Building Serv. Eng. Res. Technol. 23,2 (2002) pp. 97–106 The London Heat Island: results from summertime monitoring R Watkinsa BSc, J Palmerb BSc, MSc, CEng, MInstE, M Kolokotroni a MSc, PhD, MCIBSE, CEng, MASHRAE, and P Littlefairb BSc, PhD, CEng a Mechanical Engineering Department, Brunel University, Uxbridge, UK b Building Research Establishment Ltd, Bucknalls Lane, Garston, Watford, UK This paper reports results from a study investigating the variation of air tempera- ture across London. The paper focuses on the description of temperature measurements in summer 1999 and the analysis of the results. The intensity of the London Heat Island has been assessed using a radial grid of 68 stations recording simultaneous hourly air temperatures. The urban heat island has been found to be predominantly a nocturnal phenomenon and its intensity reaches 7 K on occasion. It was found that the thermal centre is in the City of London which is characterized by tall buildings and high anthropogenic heat emissions. The night- time temperature pro le of London tends to follow a simple pattern with contours arranged roughly circularly around the centre; it is found that 77% of the variance of the mean night-time temperature across London is related to distance from the thermal centre. The more complicated pattern in the daytime precludes such a simple relationship. 1 Introduction 1.1 Examples of heat island intensities There have been numerous eld studies of The urban environment modi es microclimate in the urban heat island effect with data collected numerous ways. In general, urban climates are over a few days to a few years. The most com- warmer and less windy than in rural areas. How- mon way of summarizing such studies is to ever, a review of the literature shows that the quote the maximum heat island intensity meas- modi cation to urban climates is highly variable, ured. Heat island intensity is the temperature dif- and depends on the particular topography, ference between an urban and rural reference regional wind speeds, urban morphology, time location. This is an important indicator of urban– of year, and many other factors. The alteration rural differences, but requires quali cation in to temperature—the urban heat island effect—is order to assess the heat island’s signi cance in particularly important, as this has a direct impact relation to energy use for cooling or heating. on fuel use: warmer temperatures reduce heating Time of day and year are particularly important, fuel use in the winter and increase air condition- and ideally either a frequency distribution of the ing use in the summer. heat island intensity or an intensity time integral for selected time-bands are needed. Few studies provide this level of detail, at least in published form. Rosenfeld makes the general statement: ‘On a clear summer afternoon, the air tempera- Address for correspondence: M Kolokotroni, Mechanical Engin- eering Department, Brunel University, Uxbridge, Middlesex, UB8 ture in a typical city is about 2.5 K hotter than 3PH, UK. E-mail: Maria.kolokotroni@brunel.ac.uk the surrounding area.’1 Ó The Chartered Institution of Building Services Engineers 2002 Downloaded from bse.sagepub.com at Templeman Lib/The Librarian on December 7, 2015 10.1191/0143624402bt031oa 98 The London Heat Island Givoni states 2 that: ‘In large cities it is com- previous examples show urban areas to be from mon to observe nocturnal air temperatures 3–5 about 1 K cooler to 15 K warmer. It is dif cult K higher than the surrounding areas, and, in to generalize about summertime urban–rural dif- extreme cases, higher by up to 8 K. During the ferences as so much depends on the particular daytime hours, however, this difference in air attributes (climate, topography, etc.) of a parti- temperature between the city and its surrounding cular location. area is smaller—only about 1–2 degrees—and often the daytime temperatures in a densely 1.2 Signi cance of urban–rural temperature built-up area are lower than in the open country.’ differences Whether the daytime urban temperature is The energy used for cooling buildings has greater or smaller than the more rural one, and not always been considered relevant to studies how this difference compares with the nocturnal of the overall effects of heat islands.7 However, difference depends on many factors. The wide in the context of global warming there is increas- variation is illustrated by the following ing interest. This is particularly so in relation to examples. climates with hot summers, or where air- In Athens, using hourly data, daytime air tem- conditioning is used extensively. It is also the peratures have been reported to be between 4 case that air-conditioning use is increasing, with and 15 K warmer in the summer than outside changing af uence or expectation. the city.3 The highest urban–rural temperature Exceptionally hot periods can lead to a dra- differences were measured in the central areas matic rise in the demand for air conditioning. at the bottom of street gorges subject to high After a series of heat waves over three years in volumes of traf c. Greece, it has been reported that annual pur- In Barcelona, using monthly means of daily chases of air conditioning units increased eight maxima, compared with its airport on the out- times in the following years.8 skirts the city was found to be cooler in the day- Air-conditioning provision is increasing in time in every month of the year, by up to 0.6 K, cars. This increases fuel consumption, although apart from in January and December when it was it may be offset by a reduction in drag in the 0.2 K warmer. Average monthly differences for absence of a sunroof. the minima indicated a nocturnal heat island of Although heat islands increase the use of between 2.5 and 3.3 K. On 10% of days the energy for cooling in the summer, they reduce urban minima were at least 5 K warmer than at the energy required for heating buildings in win- the airport, with an absolute maximum differ- ter. Landsberg compared heating and cooling ence of about 7–8 K.4 degree days for several American cities and at In Mexico City, using hourly data, daytime airports outside them.9 An amended table, (11.00–15.00) temperatures were found to be incorporating errata, was published more between 3 and 5 K warmer on average than a recently by Taha.10 See Table 1. Note that the rural site during the summer, and up to 6.5 K heating and cooling degree days are both calcu- warmer on occasion.5 lated to a base of 18.3°C. In Dallas, using transect data taken in the In the US cities studied, both Landsberg and summer, the central core of the city (with very Taha concluded that the elevation of urban tem- tall buildings) was found to be cooler at the time peratures imposes a net energy penalty for urban of maximum temperature, by about 1 K, than the areas because of increased cooling requirements. area just outside it.6 It is clear that the extent to which urban tem- 1.3 London’s heat island peratures differ from those of their surroundings In 1820, Luke Howard published an varies considerably. In the daytime, urban areas analysis of 10 years of daily temperature may be warmer or cooler than further out. The measurements which established the existence of Downloaded from bse.sagepub.com at Templeman Lib/The Librarian on December 7, 2015 R Watkins et al. 99 Table 1 Comparison of the reduction in heating days and increase in cooling degree days for urban areas compared to airport sites Heating degree-days Cooling degree-days Location Urban Airport % Diff. Urban Airport % Diff. Los Angeles 384 562 2 32 368 191 +92 Washington DC 1300 1370 2 6 440 361 +21 St Louis 1384 1466 2 6 510 459 +11 New York 1496 1600 2 7 333 268 +24 Baltimore 1266 1459 2 14 464 344 +35 Seattle 2493 2881 2 13 111 72 +54 Detroit 3460 3556 2 3 416 366 +14 Chicago 3371 3609 2 7 463 372 2 24 Denver 3058 3342 2 8 416 350 +19 Differences are shown as a percentage of the airport site. From Taha1 0 modi ed after Landsberg9 London’s heat island i.e., an area of elevated maximum values showed smaller differences, temperature.9 These data show a mean 24-h tem- with the urban centre about 1 K warmer than the perature for July in the city about 0.6 K higher surrounding country in the summertime and 0.7 and in November 1.2 K higher than in the coun- K in the winter. try. Howard also noted [in Fahrenheit] that, An indication of the distribution of the tem- ‘Night is 3.70° warmer and day 0.34° cooler in perature anomalies is also described by Chand- the city than in the country. Thus the latter has ler, for the period 1951–1960,11 using data from 4° more variation.’ (2.05 K, 0.18 K and 2.22 central London (at Kensington) and a rural site K, respectively.) Howard’s data provide the rst (at Wisley). Fifty-eight percent of summer days scienti c evidence for two important character- had urban maxima 0.6–2.2 K greater than the istics of temperature anomalies, namely, diurnal rural maxima, and about 1% of days were and seasonal variation of the size of the anomaly, between 2.8 and 4.4 K greater. For about 39% and change in sign of the anomaly (a heat island of the summer days the urban maxima were to a cold island). His reference to there being cooler than the rural maxima by up to 1.7 K. greater temperature variation in the countryside The heat island effect in London varies from underlines another important property of urban year to year. For example, the years 1933–1934 areas—that they can act as moderators of were milder and calmer than the preceding years climate. and drier than those which followed. These years Since Howard rst identi ed the heat island, were notable for having particularly strong heat the urban climate of London has been exten- islands.12 Of perhaps longer term importance is sively studied, and was described in detail in The the trend identi ed by Lee,13 looking at the tem- Climate of London.11 Meteorological Of ce tem- perature difference between St James’ Park and perature data for London (1931–1960), reported Wisley for the period 1962–1989. He found that by Chandler, show the annual mean temperature daytime heat islands have decreased over time, of central London to have been 1.4 K warmer from about 0.5 K down to 0.25 K in the summer, than the surrounding country. Daytime and night-time heat islands have increased by maximum temperatures (represented by the about 0.5 K (these are mean values). annual mean of the daily maxima) were 0.9 K An analysis of the London heat island in the warmer. Monthly mean differences showed cen- exceptionally hot and dry summer of 1976 con- tral London to be warmer by about 1.6 K in the cluded that the magnitude of the daytime heat summertime and 1.2 K in the winter. Mean island was no greater that year than normal.14 Downloaded from bse.sagepub.com at Templeman Lib/The Librarian on December 7, 2015 100 The London Heat Island Weather data from several sites in and around London were used, and isotherms drawn. On the hottest day of the year, 26 June 1976, it was found that: ‘In most country districts outside London maxima were 34–35°C. The urban areas were generally about 2 K warmer than rural areas to the north and south.’ The main heat island was displaced west, possibly as a result of afternoon sea breezes although evidence for such breezes has been reported to be scarce.15 1.4 The present study The focus of this study is the investigation of the temperature variation across London with a view to establishing its impact on cooling energy use. Since Chandler’s detailed survey, there have been three important changes to the city that might affect temperature: changes to the building stock (more taller buildings), changes Figure 1 Schematic map of London showing the position to the anthropogenic heat release (higher traf c of the 68 measuring stations arranged in eight transects levels, greater use of air-conditioning), and changes to air pollution (reduced black smoke). Given these changes, and the availability now of transects are of different length. In some cases, micro data-loggers, it was considered appropri- suitable sites could not always be found, or per- ate to reassess the temperature eld, and in mission obtained, and there are gaps in the greater spatial and temporal detail than has been transect. possible before. Results of the summertime Conventional Stevenson screens can lead to monitoring (June to August 1999) are reported erroneously high temperatures (up to 2.5 K) in here. still, sunny conditions16 because of the relatively high absorptivity of white paint, particularly to 2 Method for measuring the London infra-red radiation. For this study, it was parti- heat island intensity cularly desirable that the effects of solar radi- ation on the measurement of air temperature A multiple-transect study was adopted using a were minimized as the data were to be used for permanent set of xed sites with hourly logging. investigating the impact on air-conditioning Eight transects, or linear sampling paths, were energy use, rather than, for example, pedestrian chosen and are radial with a common focus in comfort levels. To achieve this, the temperature Bloomsbury, just north of Oxford Street, Lon- sensor and data logger were housed in an alu- don. Successive stations are positioned at a one minium can, which was then suspended between mile radial spacing along each transect, until 4 two concentric cylinders of polished stainless miles from London, when the spacing is steel—giving triple-shielding. An upper stainless increased to 2 miles. The transects are aligned steel cover was also provided with a free passage with the points of the compass. Figure 1 shows of air to the double air-gap. A diagram of the the position of the measurement stations, 68 in measuring device is shown in Figure 2. A all. Each transect extends out until a rural site is sequence of tests both in the eld and on a priv- reached—a rural reference—and because of the ate site con rmed the ef cacy of the shielding. shape of urban development around London, the Sensors were positioned on Local Authority Downloaded from bse.sagepub.com at Templeman Lib/The Librarian on December 7, 2015 R Watkins et al. 101 Figure 2 A diagramatic cross-section and a photograph of the air temperature measuring device lighting columns at a height of 6 m. A photo- eral offset of 300 mm was suf cient. Sites were graph of a typical position is shown in Figure chosen as near as possible to a radial grid point 3B. At this height, they were away from local on a transect line, but avoiding overhanging sources of heat (parked cars, etc.), but were trees, and other local anomalies. At rural sta- accessible from ground level for data transcrip- tions, in elds, identical sensor housings were tion. The sensors were offset from the lighting used, but attached to dead trees rather than light- column, which would be warmed in the sun, to ing columns. A photograph of a typical position avoid erroneous readings. A separate test at the is shown in Figure 3A. BRE, with sensors mounted at a range of offsets The data-loggers used were battery-powered from a lighting column, had con rmed that a lat- miniature loggers, 35 mm long. These have a Figure 3 Photographs of typical positions of the air temperature measuring device in (A) a rural area just outside London and (B) central London Downloaded from bse.sagepub.com at Templeman Lib/The Librarian on December 7, 2015 102 The London Heat Island quoted accuracy of 6 0.2 K with a resolution of simultaneous data the mean heat island intensity 6 0.25 K, but all loggers were inter-compared for June, July and August 1999. and individual calibrations determined so that The mean intensities are quite small: around the data from all loggers could be normalized to one degree in the daytime, and 2.5 degrees over- the mean of the distribution. Thus, mean tem- night. The standard deviations about the means perature differences have an accuracy of about show that there were periods of cool islands 6 0.1–0.2 K (a residual inaccuracy that exists (London being cooler than the countryside), because of a very small logger drift over time). particularly in June. A resolution limit of 0.25 K remains on any one The mean night-time intensity is larger than reading. As yet, no comparison to an absolute the daytime intensity in each month. There were reference has been made, but the latter adjust- times of higher intensities, and this is clearer in ment is likely to be very small, 0.2 K at the most. the frequency distribution shown in Figure 4, All data points are hourly and simultaneous. which covers the same period. The histograms Eighteen stations were operating from 1 June are shaded to distinguish periods of warm and 1999 onwards (including the rural reference and cool islands. London was at least 2 K warmer four stations at the focus), with 62 operating by than the rural reference for about 15% of the mid-July. Suitable rural reference sites proved daytime and 58% of the night-time. For 15% of dif cult to nd and a single site, at the rural end- the night-time it was more than 5 K warmer. point of the western transect, has been used for Figure 4 also shows that London was some- all the analysis. times a cool island: for 15% of the daytime, but Additional weather data such as wind only 2% of the night-time over the 3-month speed/direction, precipitation, cloud cover and period. solar radiation were provided by the Met Of ce Chandler11 reported that the difference in for the London Weather Centre site and for urban–rural maxima for July days was negative Heathrow Airport as well as from the BRE (a cool island) or zero for more than 30% of weather station at Garston. days, and for June and August negative for more than 40% of days. A similar calculation for the 3 Results June–August 1999 data shows that urban–rural maxima were negative for 9.2% of the days. 3.1 Heat island intensity Thus, for these more recent data, it would appear Heat island intensity is a parameter usually that a heat, rather than cool island prevails for used to describe the urban heat island. For the more of the time compared to the data from present study, the intensity has been calculated 1951–60 used by Chandler. from the difference between the average of three Figure 5 shows the heat island intensity by stations at the focus (in central London near the time of day for the 3-month period. All hourly British Museum), and the rural reference, 18 data are plotted, together with the mean value miles west. Table 2 summarizes from hourly and spread for each hour of the day. Time is Table 2 Mean heat island intensities for summer 1999 in K Day s Night s Overall s June 0.82 1.17 2.59 1.76 1.70 1.74 July 1.25 1.24 2.65 1.69 1.95 1.64 August 1.14 1.19 2.89 1.58 2.01 1.65 Overall 1.07 1.22 2.71 1.69 1.89 1.68 Day (08.00–19.00) and night (20.00–07.00), (s = standard deviation) Downloaded from bse.sagepub.com at Templeman Lib/The Librarian on December 7, 2015 R Watkins et al. 103 Figure 5 The variation of the heat island intensity in Lon- don by time of day for the 3-month period (June-August 1999). The mean and standard deviation for each hour are also shown 3.2 Location of thermal centre The spatial pattern of temperature is complicated in the daytime, but, after sunset, simpli es to a city-centred symetrical pattern of contours. The centre of the night-time heat island is not static, but shifts in response to wind direc- tion, with a response time of a few hours. This was observed by studying the temperatures dur- ing periods with opposite wind directions (south–west and north–east) but similar other conditions. For example on two consecutive nights in August 1999 the centre of London experienced opposite wind directions. The pre- vious days were cloudy with no rain, and the Figure 4 Frequency distribution of daytime and night- time heat island intensity (focus-rural temperature general temperature was similar. Under those difference) during the summer of 1999 in London conditions, the thermal centres were shifted, roughly in line with the change in wind direc- local British Summer Time. The plot shows tion. clearly that the London Heat Island is predomi- The second night, of the 8 August, had a very nantly a night-time phenomenon, with mean low wind speed of just over 1m/s, and by exam- urban rural temperature differences around 0.7– ination of this and other calm nights the position 0.8 K after 11.00 in the morning, and rising shar- of the thermal centre can be determined. This is ply at 19.00 to a rough plateau of 2.5–3.0 K that found to be 1.8 miles directly east of the (quasi- is maintained through the night until 06.00. arbitrary) focus of the network of stations. The Considerable scatter is evident with extremes area is the City of London, in the nancial dis- of 6 7 K. Some of these extreme values are trict, between Moorgate and Liverpool Street associated with rain showers, and short-lived, Station. This is an area characterized by a high but others are ‘true’ heat-islands and develop to density of taller of ce buildings. It is also an these values overnight. area of high anthropogenic heat release.17 Downloaded from bse.sagepub.com at Templeman Lib/The Librarian on December 7, 2015 104 The London Heat Island 3.3 Mean temperature and distance from London Mean temperature tends to reduce with increasing distance from London. The mean temperature for all night-time data in August is plotted against radial distance from London (from the effective thermal centre, 1.8 miles east of the focus) in Figure 6. A quadratic regression line has been tted to the data, although this may not be the most appropriate relationship to use. This line shows that 77% of the variance (r2 = 0.77, n = 63) of the mean night-time tem- perature across London can be accounted for by the radial distance of each location. Because the temperature pattern in the day- time is more complicated, a similar regression Figure 7 Transect pro le from west to east across Lon- don on the hottest day in August (1 August 1999) at 00:00 for mean daytime temperatures yields a very hours and 16:00 BST. All temperatures are simultaneous. poor relationship—only 25% of the variance is The dotted lines are interpolated associated with radial distance. recorded. From the western end, the temperature 3.4 Transect pro les rose quickly to a plateau at 14 miles from Lon- Transect pro les change continually, but an don, rising only very gently at 8 miles from Lon- example is given in Figure 7 of the pro le from don to a value 4 miles out the same as at the west to east across London on the hottest day in focus. On the east side the temperature fell away August (1 August 1999, maximum temperature from the focus immediately until 4 miles east 33.7°C), at 00:00 hours and 16:00 BST—a when the rate of fall slowed until 14 miles out Sunday. when the temperature decreased more quickly. At 00:00 there was practically no wind (0.5 Temperature ranged from 17.7°C in the west m/s), about 50% cloud cover and no rain to 22.3°C at the focus to 17.7°C in the east—a range of 4.6 K. At 16:00 a southerly wind had risen in the last few hours to reach 6.7 m/s. There was about 60% cloud cover and no rain recorded. From the western end, the temperature was level until about 10 miles from London, when it rose a degree, and another degree at the 6-mile point, dropping a degree down to the focus. Going east, it rose 1 degree immediately, then dropped 2.5 degrees at Bishopsgate, then rose steadily to reach a transect peak at the 8-mile point, drop- ping 3-degrees to the 12-mile point and nally reaching a minimum at the 18-mile point. Temperature ranged from 31.0°C in the west to 31.6°C at the focus to 33.7°C at 8 miles east Figure 6 The mean temperature against radial distance and 29.4°C in the east—a range of 4.3 K. (The from London for all night-time data in August 1999 (from the effective thermal centre, 1.8 miles east of the focus). station at the 16:00 peak, 8 miles east, is a large A quadratic regression line is also shown residential housing estate.) Downloaded from bse.sagepub.com at Templeman Lib/The Librarian on December 7, 2015 R Watkins et al. 105 3.4.1 Percentiles 50th and 90th with increasing distance from London, to 1.2 K Figure 8 shows the temperature differences at WW10 16 miles west (2 miles east of the ref- that were not exceeded for 50% or 90% of the erence station). time at the different stations along the western In contrast, during the daytime, the maximum transect in August 1999. These local heat island 90 percentile is 3 K and this is found both near intensities are shown for the four stations within the focus and at four miles west. The variation 400 m of the focus and the stations up to 16 of the percentiles is much smaller and all stations miles away from London. (Stations WW01 to 3 up to 14 miles west are within one degree of were not installed at the time.) Stations WW04 each other. to WW11 (the reference temperature station, D T = 0 K) are at 2-mile spacings. 4 Conclusions For 10% of the night-time, the temperature difference at the focus (focus–rural temperature) exceeded 5 K at all four stations there. This 90 The intensity of the London Heat Island has been percentile heat island intensity reduced steadily assessed using a radial grid of 68 stations rec- ording simultaneous hourly air temperatures. The intensity reaches 6 7 K on occasion. The urban heat island has been found to be predominantly a nocturnal phenomenon. The mean night-time intensity was 2.8 K in August with 5 K being exceeded for 10% of the time. The mean daytime intensity was 1.2 K with 3 K being exceeded for 10% of the time. However the temperature of the central urban areas did not show such a marked and consistent differ- ence from the western suburbs in the daytime. In June to August London was cooler than the rural reference for about 18% of the time during the day with the mean urban–rural temperature difference approaching zero at 15.00. For 10% of the night-time the centre of London was more than 5 K warmer than the rural reference station. In the daytime, the distribution was atter, and the differences smaller; 3 K was exceeded at the centre for 10% of the time. The focus of the measurement array was located in Bloomsbury, but this was found to be 1.8 miles west of the night-time thermal centre of London. This thermal centre is in the City of London and is characterized by tall buildings and high anthropogenic heat emissions. The night-time temperature pro le of London tends to follow a simple pattern with contours Figure 8 Percentiles of the hourly local heat island inten- arranged roughly circularly around the centre; it sity of different stations along an example transect, with is found that 77% of the variance of the mean respect to the rural reference station. The histogram shows the temperature differences that were not night-time temperature across London is related exceeded for either 50% or 90% of the time to distance from the thermal centre. The more Downloaded from bse.sagepub.com at Templeman Lib/The Librarian on December 7, 2015 106 The London Heat Island complicated pattern in the daytime precludes urban heat-island in Barcelona. International such a simple relationship. Journal of Climatology 1994: 14: 705–10. The elevation of temperature in the central 5 Jauregui E. Heat island development in Mexico urban areas at night reduces the potential for City. Atmospheric Environment 1997; 31: 3821– 31. night-cooling of buildings. It may therefore lead 6 Ludwig FL. Urban temperature elds. In: to increased use of air-conditioning. The poten- Symposium on urban climates and building tial impact on energy used for cooling is the sub- climatology 1968 (Brussels). World ject of current work. Meteorological Organization, 1968: 80–107. 7 Brundl W, Hoppe P. Advantages and Acknowledgements disadvantages of the urban heat-island—an evaluation according to the hygro-thermic effects. Archives for Meteorology Geophysics This work is funded by the research programme and Bioclimatology Series B—Theoretical and of the Construction Directorate of the Depart- Applied Climatology 1984; 35: 55–66. ment of the Environment, Transport and the 8 Santamouris M. Natural cooling techniques. In Regions. This work would not have been poss- Proceedings of Conference on Passive Cooling, 1990 (Ispra) EEC, DGXII, 1990. ible without the considerable help of the 33 local 9 Landsberg H, ed. The urban climate. Academic authorities in London and surrounding areas. Press, 1981. Numerous landowners have also been extremely 10 Taha H. Urban climates and heat islands: obliging in giving permission for measurement albedo, evapotranspiration, and anthropogenic stations to be set up on their land. heat. Energy and Buildings 1997; 25: 99–103. 11 Chandler TJ. The climate of London, 1st ed. Hutchinson & Co (Publishers) Ltd, 1965. References 12 Chandler TJ. City growth and urban climates. Weather 1964; 19: 170–71. 1 Rosenfeld AH, Akbari H, Bretz S, Fishman BL, 13 Lee DO. Urban warming?—An analysis of Kurn DM, Sailor D, Taha H. Mitigation of recent trends in London’s heat island. Weather urban heat islands—materials, utility programs, 1990; 47: 50–56. updates. Energy and Buildings 1995; 22: 255– 14 Lyall IT. The London heat-island in June-July 65. 1976. Weather 1997; 32: 296–302. 2 Givoni B, ed. Climate considerations in building 15 Wheeler D, Mayes J, eds. Regional climates of and urban design. Van Nostrand Reinhold, the British Isles, 1st ed. Routledge, 1997. 1998. 16 Met. Of ce. Measurement of temperature 2nd 3 Santamouris M. The Athens Urban Climate ed. London: HMSO, 1981. Experiment. In PLEA ‘98 1998 June 1998 17 Harrison R, McGoldrick B, Williams CGB. (Lisbon). James & James Science Publishers Arti cial heat release from Greater London Ltd, 1998: 147–52. 1971–1976. Atmospheric Environment 1984; 4 Moreno-Garcia MC. Intensity and form of the 18: 2291–1304. Downloaded from bse.sagepub.com at Templeman Lib/The Librarian on December 7, 2015