The seventh International Conference on Urban Climate,
29 June - 3 July 2009, Yokohama, Japan
REGIONAL CHARACTERISTICS ON DIURNAL CHANGE OF TEMPERATURE IN
THE TOKYO METROPOLITAN AREA
Ikumi Akasaka*, Haruo Ando, Hitoshi Yokoyama
*Tokyo Metropolitan Research Institute for Environmental Protection, Tokyo, Japan
Abstract
In this study, regional characteristics on diurnal changes of temperature in the Tokyo metropolitan area in summer
and winter were investigated. To clarify regional differences in the diurnal change patterns of temperature, the
cluster analysis was applied to spatial anomalies of temperature from the domain-averaged diurnal change of
temperature. The results show that temerature hardly decreases during nighttime in the center of the city
throughout the year. That is, daily minimum temperature is relatievely warm in the center of the city throughout
the year. Additionally, the area where relative warm temperature is maintained during nighttime moves
southeastward and expanded around the coastal region in winter though the area appears around the center of
the city in summer. Regional characteristics on diurnal change of temperature have seasonality in increasing and
decreasing temperature related to both urbanization, such as changes in the ground coverage, and the climate
facters, such as land-sea breeze.
Key words: temperature, diurnal change, regional difference, cluster analysis, Tokyo Metropolitan area
1. INTRODUCTION
In the Tokyo Metropolitan area, changes in thermal environment, caused by the growing UHI (the urban heat
island), receive much attention. The growing UHI is mainly related to changes in the ground coverage and
roughness, sky factor and anthropogenic heat emission with rapid urbanization. Their factors influence on diurnal
change patterns of temperature in the urban area while they are characterized by the climate factor (e.g. land-sea
breeze). Therefore, the diurnal change patterns of temperature vary in the different geographic area. It is
important to take concrete and effective measures in each area for decreasing the UHI considering the regional
characteristics on the diurnal change of temperature.
Suzuki et al. (2001) showed the regional characteristics on the diurnal change patterns of temperature in Tokyo
using temperature data for four days in summer of 1992. However, regional characteristics on the mean diurnal
change of temperature in summer have not been studied. In addition, those in winter have not been analyzed.
The purpose of this study is to clarify regional characteristics on mean diurnal change of temperature in the Tokyo
Metropolitan area. This study also aims to reveal their seasonal differences in summer and winter.
2. DATA AND METHODS
In order to reveal regional differences in diurnal change patterns of temperature, it is necessary to analyze using
high density temperature data. From July of 2002 to March of 2005, Tokyo Metropolitan Government constructed
network of weather observation, called METROS (Metropolitan Environmental Temperature and Rainfall
Observation System), in Tokyo ward area with Tokyo Metropolitan University for investigating characteristics on
the urban heat island in detail (Ando et al., 2003). Temperature was observed in the instrument screen in
elementary schools, located at about 2.5km mesh spacing, at 10 minutes interval. Therefore, the high density
temperature data that measured through METROS100 was used in this study. About 100 stations that have more
than 80% data were utilized for the analysis (Fig.1).
Hourly mean temperature data on each station was calculated for studying diurnal changes of temperature
averaged for summer (7/20-9/30) and winter (12/1-2/28) during the period July 2002-March 2005. To reveal the
mean diurnal change of temperature in the Tokyo ward area, hourly mean temperature over the domain was
averaged for 3 years. Spatial anomalies from the diurnal change of the domain-averaged temperature were also
calculated on each station to clarify regional differences in the diurnal changes of temperature.
To classify the diurnal change of temperature anomalies on each station, the cluster analysis was applied to the
spatial anomalies from the domain-averaged temperatures at about 100 stations in summer and winter using the
Euclidean distance and the Ward’s method. Additionally, spatial anomalies from the diurnal change of
temperature were averaged in each cluster.
3. REGIONAL DIFFERENCES IN DIURNAL CHANGE OF TEMPERATURE ANOMALIES
3.1. Summer
In summer, mean diurnal change patterns of temperature in Tokyo ward area have similar characteristics for the
period 2002-2004 (Fig.2a) although daily maximum and minimum temperature values varied from year to year
The seventh International Conference on Urban Climate,
29 June - 3 July 2009, Yokohama, Japan
Fig.1. Distribution of METROS100 observation stations and 23 wards in the Tokyo Metropolitan area. Suquares
indicate 106 observation stations.
a)
b)
Fig.2. Diurnal temperature variations averaged in the Tokyo Metropolitan area in a) summer and b) winter for the
period 2002-2004. Dotted lines indicate the diurnal change in the domain-averaged temperature in each year.
Red and blue lines indicate the diurnal temperature variations averaged for the period 2002-2004 in summer and
in winter, respectively.
(e.g. the cold summer of 2003 and the hot summer of 2004 in Japan). Daily minimum and maximum temperatures
appear during 5-6 a.m. and 1-3 p.m., respectively. Mean daily range of temperature for the 3 years is about 5-6C.
Based on results of the cluster analysis, diurnal cycles in spatial anomalies of temperature in summer were
classified into 6 groups (Fig.3a). A lot of stations classified into each group are distributed in a cluster. The
characteristics on diurnal change of temperature in each cluster are described below.
Cluster 1 is mainly distributed in the center and the secondary center of the city, where are highly-urbanized
region, Taitou Ward, and Sumida Ward. Positive spatial anomalies of temperature (about +0.5C) are maintained
all day and scarcely vary (Fig.3b). Thus, temperature is relatively warm all day in these areas in summer.
Cluster 2 is bordered by the southern edge of the center of the city. Positive anomalies of temperature remarkably
continue during nighttime (Fig.3c). This means that daily minimum temperature is relatively warm in this area.
Negative spatial anomalies of temperature in daytime might be related to cooling by sea breeze. Therefore,
diurnal range of temperature is relatively small compared with the northeastern part and western part of the city.
Cluster 3 is mainly distributed along the coastal region: Edogawa, Koto, and eastern part of Shinagawa Wards.
Negative spatial anomalies of temperature remarkably appear during daytime, especially afternoon (Fig.3d). This
corresponds that daily maximum temperature is relatively cold by the cooling effect of sea breeze. Although day
range of temperature in the stations classified into cluster 2 or cluster 3 is relatively small, characteristics on
diurnal change patterns of temperature differ between 2 groups. The day range of temperature in cluster 3
depends on cooling during daytime. Meanwhile, that in cluster 2 is predominantly characterized on the
maintaining warm temperature during nighttime.
Cluster 4 extends outward the center of the city. Temperature anomalies close to almost zero. That is, it indicates
that diurnal change of temperature follows the mean diurnal change of temperature (Fig.3e).
The seventh International Conference on Urban Climate,
29 June - 3 July 2009, Yokohama, Japan
Cluster 5 is particularly concentrated in Adachi Ward though some stations classified in cluster 5 are dotted with
the western part of the city. Temperature anomalies remarkably increase from morning to noon (Fig.3f). This
means that rate of increasing temperature is larger in this area in the morning than that in the other region.
Cluster 6 is distributed the western part (e.g. Nerima Ward) and the northeastern part (e.g. Katsushika Ward) of
the city. Negative spatial anomalies of temperature appear all day, especially during nighttime (Fig.3g). In cluster
5 and cluster 6, diurnal range of temperature is relatively large. However, diurnal change patterns of temperature
are characterized by variations during daytime in cluster 5 and during nighttime in cluster 6, respectively.
In daytime of summer, diurnal change patterns of tempearature indicate the contrast between the center and the
secondary center of the city and the coastal region though variations in temperature anomalies in summer are
smaller than those in winter. In nighttime, diurnal change patterns of temperature contrast sharply among the
center and the secondary center of the city, western edge and northeastern part of the Tokyo Metropolitan area.
The difference of temperature among these regions is about 1C during nighttime. In the center and the
secondary center of the city, temperature remains warmer during nighttime than that in the surrounding area.
Fig.3. Regional classification of diurnal change in spatial anomalies of temperature in summer. a) indicates
distribution of observation stations classified in 6 clusters. Squares denote observation stations, and each color
corresponds to each cluster number. b) – f) indicate the diurnal variations in spatial anomalies of temperature
averaged in each cluster. Six colors and numbers in b) – f) correspond to each cluster.
3.2. Winter
In winter, daily minimum and maximum temperatures appear during 6-7 a.m. and 2-3 p.m., respectively (Fig.2b).
Mean daily range of temperature is about 6-8C, and daily mean maximum temperature varied more largely than
daily mean minimum temperature for the 3 years. Regional differences in diurnal range of temperature remarkably
appear in winter because temperature anomalies in winter vary more largely than those in summer.
From the results of the cluster analysis, diurnal cycles in spatial anomalies of temperatures in winter were
classified into 6 groups as in summer (Fig.4a). Characteristics on the diurnal change of temperature and
distribution of station in each cluster are described below.
Cluster 1 and cluster 2 have a similar feature in spatial anomalies of temperature in winter (Fig.4b and 4c). During
nighttime, temperature is relatively warm both in cluster 1 and cluster 2, especially in cluster 1. These stations are
distributed from the east-northeastern part to the south-southwestern part of the city.
On the other hand, temperatures in cluster 3 and cluster 6 are relatively cool during nighttime (Fig.4d and 4g).
The feature remarkably appears in cluster 6 (Fig.4g). These stations classified in cluster 3 or 6 are mainly
distributed in the western part and the northeastern of the city. That is, the diurnal change patterns of temperature
between the coastal region (clusters 1 and 2), and the western and the northeastern part of the city (clusters 3
and 6) contrasts sharply. The difference in the diurnal change pattern is shown during nighttime.
The seventh International Conference on Urban Climate,
29 June - 3 July 2009, Yokohama, Japan
The stations classified in cluster 4 or 5 are distributed in the area surrounding the center of the city although there
are few stations classified in cluster 4. In cluster 4, temperature anomalies show positive values all day and
steeply increase before noon (Fig.4e). Meanwhile, temperature anomalies in cluster 5 show negative values all
day, and slightly a decrease around noon (Fig.4f).
4. SUMMARY
This study investigated regional characteristics on diurnal changes of temperature and their seasonality.
Temperature is relatively warm in the center and the secondary center of the city in all day throughout the year. In
addition, during nighttime, temperature is relatively warm in the south-southeast part of the city (e.g. southern part
of Ota Ward) in summer and around the coastal region in winter. This means that daily minimum temperature is
warmer and day range of temperature is relatively small in these areas. Moreover, the area, where relative warm
temperature is maintained during nighttime, moves southeastward and expanded around the coastal region in
winter though the area appears around the center and the secondary center of the city in summer.
During nighttime, temperature is cooler in the western pert and the northeastern part of the city both in summer
and winter than that in other area. In these areas, the day range of temperature is relatively large. Meanwhile, in
daytime of summer, temperature is relatively cool around the coastal region. This means the cool effect by sea
breeze. Thus, daily maximum temperature hardly increases, and day range of temperature is small in the coastal
region in summer. In daytime of winter, temperature slightly decreases in the areas surrounding the coastal region.
These results indicates that regional characteristics on the diurnal change of temperatures have seasonality in
increasing and decreasing temperature related to the urbanization (e.g. changes in the ground coverage) and the
climate factors (e.g. land-sea breeze).
Fig.4. Same as in Fig.3 except for winter.
References
Ando, H., Shioda, T., Morishima, W., Kojima, S., Ishii, K., Izumi, T., Mikami, T., 2003. Spatial structure of summer
temperatures over the urban area of Tokyo in 2002, Annual Report of the Tokyo Metropolitan Research Institute
for Environmental Protection 2003, 81-87.
Suzuki, T., Genchi, Y., Iizuka, Y., Komiyama, H., 2001. Extracting diurnal temperature changing patterns of Tokyo
in mid-summer-grasping heat island phenomena by means of statistical methods-, Tenki, 48, 15-23.