ARTICLE IN PRESS

Building and Environment 43 (2008) 717 www.elsevier.com/locate/buildenv

A eldwork study on the diurnal changes of urban microclimate in four types of ground cover and urban heat island of Nanjing, China
Liangmei Huanga,b, Jianlong Lia,, Dehua Zhaoa, Jiyu Zhub
a

School of Life Science, Nanjing University, Nanjing 210093, China b Guangxi Forestry Research Institute, Nanning 530001, China

Received 7 March 2006; received in revised form 16 November 2006; accepted 30 November 2006

Abstract Records of the past years showed that the climate of built-up regions differs signicantly from rural regions and one of the most obvious and important modifying effects of urbanization on local climate is the urban heat island (UHI). In this paper, four types of land cover, namely urban bare concrete cover, urban woods or the shade of trees, urban water areas and urban lawn, were selected to study their microclimate, and the UHI was also analyzed using air temperature data measured at four xed observation spots in Nanjing, China, during hot weather from July to September, 2005. Dry and wet bulb temperature data were obtained by whirling psychrometers, and wind speed data by cup anemometers. Our observed data focused on the detailed statistical analysis of the microclimate variation in the four types of land cover during the whole day. The results showed that: (1) the microclimate of these four types of land cover had signicant differences among different observation sites. In general, the air temperature of these four types of land cover complied with the order during daytime: bare concrete cover4lawn4water areas4woods or the shade of trees, with reversed order during nighttime when the air temperature of the lawn became the lowest. Compared with the bare concrete cover, the other three types of land cover showed the effect of dropping air temperature ranging between 0.2 and 2.9 1C. There were some instant dynamic characteristics in detailed temporal series among these four types of cover in the different observation sites. (2) The UHI effect could be detected obviously by the air temperature difference between the urban center area and the rural area. The average UHI intensity during the monitoring period was between 0.5 and 3.5 1C; however, there were also signicant day-to-day variations. A strong UHI effect usually occurred around midnight; while about 23 h after sunrise the UHI began to decrease till midday time; and during 13:0015:00, the UHI effect had a sudden increase and then decreased again; after sunset, a peak UHI effect was frequently observed during 18:0021:00. (3) Finally, by means of the standard deviation (SD), this paper provides a concise and comprehensive understanding for the temporal and spatial microclimatic dynamics of these four kinds of urban cover in the four observation sites. Air temperature at the height of 1.5 m in Nanjing showed that the nocturnally horizontal temperature gradient was somewhat different from that reported in other large cities, and a marked heterogeneity in a smaller ground cover scale could be detected from the microclimatic spatial pattern. There is no doubt that the analysis of these four types of land cover presents the insight into possible countermeasures to decrease the high air temperature in hot summers, and is relevant to the urban planning redevelopment. r 2007 Elsevier Ltd. All rights reserved.
Keyword: Urban ground cover types; Site specic effect; Reference points; Data normalization

1. Introduction Human settlements modify the materials, the structure and the energy balance of the surface of the Earth and the composition of the atmosphere compared with the surrounding natural terrains, although they also suffer
Corresponding author. Tel./fax: +86 25 83592715.

E-mail address: jlli2008@nju.edu.cn (J. Li). 0360-1323/$ - see front matter r 2007 Elsevier Ltd. All rights reserved. doi:10.1016/j.buildenv.2006.11.025

from some effects of human economy e.g., agriculture, forestry. These articial factors determine a distinct local climate in the cities, which is the so-called urban climate [14]. The damages on the natural and built-up environments, caused by the speculative use of urban spaces, have been taking their most serious form, and, from the climatic point of view, have been harmful to the cities and their residents, such as excessive heat storage and high concentration of air pollution. In order to prevent these

ARTICLE IN PRESS
8 L. Huang et al. / Building and Environment 43 (2008) 717

damages and create optimum bioclimatic conditions, studies on the urban climate, the urban heat island (UHI) and urban air pollution are stimulated not only by the necessity to gain knowledge of numerous secondary effects of the excessive urbanization but also by very practical needs of town planning. Studies of individuality of the urban climate have been conducted for over 150 years, as long ago as that it was observed in big cities of temperate latitudes (London, Paris) [5]. Since then, much research has been done on various interesting and practical aspects of urban climatic phenomena by means of different measurement tools and methodology, and the UHI features have also been clearly demonstrated [2,623]. Researchers have mainly paid attention to the spatial and temporal distribution, cause and effect as well as prediction model of the urban climate, the UHI and air pollution, and failed to take climatic considerations of urban design into account; therefore, data provided by climate researchers have a low impact on the urban planning process, and do not always meet the demands of urban planners and architects. While rapid changes in microclimate generated by different urban land cover inuence the comfort and health of the inhabitants as well as energy consumption and air quality, therefore, it is important to understand the causes of air temperature variation in different land use areas. A few studies have measured the temperature distribution on the exterior of a single building [12,24] and over two representative buildings having different construction characteristics both in the summer and the winter [25]. A study of a selected urban area of central city and comprehensive analysis of various urban objects over a 24-h period had been also conducted in Tel-Aviv using an infrared video radiometer [26]. However, for a diversity of urban ground cover and weather conditions, these studies were confronted with problems of spatial and temporal data substitution due to the lack of a long data record and poor spatial sites. But high temporal and spatial resolution and the use of concurrent ground truth data are necessary for accurate measurement [27]. From a climatological perspective, the UHI and its negative environmental inuence are also pronounced during the hot season. Experiments on diurnal range basis during a calm winter day, which demonstrated the most pronounced radiant temperature differences between diverse urban coverage at microscale level, have been carried out in the cities Goteborg, Hokkaido and Tel-Aviv [10,26,28]. However, few eld experiments were carried out in the hot season; therefore, it is important to carry out simultaneous eld experiments at different sites for monitoring the thermal pattern of the urban land cover more accurately in the hot season. In an attempt to address this important issue, this paper examines the effects of four different types of urban land cover on air temperature variations between day and night, different months and different observation weather situations at four observation sites in Nanjing, China, during

the hot summer period. We presumed that there would be a most representative type of ground cover at each observation site. Based on the detailed analysis of the variation of these samples of data measured from the sites, this most representative type of ground cover would be classied. This paper covered the following aspects:

   

An analysis of the air temperature differences from different types of ground cover. An analysis of the temperature differences from the different observation sites. A comprehensive and concise understanding of the deviation of the air temperature in temporal and spatial series. A discussion of the application of these results in urban planning.

2. The study area: the city of Nanjing Nanjing is one of the six ancient capitals of China, and is situated on the south bank of the lower reaches of the Yangtze River. Together with the other two cities of Chongqing and Wuhan along the Yangtze River, Nanjing is also famous as the city of the chimney place for its hot climate with maximum summer temperature of 40 1C. It is located at 32103N in the north-subtropical climatic zone with well-dened seasons and a mean annual temperature of 15 1C. Rainfall averages 1033 mm per annum, occurring mainly in summer. The zonal natural vegetation is a mixed broadleaf evergreen and deciduous forest. Sampling survey of urban trees yielded an evergreen:deciduous trees ratio of about 1:3, and the top three species were Platanus acerifolia, Juniperus chinensis and Ligustrum lucidum [29]. The citys varied topography comprises mountains, low hills, low terraces, plains and rivers (Fig. 1). Two rivers and two lakes dene the drainage system. The Yangtze River runs in the northwest, and the Qinhuai River runs southeast to northwest through the city to join Yangtze, it bifurcates into a cluster of tributaries that spread in the south city. Xuanwu Lake (3.7 km2) west of Purple Mountain and Muchou Lake (0.37 km2) west of the city wall have been developed into municipal parks. Fringe mountains surround the city, extending as low hills into built-up areas. The Purple Mountain at its eastern fringe, rising to 448 m, has the largest green cover of 29.7 km2 of semi-natural forest. In the southeast and southwest, there are green wedges of farmland which break up the otherwise monotonously contiguous urban developments. In the south, there is Yuhua Hill providing extensive green spaces for the nearby densely populated old downtown, and it is physically segregated from the inner city by the railway and the city wall. In the north, Mufu Hill is an elongated belt beside the Yangtze River, which is almost engulfed by the urban area except for the narrow riverbank which serves as a corridor to the extensive countryside beyond the city. The extensive green space penetrating into

ARTICLE IN PRESS
L. Huang et al. / Building and Environment 43 (2008) 717 9

Fig. 1. The location of the observation sites and the major landscape form of Nanjing.

the city is signicant for microclimatic amelioration and recreation. The old core of the city lies in the south near the InnerQinhuai River and large factories were established along the Yangtze River. The built-up areas increased 60% from 1986 to 2000, covering an area of 860 km2 with a population of more than 5 million. South Nanjing is more densely populated than the north, accommodating the original urban core with dense packing of buildings separated by narrow streets and scant open spaces. The three southern districts have a built-up population density reaching about 30,000 persons/km2. The northern part of the city has more institutional establishments such as government agencies and colleges as well as the industrial locations. The residential neighborhoods have a lower built-up density with green spaces interspersed between buildings. From what we have described it can be seen that this study city of Nanjing, for the reason of its geographical position, size and distribution of buildings, is an ideal testing target for research on the microclimate of land cover in a subtropical city. 3. Observations and methodology We may consider that the urban area is inuenced by a number of factors which complement each other in overall thermal behavior: physical characteristics of the various urban elements, the sky view factor (SVF), surface geometry, color and the weather conditions. Physical characteristics comprise thermal conductivity, heat capacity, thermal diffusivity and thermal admittance, whereas the SVF of a point is proportional to the area of the overlying hemisphere which is open to the sky [6]. Street

geometry, SVF and the weather conditions determine the total amount of direct solar radiation input to a horizontal surface unit. This paper describes the diurnal thermal behavior of the four types of urban land cover both in spatial and temporal series. 3.1. Study times and sites design Measurements of various parameters were taken during the hot period from July to September 2005 including different weather conditions except rainy days. Calm conditions prevailed during all of the diurnal cycles of the experiment. The daily average, minimum and maximum temperature, and the daily average relative humidity, cloud cover and wind speed measured by the Nanjing Meteorological Bureau on each observation day are listed in Table 1. The observation days selected to study urban microclimate dynamics and the heat island were chosen to represent a range of different weather conditions except rainy days. Because of the need to obtain data of high spatial and temporal resolution, taking account of our limited budget and manpower, we were only able to collect 24-h samples of data on six occasions, once in July, twice in August and three times in September. Observations were conducted simultaneously on four xed sites. The locations of the four xed sites are indicated by the % sign in Fig. 1, consisting of one urban commercial center (Fuzi Temple), one urban forest (Purple Mountain) and one urban water area (Xuanwu Lake) as well as a rural area (Pukou). The background of the commercial site is the dense buildings and heavy trafc vehicles and population, and the Pukou rural area is separated from the major city by the Yangtze River and is farmland in which the detailed observation location is an

ARTICLE IN PRESS
10 L. Huang et al. / Building and Environment 43 (2008) 717 Table 1 The weather condition of the observation days during hot period from July to September 2005 Dates Temperature (1C) Average 17 July 18 July 13 August 14 August 26 August 27 August 9 September 10 September 17 September 18 September 24 September 25 September 30.1 29.1 29.7 30.5 23.2 25.3 24.7 24.9 28.2 26.1 21.3 21.0 Max. 34.1 33.5 33.1 33.8 28.4 29.8 30.5 30.4 30.9 30.2 24.9 24.3 Min. 26.2 26.5 27.2 27.2 18.5 22.5 20.0 19.8 25.8 23.7 19.6 19.6 Relative humidity (%) Average 68 70 70 68 77 76 70 73 78 82 69 73 Cloud amount (decas) Average 5.1 4.4 7.4 3.1 3.1 8.3 1.3 1.3 3.3 9.4 9.1 9.5 Wind speed (m/s) Average 2.8 3.0 2.8 4.3 1.3 1.5 1.0 1.3 2.0 2.5 2.3 3.0

afliated campus of Nanjing University. Then at each site we selected woods, lawn, water area and bare concrete as four sample points to collect samples of microclimate data. For the similarity between the observation sites and the sample points in landscape, we dened the four observation sites as mesoscale land cover, while the four sample points as microscale land cover. This denition will be helpful for us to get a comprehensive understanding of analysis of the three types of standard deviations (SD) of air temperatures and our inference in Section 4.3. Since it was eldwork, the areas of the four types of land cover cannot be dened unanimously at two scales. We only estimate land cover type by its area of matrix 460% obviously, and the least area of the microscale land cover is 4500 m2 which can be considered as the microscale in the landscape study and microclimatic research [8,10,23,26,3032]. Using whirling psychrometers (DHM2) and cup anemometers (DEM6) which were calibrated, checked and compared before and after the experiment under the same conditions, we measured each hour the microclimatic indices of dry and wet bulb temperatures and wind speed at 1.52.0-m height in the middle of the microscale sample points from 08:00 am of one day till 07:00 am of the next day. Wind direction and velocity, in our case, were measured just at the lawn sample point. 3.2. Data process Comparison of the sites air temperatures to that of a single outside reference point such as a nearby meteorological station would lead to wrong conclusions without simultaneous measured data, even when the comparison is done for days of measurements at a single site [15]. The air temperature inside the site usually depends on the shading intensity of a partial shaded area, on the thermal properties of the soil and on the air temperature of its immediate background. The latter varies among different urban sites due to factors affecting the air temperature such as

vegetation, built-up geometry, topography, trafc density and other anthropogenic heat-release factors. Therefore, simultaneous observations were applied in the present study during a hot summer on four xed observational sites where green areas with trees, lawn, water areas and bare terrain were chosen to represent sample points and microscale land cover. And comparison was conducted as shown in the following steps to get more reliable and accurate calculation and to test our hypothesis: (a) To compare the difference of the air temperature among the sample points at each site, we select the warmest one as the reference point. In our case it refers to the bare concrete cover. (b) To compare the difference of the air temperature among the four observation sites, we also select the warmest one as the reference site. The city center is the most ideal reference in our case. (c) While to compare the deviation of the air temperature in temporal and spatial series at each site, we select the average of the four sample points as the reference point in spatial series, and the average of the 24-h air temperatures in temporal series. Then the temporal and spatial data centralization and normalization are conducted by the following equations, respectively: X X ij X i i 1; 2; 3; . . . ; P 24,
ij _

(1)

^ X ij X ij X j j 1; 2; 3; . . . ; N 4,

(2)

X ij X i  X ij q i 1; 2; 3; . . . ; P; j 1; 2; 3; . . . ; N, PN 2 i1 X ij X i =N 1

(3) where Xij is the original data of the ith hour in the jth type of ground cover,X i is the average of the four types of ground measured data, X j is the average of the 24-h _ data, X is the rst type of new data after centralization
ij

ARTICLE IN PRESS
L. Huang et al. / Building and Environment 43 (2008) 717 11

^ in spatial series, X ij is the second type of new data after  centralization in temporal series, X ij is the third type of new data after normalization in temporal and spatial series. Therefore, using these three types of new data matrixes, we can get three types of SD. Thus, at each site, a reference point is selected and site-specic effect will be also estimated and incorporated in the empirical calculation. In order to control the comparison more uniformly, P. acerifolia, aged about 50 to 70 years and planted in these xed observational sites, are also chosen to sample the air temperatures of the green woods or the shade of trees areas. To emphasize the signicant change in thermal behavior, the target of our research focuses on the thermal differences of various urban land cover at different scales. 4. Results and analysis 4.1. Detailed comparison of air temperature variation among the four types of land cover at microscale 4.1.1. Urban commercial center (the Fuzi Temple site) The Fuzi Temple site is located in the commercial center of Nanjing city (the exact location seen in Fig. 1), where dense population and heavy trafc, concrete buildings and pavements as well as skyscrapers considered as the site environment are found. For daily behavior of air temperatures in the four types of land cover during the hot summer, the thermal peaks occurred at 14:00. Bare concrete cover was found to be the warmest target during daytime, and was warmer than the other three by about 0.32.2 1C during the hot summer (Fig. 2(a)). The radiation absorbed in concrete cover, black asphalt or bare soil is relatively high due to high heat capacity and the thermal conductivity of these materials [6,26]. This radiation, converted into heat energy, causes heat uxes in a direction that warms the surface and the surrounding environment. The lowest temperatures for the lawn at night result presumably from its high humidity and exposure to the open sky, which enhances upwelling radiation loss throughout the night much more than expected. Similarly, the relatively high surface air temperatures (about 0.6 1C) at night for the concrete cover area than that for the lawn area are mostly due to their lower SVF resulting from their proximity to dense buildings and trafc. As shown in Fig. 2(a), a thermal difference was found among the four types of land cover both in daily and monthly hours. The lowest air temperature was measured in the shade of trees during hot daytime, but subtle higher air temperature than that on concrete pavement and lawn occurred during nighttime, and its frequencies increased from July to September. Thus, the trees have an effect on reducing air temperature during hot daytime, especially at midday, while retaining and even elevating air temperature during cold nighttime. The water area was the second warm target not only during daytime but also during nighttime. Due mainly to the inuence of

heavy trafc, high density of population and buildings on the microclimatic characteristics of these four types of ground cover, the differences among them were obvious. In contrast to the concrete surface, the shade of trees could lower the temperature by about 0.22 1C during the hot period, and several maxima occurred during 11:0013:00 when the solar radiation came to a peak. The lawn could lower the air temperature by about 0.41.7 1C, so could the water areas by about 0.41.0 1C. However, these three kinds of ground cover were warmer than the concrete surface sometimes, especially during nighttime; for example, the water areas were warmer by about 1.0 1C at 04:00 on August 26. 4.1.2. Urban water areas (the Xuanwu Lake site) Xuanwu Lake is close to the city center areas of Nanjing (Fig. 1), and there are three little islands in the lake connected by articial roads and bridges. The lawn selected as the observation point is surrounded by trees; the bare ground cover is the black asphalt road; and we measured the sample point of water areas just standing by the bank of Xuanwu Lake and stood in the middle of the woods to sample the data of the shade of trees. Fig. 2(b) describes the variation of air temperature for the four types of ground cover. Generally speaking, the variation trend of air temperature was similar to that of the Fuzi Temple site described above, but the temperature range (by about 0.41.9 1C) in contrast to the concrete cover was somewhat lower and more stable than that at the Fuzi site. The shade of trees reduced the air temperature signicantly during daytime, and the detailed maximum temperature ranges were 0.5 1C at 17:00 on July 17, 1.5 1C at 07:00 on August 13, 1.5 1C at 18:00 on August 26, 1.9 1C at 14:00 on September 9, 1.2 1C at 12:00 on September 17 and 1.0 1C at 12:00 on September 24. The lawn could also reduce the air temperature signicantly in contrast to concrete cover, especially at 18:00, the maximum value of reducing the air temperature was 0.4 1C at 18:00 on July 17, 0.9 1C on August 13, 1.9 1C on August 26 and 1.8 1C at 17:00 on September 10. While the water areas showed subtle reduction of the air temperature effect, the maximum range was just 0.20.5 1C. Moreover, the water area sometimes was warmer than the concrete surface even during daytime, for example, the maximum range came up to 0.8 1C at 14:00 on September 17. During the nighttime, the water areas were also the warmest target with temperature range in contrast to the concrete cover by about 0.30.7 1C, and the lawn was still cooler by about 0.3 1C than the others. 4.1.3. Urban forest (the Purple Mountain site) As an urban forest, Purple Mountain not only provides the largest greening areas but also promotes landscape beauty and atmospheric quality. It is an attractive destination for Nanjing people and the visitors from all over the world. The observation site is situated in a valley of Purple Mountain. In this area, the concrete cover

ARTICLE IN PRESS
12 L. Huang et al. / Building and Environment 43 (2008) 717

a
1 Temperature range (C) 0.5 0 -0.5 -1 -1.5 -2 -2.5

b
1 Temperature range (C) 0.5 0 -0.5 -1 -1.5

c
1 0.5 Temperature range (C) 0 -0.5 -1 -1.5 -2 -2.5 -3 -3.5

d
1 Temperature range (C) 0.5 0 -0.5 -1 -1.5 -2 -2.5 8:00 13:00 19:00 2:00 September woods 8:00 14:00 20:00 4:00 August lawn Local time (hours) 9:00 14:00 19:00 0:00 July water area 5:00

Fig. 2. The diurnal changes of air temperature ranges ( air temperature on the three types of land coverthat on the concrete cover) at the four observation sites during hot summer from July to September 2005: (a) Fuzi Temple site, (b) Xuanwu Lake site, (c) Purple Mountain site, (d) Pukou rural site.

selected as observation point is the black asphalt road with 10-m width, with trees standing on two sides, therefore, sometimes the air temperatures measured on the concrete cover areas during daytime were lower (by about

0.31.1 1C) than those on the lawn areas which is broad, and is surrounded by the trees and near to the water area (an articial lake). At nighttime, the lawn was still the lowest one in air temperature, the water areas became the

ARTICLE IN PRESS
L. Huang et al. / Building and Environment 43 (2008) 717 13

warmest one with its air temperature warmer by about 0.20.8 1C than the other three, and in the shade of the trees the air temperature was warmer up to 0.20.4 1C. Compared with the cement cover, the other types of land cover reduced the air temperature ranging from 0.5 to 2.9 1C (Fig. 2(c)). The maximum value of reducing air temperature was still from the shade of trees and mainly occurred during 11:0015:00, for example, the detailed temperature range was 2.9 1C at 17:00 on July 17, 1.5 1C at 11:00 on August 13, 2.2 1C at 15:00 on August 26, 2.0 1C at 13:00 on September 9, 1.7 1C at 13:00 on September 17 and 1.6 1C at 13:00 on September 24. Water areas could reduce air temperature signicantly during the daytime, also especially at noon, which was different to that of Xuanwu Lake which is also an attractive destination for visitors. The detailed temperature range in contrast to the concrete cover was 1.4 1C at 11:00 on August 13, 1.8 1C at 12:00 on August 26, 1.3 1C at 12:00 on September 9, 0.8 1C at 13:00 on September 17 and 0.8 1C at 13:00 on September 24. However, the lawn drop the air temperature was somewhat lower than the former ones. During the nighttime, however, the air temperature of the water areas is higher (by about 0.20.9 1C) than that of the concrete cover. The analysis of the above shows that the air temperature in the shade of trees and in the water areas are stable at microscale, while for the concrete cover and the lawn it can be easily inuenced by the current weather conditions, such as the wind speed and the solar radiation. Compared to the commercial center and the urban water areas (Xuanwu Lake) as well as the Pukou rural area, the temperature range between the four types of ground cover was larger and more obvious in the urban forest site. This is due to the fact that the site background weather condition disturbed by various anthropogenic factors in Purple Mountain is not as intensive as that in the other three observation sites; thus the instant change of microclimate is not frequent in the calm day. 4.1.4. Rural area (the Pukou rural site) The Pukou district lies in the north part of the Nanjing city, and is separated by the Yangtze River from the main city (Fig. 1). An afliated campus of Nanjing University, which is located in Pukou district, was selected as the rural site in our case. Because of certain practical difculty, there is no sample point of water area. The selected sample points of the woods, the lawn and the concrete cover at this site stand side by side, and are connected with each other. The observation points suffered from the shadow of teaching buildings at sunrise and at sunset. We could distinguish the temperature difference among these three types of ground cover obviously, and the order was that the cement cover4lawn4the shade of trees during the daytime, while it is reversed during nighttime (Fig. 2(d)). The air temperatures in the cement cover and the lawn areas are sometimes interweaved during the daytime, which rarely happened in the commercial sites where the cement cover was thronged with people and trafc. The maximum range

of air temperature was about 1.22.5 1C between the concrete cover and the shade of trees during 11:0015:00, while it was 0.81.5 1C between the lawn and the concrete cover at noon. During nighttime, the shade of woods area was occasionally warmer by about 0.21.0 1C than the concrete cover area, while the lawn was warmer by about 0.20.4 1C than the concrete cover. In general, the lawn was still the coolest one during nighttime. 4.2. UHI characteristics based on comparisons for the four sites To compare the difference of the air temperature in the four sites, we calculate the average air temperature of the four types of ground cover to represent one of the observation sites, namely site-specic effect [15]. The mean value of the 24-h air temperature at the four observation sites showed that the Fuzi Temple site was the warmest area, namely the commercial center of the city during the hot period, then the Pukou rural site, then the Xuanwu Lake site and the coolest one was the Purple Mountain site (Figs. 3 and 4). Compared with the warmest one (Fuzi Temple), the results in detail showed that the largest temperature difference occurred in the urban forest (Purple Mountain) and was 5.3 1C at 20:00 h on August 26; while the largest temperature difference in the rural area was also 5.3 1C at 20:00 h on August 26, and the largest one in Xuanwu Lake was 3.5 1C at 16:00 h on August 26. Table 1 shows that this day was calm and sunny with wind speed of 1.3 m/s and the cloud cover of 3.1; therefore, it was very suitable for monitoring the UHI effect. As in the earlier studies, the heat island effect which was shown by the temperature difference between the city center and the surrounding areas increased during the nighttime [20,33]. But for Xuanwu Lake and Purple Mountain located near the city (Fig. 1), the temperature difference decreased during 19:0022:00 h, especially for Xuanwu Lake. This is perhaps due to the inuence of the heat island on the surrounding areas during nighttime [9,21,23,34]. In fact, Xuanwu Lake is close to the city center and to the Nanjing railway station. To compare these air temperature differences more subtly, namely at the microscale land cover, the detailed corresponding results were as follows: (1) on the concrete cover areas, the Pukou rural area was warmer by about 0.41.0 1C than the city center during 11:0014:00 h, when the solar radiation in the rural area was more intensive than that in the commercial center because of the differences of their SVF and the heavy layer of urban canopy which reects a large amount of solar radiation. For the rest of the day, the city center was warmer by about 0.24.3 1C than the rural area. The largest temperature difference frequently occurred during 18:0021:00 h, when the trafc was heavy and the crowds of people going for a walk were dense on the roads of the city center; after that, this temperature difference decreased to a certain extent, but became stable during the middle of night (22:0004:00).

ARTICLE IN PRESS
14 L. Huang et al. / Building and Environment 43 (2008) 717

36 Air temperature (C)

31

26

21 Local time (hours) 16 8:00 13:00 18:00 22:00 6:00 9:00 14:00 19:00 0:00 6:00 9:00 13:00 17:00 21:00 1:00 5:00 September August July Purple Mountain Xuanwu Lake Fuzi Temple Pukou Rural

Fig. 3. The diurnal changes of the air temperature at the four observation sites during hot summer from July to September 2005.

2 1 Temperature range (C) 0 -1 -2 -3 -4

Local time (hours)

-5
8:00 13:00 18:00 22:00 6:00 9:00 14:00 19:00 0:00 6:00 9:00 13:00 17:00 21:00 1:00 5:00 September August July Purple Mountain Xuanwu Lake Pukou Rural

Fig. 4. The urban heat island intensity in Nanjing during hot summer from July to September 2005.

(2) On the lawn and the water areas, the temperature differences between the commercial center and the rural area showed a similar trend to the concrete area, but the magnitude was smaller. (3) For the wooded areas, the trees in the commercial areas are in the streets and avenues, and cars pass to and fro discharging heat which could increase the air temperature, but there were few cars passing by the observation point in the rural site where the air temperature was mainly affected by the solar radiation. Therefore, compared with the other two types of ground cover, the temperature difference (1.04.3 1C) between them was greater for most of time, especially during 17:0021:00 h. Similar phenomena occurred on the Xuanwu Lake site, where the cars frequently passed on the roads. However, besides midday time, the four kinds of ground cover were warmer by about 0.51.3 1C sometimes, especially in the

early hours after sunrise, when the near ground surface in the city center suffered from the shade of high buildings. This was different from the cases of the Pukou rural area or the Purple Mountain area. Xuanwu Lake is still in the urban area (Fig. 1) and suffered from the impact of the heat island caused by the heavy trafc, dense population, buildings and articial heat. The Purple Mountain which serves as an attractive scenery point also shows the similar temperature range (0.75.3 1C) trend as the above two cases, but the magnitude is the largest. Maybe for the altitude reason, the Purple Mountain seems to be lower by 0.31.0 1C at least over all day. During daytime, because cars and visitors crossed the roads where the shaded area of trees was selected as the observation points in our experiment, its measured temperature was also uctuating.

ARTICLE IN PRESS
L. Huang et al. / Building and Environment 43 (2008) 717 15

0.4 2.9 2.7 Standard deviation 2.5 2.3 2.1 1.9 1.7 1.5 FT 1 0.9 Standard deviation 0.8 0.7 0.6 0.5 0.4 0.3 FT XL PM PR Sites cement lawn woods Normalization XL PM PR Sites 0.1 FT XL PM PR Standard deviation Temporal series 0.35 0.3 0.25 0.2 0.15 Sites Spatial series

water area

Fig. 5. The three types of standard deviation of air temperature in the four types of land cover (FT: Fuzi Temple site, XL: Xuanwu Lake site, PM: Purple Mountain site, PR: Pukou rural site).

4.3. Comprehensive analysis of the microclimate characteristics in spatial and temporal series According to the statistical theory, the SD can manifest the variability of a group of data. After analysis of the detailed dynamic changes of air temperature in the four types of ground cover, respectively, we can see the general trend of our measured data in the spatial series and the temporal series. Fig. 5 shows three types of SD for us to understand the diurnal changes of the air temperatures in these four kinds of ground cover at two scales more conveniently. At the four observation sites, the bare concrete cover had the largest SD of air temperature in the time series, and then the lawn, the shade of trees as well as the water areas. These results coincide with the above temporal analysis of the air temperature variation of the 24-h period and some previous studies [7,10,13,30,35]. While in the spatial series, here referred to the four types of ground covers and their corresponding averages, the SD showed another case in which the shade of trees had the largest SD, and then the bare concrete cover, the water areas and the lawn. It clearly showed that the shade of trees and the concrete ground cover were two kinds of extreme targets of which one had the lowest air temperature, while the other had the top air temperature, and the lawn and the water areas were in the middle. In order to discover a comprehensive air temperature variation in the land cover at microscales, we treated our observed data with normalization and got the third type of SD. Thus in the Fuzi Temple site, the water areas had the largest SD, then the lawn and the shade of the trees, while the cement ground

cover had smallest one. According to the statistical meaning, the bare concrete cover could represent the microclimate characteristics both in spatial and temporal series at the Fuzi Temple site, and could be regarded as a representative of its land cover type in thermal behavior. So did the lawn at the Xuanwu Lake site, the water areas at the Purple Mountain and the bare concrete cover at the Pukou rural site. Combined with our initial denition of the land cover at two scales in landscape ecology, the most ideal counterpart relationship of the land cover types at two scales is that the bare concrete cover corresponds to the Fuzi Temple site, and the shade of trees corresponds to the Purple Mountain site, and so the water areas to the Xuanwu Lake site, the lawn to the Pukou site. In the light of the denition of the matrix in a geographical landscape, for example, the commercial center is mainly composed of concrete roads, buildings, shops and other types of hard cover; if only their areas 460%, they could be regarded as the matrix of the city center, and could be classied as the concrete cover roughly. Therefore, to a certain degree, both the classications of the SD of air temperature and the landscape scale could discover the same physics of the urban ground cover. Such an inference could be tested at the Fuzi Temple site; however, it seemed to be not so satised for the other three sites. Table 2 lists the daily averages and SD of air temperature, relative humidity and wind speed at the four xed observation sites. Documents of the past manifested that the temperature differences between the city center and its surrounding areas decreased with increasing wind speed and disappeared on clear days, while Shudo et al. [10]

ARTICLE IN PRESS
16 L. Huang et al. / Building and Environment 43 (2008) 717 Table 2 Mean values and standard deviations (SD) of microclimatic indices as well as their correlations at the four xed observation sites during hot period from July to September 2005 Items Fuzi Temple Ws Mean values SD Correlations 0.60 0.32 0.072 0.115 Ta 26.7 2.08 0.926** Rh 71.2 9.20 Purple Mountain Ws 0.60 0.41 0.394 0.438 Ta 24.8 2.16 0.926** Rh 82.2 8.55 Xuanwu Lake Ws 0.60 0.32 0.082 0.095 Ta 25.8 2.17 0.899** Rh 80.2 9.56 Pukou Ws 0.81 0.36 0.044 0.045 Ta 26.2 2.59 0.900** Rh 79.9 9.66

Ta Rh

Ta: air temperature (1C); Rh: relative humidity (%); Ws: wind speed (m/s); df 16; *r40.468, po0.05; **r40.590, po0.01.

and Svensson et al. [30] showed that the effect of wind speed on the temperature difference was small, and did not greatly affect the correlation between urban temperature and land use. Our data from eld observation at 1.52.0-m height which included different weather conditions except rainy days showed the temperatures were also not correlated with the wind speeds, but had a high negative correlation with humidity. 5. Discussion The air temperature differences vary signicantly in the microscale domain in the hot period of measurement. During the daytime, the highest mean temperature differences (about 0.75.3 1C) between the four types of urban coverage occurred during 12:0014:00, when the air temperatures reach their daily maximum; in the early morning hours before sunrise, the air temperatures are minimal for all the observed objects, and the thermal differences are smaller compared to the daylight hours (about 0.42.3 1C), but still relatively signicant. These results are consistent with those of previous studies of the thermal behavior of urban areas, such as those of Kim and Parlow who used satellite data [7,35] and Voogt and Oke who combined airborne and ground-based measurements [13]. The most dominant surface elements contributing heat to the city are bare concrete cover which is exposed to solar radiation. This is because there is no evaporative cooling effect over this kind of surfaces [36]. Moreover, they generate more long-wave radiation than the lawn, the water areas and the vegetated surfaces. As a result, not only do they have higher temperatures, but they are also expected to inuence the air temperatures of the nearsurface air layer. At nighttime, their temperatures are lower, whereas some parts of water areas and vegetated areas are the warmest. Vegetation areas were found to have a relatively small diurnal temperature range compared to bare concrete cover, the lawn and the water areas. The air temperature patterns of the four locations selected for this study can be seen clearly. The variation, which is due to the complexity of the weather conditions, physical characteristics of the various urban elements and the geographical site, manifests itself in development of the

UHI phenomenon. Data provided by the thermal measurements of shaded vegetation areas show that in daylight hours the vegetation shaded surfaces in the rural areas were signicantly cooler by as much as 0.55.0 1C than the same surfaces in the city center. This fact reinforces the importance of providing vegetation shade in the hot and dry areas of city, which was well documented in previous studies. The results shown in the above sections also indicate that the early morning hours before sunrise and the night hours after sunset are the best times for heat island intensity discrimination, while midday times are the ideal occasions for the discrimination between various types of urban coverage at microscale. Moreover, these differences are sufcient for thermal separation. It is important to mention that some surfaces cannot be adequately described on the basis of a single specic time period of the day. However, compared with the work of Chudnovsky et al. [26], both the diurnal temperature ranges of the observation points and sites were lower. This is due to our measurements taking place in a hot summer, and including different weather conditions, while only a winter day was studied in his case. In addition, the instruments and approaches are different, and our target data are the air temperatures measured at the height of 1.52.0 m by the calibrated whirling psychrometer at the eld sites. Shade vegetation along the streets is also an important factor for reducing the total amount of incoming solar radiation. The effect of such trees on natural ventilation, especially in warm and humid areas, such as Nanjing during the hot summer, should be taken into account. The absence of vegetation may have a negative inuence favoring the development of the UHI through an increase in the amount of direct solar radiation penetrating the surface and of upwelling thermal energy. 6. Conclusions In order to integrate urban microclimatic pattern, heat island intensity and land cover in a more comprehensive view, our study has been expanded and conducted simultaneously with four typical urban observation locations and in different weather conditions during a hot summer. The thermal and humidity data acquired by

ARTICLE IN PRESS
L. Huang et al. / Building and Environment 43 (2008) 717 17

whirling psychrometers, and wind speeds acquired by cup anemometers in the real eld measurements and the statistical results showed that our method, which integrated land cover types at two scales into a comprehensive experimental system, was capable of indicating the diurnal thermal behavior of urban land cover in real time and at low cost. Since the rapid changes in the air temperature, the wind speed and the humidity inuence the comfort and health of people as well as energy consumption and air quality in the city, this knowledge could be used not only for monitoring the UHI and microclimate, but also for improving overall decision-making for environmental planning and rehabilitation in urban areas, and could be related with environmental implications in a subtropical city along the Yangtze River. Acknowledgments The Guanxi Forestry Research Institute nancially supported this work. The authors are grateful to anonymous reviewers for their comments and suggestions on the manuscript and to Professor Arthur P. Cracknell for his great help in reviewing and correcting the manuscript. The authors wish to thank the Nanjing Bureau of Xuanwu Lake Park for providing help in eld survey. Special thanks are due to associate professor Douqi Fan from Nanjing Agricultural University for his useful comments and providing instruments. References
[1] Oke TR. Boundary layer climates. London: Methuen; 1978. [2] Landsberg HE. The urban climate. New York, NY: Academic Press; 1981. [3] Goldreich Y. Urban climate studies in Israela review. Atmospheric Environment 1995;29:46778. [4] Gomez F, Gaja E, Reig A. Vegetation and climatic changes in a city. Ecological Engineering 1998;10:35560. [5] Howard L. The climate of London, vols. IIII. London: Harvey and Dorton; 1883. [6] Oke TR. The energetic basis of the urban heat island. Atmospheric Environment 1982(7):76979. [7] Kim HH. Urban heat island. International Journal of Remote Sensing 1992;13(12):231936. [8] Sakakibara Y. A numerical study of the effect of urban geometry upon the surface energy budget. Atmospheric Environment 1996;30(3):48796. [9] Tso CP. A survey of urban heat island study in two tropical cities. Atmospheric environment 1996;30(3):50719. [10] Shudo H, Sugiyama J, Noriyoshi Y, et al. A study on temperature distribution inuenced by various land uses. Energy and Buildings 1997;26:199205. [11] Voogt JV, Viau AA, Paquet F. Mapping regional air temperature elds using satellite-derived surface skin temperatures. International Journal of Climatology 1997;17:155979. [12] Voogt JA, Oke TR. Complete urban surface temperatures. Journal of Applied Meteorology 1997;36(9):111732. [13] Voogt JA, Oke TR. Effects of urban surface geometry on remotelysensed surface temperature. International Journal of Remote Sensing 1998;19(5):895920. [14] Voogt JA, Oke TR. Thermal remote sensing of urban climates. Remote Sensing of Environment 2003;86:37084.

[15] Shashua-Bar L, Hoffman ME. Vegetation as a climatic component in the design of an urban street: an empirical model for predicting the cooling effect of urban green areas with trees. Energy and Buildings 2000;31:22135. [16] Shashua-Bar L, Hoffman ME. The Green CTTC model for predicting the air temperature in small urban wooded sites. Building and Environment 2002;37:127988. [17] Shashua-Bar L, Hoffman ME. Quantitative evaluation of passive cooling of the UCL microclimate in hot regions in summer, case study: urban streets and courtyards with trees. Building and Environment 2004;39:108799. [18] Shashua-Bar L, Hoffman ME. Geometry and orientation aspects in passive cooling of canyon streets with trees. Energy and Buildings 2003;35:618. [19] Pauleit S, Ennos R, Golding Y. Modeling the environmental impacts of urban land use and land cover changea study in Merseyside, UK. Landscape and Urban Planning 2005;71:295310. [20] Fehrenbach U, Scherer D, Parlow E. Automated classication of planning objectives for the consideration of climate and air quality in urban and regional planning for the example of the region of Basel/ Switzerland. Atmospheric Environment 2001;35:560515. [21] Noto K. Dependence of heat island phenomena on stable stratication and heat quantity in a calm environment. Atmospheric Environment 1996;30:47585. [22] Klysik K. Spatial and seasonal distribution of anthropogenic heat emissions in Lodz, Poland. Atmospheric Environment 1996;30(20): 3397404. [23] Giridharan R, Ganesan S, Lau SSY. Daytime urban heat island effect in high-rise and high-density residential developments in Hong Kong. Energy and Buildings 2004;36:52534. [24] Tanimoto J, Hagishima A, Chimklai P. An approach for coupled simulation of building thermal effects and urban climatology. Energy and Buildings 2004;36:78193. [25] Hoyano A, Ilno A. Analysis of the sensible heat ux from exterior surface of buildings using time sequential thermography. Atmospheric Environment 1999;33:394151. [26] Chudnovsky A, Ben-Dor E, Saaroni H. Diurnal thermal behavior of selected urban objects using remote sensing measurements. Energy and Buildings 2004;36:106374. [27] Streutker DR. Satellite-measured growth of the urban heat island of Houston, Texas. Remote Sensing of Environment 2003;85:2829. [28] Eliasson I. Urban nocturnal temperatures, street geometry and land use. Atmospheric Environment 1996;30:37992. [29] Jim CY, Chen SS. Comprehensive green space planning based on landscape ecology principles in compact Nanjing city, China. Landscape and Urban Planning 2003;65:95116. [30] Svensson MK, Eliasson I. Diurnal air temperatures in built-up areas in relation to urban planning. Landscape and Urban Planning 2002;61:3754. [31] Mueller EC, Day TA. The effect of urban ground cover on microclimate, growth and leaf gas exchange of oleander in Phoenix, Arizona. International Journal of Biometeorology 2005;49(4):24455. [32] Saaroni H, Ben-Dor E, Bitan A, et al. Spatial distribution and microscale characteristics of the urban heat island in Tel-Aviv, Israel. Landscape and Urban Planning 2000;48:118. [33] Mills G. Urban climatology and urban design. In: de Dear RJ, Kalma JD, Oke TR, Auliciems A, editors. Biometeorology and urban climatology at the turn of the millennium. Proceedings of the conference ICB-ICUC99, WMO/TD No. 1026, Sydney, 1999. p. 5414 [selected papers]. [34] Klysik K, Fortuniak K. Temporal and spatial characteristics of the urban heat island of Lodz, Poland. Atmospheric Environment 1999;33:388595. [35] Parlow E. Analysis of urban climates through remote sensing methods. Geographische-Rundschau 1998;50(2):8993. [36] Kjelgren R, Montague T. Urban tree transpiration over turf and asphalt surfaces. Atmospheric Environment 1997;32(1):3541.