Atmospheric Environment 40 (2006) 4308–4315 Vertical atmospheric structure estimated by heat island intensity and temporal variations of methane concentrations in ambient air in an urban area in Japan Masahide Aikawa à , Takatoshi Hiraki, Jiro Eiho Hyogo Prefectural Institute of Public Health and Environmental Sciences, 3-1-27 Yukihira-cho, Suma-ku, Kobe, Hyogo 654-0037, Japan Received 18 January 2006; received in revised form 28 March 2006; accepted 31 March 2006 Abstract The vertical atmospheric structure was studied and evaluated based on the distribution and variation of the air temperature in an urban area in Japan. A difference was observed in the annual mean diurnal variation of the air temperature between the urban site and a suburban site. The maximum and minimum temperatures were 1:64  C at 1:00 and 1:17  C at 15:00, respectively, resulting in an estimated intrinsic heat island intensity of 0.47 ð¼ 1:6421:17Þ  C. The height of the temperature inversion layer was approximately 90 m above the ground, based on the intrinsic heat island intensity in an area where no vertical air temperature was available. The temporal variations of the methane concentrations in ambient air and the contribution of automobile emissions were estimated and well accounted for by the postulated temperature inversion layer. r 2006 Elsevier Ltd. All rights reserved. Keywords: Urban heat island; Air temperature; Air pollution; Methane; Urban area; Japan 1. Introduction The urban heat island phenomenon has been studied all over the world with the objective of limiting increases in air temperature (e.g., Oke, 1973; Oke and Maxwell, 1975; Gotoh, 1993; Saitoh et al., 1996; Yamashita, 1996; Oke et al., 1999). Some studies have demonstrated that urban air temperatures increase more on their own than they do as a result of climate change and that the rapid development of urban areas influences the magnitude and patterns of heat islands (Hinkel et al., 2003; Zhou et al., 2004; Weng and Yang, 2004; Fujibe, 2004). On the other hand, the urban heat island phenomenon has been studied in terms of vertical atmospheric structure (e.g., Bornstein, 1968; Bornstein and Azie, 1981; Draxler, 1986; King and Russell, 1988; Saitoh et al., 1996; Shahgedano- va et al., 1997). The vertical atmospheric structure is closely related to air pollution (Aikawa et al., 1996; Sahashi et al., 1996). In the present study, data sets including air tempe rature and concentrations of air pollutants such as methane were analyzed to investigate the relationship of the air temperature with concentrations of air pollutants and to identify any factors which control temporal variations of air pollutant concentrations in the atmosphere in the ARTICLE IN PRESS www.elsevier.com/locate/atmosenv 1352-2310/$ - see front matter r 2006 Elsevier Ltd. All rights reserved. doi:10.1016/j.atmosenv.2006.03.044 à Corresponding author. Tel.: +81 78 735 6930; fax: +81 78 735 7817. E-mail address: Masahide_Aikawa@pref.hyogo.jp (M. Aikawa). Hanshin area, which is a 10  10 km area between the cities of Osaka and Kobe, two of the largest cities in Japan. The findings are reported below. 2. Experimental 2.1. Survey sites Air temperature and methane concentrations were measured at two environmental monitoring stations in twocitiesintheHanshinarea:Amagasaki(StationA: 135  24 0 58 00 E; 34  43 0 19 00 N) and Nishinomiya (Station N: 135  21 0 18 00 E; 34  45 0 54 00 N). The two stations are in a10 10 km area. The locations o f the stations are shown in Fig. 1. The Hanshin area is between Osaka City (population 2; 634; 000=222 km 2 ) and Kobe City (population 1; 520; 000=551 km 2 ). Th e H an shin area is characterized by intensive industrial development and dense populations. Mt. Rokko (altitude 931 m), which runs east and west, is located in Kobe City. Station N is at the east end of the mountain range. Station A is in an urban area, whereas Station N is at a suburban ARTICLE IN P RESS Fig. 1. Location of environmental monitoring stations. Station A and Station N are located in Amagasaki City and Nishinomiya City, respectively. M. Aikawa et al. / Atmospheric Environment 40 (2006) 4308–4315 4309 site. The two stations are located less than 50 m above sea level (a.s.l.). 2.2. Air temperature and air pollutant concentrations The conditions for the measurement of air temperature and methane concentrations are sum- marized as follows: Station A. The air temperature was measured on the grass-covered roof of a five-story building (about 19 m above the ground), where a thermometer shelter was installed. The methane concentrations were measured in the same building by using a non-methane hydro- carbon monitor (HCM-4A, Shimadzu C orp., Kyoto, Japan). The air inlet was about 15 m above the ground. Station N. T he air temperature was measured on the concrete roof of a two-story building (about 8 m above the grou nd) by using a forcibly aspirated shelter. The methane concentrations were measured in the same b uilding by using a non -methane hydro- carbon monitor (HCM-4A, Shimadzu Corp., Kyo to, Japan). The air inlet was about 8 m above the ground. 2.3. Survey period and data acquisition The data measured in 2004 were used for analyses. All of the parameters were measured hourly. 3. Results and discussion 3.1. Temporal variation of methane concentrations in ambient air Methane in ambient air is one of the main gases related to climate change. The lifetime of methane in ambient air is approximately 10 years (IPCC, 2001), which is longer than those of other air pollutants such as NO þ NO 2 ðNO x Þ (1 day) and CO (65 days) (Seinfeld, 1986). NO x and CO are among the most important air pollutants in urban areas because they are emitted by automobiles. Sahashi et al. (1996) demonstrated the nitrogen- oxide layer over a heat island. However, when considering kinetic behaviors of air pollutants in ambient air, air pollutants with longer lifetimes are advantageous because complicated atmospheric chemical reactions can be avo ided, suggesting that methane is more favorable for study purposes than NO x and CO. In addition, natural sources ac- counted for approximately 40% of total methane sources (IPCC, 2001), and trans portation contrib- uted 1.1% of methane emissions in Japan (CGER/ NIES, 2004), indicating that there is a smaller influence of methane emissions from mobile sources on methane concentrations in ambient air compared with other air pollutants such as NO x and CO. Fig. 2 sh ows the annual average t emporal varia- tions of methane concentrations in ambient air at Station A and Station N. In general, the methane concentrations in ambient air were low during the daytime and high a t night, and the lowest and the highest concentrations appeared at 16:00 and 7:00–8:00, respectively. Fig. 3(a)and(b)showthe seasonally average t emporal variation s o f m ethane concentrations in ambient a ir at St ation A and Station N, respectively. The methane concentration in ambi- ent air in winter (December–February) showed two maximum peaks in the temporal variations at mid- night (23:00–1:00) a nd in the morning (7:00–9:00), that in summer (June–August) showed one maximum ARTICLE IN PRESS 174 176 178 180 182 184 186 1:00 2:00 3:00 4:00 5:00 6:00 7:00 8:00 9:00 10:00 11:00 12:00 13:00 14:00 15:00 16:00 17:00 18:00 19:00 20:00 21:00 22:00 23:00 Time CH 4 concentration /ppm StationA StationN Fig. 2. Annual average temporal variations of methane concentrations in ambient air at Station A and Station N. M. Aikawa et al. / Atmospheric Environment 40 (2006) 4308–43154310 peak in the morning (7:00–9:00), and t hose in spring (March–May) and in a utumn (September–October) showed transitional temporal variations between winter and summer. Aikawa et al. (1996) reported similar annual and monthly/seasonal temporal varia- tions of th e methane co ncentration s in Nago ya City (population 2; 202; 000=326 km 2 ), Japan, and dis- cussed the temporal variations in relation to the stability of the atmosphere, suggesting that atmo- spheric stability i s r elated to the temporal variatio ns o f the methane concentrations in ambient air in the current study area. ARTICLE IN P RESS Spring 170 175 180 185 190 1 3 5 7 9 11131517192123 Time CH 4 concentratin /ppm Summer 170 175 180 185 190 1 3 5 7 9 11131517192123 Time CH 4 concentratin /ppm Autumn 170 175 180 185 190 1 3 5 7 9 11131517192123 Time CH 4 concentratin /ppm Winter 170 175 180 185 190 1 3 5 7 9 11131517192123 Time CH 4 concentratin /ppm (a) Spring 170 175 180 185 190 1 3 5 7 9 11131517192123 Time CH 4 concentratin /ppm Summer 170 175 180 185 190 1 3 5 7 9 11131517192123 Time CH 4 concentratin /ppm Autumn 170 175 180 185 190 1 3 5 7 9 11131517192123 Time CH 4 concentratin /ppm Winter 170 175 180 185 190 1 3 5 7 9 11131517192123 Time CH 4 concentratin /ppm (b) Fig. 3. Seasonally average temporal variations of methane concentrations in ambient air at Station A (a) and Station N (b). M. Aikawa et al. / Atmospheric Environment 40 (2006) 4308–4315 4311 3.2. Mean air temperature and diurnal variation The annual mean values of the daily mean air temperatures at Station A and Station N in 2004 are summarized in Table 1. The annual mean air temperature at Station A ð17:6  CÞ was higher than that at Station N ð15:8  CÞ. One reason is the difference in elevation. Therefore, the annual mean air temperature had to be corrected to consider the effect of the elevation of the location. A moist- adiabatic lapse rate of 0:6  C=100 m was considered. The result of the correction is shown in Table 1. The difference after the correction was 1:5  C. Aikawa et al. (2006) demonstrated that the difference in mean air temperatures at Station A and Station N from 1990 to 2003 was approximately 0:4  C, smaller than that in 2004. Japan Meteorological Agency measured the air temperatures in Osaka: representative urban site ð135  31:1 0 E; 34  40:9 0 NÞ, approximately 9 km east–southeast from Station A and in Sanda: suburban site ð135  12:7 0 E; 34  53:6 0 NÞ, approximately 16 km northwest from Station N. The difference of the annual mean air temperature between Osaka and Sanda was 3:04  C: mean and 3:10  C: median for the duration of 1990–2003. On the other hand, the difference of the annual mean air temperature between Osaka and Sanda in 2004 was 3:40  C. The difference in 2004 was also larger than that in the duration of 1990–2003 in the survey by Japan Meteorological Agency, similar to the current resul ts. Fig. 4 shows the diurnal variations in the air temperature as illustrated by the corrected hourly air temperature at each station. The differences in the diurnal variations between Station A and Station N were maximum and minimum at 1:00 ð1:64  CÞ and 15:00 ð1:17  CÞ, respectively, with a mean difference of 1:47  C. Assuming that urban heat island intensity is defined as the difference in the air temperatures at Station A and Station N, the urban heat island intensity was strongest at 1:00 and weakest at 15:00. 3.3. Height of postulated temperature inversion layer Bornstein (1968) demonstrated that a tempera- ture inversion layer covered New York City at approximately 310 m above the ground. Aikawa et al. (1996) reported that a temperature inversion layer was formed at approximately 60 m above the ground in Nagoya City, Japan. Bornstein (1968) measured the vertical air temperature profile by helicopter, while Aikawa et al. (1996) showed the vertical air temperature profile based on measure- ments taken at the Nagoya TV Tower, 180 m above the ground. In the current study, no air temperature data were availab le to evaluate a vertical air temperature profile. Therefore, the height of a postulated temperature inversion layer was calcu- lated by using the distribution of the air temperature ARTICLE IN PRESS Table 1 Measured and corrected annual mean values of daily mean air temperatures in 2004 Station A Station N Measured ð  CÞ 17.6 15.8 Corrected ð  CÞ 17.6 16.1 12 13 14 15 16 17 18 19 20 21 1:00 2:00 3:00 4:00 5:00 6:00 7:00 8:00 9:00 10:00 11:00 12:00 13:00 14:00 15:00 16:00 17:00 18:00 19:00 20:00 21:00 22:00 23:00 0:00 Time Air temperature /°C Station A StationN Fig. 4. Diurnal variations of air temperature corrected by elevation at Station A and Station N. M. Aikawa et al. / Atmospheric Environment 40 (2006) 4308–43154312 shown in Section 3.2. The following assumptions were made: Assumption (i). A temperature inversion layer would exist in the current study area. Assumption (ii). A temperature inversion layer with a maximum height would be formed in the air at the site of Station A. Assumption (iii). A temperature inversion layer would be formed on the ground condition at the site of Station N. In general, solar radiation is strong in the daytime, leading to an active vertical mixing of the air by convection. In contrast, there is no solar radiation at night, resulting in the formation of a temperature inversion layer by radiation cooling. Therefore, it is appropriate to discuss the height of the temperature inversion layer based on the distribution of the nighttime air temperature. In the current study area, even in the daytime, when there should have been active vertical mixing, a difference in the air temperature ð1:17  CÞ was observed between Station A (urban area) and Station N (suburban area) as shown in Section 3.2. The 1:17  C difference in the air temperature was presumably due to an intrinsic difference caused by the characteristics of the sites. Therefore, when the height of the postulated temperature inversion layer in nighttime is discussed, the 1:17  C difference in the air temperature in the daytime should be subtracted from the difference in the nighttime. Bornstein (1968) clarified that the height of the base of the crossover layer was 310 m and the average intensity of the urban heat island , as measured by the magnitude of the temperature difference between urban and rural sites, was 1:6  C. Saitoh et al. (1996) observed and simulated the heat island intensity ð5  CÞ and the height of the cross- over phenomenon (1000 m) in metropolitan Tokyo. Shahgedanova et al. (1997) showed the heat island intensity ð123  CÞ and the height of the urban boundary layer (85–128 m) in Moscow. The statis- tics shown by Bornstein (1968) were used for the following calculation since the studies of Bornstein and Saitoh et al. (1996) yielded similar calculation results. Taking into account the statistics shown by Bornstein and the above-mentioned essential differ- ence in the corrected hourly air temperatures at Station A and Station N ð1:6421:17 ¼ 0:47  CÞ, the height of the postulated temperature inversion layer in the current study area can be calculated as follows: height of postulated tempe rature inversion layer ¼ 0:47=ð 1 :6 = 310Þ¼91 m. 3.4. Relationship of temporal varia tions of methane concentrations with postulated temperature inversion layer Aikawa et al. (1996) reported the relationship between the temporal variations of methane con- centrations and the lifted temperature inversion layer, and they accounted for the temporal varia- tions of methane concentrations by the formation and disappearance of the lifted temperature inver- sion layer. The seasonally average temporal varia- tions of the methane concentrations in the current study were so similar to those found in the study by Aikawa et al. (1996) that the temporal variations of methane concentrations could be also accounted for by the lifted postulated temperature inversion layer introduced in Section 3.3. On the other hand, in the seasonally average temporal variations of the methane concentrations in ambient air during the winter season, as shown in Fig. 3(a) and (b), a small shoulder was observed in the morning (6:00–10:00). The contribution of methane from automobiles to the total emission of methane was generally not large (IPCC, 2001). However, considering a relatively small and urba- nized area such as that in the current study, the contribution of automobile emissions would not be negligible. The small shoulder would appear in the morning as a result of a combination of automobile emissions and the formati on of the lifted postulated temperature inversion layer found in the current study area. Fig. 5 shows the average temporal variations of methane concentrations on weekdays and weekends. The shoulders on weekdays were larger than those on weekends at both sites, strongly suggesting that the shoulder results from the contribution of automobile emissions. Seasonal differences in the air temperature at midnight (23:00–1:00) and in the early morning (7:00–9:00) between Station A and Station N are summarized in Table 2. The corrected seasonal differences in the air temperature at midnight and in the early morning in winter (0.60 and 0:22  C, respectively) were larger than those (À0:04 and 0:05  C) in summer, suggesting that the heat island phenomenon was observed in winter both at mid- night and in the early morning in the current study area, while the heat island intensity was small or nonexistent in summer both at midnight and in the early morning. Aikawa et al. (1996) also reported the temperature inversion layer almost never formed in Nagoya City during the relevant time in ARTICLE IN P RESS M. Aikawa et al. / Atmospheric Environment 40 (2006) 4308–4315 4313 summer. The seasonal variation of the heat island phenomenon and the formation of the postulated temperature inversion layer would result in the small shoulder in the morning in winter. 3.5. Estimation of contribution of automobile emissions The contribution of automobile emissions to the temporal variations in methane concentrations was estimated based on the road traffic census data taken in 1997. Fig. 6 shows an outline of the traffic volume on the main roads on weekdays (7:00–19:00). To estimate the contribution in the morning, one-third of the traffic volume shown in Fig. 6 was distributed as the morning traffic volume. The traffic volume shown in Fig. 6 included all types of vehicles, including gasoline- and diesel-powered passenger cars as well as small and large diesel- powered freight vehicles. The methane emission factor was calculated by taking into account the types of vehicles and constituent ratio of vehicles. It was assumed that the travel distance of vehicles in the current study area was 10 km. It was also assumed that the diffusion volume was one-fourth of the volume of a circular cone with a 10 km radius and 90 m height. Under these assumptions, the estimated concentration of methane for the shoulder peak in the morning was 2.6 ppm. In contrast, the observed shoulder methane concentra- tion on weekdays was approximately 3 ppm, which shows that the shoulder methane concentration was estimated fairly well. ARTICLE IN PRESS 178 180 182 184 186 188 190 1 7 9 1011121314151617181920212223 Time CH 4 concentration /ppm Weekend inStation A Weekdayin Station A Weekend inStation N Weekdayin Station N 654328 Fig. 5. Average temporal variations of methane concentrations on weekdays and weekends at Station A and Station N. Table 2 Seasonal difference of air temperature between Station A and Station N in midnight and early morning a Difference of air temperature Summer Winter 23.00–1.00 7.00–9.00 23.00–1.00 7.00–9.00 Measured ð  CÞ 1.13 1.22 1.77 1.39 Corrected b ð  CÞÀ0.04 0.05 0.60 0.22 a Midnight and early morning means 23.00–1.00 and 7.00–9.00, respectively. b Measured values are corrected by subtracting intrinsic difference ð1:17  CÞ. 10 km 10 km R171 R2 R43 H.H. M.H. 25,000 26,000 45,000 52,000 33,000 42,000 36,000 63,000 Fig. 6. Outline of traffic volume on main roads in the daytime (7:00–19:00) on weekdays in the current study area. R2, R43, R171, M.H., and H.H. show National Routes No. 2, No. 43, No. 171, Meishin Expressway, and Hanshin Expressway, respectively. The numerals show the traffic volume of vehicles (vehicles/ daytime (12 h)). M. Aikawa et al. / Atmospheric Environment 40 (2006) 4308–43154314 4. Conclusions The temporal variations of methane concentrations in ambient air in an urban area were studied in relation to air temperature distribution and the estimated p ostulated vertical a tmospheric structure. The average temp oral variations in methane concen- trations showed seasonal c haracteristics. In w inter, th e observed sho ulder peak was due to the contribution of automobile emissions. T he shoulder peak concen tra- tion could be estimated fairly well by considering the contribution of automobile emissions and the postu- lated vertical a tmospheric structure estimated b y the measured air temperature distribution. References Aikawa, M., Yoshikawa, K., Tomida, M., Haraguchi, H., 1996. Characteristic behaviors of the methane concentrations in urban atmosphere of Nagoya with relation to atmospheric stability. Environmental Sciences 9 (2), 201–210. Aikawa, M., Hiraki, T., Sumitomo, S., Eiho, J., 2006. Distribu- tion and variation of the air temperature from 1990 through 2003 in urban areas of Hyogo prefecture from the aspect of heat island phenomenon. Bulletin of the Hyogo Prefectural Institute of Public Health and Environmental Sciences 2, 1–9. Bornstein, R.D., 1968. Observations of the urban heat island effect in New York City. Journal of Applied Meteorology 7, 575–582. Bornstein, R.D., Azie, I., 1981. Urban effects of sodar-derived inversion patterns in San Jose. Fifth Symposium on Turbulence, Diffusion, and Air Pollution 1981, pp. 187–188. Center for Global Environmental Research, National Institute for Environmental Studies, Japan, 2004. National greenhouse gas inventory report of Japan 2004. CGER-REPORT CGER- I059-2004, p. 2.8. Draxler, R.R., 1986. Measured influence of the nocturnal urban heat island on pollutant transport. Joint Conference on Applications of Air Pollution Meteorology 6, 243–245. Fujibe, F., 2004. Features of extremely high summertime temperatures in Japan in recent years. Geographical Review of Japan 77 (3), 119–132 (in Japanese with English abstract). Gotoh, T., 1993. Relation between heat islands and NO 2 pollution in some Japanese cities. Atmospheric Environment 27B, 121–128. Hinkel, K.M., Nelson, F.E., Klene, A.E., Bell, J.H., 2003. The urban heat island in winter at Barrow, Alaska. International Journal of Climatology 23, 1889–1905. IPCC (Intergovernmental Panel on Climate Change), 2001. In: Houghton, J.T., Ding, Y., Griggs, D.J., Noguer, M., van der Linden, P.J., Dai, X., Maskell, K., Johnson, C.A. (Eds.), The Scientific Basis. Cambridge University Press, Cambridge, U.K. King, C.W., Russell, C.A., 1988. Temperature structure effects on pollutant distribution in the Denver metropolitan area. Joint Conference on Applications of Air Pollution Meteorol- ogy 6, 243–245. Oke, T.R., 1973. City size and the urban heat island. Atmo- spheric Environment 7, 769–779. Oke, T.R., Maxwell, G.B., 1975. Urban heat island dynamics in Montreal and Vancouver. Atmospheric Environment 9, 191–200. Oke, T.R., Spronken-Smith, R.A., Jauregui, E., Grimmond, C.S.B., 1999. The energy balance of central Mexico City during the dry season. Atmospheric Environment 33, 3919–3930. Sahashi, K., Hieda, T., Yamashita, E., 1996. Nitrogen-oxide layer over the urban heat island in Okayama City. Atmo- spheric Environment 30, 531–535. Saitoh, T.S., Shimada, T., Hoshi, H., 1996. Modeling and simulation of the Tokyo urban heat island. Atmospheric Environment 30, 3431–3442. Seinfeld, J.H., 1986. Atmospheric Chemistry and Physics of Air Pollution. Wiley, New York, p. 8. Shahgedanova, M., Burt, T.P., Davies, T.D., 1997. Some aspects of the three-dimensional heat island in Moscow. International Journal of Climatology 17, 1451–1465. Weng, Q., Yang, S., 2004. Managing the adverse thermal effects of urban development in a densely populated Chinese city. Journal of Environmental Management 70, 145–156. Yamashita, S., 1996. Detailed structure of heat island phenomena from moving observations from electric tram-cars in metro- politan Tokyo. Atmospheric Environment 30, 429–435. Zhou, L., Dickinson, R.E., Tian, Y., Fang, J., Li, Q., Kaufmann, R.K., Tucker, C.J., Myneni, R.B., 2004. Evidence for a significant urbanization effect on climate in China. Proceed- ings of the National Academy of Sciences of the United States of America 101, 9540–9544. ARTICLE IN P RESS M. Aikawa et al. / Atmospheric Environment 40 (2006) 4308–4315 4315 . Atmospheric Environment 40 (2006) 4308–4315 Vertical atmospheric structure estimated by heat island intensity and temporal variations of methane concentrations. observed and simulated the heat island intensity ð5  CÞ and the height of the cross- over phenomenon (100 0 m) in metropolitan Tokyo. Shahgedanova et al. (1997) showed the heat island intensity ð 123  CÞ. Relation between heat islands and NO 2 pollution in some Japanese cities. Atmospheric Environment 27B, 121–128. Hinkel, K.M., Nelson, F.E., Klene, A.E., Bell, J.H., 2 003. The urban heat island in winter