as explained in thesection Spatial modelling alters the spatial patterns of Tmrtconsiderably (Fig. 3b to d). When G is reduced to 80 %, theareas close to building walls are no longer the areas with thehighest Tmrt. Instead more open settings show the highestvalues (Fig. 3b). This pattern becomes more pronounced asG is further reduced (Fig. 3c and d). The areas with the

highest values of Tmrt are now found at the most openlocations, such as building rooftops. The dense settings withnarrow street canyons become cooler as G is reduced. Clearweather situation (Fig. 3a) results in overall higher values ofTmrt at street level (Av=48.3 °C). As G is reduced Tmrt isalso reduced. The lowest average value of Tmrt at street level(34.2 °C) is found when G is reduced to 40 % of its originalvalue (Fig. 3d). When G is reduced, there are still areas withhigh values of Tmrt. For example, the open areas and roof-tops in Fig. 3b (G=80 %) have Tmrt almost as high aslocations close to sunlit building walls in Fig. 3a (G=100 %). Figure 3e–g shows the differences in the spatialdistribution of Tmrt for various levels of cloudiness. Thedifference maps are generated using the clear weather situ-ation (Fig. 3a) as a constant. The average Tmrt at street levelfor a specific cloudiness situation is subtracted from thatimage before the difference map is crated. ExaminingFig. 3e–g, a number of interesting features can be seen.

Fig. 4 Spatial variations of various radiation components during clearand warm weather in the city centre of Göteborg, Sweden at 1500 hLST on 6 June 1997. All values are given in Wm−2. a Direct shortwave

radiation. b Diffuse shortwave radiation. c Reflected shortwave radia-tion. d Outgoing longwave radiation. e Incoming longwave radiation. fAverage longwave radiation from the four cardinal points

Int J Biometeorol

For example, the effect of lower Tmrt in shadowed locationsis reduced as cloudiness increases. Also, a small proportionof sunlit areas close to building walls are colder when asmall increase in cloudiness is used (Fig. 3e). This areaincreases as cloudiness increases (Fig. 3f–g). The alteredspatial pattern of Tmrt during cloudy weather situations has anumber of explanations. As the sky becomes cloudier, thetemperature of the sky vault increases due to an increase inemissivity which favors higher Tmrt values at open locations.Furthermore, the surface temperatures of sunlit buildingwalls are reduced as less incoming radiation is received bythese surfaces. Increased cloudiness also increases the im-portance of the shortwave diffuse radiation componentwhich originates from all positions throughout the sky vault.In the current version of the SOLWEIG model (2.3) thediffuse component is considered to be isotropic, althoughthis is not the case. Especially during clear weather situa-tions an anisotropic pattern is present where a higher pro-portion of the diffuse irradiation originates from the positionaround the sun (McArthur and Hay 1981). Further develop-ment of the SOLWEIG model is currently being conducted,and we plan to incorporate a simplified version of the skymodel presented by Perez et al. (1993). When G is reducedto 40 % of its original value, the shadow patterns are hard toidentify due to the low proportion of direct radiation present(24 Wm−2 at 3 pm LST). Now the areas with the lowest Tmrtare found in very dense urban structures such as courtyards.

Regarding the effect of vegetation on the spatial patternof daytime Tmrt for different levels of cloudiness, trees andbushes lower Tmrt within their close proximity. However, thereason for this has different explanations based on the levelof cloudiness. During clear weather the shadow from thevegetation lowers Tmrt by blocking direct solar radiation. Ascloudiness increases the shadow effect is reduced, and in-stead, the low SVF under the trees block the relativelywarmer sky which is found during cloudy weather situa-tions. An interesting feature found during clear weathersituations is the relatively high values of Tmrt for sunlit areasunderneath the vegetation canopies. This is caused by thevegetation canopies that block the relatively colder skyresulting in increased incoming longwave radiation origi-nating from vegetation rather that from sky. Even thoughTmrt have local maxima at these locations, they are probablynot prone to enhanced heat stress because areas underneathtrees are well-ventilated compared to areas close to a sunlitbuilding wall. Using a thermal index, e.g. physiologicalequivalent temperature (PET) (Mayer and Höppe 1987),would probably demonstrate this. Also since Tmrt is estimat-ed for the center of gravity of a human (1.1 m agl), the upperpart of a standing body could be in shadow from vegetationwhile values of Tmrt are still estimated as high. Furthermore,episodes of heat stress usually occur in the summer monthswhen the Sun’s altitude angles are high and the area under-neath the vegetation canopy is shaded.

Fig. 5 Continuous sky view factors (SVF) for the study area including both buildings and vegetation units. Black represents SVF=0 and whiterepresents SVF=1

Int J Biometeorol

Another interesting pattern is found when comparingTmrt and SVF. Figure 6 shows scatterplots between Tmrtand SVF under the four different levels of G. Duringclear weather (Fig. 6a) two clusters of data points areevident. These two clusters originate from sunlit orshaded locations within the model domain (red and bluein Fig. 6). Furthermore, the peak in Tmrt for both theshaded and sunlit points are found in the more denseurban structures where low SVF values are found. Ascloudiness increases (G=80–60 %) the two clustersmoves closer together (Fig. 6b and c) and for G(40 %) (Fig. 6d) only one cluster is present. The peakin Tmrt for G (80–40 %) is found in the most opensettings with high SVF-values.

Quantitative analysis of Tmrt

Scatterplots between Tmrt and sky cloudiness (here repre-sented by D/G) for fictitious sunlit points with differentSVFs in Göteborg between 1983 and 2006 are shown inFig. 7. Only daytime situations are considered (G>0). A lowvalue of D/G represents clear weather conditions and a highvalue (close to 1) represents overcast weather conditions. Anumber of interesting features are shown. First, the shape ofthe data point cloud changes as the SVF is altered. Thehighest value of Tmrt (71.4 °C) was found during semi-cloudy conditions in the fictitious open space (SVF=1.00).The reason for this is the high diffuse radiation (475 Wm−2)in combination with high value of global radiation (749 W

Fig. 6 Hourly values of Tmrt (standing man) versus sky view factor (SVF)for the different levels of global radiation for the model domain asshown in Figs. 4 and 6. 100 % (a), 80 % (b), 60 % (b), and 40 % (d) of

the original global radiation value. Only ground pixels are considered(n=173842). Red data points are sunlit and blue are data points inshadow

Int J Biometeorol

Fig. 7 Hourly daytime values of Tmrt (standing man) versus cloudiness (diffuse/global) at a sunlit location for different urban geometries (SVFs)

Int J Biometeorol

m−2) which increase Tmrt as the whole sky vault is visible.Also, the longwave radiation from the sky vault is relativelyhigh (316 Wm−2) due to the presence of clouds, and air andsurface temperature are also high (26 °C and 35.5 °C,respectively) resulting in outgoing longwave radiation of489 Wm−2. The lowest value of daytime Tmrt is also foundat SVF=1.00 (Tmrt=−32.7 °C) during a totally overcastsituation with extremely low Ta (−21.9 °C) and almost noshortwave radiation (I=18.1 Wm−2). This gives very smalloutgoing and incoming longwave radiant fluxes of 212 Wm−2 and 119 Wm−2, respectively. As SVF is lowered, i.e.moving towards more dense urban settings, the highestvalues are instead found during clear weather situations asbuilding walls can contribute to the longwave radiationfluxes and hence influence Tmrt at the fictitious point. AsSVF is reduced to 0.2 and 0.1, the dense urban structureswill have the highest values of Tmrt within the urban setting.The lowest mean values of daytime Tmrt for the period 1983to 2007, are found at SVF=0.60 (Tmrt=17.2 °C) and thehighest at SVF=0.10 (Tmrt=18.4 °C). The scattering ishighest at SVF=1.00 (σTmrt=20.4 °C) and the lowest atSVF=0.10 (σTmrt=15.0 °C). If nocturnal values are includ-ed, the lowest value of Tmrt is found at SVF 1.00 (Tmrt=7.25°) with increasing mean values as SVF is lowered (notshown).

Regional comparison of Tmrt

Figure 8 shows monthly averages of Tmrt, Ta, global radia-tion and cloudiness for Göteborg, Luleå and Stockholmbetween 1983 and 2007. The distribution of hourly data inthe form of boxplots is shown in Fig. 9. Summer tempera-tures are relatively similar between the three cities whereaswinter temperatures are considerably lower in Luleå due toits more northern location. Incoming solar radiation is rela-tively similar but differences are also found between thethree cities. The highest level of monthly average solar

radiation during summer is found in Luleå due to the longdays. However, the peak values of summer time solar radi-ation are found in Göteborg due to its location furthest south(Fig. 9). During the winter months, the incoming solarradiation in Luleå is very low and during January andDecember there is almost no sunlight present at all.Göteborg and Stockholm are very similar with regards toincoming solar radiation. This is because they are bothcoastal cities with similar latitude. Midday cloudiness forthe three cities shows that all three cities experience moreclear weather during the summer compared to winter.Stockholm has the overall lowest values of cloudiness,especially during spring (April and May).

The monthly averages of Tmrt for a standing man at afictitious sunlit point with a SVF of 0.60 representing atypical urban environment reveals a number of interestingfeatures (Fig. 8). Regardless of the location, the level of Tmrtduring the summer months (June and July) is very similarfor the three cities. The most northern location even has aslightly higher median value (19.4 °C in June) compared tothe two more southern locations where Göteborg has thelowest (17.0 °C in June) median value (Fig. 9). The peakvalues of Tmrt during summer are similar between the threecities (Fig. 9). The highest values of Tmrt for Göteborg andStockholm occur in July (62.1 °C and 63.6 °C, respectively).Tmrt in Luleå peaks in June (62.1 °C). This could foremost beexplained by the difference in sun altitude. Even though thepeak in incoming short wave radiation is smaller in Luleåcompared to Stockholm and Göteborg (Fig. 9), a standingperson will receive a larger fraction of shortwave radiationdue to the posture of the vertical, cylinder body representing astanding man (Mayer and Höppe 1987). Figure 10 shows the

Fig. 8 Monthly averages of Tmrt, air temperature, global radiation, and cloudiness for Göteborg, Luleå and Stockholm between 1983 and 2007.Cloudiness is calculated for a midday period between 10 AM and 2 PM

�Fig. 9 Distribution of hourly values of air temperature, global radia-tion, cloudiness and hourly values of mean radiant temperature for astanding man from a fictitious sunlit point (SVF=0.60) for Göteborg,Luleå and Stockholm between 1983 and 2007. Cloudiness is calculatedfor a midday period between 10 AM and 2 PM. Circles represents meanvalues

Int J Biometeorol

Int J Biometeorol

different radiative loads of shortwave and longwave radiationduring the peaks of Tmrt for the three different cities. Theincoming shortwave radiation for a spherical body (Fig. 10a)is lowest for Luleå (99 Wm−2). However, the lower solaraltitude results in the highest shortwave flux from the side(cardinal points) (188 Wm−2). On the other hand, thelongwave fluxes are the lowest in Luleå (340 Wm−2).Adding the view factors of a standing man to the fluxes(Fig. 10b) increases the importance of the sideway fluxeswhile reducing the fluxes from above and below (Fig. 10a).

The inter-quartile range of summertime values of Tmrt issmaller in Göteborg compared to the other two cities (Fig. 9).This is probably because of its maritime location on the westcoast of Sweden which is exposed to the westerly winds thatpredominate the overall climate in Sweden. The location ofGöteborg is also evident when examining the wintertimevalues of Tmrt where Stockholm has lower values. Luleåshows much lower values during the winter months due tothe colder temperatures and lack of solar radiation. Alteringthe urban density by changing SVF does not change thepatterns found between the three cities (not shown).However, the distribution of Tmrt in relation to clearness

throughout the data period (1983–2007) changes forStockholm and Luleå according to the patterns found inGöteborg (Fig. 9).

Implications for climate sensitive planning

As shown in Fig. 3a, the highest Tmrt values are found onclear and warm days close to sun-lit walls. This implies thatshading should be maximized in these areas in order toreduce outdoor heat stress as well as heating of buildingsduring periods of extensive heat. During semi-cloudy con-ditions, however the highest values are found in open areas,such as roofs and squares (Fig. 3b to d). Although Tmrt areless in magnitude during semi-cloudy conditions than dur-ing clear conditions, the results suggest that shading shouldalso be provided in open areas. Also, even though the hourswhen Tmrt is highest occurs during semi cloudy conditionsin open areas (Fig. 7, SVF=1.00), dense urban areas showclusters of points of high Tmrt during clear weather situations(Fig. 7, SVF<1).

Although dense urban settings are generally warmer on adaily average than open settings, dense settings mitigate the

Fig. 10 Radiative loads as calculated in SOLWEIG during peak values of Tmrt at a fictitious location (SVF=0.60) in Göteborg (GBG), Luleå (LUL) andStockholm (STK) for a spherical body (a) and a cylindrical body (standing man) (b). Components “L” are longwave, and components “K” shortwave

Int J Biometeorol

swings in Tmrt (Fig. 7), i.e. reducing the heat stress as well ascold stress. These results are in line with previous studies(Thorsson et al. 2011).

Tmrt is similar throughout Sweden in summer (Figs. 8 and9). This implies that the same design guidelines to reduceheat stress can be applied in all three cities. However,latitudinal differences result in different sun altitudes, whichneeds to be considered in terms of shadows from buildingand vegetation. In high latitude areas, there is always a greatchallenge for architects and urban planners to create outdoorurban settings which are both able to mitigate heat during afew hazardous occasions but also to be pleasant during therest of the year. During heat waves, shadowing is one themost effective measures for mitigating high Tmrt . However,extensive areas of shadow are usually not desired in theoutdoor environments of high latitude cities. One solutionis to create diverse outdoor urban spaces in terms of shadow,ventilation, etc. This gives opportunities of various choiceswithin short distances to regulate the thermal environmentof the human body based on personal needs and desires(Thorsson et al. 2004). Another effective mitigation of heatstress is to maximize shading in areas that tend to heat up,i.e. sunlit areas, especially those close to buildings or indensely built structures.

Besides altering the urban geometry, vegetation hasshown to be an effective and relatively simple measure tomitigate heat stress during periods of extensive temperatures(Lindberg and Grimmond 2011a; Bowler et al. 2010;Shashua-Bar et al. 2011; Picot 2004; Mayer et al. 2008).In order to maximize the cooling effect type, arrangementsand location of vegetation need to be considered. In terms oftype of vegetation, trees are to be preferred over lowervegetation (bushes, ground vegetation) due to their abilityto provide shade (Lindberg and Grimmond 2011b;Konarska et al. 2012). Furthermore, species that thrive inrelatively warm and dry environments should be selected inorder to provide maximum growth and thus maximumcooling (Sjöman and Busse Nielsen 2010). During the daythe cooling effect of trees is mainly due to shading(Konarska et al. 2012; Mayer et al. 2008), which impliesthat vegetation should be placed near surfaces that tend toheat up, i.e. trees near north facing walls have a limitedcooling effect on its surroundings. By planning vegetationin different layers, i.e. a mix of trees, shrubs and groundvegetation, the shading as well as the cooling effect can beincreased (Shashua-Bar et al. 2011). Adding vegetation inareas with no or little vegetation has been shown to give alarger cooling effect than adding vegetation in already high-ly vegetated areas (Loridan and Grimmond 2012).

Altering the material of the surrounding surfaces couldalso be an option to regulate the outdoor thermal environ-ment. However, these measures usually have a minor effect.Erell (2012) showed that although high-albedo surfaces may

reduce air temperature to which pedestrians are exposed, thechange in temperature has only a small effect on theirthermal balance with the environment. Reduction in surfacetemperatures, leading to a reduction in long wave radiationfluxes, is counterbalanced by increased reflection of short-wave radiation. The net effect of increasing the albedo ofurban surfaces may thus have a minor effect on the thermalenvironment in outdoor urban settings. However, choice ofmaterials will affect features such as heat storage and indoortemperatures.

As stated in the section Spatial modelling, wind isnot considered when calculating Tmrt. To estimate hu-man thermal comfort, a more comprehensive approachthat takes into consideration both personal factors andalso wind is required. To estimate the spatial patterns ofwind fields at pedestrian heights, a more complex modelsuch as ENVI-met (Bruse and Fleer 1998) could beused. Even though Tmrt alone cannot be used to calcu-late physiological equivalent temperature (PET) or theuniversal thermal climate index (UTCI), it has beenproven that Tmrt is one of the most important parametersfor estimating outdoor thermal comfort, especially onhot and sunny days when heat stress is likely to occur(Mayer et al. 2008). Furthermore, during these heatstress situations, high pressure systems are common,resulting in low wind speeds which strengthen the rela-tionship between thermal comfort and Tmrt.

Conclusions

Increasing the knowledge on how Tmrt varies based onfactors such as location, climate and urban setting contrib-utes to the practice of climate sensitive planning. This paperexamines various aspects on how Tmrt varies spatially withina high latitude urban setting based on cloudiness.Furthermore, long-term variations of Tmrt in different re-gional settings within Sweden are investigated.

Results show a clear change in the spatial pattern ofTmrt between clear and cloudy weather situations.Besides the large variations between sunlit andshadowed areas during clear weather, the highest valuesof Tmrt are found close to sunlit walls. Increased cloud-iness alters the spatial pattern by reducing the differ-ences between sunlit and shaded areas as well asmoving the warmest areas of Tmrt to open areas wherethe proportion of shortwave diffuse radiation from thesky vault is high. Vegetation lowers Tmrt through bothshadow generation and by lowering surface tempera-tures compared to building walls. From a regionalcomparison of three Swedish coastal cities, it is shownthat Tmrt during summer is similar regardless of latitu-dinal location. This could foremost be explained by the

Int J Biometeorol

difference in sun altitude. Even though the peak inincoming short wave radiation is smaller in Luleå com-pared to Stockholm and Göteborg, a larger fraction ofshortwave radiation will be received by a person due tothe posture of the vertical, cylinder body representing astanding man. On the other hand, large differences inTmrt during winter are found.

Regarding climate sensitive planning, the spatial pat-tern of Tmrt should be considered both from a heatstress point of view as well from an all-year perspec-tive. Shadow, both from buildings and vegetation, is themost effective measure to reduce extreme values of Tmrt.However, extensive areas of shadow are usually notdesired within outdoor urban environments at high lati-tude cities. One solution is to create diverse outdoorurban spaces with a variety of shadow and ventilationpatterns. This gives opportunities of various choiceswithin short distances to regulate the thermal environ-ment of the human body based on personal needs anddesires. The summer values of Tmrt are more or less thesame in the three high latitude cities included in thisstudy. This implies that the same design guidelines toreduce heat can be applied in all three cities. However,latitudinal differences result in different sun altitudeswhich need to be considered in terms of shadows frombuildings and vegetation.

Acknowledgements This work is financially supported byFORMAS—the Swedish Research Council for Environment, Agricul-tural Sciences and Spatial Planning within the European Commissionprogramme Urban-Net.

References

ASHRAE (2001) ASHRAE fundamentals handbook 2001 (SIEdition). American Society of Heating, Refrigerating, andAir-Conditioning Engineers, USA

Bowler DE, Buyung-Ali L, Knight TM, Pullin AS (2010) Urbangreening to cool towns and cities: a systematic review of theempirical evidence. Landsc Urban Plann 97(3):147–155

Bruse M, Fleer H (1998) Simulating surface-plant-air interactionsinside urban environments with a three dimensional numericalmodel. Environ Model Softw 13(3–4):373–384

Erell E (2012) Effect of high-albedo materials on pedestrian thermalcomfort in urban canyons. In: International Conference on UrbanClimate, Dublin

Höppe P (1992) A new procedure to determine the mean radianttemperature outdoors. Wetter und Leben 44:147–151

IPCC (2007) AR4 synthesis report. Full report. IntergovernmentalPanel on Climate Change, http://www.ipcc.ch/pdf/assessment-report/ar4/syr/ar4_syr.pdf. Accessed 15 February 2012

Konarska J, Larsson A, Lindberg F, Holmer B, Thorsson S (2012)Transmissivity of direct solar radiation through the crowns ofsingle urban trees. In: The 8th International Conference onUrban Climate, Dublin, Ireland

Lindberg F (2005) Towards the use of local governmental 3-D datawithin urban climatology studies. Mapping Image Sci 2:32–37

Lindberg F (2012) The SOLWEIG-model. http://www.gvc.gu.se/Forskning/klimat/stadsklimat/gucg/software/solweig/. Accessed12 February 2012. Gothenburg University, Sweden

Lindberg F, Grimmond CSB (2010) Continuous sky view factor mapsfrom high resolution urban digital elevation models. Clim Res42:177–183. doi:10.3354/cr00882

Lindberg F, Grimmond CSB (2011a) The influence of vegetation andbuilding morphology on shadow patterns and mean radiant tem-peratures in urban areas: model development and evaluation.Theor Appl Climatol 105(3–4):311–323. doi:10.1007/s00704-010-0382-8

Lindberg F, Grimmond CSB (2011b) Nature of vegetation and buildingmorphology characteristics across a city: influence on shadowpatterns and mean radiant temperatures in London. UrbanEcosyst 14(4):617–634. doi:10.1007/s11252-011-0184-5

Lindberg F, Holmer B, Thorsson S (2008) SOLWEIG 1.0—Modellingspatial variations of 3D radiant fluxes and mean radiant temper-ature in complex urban settings. Int J Biometeorol 52(7):697–713.doi:10.1007/s00484-008-0162-7

Loridan T, Grimmond CSB (2012) Characterization of energy fluxpartitioning in urban environments: links with surface seasonalproperties. J Appl Meteorol Climatol 51(2):219–241.doi:10.1175/jamc-d-11-038.1

Matzarakis A, Rutz F, Mayer H (2009) Modelling radiation fluxes insimple and complex environments: basics of the RayMan model.Int J Biometeorol 54(2):131–139

Mayer H, Höppe P (1987) Thermal comfort of man in different urbanenvironments. Theor Appl Climatol 38:43–49

Mayer H, Holst J, Dostal P, Imbery F, Schindler D (2008) Humanthermal comfort in summer within an urban street canyon inCentral Europe. Meteorol Z 17(3):241–250. doi:10.1127/0941-2948/2008/0285

McArthur LJB, Hay JE (1981) A technique for mapping the distribu-tion of diffuse solar radiation over the Sky hemisphere. J ApplMeteorol 20(4):421–429

McCarthy MP, Best MJ, Betts RA (2010) Climate change in cities dueto global warming and urban effects. Geophys Res Lett 37(9):L09705. doi:10.1029/2010gl042845

Meehl GA, Tebaldi C (2004) More intense, more frequent, and longerlasting heat waves in the 21st century. Science 305(5686):994–997. doi:10.1126/science.1098704

Mochida A, Lun IYF (2008) Prediction of wind environment andthermal comfort at pedestrian level in urban area. J WindEng Ind Aerodyn 96(10–11):1498–1527. doi:10.1016/j.jweia.2008.02.033

Nikolopoulou N, Baker N, Steemers K (1999) Improvements to theglobe thermometer for outdoor use. Archit Sci Rev 42:27–34

Pascal M, Laaidi K, Ledrans M, Baffert E, Caserio-Schönemann C, LeTertre A, Manach J, Medina S, Rudant J, Empereur-Bissonnet P(2006) France’s heat health watch warning system. Int JBiometeorol 50(3):144–153

Perez R, Seals R, Michalsky J (1993) All-weather model for skyluminance distribution—preliminary configuration and valida-t ion . Sol Energ 50(3) :235–245. doi :10 .1016/0038-092x(93)90017-i

Picot X (2004) Thermal comfort in urban spaces: impact of vegetationgrowth—case study: Piazza Della Scienza, Milan, Italy. EnergBuild 36(4):329–334. doi:10.1016/j.enbuild.2004.01.044

Reindl DT, Beckman WA, Duffie JA (1990) Diffuse fraction correla-tion. Sol Energ 45(1):1–7

Shashua-Bar L, Pearlmutter D, Erell E (2011) The influence of treesand grass on outdoor thermal comfort in a hot-arid environment.Int J Climatol 31(10):1498–1506. doi:10.1002/joc.2177

Int J Biometeorol

Sjöman H, Busse Nielsen A (2010) Selecting trees for urban pavedsites in Scandinavia—a review of information on stress toleranceand its relation to the requirements of tree planners. Urban ForUrban Green 9(4):281–293. doi:10.1016/j.ufug.2010.04.001

Thorsson S, Lindqvist M, Lindqvist S (2004) Thermal bioclimaticconditions and patterns of behaviour in an urban park inGoteborg, Sweden. Int J Biometeorol 48(3):149–156

Thorsson S, Lindberg F, Eliasson I, Holmer B (2007) Different methodsfor estimating the mean radiant temperature in an outdoor urbansetting. Int J Climatol 27(14):1983–1993. doi:10.1002/joc.1537

Thorsson S, Lindberg F, Bjorklund J, Holmer B, Rayner D (2011)Potential changes in outdoor thermal comfort conditions inGothenburg, Sweden due to climate change: the influence ofurban geometry. Int J Climatol 31(2):324–335. doi:10.1002/joc.2231

Thorsson S, Rocklöv J, Rayner D, Konarska J, Lindberg F,Holmer B (2012) Mean radiant temperature—a measurefor evaluating the impact of climate(change) on health.The 8th International Conference on Urban Climate, Dublin,Ireland

Int J Biometeorol