Journal of Weather Changes
ISSN: Coming Soon
Current Issue
Volume No: 1 Issue No: 1
share this page

Review Article | Open Access
  • Available online freely | Peer Reviewed
  • The Microclimate in Hiroshima. A Model to Mitigate the Urban Heat Island Effects

    Mutani Guglielmina 1       Todeschi Valeria 1     Matsuo Kaoru 2    

    1 Department of Energy, Politecnico di Torino, Corso Duca degli Abruzzi 24, 10129 Torino, Italy.

    2 Department of Urban Engineering, University of Tokyo, Japan.

    Abstract

    The overheating phenomena in urban areas is an increasingly important problem for the quality of life and public health. The Urban Heat Island (UHI) effect reduces the dispersion of atmospheric pollution, worsens the urban comfort, increases the energy costs for space cooling and amplify the heatwaves, with sometimes health complications.

    This paper proposes a simple model, which was integrated with a Geographic Information Systems (GIS), in order to analyse a microclimate of outdoor spaces, using the relationship between air temperature and the characteristics of urban environment. In this work, the main characteristics of the urban environment that mainly influence the UHI have been considered: density of population and buildings, urban morphology (aspect ratio H/W and buildings height, distance from the sea and altitude, presence of green areas and vegetation). Remote sensing data and satellite images (Landsat 7 and 8) were also used to evaluate the presence of vegetation and the type of surfaces in the urban space (albedo). Through the construction of linear regression models, the main influent variables were identified for a typical summer day. From this study it is emerged that the UHI effect decreases proportionally with the presence of vegetation and higher values of: the albedo of urban surfaces, the altitude and the distance from the sea; instead the UHI effect increases proportionally with higher values of: the H/W ratio, the building density and the Land Surface Temperature. These models are useful in urban planning to support policy makers to plan an optimal urban form in order to mitigate the overheating and UHI effects in metropolitan areas.

    Received 02 Aug 2017; Accepted 27 Aug 2018; Published 31 Aug 2018;

    Academic Editor:Mohamed Amine Boutabba, Assistant Professor (Maître de Conférences) of Economics, Université ParisSaclay, Univ Evry, France

    Checked for plagiarism: Yes

    Review by: Single-blind

    Copyright©  2018 Mutani Guglielmina. et al.

    License
    Creative Commons License    This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.

    Competing interests

    The authors have declared that no competing interests exist.

    Citation:

    Mutani Guglielmina, Todeschi Valeria, Matsuo Kaoru (2018) The Microclimate in Hiroshima. A Model to Mitigate the Urban Heat Island Effects. Journal of Weather Changes - 1(1):1-26.
    Download as RIS, BibTeX, Text (Include abstract )
    DOIComing Soon

    Introduction

    Global warming and rapid urbanization has significantly increased the urban heat island (UHI) 1, and then the UHI intensity has become a key aspect to characterize the thermal environment of urban areas 2. The UHI phenomenon is defined by the temperature rise in dense city centres compared with the surrounding countryside 3. In recent years, UHI phenomena caused by land cover change and an increase of anthropogenic heat release are occurring in many cities throughout Japan 4 and other countries; consequently, air temperatures rise in urban areas. UHI effect and global warming have caused adverse effects on human health and urban ecosystems in addition to uncomfortable outdoors environments and an increase in the energy consumed for space cooling. Therefore, in order to improve the livability and urban comfort in cities it is necessary to identify mitigation measures including improvements in land cover and ventilation as well as reductions in anthropogenic heat release 5. The Japanese local government has established guidelines concerning the UHI mitigation. In particular, five general actions have been identified: reduction of anthropogenic heat emissions, improvement of urban surfaces and structures, improvement of lifestyles, and promotion of adaptation (“The policy framework to reduce urban heat island effects”, 2004).

    The urban heat island and the microclimate influence each other as the microclimate is influenced not only by the urban characteristics of the different areas within a city but also by the presence of the UHI that also depends on the shape, the orographic characteristics and the size of the city. In order to evaluate the UHI intensity, the different microclimates, as urban and rural ones, are used and a good distribution of weather stations is fundamental in the calculation of UHI indicators. For example, the presence of vegetation and water cause a decrease in air temperatures and hence mitigates the UHI effects with sensible consequences on the thermal comfort of inhabitants 6.

    In a previous research the microclimate of outdoor spaces was investigated in the Metropolitan City of Turin (Italy) considering the different outdoor air temperatures registered by various weather stations. The air temperature variations were correlated with the built urban morphology, the solar exposure of urban spaces, the albedo coefficients of outdoor surfaces, the presence of vegetation and water (Normalized Difference Vegetation Index, NDVI), the distance from the town centre and the Land Surface Temperature (LST). A GIS-based method was used to calculate the various parameters influencing the air temperature variation 7, 8.

    The aim of this work is to improve a methodology used to mitigate the Urban Heat Island (UHI) effects 7, and Hiroshima was selected as the case study. In particular, in the first part of work the method for the evaluation of the air temperature variations and the UHI effect for the City of Hiroshima is presented. Moreover, the data and the variables useful for the construction of the models are indicated: satellite images, weathers stations data and their localization, indicators used in order to implement the UHI models. In the second part, the case study of Hiroshima and the evaluation of microclimate conditions in the urban context (air temperature, wind speed and wind direction) are presented. After the application of the models, the results are shown with the spatial distributions of the air temperature and maps with the support of a GIS tool.

    Material and Methods

    Starting from two previous researches 7, 9, the UHI models were applied to the case study of Hiroshima. In particular, the aim of this work is to obtain a simple model for the simulation of the hourly air temperature in the City of Hiroshima with the use of a linear regression 7. In Figure 1, the methodology used to evaluate the outside air temperature is represented. All variables were identified on the territorial unit of buildings block using a GIS tool. Then the variables were correlated with the outside air temperature to identify the main influencing variables with a linear regression model. More models were reported as a function of a different number of variables.

    Figure 1. Flowchart of methodology.
    Figure 1.

    Finally, the hourly variation of the outside air temperature was evaluated classifying the territory in areas in the mountain, in the plain area near the sea and in the plain area not near the sea. This classification will influence the air temperatures and their daily amplitude.

    The main variables influencing the outdoor air temperature, used to analyzed the UHI phenomenon, are the following urban variables 6, 10, 11, 12: altitude (masl), distance from the sea (Dsea), albedo of outdoor surfaces (ANIR), normalized difference vegetation index (NDVI), land surface temperature (LST), building coverage ratio (BCR), building density (BD), buildings’ height (BH or relative height H/Havg), height to distance ratio or aspect ratio (H/W) and main orientation of streets (MOS). In particular, MOS was evaluated with a variable varying from 0 to 1 with 0 corresponding to the North-South direction and 1 to West-East direction.

    For each weather station, these variables have been calculated at block of buildings scale considering the blocks into a buffer of 300 m from the weather station.

    To calculate the values of albedo ANIR, LST and NDVI, satellite images (Landsat 7 and 8) were used referring to the typical summer day of August 19th, 2013 at 1:49 a.m.. These satellite images were chosen because they were in the period of the measurements (from July 20th to September 23rd in 2013) and with clouds (i.e. 3.8 %). Then the resulting linear regression model refer to August 19th, 2013 at 1:49 a.m.. Knowing also the hourly distribution of air temperature, it was possible to define the typical summer day. The definition of the models have been carried out comparing the calculated data with the measured ones and reducing their differences and relative differences.

    The Air Temperature Model

    As already mentioned, the main influential variables on the outside air temperature were identified considering the correlations between variables and outside air temperature. Then the linear regression models of the air temperature were performed considering all variables or a limited number of variables. As already mentioned, in this work a linear regression model was created using the reference day of August 19th, 2013 (typical summer day) at 1:49 a.m.:

    (1)

    where:

    I-is the intercept;

    α1- αn-are coefficients used to estimate the influence of the various variables X on the outdoor air temperature;

    X-are the independent variables.

    For the evaluation of the different weights of the variables on the air temperature, the normalization of the variables was also introduced. Normalized variables were obtained with this equation:

    (2)

    Then normalized variables were also dimensionless varying from 0 to 1. The accuracy of the models was assessed with the absolute error │ε│ and the relative error εr.

    The Hourly Air Temperature Model

    The hourly trend of a typical summer day has been analyzed with eq. 3 as a function of the minimum, maximum, and daily distribution of air temperature. These values were mainly influenced by the altitude, the distance from the sea and the presence of vegetation and water (characterized by the NDVI index). Then, in this work, the territory was classified in mountain area, plain area near the sea and plain area not near the sea. For each area, an air temperature hourly-distribution factor f(t) was calculated to reduce the relative error εr between the measured and calculated hourly air temperature:

    (3)

    where:





    .

    Case Study

    Hiroshima is the capital of Hiroshima ken (prefecture) and it is located in the southwestern part of Honshu, in Japan. It has a rich topography with the islands and the clear waters of the Seto Inland Sea in the south and the impressive Chugoku Mountains to the north; in particular, Hiroshima is located on the delta of the Ōta River, whose six channels divide it into several islets.

    In 2017, Hiroshima had an estimated population of 1,195,327; the total area of the city was 905.08 km2, with a population density of 1,321 inh/km². Hiroshima has a humid subtropical climate characterized by cool to mild winters and hot humid summers. Like much of the rest of Japan, the warmest month of the year is August. Precipitation are present all year, with rainfall peaks in June and July, while August experiencing sunnier and drier conditions.

    Figure 2 shows the average air temperature of Hiroshima WS, which was located in the urban center of the city with an altitude of 3.7 m a.sl. and BD of 6.7 m3/m2. The measured period is from 2007 to 2017 (average, minimum and maximum Tair) and the dotted line indicates the average data for the year 2013. In Hiroshima, considering the last decade, the average annual air temperature is 16.51 °C with lower temperatures of 5.15 °C in January (the minimum Tair is 1.72 °C) and higher temperatures in August with 28.58 °C (the maximum Tair is 32.91 °C). In this study, a typical summer day of 2013 was chosen (August 19th 2013), because the year 2013 is similar to the average trend of the last 10 years (average annual air temperature is 16.58 °C).

    Figure 2. Average air temperature of Hiroshima weather stations, the continuous line shows average data from 2007 to 2017, the dotted line the average data for 2013.
    Figure 2.

    Data Collection

    In this work, the data used to create UHI models have been organized with the use of the GIS tool, data refer to (Table 1):

    Satellite images (August 19th 013 at 1:49 a.m.) used to evaluate LST, NDVI, albedo ANIR;

    Municipal Technical Map useful to evaluate buildings variables and territorial characteristics at buildings blocks scale;

    Different kind of weather stations data: 60 schools WS (with hourly air temperature data from July 20th to September 23rd, 2013); 7 municipal WS (with hourly wind speed and wind direction data from July 20th to September 30th, 2013); 2 municipal WS (with hourly air temperature data for the years 2007 and 2013); Hiroshima WS (monthly air temperature data from 2007 to 2017).

    Table 1. Description of data collection.
    Type of Data Reference Format Variables
    Satellite images Landsat 8 raster ANIR, NDVI, LST
    Buildings variables Municipal Technical Map shape BCR, BD, BH, H/W, MOS, blocks of buildings units
    Weather data 7 municipal and 60 schools weather stations numeric format T, vwind, vdirection
    Territorial characteristics Digital Elevation Model (DEM) at 10 m raster masl
    Municipal Technical Map shape Land cover (residential, commercial, industrial, tertiary)

    Figure 3 and Figure 4 show an example of GIS available database. Figure 3 classifies the territory based on the presence of water/vegetation and on the built-up areas considering the land cover with: residential, commercial, industrial, tertiary use. Figure 4 shows the NDVI index evaluated with the use of satellite images, highlighting the distribution of vegetation, water and built-up areas; it is possible to observe that, in the urban context, there are some green areas influencing the microclimate and the UHI effect, the parks: Ogonzan, Futabayama Ryokuchi, Ushita Ryokuchi, Hijiyama and Hiuna Ryokuchi.

    Figure 3. Land Cover of Hiroshima and localization of weather stations.
    Figure 3.

    Figure 4. NDVI index evaluated with the support of ArcGIS from satellite images Landsat 8, August 19th 2013 at 1:49 a.m.
    Figure 4.

    Classification of Weather Stations

    Mitigation measures depend on different urban variables (microclimate, altitude, urban density, distance from the center of the city), and for coastal cities it is necessary to consider also the effect of sea proximity and breezes 13. In Hiroshima, sea breeze affects the local climate in coastal urban areas as much as the ground surface condition; therefore, in this work to create UHI models the weather stations were classified considering the altitude and the distance from the sea.

    The model was create using weather data, distinguishing temperature stations and wind stations. The temperature distribution was analyzed with the use of 60 observation points, which were performed by setting 60 temperature sensors with instrument screens outside schools. The observation period was from July 20th 2013, to September 23rd 2013, with an observation interval of 1 hour. The wind direction and wind speed were analyzed using seven municipal weather stations: hourly wind direction 0 from North direction and hourly speed data m/s are known for the same period.

    The temperature stations were classified in three clusters considering the altitude and the distance from the sea (Figure 5):

    The altitude m a.s.l. was used to define weather stations in mountain and plain areas; in particular mountain stations have an altitude higher than 50 m a.s.l., instead plain stations have an altitude lower than 50 m a.s.l..

    The distance from the sea m was used to define the stations that are localized near or not near the sea; in particular the stations “near the sea” have a distance lower than 6,000 m from the sea (with average altitude of 6.03 m a.s.l.) and stations “not near the sea” have a distance higher than 6,000 m from the sea (with average altitude of 22.70 m a.s.l.).

    In order to evaluate the influence of altitude on air temperature, the correlation factor “d” °C/m was calculated with the following equation:

    (4)

    In the mountain area, considering the higher differences in altitude, the temperature-altitude coefficient “d” was estimated with an average value of 166.1 °C/m.

    In Table 2 the air temperature characteristics for the above three areas are reported. It is possible to observe that:

    the average air temperature in the mountains is lower than the ones in the plain areas, because the temperature is influenced by the altitude (with also a higher standard deviation);

    the air temperature amplitude (∆T) is lower near the sea thanks to the mitigating effect of the large mass of water (with an altitude less than 11 m a.s.l.);

    the average air temperatures are similar for the 3 areas.

    Figure 5. Land Surface Temperature (from satellite images Landsat 8, August 19th 2013 at 1:49 a.m.) at buildings block scale and localization of weather stations in Hiroshima considering 7 “wind stations” (light blue) and 60 “temperature stations” (violet: mountain; light violet: plain area not near the sea; pink: plain area).
    Figure 5.

    Table 2. Description of weather data.
    Weather Stations ∆T °C T min °C T max °C T avg °C St.Dev .
    Plain near the sea 8.5 27.1 35.6 31.0 0.4
    Plain not near the sea 11.0 25.7 36.8 31.0 0.4
    Mountain 11.7 24.3 35.9 29.7 0.7

    Table 3 shows for each weather stations: the distance from the sea, the altitude, the temperature data which refer to August 19th, 2013, NDVI, Albedo ANIR and LST which refer to satellite images (Landast 8, August 19, 2013 at 1:49 a.m.) and urban variables. The variables were calculated at buildings block scale for each weather station and an average value has been identified considering a circular buffer of 300 m around the WS. For 5 weather stations the urban variables have not been calculated, either because the information of building blocks (WS: 12, 17) were missing or because the weather stations were located in non-built areas (WS: 8, 16, 36) or in a cloudy zone (WS: 16 and 50) in the satellite images.

    Table 3. Weather stations characteristics: classification in three clusters.
    WS ID Dseam maslm a.s.l. Tavg°C Tmin°C Tmax°C ∆T °C NDVI -1;1 ANIR0;1 LST °C MOS 0;1 BCR [m2/m2] BD [m3/m2] BH m H/Havg- H/W -
    Mountain Weather Stations (n. 17)
    3 5,680 126 30.72 25.90 36.80 10.90 0.49 0.24 24.67 0.35 0.08 0.50 6.16 1.02 0.13
    7 6,233 142 29.98 25.50 35.40 9.90 0.43 0.25 25.03 0.20 0.30 1.82 6.26 1.05 0.23
    8 21,163 99 28.72 23.30 34.80 11.50 0.49 0.25 24.66            
    10 16,326 236 29.32 23.70 35.60 11.90 0.39 0.25 25.25 0.62 0.19 1.07 5.97 1.03 0.15
    11 11,015 80 29.42 24.20 35.00 10.80 0.42 0.23 24.87 0.77 0.11 0.62 6.58 0.92 0.37
    12 18,685 82 28.68 23.20 34.30 11.10 0.50 0.19 23.56            
    14 18,233 60 29.87 24.60 35.70 11.10 0.39 0.21 25.39 0.51 0.17 0.92 5.87 1.07 0.13
    16 9,919 179 29.20 22.80 36.20 13.40                  
    25 10,318 77 29.94 23.80 38.30 14.50 0.51 0.25 23.92 0.48 0.16 0.87 5.28 1.02 0.10
    26 9,748 152 29.46 23.60 35.20 11.60 0.49 0.29 26.03 0.48 0.14 0.74 5.66 1.00 0.14
    36 18,808 170 28.43 22.40 35.50 13.10 0.63 0.25 22.73            
    37 14,026 126 30.14 25.20 35.70 10.50 0.47 0.28 23.59 0.35 0.24 1.72 7.08 1.10 0.20
    44 6,950 134 30.48 25.30 37.30 12.00 0.59 0.27 23.76 0.49 0.19 1.97 13.79 1.82 0.19
    48 13,672 148 30.43 25.40 36.50 11.10 0.33 0.25 26.47 0.56 0.24 1.49 6.41 1.06 0.18
    49 9,265 150 29.58 24.60 35.00 10.40 0.36 0.27 25.21 0.46 0.27 1.53 5.77 1.01 0.20
    50 7,109 104 30.73 25.30 37.10 11.80       0.55 0.26 1.39 5.59 1.06 0.18
    61 19,473 63 30.28 24.00 36.70 12.70 0.38 0.22 25.34 0.51 0.22 1.19 5.76 1.07 0.14
    Plain Weather Stations near the sea (n. 24)
    1 1,131 3 30.27 27.80 33.50 5.70 0.20 0.21 26.58 0.48 0.30 2.31 9.14 1.31 0.21
    2 3,711 2 30.93 27.40 35.20 7.80 0.15 0.19 27.27 0.38 0.27 1.97 8.74 1.28 0.24
    4 274 3 30.34 27.10 33.60 6.50 0.17 0.23 26.24 0.28 0.34 2.60 9.07 1.40 0.23
    5 4,440 31 30.74 27.00 35.40 8.40 0.41 0.24 25.68 0.55 0.26 1.56 6.63 1.18 0.18
    6 3,393 8 30.55 25.60 36.80 11.20 0.23 0.22 26.25 0.60 0.30 1.76 6.46 1.13 0.21
    9 1,917 9 30.67 27.00 35.50 8.50 0.25 0.21 26.38 0.43 0.35 2.15 6.69 1.15 0.27
    21 1,740 17 30.62 25.60 35.90 10.30 0.34 0.23 26.25 0.57 0.27 1.49 5.91 1.10 0.19
    22 1,444 4 31.21 26.50 36.50 10.00 0.17 0.19 27.59 0.37 0.35 2.01 6.38 1.13 0.23
    23 2,001 3 31.38 26.80 36.40 9.60 0.17 0.19 26.91 0.40 0.32 2.25 8.99 1.42 0.23
    28 2,990 4 31.20 26.90 36.20 9.30 0.06 0.17 25.90 0.37 0.21 1.84 11.59 1.81 0.17
    30 611 3 30.60 26.90 34.20 7.30 0.14 0.22 27.54 0.44 0.31 2.35 8.73 1.28 0.28
    34 1,397 5 31.27 25.90 36.50 10.60 0.19 0.21 27.18 0.43 0.33 1.96 6.47 1.14 0.22
    38 4,279 3 31.51 27.90 35.00 7.10 0.10 0.18 26.03 0.42 0.39 6.84 21.38 1.52 0.53
    43 3,988 3 31.74 27.60 36.50 8.90 0.11 0.17 26.76 0.48 0.33 2.80 10.05 1.36 0.31
    45 876 14 31.13 27.30 35.40 8.10 0.31 0.25 26.61 0.35 0.22 1.64 7.67 1.30 0.15
    47 4,853 8 31.04 26.50 36.00 9.50 0.20 0.20 26.01 0.40 0.31 2.32 9.31 1.35 0.26
    51 2,541 4 31.59 27.90 36.10 8.20 0.13 0.20 27.29 0.47 0.33 2.37 9.04 1.26 0.29
    52 2,627 2 30.92 27.90 35.40 7.50 0.16 0.21 26.99 0.48 0.25 1.93 9.76 1.50 0.19
    53 1,342 1 30.92 27.90 34.70 6.80 0.19 0.22 27.52 0.53 0.30 2.08 7.84 1.23 0.25
    54 5,217 3 31.07 27.40 36.20 8.80 0.11 0.16 26.28 0.38 0.34 3.51 13.75 1.56 0.34
    55 3,811 2 31.18 27.10 36.50 9.40 0.11 0.17 25.46 0.54 0.30 3.54 15.28 1.61 0.33
    56 1,835 3 30.98 28.10 34.90 6.80 0.01 0.16 25.09 0.45 0.35 2.45 7.83 1.18 0.32
    57 5,495 4 31.09 27.60 35.40 7.80 0.17 0.21 26.35 0.36 0.30 3.07 12.45 1.38 0.29
    58 2,239 5 30.57 28.00 34.10 6.10 0.07 0.17 26.04 0.43 0.32 2.32 8.11 1.17 0.30
    Plain Weather Stations not near the sea (n. 18)
    13 15,593 16 31.21 25.60 38.30 12.70 0.28 0.23 26.03 0.46 0.15 0.95 7.89 1.27 0.13
    15 12,399 41 30.60 24.90 36.20 11.30 0.31 0.23 24.96 0.43 0.19 1.23 6.66 1.07 0.16
    17 13,868 57 30.21 23.80 38.10 14.30 0.62 0.26 23.41            
    18 8,101 25 30.28 26.30 34.20 7.90 0.39 0.25 25.58 0.51 0.29 1.72 6.27 1.09 0.23
    19 13,205 26 30.54 24.30 37.20 12.90 0.30 0.17 24.76 0.46 0.16 0.89 5.70 1.05 0.13
    24 9,650 8 31.29 27.00 36.00 9.00 0.17 0.21 27.42 0.51 0.30 1.90 7.24 1.18 0.21
    27 8,424 10 31.13 26.20 36.60 10.40 0.17 0.20 26.54 0.40 0.33 2.25 7.55 1.21 0.24
    29 6,437 34 30.25 24.50 36.90 12.40 0.38 0.26 25.96 0.43 0.22 1.20 5.95 1.09 0.16
    32 12,405 19 31.16 25.80 36.70 10.90 0.28 0.23 26.62 0.53 0.29 2.17 8.08 1.23 0.21
    33 9,567 7 31.55 26.20 37.70 11.50 0.28 0.24 26.59 0.58 0.17 1.03 6.27 1.11 0.15
    35 13,216 43 30.85 25.10 37.30 12.20 0.38 0.22 25.71 0.45 0.24 1.41 6.31 1.09 0.19
    39 13,159 19 31.28 25.80 37.10 11.30 0.27 0.22 26.60 0.46 0.21 1.31 6.43 1.12 0.17
    40 11,853 44 31.18 25.40 36.40 11.00 0.26 0.21 26.76 0.45 0.30 1.79 6.05 1.03 0.21
    41 8,381 9 31.40 26.70 36.80 10.10 0.21 0.19 26.02 0.44 0.31 2.27 8.57 1.36 0.23
    42 11,493 9 31.30 26.20 37.00 10.80 0.22 0.22 27.37 0.48 0.22 1.35 6.68 1.14 0.18
    46 12,680 25 31.12 25.90 36.70 10.80 0.32 0.21 26.06 0.32 0.20 1.30 7.26 1.21 0.16
    59 10,136 8 31.33 26.60 36.50 9.90 0.23 0.21 27.13 0.52 0.24 1.77 8.15 1.26 0.22
    60 8,113 9 30.74 26.60 35.80 9.20 0.24 0.21 26.38 0.53 0.30 2.02 7.45 1.21 0.24

    The urban climate of Hiroshima was analyzed with the use of 60 weather stations of elementary schools for the year 2013 from July 20th, to September 23rd. Also, the wind characteristics were investigated with the data of 7 municipal weather stations available for the same period. The weather stations were classified in mountain, plain near the sea and plain not near the sea WS. The evaluation of the weather stations data showed that for the 60 analyzed WS, the minimum air temperature is almost always at 6:00 a.m. (97% of WS), whereas the maximum temperature is at 3:00 p.m. (57% of WS; 92% of WS considering also 2:00 p.m.).

    Figure 6a shows the average hourly temperature value for each observations point considering the month of August, the red dotted line refers to August 19th 2013. In this work, August 19th in 2013 was chosen like the typical summer day because the daily trend is regular and with temperatures higher then the 90% of the data in August. This typical day corresponds to a hot summer day (excluding the hottest days).

    Figure 6. (a) Average hourly temperature value for each observations point considering the month of August, the red dotted line refers to August 19th 2013; (b) Identification of the daytime (from 9:00 a.m. to 7:00 p.m.) and the nighttime (from 8:00 p.m. to 8:00 a.m.) for the typical summer day (August 19th 2013) distinguishing mountain, plain near the sea and plain not near the sea areas WS.
    Figure 6.

    Wind data were analyzed considering daytime and nighttime in August 19th 2013; the daytime is from 9:00 a.m. to 7:00 p.m. and the nighttime is from 8:00 p.m. to 8:00 a.m. (Figure 6b). Then, the hourly wind direction and hourly wind speed were explored considering 7 municipal wind stations, the average hourly values from July 20th, 2013 to September 30th, 2013 (Figure 7a) were compared with the hourly values of August 19th, 2013 (Figure 7b). Figure 7 show direction and wind speed on 19th August 2013 of municipal weather station number 7 (localized in mountain area), distinguishing the daytime (Figure 8a) and the nighttime (Figure 8b); during the daytime the main wind direction is South, like the typical descending mountain breezes, with an average value of speed equal to 3.7 m/s (higher than the nighttime with 1.3 m/s); instead in nighttime the main wind direction is North opposite to the daytime, due to the presence of the sea and the orientation of mountains: the ascending valley breezes.

    Figure 7. Comparison between wind direction from July 20th 2013 to September 30th 2013 (a) and a typical summer day August 19th 2013 (b), considering daytime (from 9:00 a.m. to 7:00 p.m.) and nighttime (from 8 p.m. to 8 a.m.) for 7 municipal weather stations.
    Figure 7.

    Figure 8. (a) Wind rose diagram: daytime of August 19th 2013 (municipal WS number 7); (b) Wind rose diagram: nighttime of August 19th 2013 (municipal WS number 7).
    Figure 8.

    Results and Discussion

    The results shown below are divided into 3 sections. In the first part is on the linear regression model used to create the air temperature model for the typical summer day (19th August 2013). In the second part the results of hourly air temperature distribution model is presented, distinguishing the weather stations located in the mountains area, plain area not near the sea and plain area near the sea. In the last part, assessments of the urban heat island intensity were made using the UHI-driven indicators (Q1 and Q2) and land-cover-driven indicators (Q3); also, heatwaves and cold waves were analyzed for the year 2007 and 2013.

    The Air Temperature Model

    In order to identify the main influential variables on the air temperature, the correlations between variables and air temperature have been evaluated (Figure 9). For example, the altitude has a negative correlation because the air temperature decreases if the altitude increases; also the presence of vegetation and water reduce the air temperature and then NDVI has a negative correlation. While positive correlations can be observed with the BD, BCR, H/Havg and H/W.

    Figure 9. Correlations between variables and air temperature.
    Figure 9.

    Only not-dependent variables on each other were used, with the higher correlation factor; BD and BCR are dependent, then only BCR was used for the model; also BH and H/Havg are dependent, then only H/Havg was used for the model.

    Below the linear regression models of the air temperature are presented, distinguishing models with non-normalized variables (in Figure 10a):

    Eq. 5: Linear regression model with all non-normalized variables;

    Eq. 6: Linear regression model with non-normalized variables and without LST;

    and with normalized variables (in Figure 10b):

    Eq. 7: Linear regression model with all normalized variables;

    Eq. 8: Linear regression model with normalized variables and without LST;

    Eq. 9: Linear regression model with normalized variables and without LST and NDVI.

    Figure 10. (a) Air temperature model with all non-normalized variables; (b) Air temperature model with all normalized variables.
    Figure 10.

    The best results, with the higher coefficient of determination R2, were provided by eq. 5 (all non-normalized variables) and eqq. 7 and 8 (all normalized variables and all normalized variables without LST). In particular, Table 5 reports the R2, which shows how much of the variation in air temperature can be explained by the regression model as function of the selected variables (eqq. 6 and 8 without LST and eq. 9 without LST, NDVI).

    Table 5. The coefficients, the relative error │εr│ and the coefficient of determination R2 for the air temperature model.
    Eq. I αD,sea α m,asl αNDVI αA,NIR αLST αMOS αBCR αH/Havg αH/W │εravg R2
    (5) 21 0.0000212 -0.0101 -4.685 11.332 0.111 0.040 5.785 1.616 -0.188 1.4% 0.78
    (6) 24 0.0000124 -0.0095 -6.425 13.556 - 0.038 5.303 1.399 -0.353 1.5% 0.77
    (7) 27 -0.015 -2.125 -2.711 1.541 0.394 0.035 1.534 1.709 -0.343 1.4% 0.79
    (8) 27 -0.015 -2.098 -3.241 1.819 - 0.35 1.633 1.617 -0.356 1.5% 0.78
    (9) 27 -0.015 -2.419 - -0.147 - 0.036 2.639 1.639 -0.761 1.7% 0.73
    (10) 28 -0.015 -1.517 -1.905 - - 0.035 1.927 1.553 -0.705 1.6% 0.75

    The relative error εr, that is the ratio of the absolute error, between measured and calculated values of air temperatures, and the measured value was used to describes the accuracy of the models; it has low values in all linear regression models and tends to increase slightly excluding some variables like LST and NDVI. Moreover, NDVI and ANIR are dependent variables, with a correlation coefficient of 0.76, therefore in the models the weight of the variables should be negative, but the ANIR tends to be positive for the presence of the NDVI (compensatory effect); in addition, for this case study, ANIR is quite constant with an average value equal to 0.22 with a low standard deviation of 0.04.

    Then, the models were applied to the territory of Hiroshima with the use of GIS tool; Figure 11 shows the air temperature simulated for August 19th 2013 at 1:49 a.m. using model with all the normalized variables (Eq.7). In urban areas, the air temperature is higher than in the peripheral plain and mountain areas, where the temperature is mitigated by the altitude, the presence of vegetation and lower buildings density.

    Figure 11. Air temperature model with all normalized variables (Eq.7).
    Figure 11.

    In Figure 12, the range of variability of the different variables multiplied by the relative coefficients α are indicated. The main influent variables (in green non-normalized variables, in blue normalized variables) are: the altitude, the presence of vegetation, the characteristics of the outdoor surfaces (ANIR), the buildings density (BCR) and the relative buildings height (H/Havg). The LST is present in two equations (5 and 7), the ANIR coefficient becomes negative when the NDVI is not inserted in the model (Eq. 9). The distance from the sea and the altitude are uncontrollable variables, therefore to improve the microclimatic conditions and to mitigate the air temperature it is necessary to intervene on the others variables. For example, in newly built areas (so there is an increase in BCR and H/Havg, and consequently an increase in air temperature) the share of green areas can be improved (NDVI is inversely proportional to the air temperature) to compensate the UHI effect.

    Figure 12. The range of variability of the different variables multiplied by the coefficients α (on the air temperature) with the equations 5, 6, 7, 8, 9 and 10.
    Figure 12.

    The eqq. 9 and 10 confirm the correlation between ANIR and NDVI, in the eq. 9 NDVI is inversely proportional to the Tair, the same relation is with the ANIR in the eq. 10 without NDVI.

    The Hourly Air Temperature Model

    The outside air temperature was simulated using the equations reported in Table 6, in which are indicated the intercept “I”, the weight coefficient of the variables “α” (distance from the sea, altitude and presence of vegetation and water):

    Eq. 10 refers the whole territory;

    Eq. 11 refers to the mountain areas;

    Eq. 12 refers to the plain areas near the sea;

    Eq. 13 refers to the plain areas not near the sea and.

    Figure 13a and Figure 13b show the equations used to evaluate the daily minimum air temperature, Tmin, and the daily amplitude of air temperature, Delta (t), in a hot summer day. Then, with the eq. 3, the hourly air temperature have been simulated for the typical hot summer day (August 19th 2013). In these models, the relative error εr is higher and the coefficient of determination R2 is lower than in the others models, because the Tair does not depend only from these variables; consequently there is a greater dispersion of data from the average value (this trend is more evident in the case of the Delta (t) with an average εr of 10.8%).

    Figure 13. Equations used to evaluate (a) the minimum air temperature (Tmin) and (b) the amplitude of air temperature (Delta(T)).
    Figure 13.

    Table 6. The coefficients, the relative error │εr│ and the coefficient of determination R2 for the hourly air temperature model (global, mountain, plain near the sea and plain not near the sea areas).
    Eq. Tmin ∆T │εravg Area
    I αD,sea αm,asl αNDVI I αD,sea αm,asl αNDVI
    (10) 28.36 -0.00009 -0.009 -4.45 6.41 0.00022 -0.002 6.05 2.2 % Global
    (11) 27.20 -0.00010 -0.003 -2.97 9.69 0.00005 0.005 0.50 2.8 % Mountain
    (12) 27.89 0.00006 0.036 -5.44 6.10 0.00040 -0.004 5.55 2.0 % Plain near the sea
    (13) 28.49 -0.00009 -0.022 -3.92 5.82 0.00025 0.004 7.64 1.5 % Plain not near the sea

    The Figure 14a, Figure 14b and Figure 14c) show the hourly trends of three weather stations for each area: mountain, plain near the sea and not near the sea areas. The measured data (continuous line) were compared with the calculated one (dashed line). It is possible to observe that in mountain area the variation on air temperature at 6 a.m. is higher than in the plain area, due to the differences of altitude.

    Figure 14. (a) Hourly air temperature distribution for August 19th 2013 in the mountain area; (b) Hourly air temperature distribution for August 19th 2013 in the plain area near the sea; (c) Hourly air temperature distribution for August 19th 2013 in the plain area not near the sea.
    Figure 14.

    Figure 15 and Figure 16 show the results of the hourly model (Eq. 10) with the spatial distribution of the air temperature at 6:00 a.m. with an average air temperature of 26.2 °C and at 3:00 p.m. with an average air temperature that reaches 29.3 °C.

    Figure 15. Application of hourly air temperature model at 6 a.m. at buildings block scale (Eq. 10).
    Figure 15.

    Figure 16. Application of hourly air temperature model at 3 p.m. at buildings block scale (Eq. 10).
    Figure 16.

    The UHI-Driven and Land-Cover-Driven Indicators

    The UHI intensity (UHII) is an indicator that can measure the daily amplitude of air temperature and the temperature gradient between the urban and the surrounding rural areas. There are two types of indicators useful to evaluate the different microclimate conditions: UHI-driven (Q1 and Q2) and land-cover-driven (Q3) 14, 15, 16:

    Q1. The UHI-driven indicator “magnitude” equal to the maximum temperature minus the average daily air temperature, distinguishing daytime and nighttime;

    Q2. The UHI-driven indicator “range” equal to the difference between the maximum and the minimum daily air temperature, distinguishing daytime and nighttime;

    Q3. The land cover-driven indicator “urban-rural” describes the difference between the hourly air temperature in the urban and surrounding areas.

    These three indicators were calculated with the hourly data for the years 2007 and 2013 for two municipal weather stations: WS number 7 located (altitude of 26.7 m a.s.l.) in surrounding rural area and WS number 3 (altitude of 5.8 m a.s.l.) located in urban area (see Figure 5).

    Figure 17 show the trends of the indicators for the years 2007 and 2013 considering the annual average quantitative values of UHI intensity at different times. It is possible to see that the values of Q1 and Q2 remains almost stable and from midnight to 6 a.m., the values decrease, then there is a slight increase and after 4 p.m. the UHI effect tends to decrease (this trend is similar for both years); Q3 decreases during the day and tends to increase after 6 p.m.. Figure 18, Figure 19 and Figure 20 show the moving average times series of the previous 30 days of the values of UHII at times 0:00 a.m., 6:00 a.m., 10:00 a.m. and 2:00 p.m.. There are different trends: Q1 and Q2 have, in winter, daytime values are lower than nighttime ones; while, in summer the daytime values are higher than the nighttime values. Also, in winter, the land-cover-driven indicator Q3 has a higher values due to higher thermal excursion during nighttime (this effect is due to lower air temperatures); in summer, Q3 is higher in the daytime due to the higher solar irradiation. The urban station has values always smaller than the suburban station because the Tair is influenced by the urban density, with higher air temperature values. In fact, WS number 3 has a BCR of 0.33 and BD of 3.05, higher than WS number 7 (BCR equal to 0.23 and BD equal to 1.64).

    Figure 17. The UHI-driven and land-cover-driven indicators for the year 2007 (a) and 2013 (b): annual average values at different times.
    Figure 17.

    Figure 18. The UHI-driven indicator Q1 for the year 2007 (a) and 2013 (b): 30 per moving average times series of the UHI intensity at the time 00:00 a.m. (in blue), 06:00 a.m. (in green), 10:00 a.m. (in yellow) and 02:00 p.m. (in orange).
    Figure 18.

    Figure 19. The UHI-driven indicator Q2 for the year 2007 (a) and 2013 (b): 30 per moving average times series of the UHI intensity at the time 00:00 a.m. (in blue), 06:00 a.m. (in green), 10:00 a.m. (in yellow) and 02:00 p.m. (in orange).
    Figure 19.

    Figure 20. The UHI-driven indicator Q3 for the year 2007 (a) and 2013 (b): 30 per moving average times series of the UHI intensity at the time 00:00 a.m. (in blue), 06:00 a.m. (in green), 10:00 a.m. (in yellow) and 02:00 p.m. (in orange).
    Figure 20.

    Heatwaves and cold-waves for the years 2007 and 2013 have been evaluated using the same weather stations data (municipal WSs number 3 and 7). The heatwaves were defined as event with temperature over the 97.5th percentile and cold-waves as events with temperature under the 2.5th percentile 17. For the years 2007 and 2013, in Hiroshima, there are 10 hot days a year with an average air temperature of 30.43 °C in 2007 and 31.39 °C in 2013, heatwaves have air temperature higher than 29.80 °C for the year 2007 and 31.01 °C for the year 2013. Also, there are 9 cold days a year with an average air temperature of 3.36 °C in 2007 and 1.37 °C in 2013, cold-waves have air temperature lower than 4.24 °C for the year 2007 and 2.48 °C for the year 2013.

    Figure 21 show the average hourly variability in UHI during summer months (June-August) for Hiroshima. Data from the summer months for the years 2007 and 2013 are used to analyze the UHI variability trend. In the Figure 21, the shaded area represents the average hourly standard deviation and the dashed line with the sun and the moon symbols represent the approximate local sunrise (05:00 a.m.) and sunset (07:00 p.m.) times. The UHII increases between hours 3:00 p.m.:01:00 a.m. and decreases during the period 03:00 a.m.:3:00 p.m.; the highest UHI value is of 1.5 °C and it is observed at 01:00 a.m. (average value for the years 2007 and 2013). This trend due to the influence of coastal winds and it is similar to the City of Seattle located near the coast 2. In fact, in addition to the surface characteristics, wind effects play a crucial role in influencing the UHII; as wind speed increases, the volume of relatively cooler air brought from the surrounding rural areas, reduces the urban air temperature. These circulations play a crucial role in reducing the horizontal temperature gradient between urban and rural areas 2.

    Figure 21. Mean daily variability of UHI intensity for summer months (June-August) for the years 2007 (a) and 2013 (b). The two black dotted lines indicate sunrise and sunset times. The shaded region indicates the standard deviation.
    Figure 21.

    Conclusion

    In Hiroshima, the issue of urban heat island phenomenon reportedly engenders high temperatures. Therefore, urban planning incorporating mitigation of the UHI phenomenon is needed.

    This work showed that the UHI effects are influenced by the built-up areas, the presence of water and vegetation, the speed and direction wind, the distance from the sea and the altitude. The air temperature model and the hourly air temperature model have the objective to investigate which are the main variables that influence the UHI effect and understand how, through the compensative method, it is possible to improve urban comfort and microclimatic conditions. For example, an increase of 20% in NDVI determines an average decrease of air temperature of about 0.2 °C. The reliability of the models depends on the amount of available data and their quality. In these UHI models the errors are also influenced by the variability of the indicators; for example, in Hiroshima, the values of albedo are very low with a standard deviation of 0.04; moreover, the NDVI index has only positive values (Table 4). In addition, the choice of the typical summer day was conditioned by the availability of satellite images with a cloud coverage above 4%.

    Table 4. Main variables in the weather stations’ zones.
    WS id Dseam maslm Tavg°C Tmin°C Tmax°C ∆T°C NDVI-1;1 ANIR0;1 LST°C MOS0;1 BCR[m2/m2] BD[m3/m2] BHm H/Havg- H/W-
    min 274 1.00 28.43 22.40 33.50 5.7 0.01 0.16 22.7 0.20 0.08 0.50 5.3 0.92 0.10
    max 21,163 236.0 31.74 28.10 38.30 14.5 0.63 0.29 27.6 0.77 0.39 6.84 21.4 1.82 0.53
    median 8,101 16.0 30.74 25.90 36.20 10.4 0.27 0.22 26.0 0.46 0.27 1.81 7.2 1.18 0.21
    avg 8,126 45.4 30.62 25.89 36.01 10.1 0.29 0.22 25.9 0.46 0.26 1.88 8.0 1.22 0.22
    st dv. 5,598 57.9 0.76 1.48 1.09 2.1 0.15 0.03 1.1 0.09 0.07 0.95 2.9 0.19 0.07

    The UHI indicators were used to describe the UHI intensity and the results show that: the trend of UHI-driven indicators (Q1 and Q2) is always positive and constant; the land-cover-driven indicator (Q3) shows higher air temperatures in urban area then in suburban area, it is due to the urban activities and buildings density. From 2007 to 2013 the average air temperature of heatwaves increases from 30.43 °C to 31.39 °C and the average air temperature of cold-waves decreases from 3.36 °C to 1.37 °C.

    With the application of these models, the planning of urban areas can be improved for future urban developments and, for consolidated urban environments, it is possible to identify the main critical areas to pilot the mitigation measures.

    Authors' Contributions

    Guglielmina Mutani and Valeria Todeschi contributed equally to the definition and analysis on the models and on the writing of this manuscript. Kaoru Matsuo has provided the data of the Hiroshima case study. All authors have read and approved the final version of the manuscript.

    Nomenclature

    ANIR - Albedo of outdoor surfaces: Near InfraRed

    MOS -Main orientation of streets

    BD - Building density, m3/m2

    NDVI - Normalized difference vegetation index

    BCR - Building coverage ratio, m2/m2

    H/Havg - Relative height

    BH - Buildings’ height, m

    H/W - Height to distance ratio or aspect ratio

    Dsea - Distance from the sea, m

    t - Time, hour

    LST - Land surface temperature, °C

    T - Air temperature, °C

    Masl - Altitude, m

    WS - Weather station

    References

    1.Li G, Zhang X, P A Mirzaei, Zhang J, Zhao Z. (2018) Urban heat island effect of a typical valley city in China: Responds to the global warming and rapid urbanization, Sustainable Cities and Society. 38, 736-745.
    2.Ramamurthy P, Sangobanwo M. (2016) Inter-annual variability in urban heat island intensity over 10 major cities in the United States, Sustainable Cities and Society. 26, 65-75.
    3.C M Nakata-Osaki, Souza L C L, D S Rodrigues. (2018) THIS – Tool for Heat Island Simulation: A GIS extension model to calculate MARK urban heat island intensity based on urban geometry. , Computers, Environment and Urban Systems 67, 157-168.
    4.Ichinose T, Suzuki K, Suzuki K, Seino S. (2009) Research on effect of urban thermal mitigation by heat circulation through Tokyo Bay. The 7thInternational Conference on Urban Climate .
    5.Yamamoto Y. (2006) Measures to mitigate urban heat island. Science & Technology Trends Quarterly Review - NISTEP Science & Technology Foresight Center .
    6.Mutani G, Cristino V, Bullita M. (2018) The urban heat island of the Metropolitan City of Turin. Strategies for a sustainable urban planning. Proceedings of the XXIII A.I.P.T. Conference in press .
    7.Mutani G, Fiermonte F. (2016) The Urban Microclimate and the Urban Heat Island. A model for a sustainable urban planning, Chapter of: Topics and Methods for Urban and Landscape Design. From the river to the project, Editors R Ingaramo and A Voghera, Urban and Landscape Perspectives. doi:, 10-1007.
    8.Matsuo K, Tanaka T, Urban. (2014) Environmental Climate Maps for Urban Planning Considering Urban Heat Island Mitigation in Hiroshima. , Journal of Heat Island Institute International vol, 9-2.
    9. (2015) Mirzaei PA.Recent challenges in modeling of urban heat island. , Sustainable Cities and Society 19, 200-206.
    10.M A Bakarman, J D Chang. (2015) The Influence of Height/width Ratio on Urban Heat Island. in Hot-arid Climates, Procedia Engineering 118, 101-108.
    11.Yang J, Wang Z, Kaloush K E. (2015) Environmental impacts of reflective materials: Is high albedo a ‘silver bullet’ for mitigating urban heat island?, Renewable and Sustainable Energy Reviews. 47, 830-843.
    12.Mutani G. (2016) Urban planning for the liveability and thermal comfort of outdoor spaces, 11thConference on Sustainable of Energy, Water and Environmental Systems. , Lisbon
    13.Sasaki K, Mochida A, Yoshida H, Watanabe H, Yoshida T. (2008) A new method to select appropriate countermeasures against heat island effects according to the regional characteristics of heat balance mechanism. , J. Wind Eng. Ind. Aerodyn 96, 1629-1639.
    14.Sheng L, Tang X, You H, Gu Q, Hu H. (2017) Comparison of the urban heat island intensity quantified by using air temperature and Landsat land surface temperature in Hangzhou, China. , Ecological Indicators 72, 738-746.
    15.Li H, Zhou Y, Li X, Meng L, Wang X et al. (2018) A new method to quantify surface urban heat island intensity. , Science of the Total Environment 624, 262-272.
    16.Zhao S, Zhou D, Liu S. (2016) Data concurrency is required for estimating urban heat island intensity. , Environmental Pollution 208, 118-124.
    17.Pyrgou A, V L Castaldo, A L Pisello, Cotana F, Santamouris M. (2017) Differentiating responses of weather files and local climate change to explain variations in building thermal-energy performance simulations. , Solar Energy 153, 224-237.