SOLA, 2012, Vol. 8, 029−032, doi:10.2151/sola.2012-008 29

Effects of Land Reclamations in Bay on the Regional Climate

Rui Ito1, Shozo Yamane2, and Takehiko Satomura3 1Graduate School of Engineering, Doshisha University, Kyotanabe, 2Department of Environmental Systems Science, Doshisha University, Kyotanabe, Japan 3Graduate School of Science, Kyoto University, Kyoto, Japan

that used by Arakawa (1938). They observed that the DTR does Abstract not decrease monotonously, but has decadal and multi-decadal variation: DTR even increased in some periods. The results are The effects of land reclamations in the Osaka Bay area on the different from the well-known feature of urban warming, i.e., the climate in the region are investigated using a numeri- monotonic decrease in the DTR. Therefore, to understand the cal model. It is observed that reclamations in Osaka Bay lead to urban climate in Osaka, one needs to explain the cause of the DTR a significant increase in the surface air temperature (SAT) over increase in addition to causes of the urban climate causing the the inland during the daytime and a decrease from midnight to DTR decrease, such as the increase in artificial materials, high- the morning. This indicates that reclamations increase the diurnal rise buildings, and anthropogenic energy consumption. temperature range. The decrease in the SAT in the morning, which In this study, the effects of the reclamations in the Osaka Bay is in contrast to urban warming, is due to the decreased surface area on the regional climate are investigated by numerical experi- wind speed caused by the reclamations and is accompanied by re- ments to discuss the causes of the increase in the DTR. Using old duced vertical heat exchange. maps, we reproduced the coastline that existed about a century ago. We can estimate the realistic effects of the reclamations by 1. Introduction using the actual past coastline, unlike the idealized coastlines used in the previous studies. This study focused on the decreasing SAT Reclaimed areas continue to spread in Japan. Large-scale rec- in the early mornings, which is indicated in Kimura and Takahashi lamations have been conducted around the coastal areas near big (1991), and Kondo et al. (1996), but is not yet extensively studied. cities. Over the last few centuries, reclaimed areas have expanded Kitao et al. (2010) showed the effects of urban surface change by on the coast of Osaka Bay. Even now, reclaimed areas continue to investigating the differences in SAT and surface wind between the expand and the coastline is moving further away. Simultaneously, present and potential natural vegetation cases in the Osaka region in the inland, some cities have developed owing to the change in using a mesoscale model, WRF. In the potential natural vegeta- land use and rising of new buildings. tion case, which is used as the past condition, the land use in a The effects of reclamations on the urban climate have been in- part of the present reclaimed areas is changed to the sea. Thus, the vestigated using numerical models under ideal conditions. Kimura differences included the effects of the reclamation. But, the only and Takahashi (1991) examined the effects on the central part of effects of reclamation can not be distinguished from the effects of Tokyo for typical winter and summer days using the Boussinesq the land use change. In addition, the decreasing SAT in the early hydrostatic model. Three cases were examined: the northern half mornings is not shown. We also discuss the cooling effect of the of Tokyo Bay, the part of the bay shallower than 20 m depth, and reclamations. the entire bay. Temperature changes in the central part of Tokyo due to the reclamations are not strongly dependent on these cases. 2. Methods The surface air temperature (SAT) after reclamations increased about 1.5°C in the late afternoons, about 0.4°C at midnight, and The non-hydrostatic mesoscale model developed by the Japan decreased less than 0.1°C in the early mornings compared to those Meteorological Agency (JMA-NHM) (Saito et al. 2006) is used before the reclamations. Kondo et al. (1996) studied the effects of to reproduce the regional climate. The improved Mellor-Yamada reclamations between the islands in Pusan, Korea, with a hydro- level-3 scheme for the planetary boundary layer (Nakanishi and static model. The SAT was estimated using a heat budget model Niino 2006), short- and long-wave radiation scheme based on the developed by Deardorff (1978), except for the calculations of heat GSM radiation scheme (Kitagawa 2000; Yabu et al. 2005), and the and moisture transfer coefficients. The SAT over the reclaimed four-layer thermal diffusion scheme as a land surface (Saito et al. island increases about 1.0°C at 16 local standard time (LST) and 2001) are employed in this study. decreases about 0.1°C at 4 LST. These studies indicate that the di- The JMA-NHM is configured with 40 vertical layers from the urnal temperature range (DTR) increases because of the reclama- surface to an altitude of about 16 km. The thickness of the lowest tions. layer is 40 m and that of the top layer is 802 m. We used two In contrast, urban warming is known to decrease the DTR, that nested domains as shown in Fig. 1. The outer domain is 333 km × is, increasing the annual mean of daily minimum temperature (Tmin) 333 km with a 3-km horizontal resolution, and the inner domain is rather than that of daily maximum temperature (Tmax). Kawamura 111 km × 111 km with a 1-km resolution. (1979) and Mikami (2003) showed that the increasing rate of Tmin Initial and boundary conditions for the outer domain are gen- was greater than that of Tmax from the data at the Tokyo observa- erated from a 33-h mesoscale model forecast provided by JMA tory from 1876 to 1970 and from 1900 to 2000, respectively, (GPV-MSM). The horizontal resolution of GPV-MSM is 5 km. thus indicating a decrease in DTR by the effects of urbanization. Conditions for the inner domain are from the hourly outputs of the Kondo (2009) estimated the effects of urbanization on tempera- outer domain. The topography and land-sea distributions in the ture by using a simple one-dimensional boundary layer model. He models are based on the global digital elevation (GTOPO30) and showed that the DTR decreased by 2.4°C because of the change the global land cover characteristics datasets (GLCC), obtained of thermal conductivity from a dry bare area to an urban area. from the United States Geological Survey, with a horizontal grid Ito et al. (2008) also studied the decrease in the DTR by using spacing of 30 arc seconds (approximately 1 km). the data over a period from 1883 to 2007, which is longer than The land surface types are defined by the digital national land information with a horizontal grid of 100 m (version 1997) pro- duced by the Geographical Survey Institute of Japan. It has eleven Corresponding author and present affiliation: Rui Ito, Graduate School of Science, Kyoto University, Kitashirakawa, Oiwake, Sakyo, Kyoto 606- categories: rice field, other agricultural field, forest, barren, build- 8502, Japan. E-mail: [email protected]. ©2012, the Meteorologi- ing, traffic field, vacancy, river field, seashore, sea, and golf field. cal Society of Japan. The outer model is integrated for 33 h from 18 LST of the previ- 30 Ito et al., Effects of Land Reclamations in Osaka Bay on the Regional Climate

was greater than 0.9 over the land, and the wind speed was greater than 0.75 over most areas of the Osaka Plain. The calculation seems to agree well with the observed data. The following results shown are based on the inner model out- puts. The SAT and surface wind denote the temperature and wind at the lowest model level (20 m above the ground). The effects of the reclamations on the SAT over the reclaimed areas are clearly noticed as a cold anomaly in the early mornings and a warm anomaly in the afternoons in both summer (JJA) (Fig. 2) and winter (DJF) (not shown), as reported in the previous stud- ies (Kimura and Takahashi 1991; Kondo et al. 1996). Sato et al. (2004) showed the effect on the SAT over the land-use changed Fig. 1. Calculated domain for the a) outer and b) inner models. A square in areas, and the same results are obtained for the afternoon. The (a) represents the inner domain, and square in (b) shows the area of Fig. 6. warm anomaly extends over the inland during the daytime in all Labeled color lines in (a) represent the topography. Black solid lines and a seasons. These studies clearly show that the reclaimed areas act as black long dash line in (b) represent the present coastlines and past coast- line, respectively. Black thin lines in (b) represent contours of the topogra- a new heat source in the daytime. phy at 100-m height. A closed circle ● in (b) represents the Osaka District Figure 3 shows the wind in the present case and the anomaly Meteorological Observatory. of the wind speed at 12 LST in summer. The sea-breeze flows from Osaka Bay to the inland, and compared to the sea-breeze speed in the past, the speed is higher over Osaka Bay and lower ous day to 3 LST of the next day, and the inner model for 30 h over the Osaka Plain in the present conditions. The weakened from 21 LST. sea-breeze speed delays an arrival time of cool marine air at an Fifty-nine experimental days, five days within the same inland point, and it contributes to the increase in the inland SAT months for two years (about 60 days), are selected from June 2007 during the daytime (not shown). The SAT over central Osaka city to May 2009 (except July, in which only four days are selected). increases from 0.3°C to more than 1.0°C in the afternoons. Selected days satisfy the following three weather conditions at the The most extended cold anomaly over the Osaka Plain for Osaka observatory: (1) total sunshine for more than 70% of the each season is shown in Fig. 4. The cold anomaly extends over day, (2) daily mean surface wind speed of less than 3 m s−1, and both Osaka Bay and the inland in the mornings. The area of cold (3) no daily precipitation. These conditions are chosen referred to Kusaka et al. (2000) and Ohashi and Kida (2002). These weather conditions are satisfied about 20% of the days in the two years. The four or five longest sunshine days of the months are chosen for this experiment. Numerical experiments are conducted under two different surface conditions: the past coastline condition and the pres- ent coastline condition (Fig. 1b). The past coastline is estimated from the old maps from 1886 (Hiraoka and Noma 2006), 1908, and 1909 (obtained from the Geospatial Information Authority of Japan), and the literature on the obtained from As- sociation for the Environmental Conservation of the Seto Inland Sea (2001). By comparing old and new maps, it is found that the coastline in some areas has moved seaward by 8 km. The land-use in the past sea is estimated by replacing the pres- ent reclaimed areas with the sea. The surface height of the areas is set to 0 m. The surface parameters in the other areas remain the Fig. 2. Warm anomaly at 12 LST, averaged over the 14 summer days. same as those in the present because the purpose is to assess only Light and dark orange shaded areas are positive anomalies for the SAT the effects of the reclamations on the surrounding climate. In other difference at significance levels of more than 95% and 99%, respectively. Red contours are 0.1°C and 1°C. Black thick lines represent the present words, there is no change for the areas except the reclaimed areas. coastline. Black thin lines represent contours of the topography at 100-m The sea surface temperature for the outer model is derived from height. the near-surface air temperature at the initial time, assuming neu- tral stratification, and is constant during the integration. Anthropo- genic heat release is not taken into account. The effects of the reclamations are quantitatively estimated by subtracting the values in the past from those in the present. Statis- tical significance of the difference is examined by Student’s t-test. Hereafter, the solution obtained by the subtraction is referred to as the difference, and the statistically significant areas of difference as the anomaly. For example, a warm anomaly means the areas in which the SAT in the present is higher than in the past at a signifi- cance level of more than 95%. 3. Results and discussion The result was compared to the GPV-MSM to evaluate the model’s ability to reproduce the actual condition. The correlation coefficient of the temporal change for temperature and wind speed at 10 m height over 0−24 LST was calculated between the outer Fig. 3. Surface wind in the present case (vectors) and the anomaly for the model in the present and GPV-MSM. The correlation coefficient surface wind speed (shaded) at 12 LST, averaged over the 14 summer of the temperature during the 59 experimental days was greater days. Green and yellow shaded areas represent negative and positive anomalies, respectively, for the surface wind speed difference at a sig- than 0.9 over the land; the wind speed was also greater than 0.9 nificance level of more than 90%. Black thick lines represent the present over most of the land, and was greater than 0.95 over the Osaka coastline. Black thin lines represent contours of the topography at 100-m Plain. The temperature in winter, which we mainly focused on, height. SOLA, 2012, Vol. 8, 029−032, doi:10.2151/sola.2012-008 31

Fig. 4. Most extended cold anomaly over the Osaka Plain in a day in a) spring (8 LST), b) summer (7 LST), c) autumn (9 LST), and d) winter (10 LST), averaged over each seasonal days. Light and dark blue (orange) Fig. 5. Temporal variation of the surface wind in the present case (vectors) shaded areas are negative (positive) anomalies for the SAT difference at and the anomaly (shaded) averaged over the 15 winter days at a) 5 LST, b) significance levels of more than 95% and 99%, respectively. Red contour 7 LST, c) 9 LST, and d) 11 LST. Light and dark blue (orange) shaded areas is 0.1°C. Blue contours are −0.01°C and −0.1°C. Black thick lines show are negative (positive) anomalies for the SAT difference at significance the present coastline. Black thin lines show contours of the topography at levels of more than 95% and 99%, respectively. Squares in (a), (b), and 100-m height. (c) represent the middle of the cold anomaly over the Osaka Plain at each time. anomaly is the largest in winter, and smallest in summer. The cold anomaly in winter extends over most of the Osaka Plain with a difference of −0.01°C or less. The areas of cold anomaly in spring and autumn reach over half the plain. Though the difference is small, it is statistically significant at the level of 99%. The winter cold anomalies from 5 LST to 11 LST are shown in Fig. 5. The squares in Fig. 5a, 5b, and 5c are the middle of the cold anomaly over the Osaka Plain and the urbanized area, and are used in the subsequent discussion below. The cold anomaly, which is generated over the reclaimed area during the night, starts extending over the near-coast areas of the Osaka Plain from mid- night (not shown). The SAT decreases more than 0.1°C over the reclaimed areas and the near-coast areas. The cold anomaly ex- tends slowly to the inland in the early mornings (Fig. 5a and 5b). The extension over the plain is notable around 9 LST (Fig. 5c). After 9 LST, the anomaly proceeds further inland over 3−4 h (Fig. 5d). Figure 6 shows the surface wind in the present case at 9 LST over the warm and cold anomalies, and the surface wind differ- Fig. 6. a) Surface wind in the present case at 9 LST (vectors) over the ence with the present potential temperature at 9 LST in winter. It warm and cold anomalies (shaded), and b) surface wind difference (vec- is the time when the cold anomaly notably starts extending over tors) over both the anomalies with the present potential temperature (K) the plain. The prevailing surface wind in Fig. 6a flows from the (contours) at 9 LST, for the averaged 15 winter days. Light and dark blue inland toward the bay: seasonal northerly wind and land-breeze (orange) shaded areas in (a) are the negative (positive) anomalies for the flow over the plain. This surface wind direction is opposite to the SAT difference at significance levels of more than 95% and 99%, respec- tively. The red contour interval is 0.3 K in (b). A black thick line shows direction of the cold anomaly extension. Area A in Fig. 6a and 6b the present coastline. Black thin lines show contours of the topography is the middle of the cold anomaly over the Osaka Plain and the ur- at 100-m height. Squares in both figures denoted by area A are the same banized area, which is the same area squared in Fig. 5c. The wind square shown in Fig. 5c. difference directs to the inland in area A (Fig. 6b): the surface wind weakens because of the reclamations. To discuss the cause of the extension of cold anomaly to the for the present and past case, where θ is the potential temperature, inland against the surface wind, the time tendency of potential x (west−east) and y (south−north) are the horizontal directions, z is temperature is defined as the vertical direction, and u, v, and w are the wind velocities in the x, y, and z directions, respectively. Q is the diabatic heating rate by æ ö ¶qç ¶ q ¶q¶ q÷ Q radiation, Cp is the isobaric specific heat, � is the exner function, = -çu + v + w ÷+ ¶t èç ¶x ¶y ¶z ø÷ C p and KH and KV are the horizontal and vertical eddy diffusivities of p heat, respectively. ¶ ¶q¶ ¶ q ¶ ¶q + K + K + K , (1) To calculate ∂θ/∂t in Eq. (1), variables are averaged over 8−10 ¶x H ¶x¶ y H ¶y¶ z V ¶z LST, and then averaged over the area A. Table 1 shows the calcu- 32 Ito et al., Effects of Land Reclamations in Osaka Bay on the Regional Climate

Table 1. Horizontal advection, radiation, and vertical diffusion terms [K s−1] Japan. The reclamations, however, affect the surrounding surface in time tendency equation of potential temperature (Eq. 1) averaged over air temperature regardless of the development stage of cities and the area A in Fig. 6 for the present and past case, and their differences (present-past). Variables used for the calculations are averaged for 8− their sizes, and thus, the increase in DTR may be observed in all 10 LST in the 15 winter days. areas around the reclaimed areas. advection radiation diffusion Acknowledgments present −4.4 × 10−5 3.5 × 10−5 2.8 × 10−2 past −4.7 × 10−5 3.5 × 10−5 2.8 × 10−2 We are deeply grateful to Prof. Fujio Masuda of Doshisha −6 −8 −5 University for contributing to this study and supports. difference 3.2 × 10 1.7 × 10 −1.4 × 10 Thanks are due to the Japan Meteorological Agency for using the JMA-NHM and providing various meteorological data used in this study. lated results for each term. The vertical advection and horizontal Reference diffusion terms are neglected because their orders are too small. The orders of the horizontal advection and radiation terms in both Arakawa, H., 1938: Increasing mean daily minimum temperature in large, present and past are 10−5. The difference between the horizontal developing cities. J. Meteor. Soc. Japan, Ser. II, 16, 379−380 (in advection terms is 10−6 and between the radiation terms is 10−8. In Japanese). contrast, the orders of the vertical diffusion terms are 10−2 and the Association for the environmental conservation of the Seto Inland Sea, −5 2001: The environmental conservation of the Seto Inland Sea 2001. difference is 10 . The orders of the vertical diffusion are clearly Retrieved from http://www.emecs.or.jp/01cd-rom/section3/seto/ bigger than the other two terms. This indicates that both the ∂θ/∂t sisaku/umeta1.htm (in Japanese). and the difference of ∂θ/∂t are determined by the vertical diffusion Deardorff, J. W., 1978: Efficient prediction of ground surface temperature term. The vertical diffusion term in the present is smaller than that and moisture, with inclusion of a layer of vegetation. J. Geophys. in the past. In other words, the vertical heat exchange between the Res., 83, 1889−1903. lower and upper layers of air is reduced because the surface wind Hiraoka, A., and H. Noma, 2006: Kinki II Chizu de Yomu Hyakunen − is weaker in the present than in the past due to the reclamations Osaka, Hyogo, Wakayama−. Kokinsyoin, 121 pp (in Japanese). (Fig. 6). Because of the formation of a stable inversion layer by 9 Ito, R., K. Masuda, and F. Masuda, 2008: Urban warming in the Kinki region. J. Environ. Eng., 37, 428−433 (in Japanese). LST, the surface air is not as warmed from the upper layer by the Kawamura, T., 1979: Urban Atmospheric Environment. University of vertical heat exchange compared to the past, resulting in the po- Tokyo Press, 185 pp (in Japanese). tential temperature decreases at the level of 20 m. Kimura, F., and S. Takahashi, 1991: Climatic effects of land reclamation in The weakened vertical diffusion is also verified at other times, Tokyo Bay −numerical experiment. Energy and Buildings, 15, 147− 5 LST and 7 LST. The direction of the surface wind is seaward, 156. and its difference is landward over the cold anomaly, which ap- Kitagawa, H., 2000: Radiation process. NPD Report No.46, Numerical pears over the near-coastal area of the Osaka Plain in the early Prediction Division, JMA, 16−31 (in Japanese). mornings. The time tendency of potential temperature is calculat- Kitao, N., M. Moriyama, T. Tanaka, and H. Takebayashi, 2010: Analy- sis of urban heat island phenomenon in Osaka region using WRF ed in Eq. (1) for the squares of Fig. 5a over 4−6 LST as 5 LST and model: To estimate the effects of the urbanization using the concept Fig. 5b over 6−8 LST as 7 LST. These represent the middle of the of potential natural vegetation. J. Environ. Eng., 75, 465−471 (in cold anomaly over the Osaka Plain and the urbanized area. The Japanese). diffusion terms and the difference are also the biggest at 5 LST Kondo, A., K. Yamaguchi, and H. K. Ahn, 1996: Simulation of climatic and 7 LST, just as the result at 9 LST. The diffusion difference is effects by construction of reclaimed island in Pusan, Korea. Atmos. −5.2 × 10−4 K s−1 at 5 LST and −2.2 × 10−5 K s−1 at 7 LST. Hence, Environ., 30, 2437−2448. it is clear that the reduced vertical heat exchange by the weakened Kondo, H., 2009: The meteorology in a human scale. Asakura Publishing, surface wind is a common cause of the SAT decrease over the 156 pp (in Japanese). inland during the early mornings. Kusaka, H., F. Kimura, H. Hirakuchi, and M. Mizutori, 2000: The effects of land-use alteration on the sea breeze and daytime heat island in the Tokyo metropolitan area. J. Meteor. Soc. Japan, 78, 405−420. 4. Conclusion Mikami, T., 2003: Actual condition of urban heat islands in Japan. Envi- ronmental Information Science, 32, 32−36 (in Japanese). Numerical simulation was conducted to investigate the effects Nakanishi, M., and H. Niino, 2006: An improved Mellor-Yamada level-3 of the reclamations in Osaka Bay. model: Its numerical stability and application to a regional predic- The statistically significant warm difference appears over the tion of advection fog. Bound.-Layer Meteor., 119, 397−407. reclaimed areas in the daytime and the cold difference appears in Ohashi, Y., and H. Kida, 2002: Effects of mountains and urban areas the nighttime, in all seasons. The warm anomaly extends over the on daytime local-circulations in the Osaka and Kyoto regions. J. Meteor. Soc. Japan, 80, 539−560. inland during the daytime. The SAT increases by more than 1.0°C Saito, K., T. Kato, H. Eito, and C. Muroi, 2001: Documentation of the Me- over central Osaka city in the afternoons in spring and summer teorological Research Institute/Numerical prediction division unified due to the reclamations. nonhydrostatic model. Tech. Rep. MRI, 42, 133 pp. The cold anomaly appears over the reclaimed areas at night Saito, K., T. Fujita, T. Fujita, Y. Yamada, J. Ishida, Y. Kumagai, K. Ara­ and extends over Osaka Bay and inland. The cold anomaly ex- nami, S. Ohmori, R. Nagasawa, S. Kumagai, C. Muroi, T. Kato, H. tends over the largest area of the inland in winter and covers Eito, and Y. Yamazaki, 2006: The operational JMA Nonhydrostatic almost all the areas of the Osaka Plain. The anomaly starts extend- Mesoscale Model. Mon. Wea. Rev., 134, 1266−1298. ing over the near-coastal area of the Osaka Plain from midnight, Sato, T., S. Murakami, R. Ooka, S. Yoshida, K. Harayama, and H. Kondo, and moves further along the inland for several hours in the morn- 2004: Numerical study on effects of countermeasures for urban heat island in summer and winter: The characteristic of urban climate and ings. effects of greening and high albedo surface. J. Environ. Eng., 577, Cold anomaly is formed over the inland area during the early 55−62 (in Japanese). mornings. This is because, at first, the surface wind weakens due Yabu, S., S. Murai, and H. Kitagawa, 2005: Clear-sky radiation scheme. to the reclamations; furthermore, the vertical heat exchange reduc- NPD Report No.51, Numerical Prediction Division, JMA, 53−64 (in es at the boundary layer. Therefore, the SAT in the present case Japanese). becomes colder than that in the past. The reclamations in Osaka Bay bring about the extension of Manuscript received 18 December 2011, accepted 13 March 2012 the warm and cold anomalies to the inland in the daytime and SOLA: http://www.jstage.jst.go.jp/browse/sola nighttime, respectively, and lead to an increase in Tmax and a de- crease in Tmin in Osaka. Thus, they increase the DTR in Osaka. In this study, we focused on Osaka, one of the biggest cities in