Ecotoxicology and Environmental Safety 237 (2022) 113521

Contents lists available at ScienceDirect

Ecotoxicology and Environmental Safety

journal homepage: www.elsevier.com/locate/ecoenv

Spatial and temporal analysis of the increasing effects of large-scale

infrastructure construction on the surface urban heat island
Jikang Wan a, b, Bin Yong a, *, Xiaofeng Zhou b
a
State Key Laboratory of Hydrology-Water Resources and Hydraulic Engineering, Hohai University, Nanjing 211100, China
b
College of Computer and Information, Hohai University, Nanjing 211100, China

A R T I C L E I N F O A B S T R A C T

Edited by Professor Bing Yan With the rapid development of China’s economy and the continuous improvement of people’s living standards,
large-scale infrastructure constructions (LSICs) are also increasing rapidly. LSICs with impervious surfaces have
Keywords: increasingly resulted in replacing natural landscapes, altering surface radiation, thermal properties, and hu­
Large-scale infrastructure construction midity in urban areas. To study the environmental thermal changes of Beijing Daxing International Airport
Land surface temperature
before and after its construction and operation, four Landsat-8 images (from the year of 2014, 2017, 2019, and
Remote sensing
2021) were used to calculate the land surface temperature (LST). Then the LST values of four images covering the
Urban heat island ratio index
Land classification types study area were compared and analyzed using the urban heat island ratio index (URI). Results show that the URI
Environmental safety value of this area increased from 0.120 of 2014 to 0.185 of 2017 after the construction of Daxing Airport,
indicating that the urban surface heat island effect in this area greatly increased. Additionally, the URI value of
this area increased from 0.153 of 2019 to 0.206 of 2021 after the operation of Daxing Airport, indicating that the
surface urban heat island effect in this area further increased. Therefore, we infer that this effect is closely related
to airport construction and operation. Afterward, the random forest classification algorithm is used to classify
land types based on pixels, and then the relationship between URI and land classification types was discussed. It
is found that before the construction of Daxing Airport, both dark buildings and bare land contribute significantly
to the thermal environment of the airport. After the completion of Daxing Airport, the contribution model was
changed to high-reflectivity buildings and bright soil. The thermal pollution generated by the airport has a
greater impact on the ground objects within the range of 7.5 km, and a relatively smaller impact on the ground
objects outside the range of 9.5 km. Our results can provide a valuable reference for the study of the thermal
environment caused by human activities.

1. Introduction 2021). From 2016 to now, China has added 74 large-scale airports, and
about 400 airports will be planned by 2035. However, LSICs change the
The rapid urbanization of human society results in the increasing surface thermal radiation and aggravate the severity of the urban ther­
replacement of natural landscapes by impervious surfaces, which can mal environment (Schwarz et al., 2011; Shen et al., 2016; Wan and Zhu,
alter surface radiation, thermal properties, and humidity over urban 2021; Ye et al., 2011; Zhu et al., 2019). Therefore, it is highly necessary
areas (Cai et al., 2021; Du et al., 2021; Jiang et al., 2021; Li et al., 2021a, to accurately and rapidly analyze the impact of LSICs on the urban
2021b). Anthropogenic waste heat generated through energy con­ thermal environment (Meng et al., 2018; Wan and Zhu, 2021; Zhou
sumption by industry, traffic, and buildings has long been recognized as et al., 2018).
a contributor to the urban heat island (UHI) (Boehme et al., 2015; Jiang Thermal pollution can be evaluated by air temperature (AT) and
et al., 2021; Li et al., 2021a, 2021b; Weng, 2012; Zhang et al., 2021). satellite land surface temperature (LST) measurements (Imhoff et al.,
Recently, with the rapid development of China’s economy, large-scale 2010; Seto and Christensen, 2013). The two indicators are corresponded
infrastructure constructions (LSICs) have been increasing. For to each other because the AT on the Earth’s surface is controlled by the
example, by the end of 2020, China’s high-speed railway reached 37, thermal properties of the underlying surface. AT is typically measured at
900 km, ranking first in the world (Li et al., 2021a, 2021b; Li and Rong, approximately 5 feet (1.5 m) above the ground by weather stations with

* Corresponding author.
E-mail address: yongbin@hhu.edu.cn (B. Yong).

https://doi.org/10.1016/j.ecoenv.2022.113521
Received 20 February 2022; Received in revised form 3 April 2022; Accepted 11 April 2022
Available online 18 April 2022
0147-6513/© 2022 The Author(s). Published by Elsevier Inc. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-
nc-nd/4.0/).
J. Wan et al. Ecotoxicology and Environmental Safety 237 (2022) 113521

high temporal resolution (Feyisa et al., 2014a, 2014b; Li et al., 2013; Lin (1) URI changes caused by the construction and operation of Daxing
et al., 2018; Parker, 2010; Velazquez-Lozada et al., 2006). However, Airport.
spatially continuity analysis is difficult due to the sparse distribution of (2) The relationship between URI change and land classification
observation stations (Meng et al., 2018; Yao et al., 2021). Satellite data types.
has the spatial-temporal characteristics of continuous wide coverage (3) The impact of thermal pollution from Daxing Airport on the
(Seto and Christensen, 2013). Therefore, LST data derived from thermal surrounding environment.
infrared remote sensing have become the commonly used indicator for
heat island analysis (Shen et al., 2016). 2. Study area and data compilation
The urban thermal environment is not only affected by natural fac­
tors (such as the climate type of the region where the city is located, the Daxing Airport is located between the Daxing district of Beijing and
terrain, altitude, whether the city is near the sea, etc.) (Giridharan and the Langfang City of Hebei Province, 46 km away from Tian’anmen Gate
Emmanuel, 2018; Manoli et al., 2019), but also by urban factors (such as of Beijing and 67 km away from Beijing Capital Airport (Fig. 1). The
urban landscape layout, urban land cover, urban development level, climate type is the warm temperate sub-humid continental monsoon
population data, etc.) (Boehme et al. 2015; Cai et al., 2019; Cai et al., climate. The construction of Daxing Airport began on December 26,
2018; Feng et al., 2019; Li et al., 2016; Sun et al., 2019). Among them, 2014, and officially opened to the public on October 27, 2019. By
the influence of natural factors on the urban thermal environment is the February 2021, the terminal area of Daxing Airport is 780,000 m2. We
most intuitive and the greatest, but the natural factors can not be began to pay attention to the thermal environment changes in this study
regulated. Social factors are adjustable and emulated, so many scholars area from the initial planning of the airport to the construction and then
have explored urban thermal environment in terms of social factors and to the operation after the completion of the construction, which took 8
improved urban thermal environment based on urban landscape design years. With the help of remote sensing images, we recorded the surface
(Manoli et al., 2019; Sun et al., 2019), such as providing more green changes and thermal environment changes in each period of the study
space or water (Chibuike et al., 2018), optimizing urban spatial form area, and deeply explored the impact of large-scale infrastructure on the
(Aboelata, 2020; Natanian et al., 2019; Yu et al., 2020), and designing surrounding thermal environment from planning to construction to
cooling road surface (Mohajerani et al., 2017). Install low albedo operation. Quantitative evaluation is made on the thermal environment
building materials (Aflaki et al., 2017). Scholars have also proposed caused by the construction of large-scale transportation hubs in China.
some useful methods to connect urban physical characteristics and Data processing in this study was conducted on Google Earth Engine.
urban thermal environment, such as the Local Climatic Zones (LCZ) The four remote sensing Landsat-8 images (Fig. 2) were acquired from
(Oke and Stewart, 2012). In recent years, scholars have expanded the the United States Geological Survey (USGS) published Level 2 product
scope of LCZ and applied it to other issues, such as urban health (Brousse dataset (Table 1). The Landsat-8 satellite was successfully launched by
et al., 2019) and carbon emissions (Manoli et al., 2019; Wu et al., 2018). the National Aeronautics and Space Administration (NASA) on February
The temperature dynamics and spatial variability in temperature 11, 2013. It carries two sensors, namely the Operational Land Imager
values in the atmospheric urban thermal environment may greatly differ (OLI) and the Thermal Infrared Sensor (TIRS). The satellite has a total of
from those of the urban thermal environment (Wan et al., 2021). eleven bands, of which seven multispectral bands have a spatial reso­
Numerous studies have examined the relationship between urban ther­ lution of 30 m, and two thermal infrared bands have a spatial resolution
mal environment and urban planning (Aboelata, 2020; Bouyer et al., of 100–30 m by resampling, and the revisit period is 16 days for global
2011; Lin et al., 2017; Shafaghat, 2015; Yusuf et al., 2014). Anthropo­ coverage. Since the surface reflectance is affected by the atmospheric
genic buildings were found to have pronounced UHI effects on a rapidly conditions at the time of image acquisition, and it is also a potential
urbanizing arid region compared with an adjacent rural region (Salvati factor for studying the surface thermal model, when screening the data,
et al., 2017). It has been shown that the land usage can influence the we select the imaging data under good weather conditions, which was a
SUHI, industrial, commercial, and residential areas, and airports and clear day with less than 2% cloud cover and use the thermal band data to
parks, exhibit high daytime temperatures (Kato and Yamaguchi, 2005). retrieve the LST conduct research. The Landsat-8 surface reflectance
Temperatures differ between industrial and commercial areas with grass (SR) products are created with the Land Surface Reflectance Code. All
and impervious surfaces (Kardinal Jusuf et al., 2007). However, there collection 2 LST products are created with a single-channel algorithm
are two main strands of previous SUHI studies: the first focuses on the jointly created by the Rochester Institute of Technology and National
comparison of SUHI variations among a great number of cities distrib­ Aeronautics and Space Administration (NASA) Jet Propulsion
uted within an extensive region (Connors et al., 2012; Clinton and Gong, Laboratory.
2013; Peng et al., 2012; Yao et al., 2019), while the second focuses on
changes in intra-city SUHI within individual cities (Li et al., 2017; Lin 3. Methods
et al., 2018; Kim et al., 2017; Peng et al., 2018; Zhao et al., 2014). Within
an extensive region, cities across countries and climates are character­ The land surface temperature data covering the study area are
ized by significant inter-city SUHI variations. Whereas few consider­ normalized to make the temperature between different years compara­
ations have been taken to the extent and scope of the influence of each ble. Then, natural classification was carried out based on normalized
process from birth, growth, and operation of a particular functional area Surface temperature, and the heat island ratio values of the study area in
on the surrounding thermal environment. 2014, 2017, 2019, and 2021 were calculated using the classification
This study was to analyze the impact of airport construction and data. Secondly, the random forest (RF) method based on pixels was used
operation on the thermal environment using Landsat data of the Beijing to classify the land features in the study area, and the classification re­
Daxing International Airport. The urban heat island ratio index (URI) sults were combined with the temperature classification data to explore
was used to compare LST values from remote sensing data of different the relationship between the change in land surface classification and
periods (Zhang et al., 2017). Then the temporal and spatial analysis of land cover change in the study area. Finally, the thermal environment in
the land classification types of the airport is carried out to detect the the study area is quantitatively calculated by concentric circles with a
changes of thermal pollution during the construction and operation of specific radius to explore the specific impact range of the construction of
the airport. Finally, the relationship between land classification types Beijing Daxing International Airport on the thermal environment. The
and URI was studied to determine the cause of heat pollution intensifi­ research process is shown in Fig. 3.
cation after the construction and operation of Daxing Airport. Three
quantitative questions are mainly explored:

2
J. Wan et al. Ecotoxicology and Environmental Safety 237 (2022) 113521

Fig. 1. Beijing Daxing International Airport.

3.1. Urban heat island ratio index level i with respect to the entire area, which can be retrieved in Table 2.

It is necessary to normalize the LST data to ensure the comparability 3.2. Random forest classification
in different periods. The LST images were rescaled to the same level
between 0 and 1 using the maximum and minimum values of those With the help of the Google Earth Engine (GEE) platform, the
images. The computation is as Eq. (1): random forest method is used to classify land use in the study area. We
considered land use types related to the thermal environment in the
Ts − Tmin
Tk = (1) study area. The land use in the study area is divided into 6 categories,
Tmax − Tmin
including water, vegetation, High reflectivity building, dark buildings,
where Tk is the normalized value of a specified pixel, Ts is the LST of the bright soil, and dark soil. The references for this classification are from
related pixel, Tmax is the maximum value of that image, and Tmin is the literature (Gudina et al., 2014; Wan and Zhu, 2021). The two classifi­
minimum value of that image. cation criteria of this study are the identifiability of 30 m resolution
In this study, URI (Zhang et al., 2017) was used to quantitatively ground objects; The close relationship between ground feature category
assess the differences of SUHI in different periods. The larger the URI and urban thermal environment change.
value, the more severe the SUHI. The URI is expressed as Eq. (2): Since the period is too large and the ground object monitoring
changes too much, we separately select training samples and classifi­
1 ∑ cation for each scene image. Sample points were used to pick up ei­
n
URI = Wi Pi (2)
100m 1 genvalues for model training and validation. The samples we selected
and the training process, as well as the iteration results, can be seen on
where m is the number of divided temperature levels, in this study m the shared website (https://code.earthengine.google.com/
= 5. n is the number of temperature levels that are higher than those in b80e32e52d090167e6617527d68288de).
surrounding areas, levels 4 and 5 of the top 2 are selected in this article,
n = 2. Wi is the weight determined using the value of the corresponding
level i, in this study W1 = 4, W2 = 5. Pi is the percentage of the area of

3
J. Wan et al. Ecotoxicology and Environmental Safety 237 (2022) 113521

Fig. 2. True color of four remote sensing data. (a) 2014, (b) 2017, (c) 2019, (d) 2021.

4. Results and analysis

Table 1
The detailed information of the retrieved four remote sensing images.
4.1. Analysis for the change of URI and RF classification types
Collection Name Data (Year-Month- Column Cloud
Day) Number Cover
The normalized LST image was divided into five temperature levels
LANDSAT/LC08/C02/ 2014–09–04 123,033 Less than by the natural classification method, with the lowest value assigned to
T1_L2 2%
level 1 and the highest to level 5. The area changes of each temperature
LANDSAT/LC08/C02/ 2017–09–28 123,033 Less than
T1_L2 2%
layer before and after the construction and operation of Daxing Airport
LANDSAT/LC08/C02/ 2019–09–02 123,033 Less than were analysed in Table 2.
T1_L2 2% Accordingly, the URIs for 2014, 2017, 2019, and 2021 were calcu­
LANDSAT/LC08/C02/ 2021–09–07 123,033 Less than lated as follows:
T1_L2 2%
1
URI 2014 = (5 ∗ 3.54 + 4 ∗ 13.07) = 0.120
100 ∗ 5
3.3. Influence area estimation
1
URI 2017 = (5 ∗ 5.08 + 4 ∗ 16.81) = 0.185
To explore the influence range of Daxing Airport on the surrounding 100 ∗ 5
thermal environment, firstly, the data of different periods are normal­
ized according to Formula (1). Then take the whole airport as a circle 1
URI 2019 = (5 ∗ 3.40 + 4 ∗ 14.85) = 0.153
(the outermost three points are tangent, the radius is about 5 km, we 100 ∗ 5
take an integer) to obtain the center of the circle. Finally, concentric 1
circles were obtained by increasing the radius by 0.5 km with the same URI 2021 =
100 ∗ 5
(5 ∗ 2.29 + 4 ∗ 22.89) = 0.206
center, and the normalized mean temperature in each concentric circle
was calculated respectively, the data of 2017, 2019, 2021, and 2014 As can be seen, the URI of the study area was 0.12 in 2014. When the
were compared (See Fig. 4). The main purpose is to explore the scope of airport was under construction in 2017, the URI of the study area was
influence on the surrounding thermal environment in the initial stage of 0.185, 0.065 higher than that in 2014. After the completion of the
airport construction, the completion of airport construction, and the airport in 2019, the URI of the study area was 0.153, 0.033 higher than
operation of the airport. that in 2014. After the airport is operational in 2021, the URI of the
study area is 0.206, an increase of 0.086 from 2014 and 0.053 from
2019. The URI changes in the study area are shown in Fig. 5.
We note that the URI value for 2019 is lower than that for 2017,
mainly due to the data imaging date. As can be seen from the data in

4
J. Wan et al. Ecotoxicology and Environmental Safety 237 (2022) 113521

Fig. 3. The basic flow chart of this study. The process is top-down, with data processing at the top, URI calculation at the left, relationship exploration between URI
and RF classification types in the middle, and thermal pollution influence calculation at the right.

Table 2
The Temperature Levels.
Levels 2014 2017 2019 2021

Area Area Area Area

Km2 % Km2 % Km2 % Km2 %

1 145.41 25.62 83.49 14.72 126.76 22.31 150.44 26.52

2 188.77 33.29 190.54 33.59 187.54 33.09 208.74 36.78
3 138.78 24.48 169.18 29.80 149.48 26.35 65.33 11.52
4 74.20 13.07 95.34 16.81 84.19 14.85 129.77 22.89
5 20.11 3.54 28.72 5.08 19.28 3.40 12.99 2.29
Total 567.27 100.0 567.27 100.0 567.27 100.0 567.27 100.0

statistics in Table 2, the proportion of the first temperature level (mainly

vegetation areas) in 2019 was higher than that in 2017, and the pro­
portion of the fourth and fifth temperature levels was slightly lower than
that in 2017. However, the decrease in the proportion of level 4 and 5
temperature was not significant, indicating that the construction of the
airport was the main contributor to the high level temperature area in
the study area. We only used levels 4 and 5 in our calculations, so the
URI values for 2019 are lower than those for 2017. This also indicates
that the study of urban thermal environment is affected by phenology, so
it is more necessary to carry out data normalization processing when
comparing surface temperature in different periods.
The land use classification results of the study area by random forest
method are shown in Fig. 6. The overall classification accuracy of each
data is more than 80%. The distribution and number of training samples
as well as the data of the mixology matrix can be found on our shared
website. The percentage of statistics for each classification type is shown
Fig. 4. The influence range of different radii: 5–9.5 km respectively. in Table 3.
According to Figs. 2, 5, and 6, the lower right part of the URI 2014
Table 1 and Fig. 2, the imaging date of the 2019 image is September 2, study area belongs to Level 1, mainly consisting of water and vegetation,
when vegetation was planted in the farmland of the study area (ac­ while the lower left part of the URI of dark buildings and bright soil
cording to the field investigation, corn was planted in this area in this belongs to levels 4 and 5, major contributors to the thermal environ­
season). The date of the 2017 image was September 28, when no crops ment. Since the construction of the airport began in 2017, the land
were planted in the farmland of the study area and it was in the bare classification type of the study area center has changed from dark soil
field stage. At the same time, according to the temperature level

5
J. Wan et al. Ecotoxicology and Environmental Safety 237 (2022) 113521

Fig. 5. LST levels distribution of Daxing Airport area in different periods: (a) 2014, (b) 2017, (c) 2019, (d) 2021.

and vegetation to high reflectivity buildings and bright soil. At the same effects, vegetation increased by − 29%, with a growth rate of − 47%.
time, the central airport in the study area has a URI of Level 4 and level 5 High reflectivity buildings showed a significant increase, increasing by
respectively, initially replacing the dark building on the lower left as the 0.4% and increasing by 75.3%. Dark building types also remained sta­
main contributor to the thermal environment. It is worth noting that the ble, increasing by 4–24.8%. The pixels in the bright soil accounted for
URI grade of bright soil is classified as grade 4 and grade 5, which in­ less, but the change was the most drastic, increasing by 3.3% with a
dicates that the contribution of bright land to the thermal environment growth rate of 523%. The change in the dark soil was relatively stable,
of the city is higher than that of ordinary dark buildings. In 2019, the increasing by 22.5% with a growth rate of 78.1%. During this process,
area of high reflectivity buildings and bright soil increased again at the the change of land classification type is mainly influenced by natural
Study Regional Central airport, which has URI ratings of 4 and 5, phenology, vegetation is greatly reduced, and dark land is greatly
respectively. Except for the industrial buildings on the lower left, the increased. The construction of airports has led to an increase in high-rate
thermal contribution of the airport becomes the main contributor to the construction.
thermal environment. At the same time, it can be seen that the heat Compared with 2014, the change of water body type in 2019 was
distribution in the study area is mainly concentrated in the airport relatively stable, with an increase of − 2.3% and a growth rate of
center, which indicates that the contribution of buildings with high − 77%. Vegetation increased by − 16.7% with a growth rate of
reflectivity to the thermal environment is higher than that of bare land. − 26.9%. High reflectivity buildings increased significantly, by 1.7%, or
After the airport is put into operation in 2021, the distribution of high- 305%. Dark building types also remained stable, increasing by
level URIs in the study area is more concentrated in the airport area and 7.1–58.7%. The pixels of bright soil accounted for less, but the change
the lower left industrial zone. However, the area proportion of low- was also the most drastic, increasing by 4.2%, or 655%, while the
grade URIs is also increasing, mainly because residents are living in a change of dark soil was relatively stable, increasing by 6%, or 22.3%. In
more concentrated way, which makes dark buildings turn into vegeta­ this process, the change of land classification type was not affected by
tion (as can be seen from the comparison of Fig. 6(C) and (D), some phenology, but only by airport construction, which also resulted in a
residential buildings at the lower right corner of the airport are removed large reduction of vegetation, and a significant increase in the number of
and turned into vegetation). high-reflectivity buildings and bright soil.
According to statistics (Fig. 7), the mainland classification type in the Compared with 2014, the change of water body type in 2021 is
study area is vegetation, with a pixel above 2.0 × 105. The second is relatively stable, with an increase of − 2.2% and a growth rate of
dark soil, the pixel is more than 1.6 × 105. Water, high reflectivity − 72%. Vegetation increased by − 12.2% with a growth rate of
buildings, and bright soil have fewer pixels. In 2017, compared with − 20.4%. High reflectivity buildings increased significantly, by 2.3%, or
2014, the change in water body type was relatively stable, with an in­ 449%. Dark building types were also stable, with a 5% increase and a 7%
crease of − 0.8% and a growth rate of − 33.4%. Due to phenological increase. The pixels of bright soil accounted for less, but the change was

6
J. Wan et al. Ecotoxicology and Environmental Safety 237 (2022) 113521

Fig. 6. LCZ classification results of the four data: (a) 2014, (b) 2017, (c) 2019, (d) 2021.

Table 3
The percentage of categories in the classification.
Types 2014 2017 2019 2021

Water 0.028 0.020 0.005 0.006

Vegetation 0.574 0.280 0.407 0.452
High reflectivity building 0.004 0.008 0.021 0.027
Dark building 0.093 0.133 0.164 0.098
Bright soil 0.005 0.038 0.047 0.026
Dark soil 0.296 0.521 0.356 0.391
Total 1.000 1.000 1.000 1.000

also the most drastic, increasing by 2.1%, or 304%, while the change of
dark soil was relatively stable, increasing by 9.5%, or 35.6%. In this
process, the change of land classification type was not affected by
phenology, but only by airport construction, which also resulted in a
large reduction of vegetation, and a significant increase in the number of
high-reflectivity buildings and dark soil.
Compared with 2019, the changes of water body and high reflectivity
buildings in 2021 are more stable. The pixels of dark buildings and
bright soil decreased significantly, while the pixels of vegetation and
dark soil increased significantly. This phenomenon shows that after the
Fig. 7. Variation between Land classification types. The line chart shows the
stable operation of the airport, the surrounding residents move to live in
change rate relative to 2014. 1. Water; 2. Vegetation; 3.High refectivity
a more concentrated way, and the original building land is changed into
building; 4. Dark building; 5. Bright soil; 6. Dark soil.
vegetation area and dark soil area. We believe that this is closely related
to the concept of carbon neutrality and ecological restoration in China in
Daxing Airport, URI in the study area has been increasing. The main
recent years.
factor leading to the increase of URI is the construction of the airport,
To sum up, from the beginning of construction to the operation of
which greatly increases levels 4, and 5 of URI in the study area. The

7
J. Wan et al. Ecotoxicology and Environmental Safety 237 (2022) 113521

construction of the central airport in the study area, replacing the type of high-reflectivity buildings at the airport’s completion in 2019. It
original water body and vegetation, becomes the main contributor to the is worth noting that the URI2 type decreased significantly from phase
thermal environment. 2014–2019. The phenology of the two data images is consistent, so we
believe that the main reason for this phenomenon is that the construc­
tion of the airport brings about great changes in land classification types.
4.2. Relationship between URI change and land classification types
The proportion of 2021 land classification type minus the proportion
of 2014 land classification type, the largest changes in the URI4 and
According to the superposition analysis of land classification type
URI5 areas are high reflectivity buildings and dark buildings. The
and URI (See Fig. 8), the proportion of land classification types in 2017
change ratio of high-reflectivity buildings is 15.3%, and that of dark
subtracts the proportion of land classification types in 2014. In URI4 and
buildings is 8.1%, indicating that compared with 2019, after the oper­
URI5 regions, the change ratio of bright soil and high reflectivity
ation of the airport in 2021, the increase of high-reflectivity buildings
buildings is 3.2%, and the change ratio of bright soil is 8.5%. It indicates
will increase again, and the increase of bright soil will be alleviated
that compared with 2014, the construction of Daxing Airport in 2017 led
again, and dark buildings will become a feature with a large contribu­
to the increase of high-reflectivity buildings and bright soil, which led to
tion to the thermal environment.
a sharp increase in URI index. It is worth noting that the URI1 type
decreased significantly from 2014 to 2017. This phenomenon may be
4.3. 4.3 Analysis of influence area
caused by the phenological influence on the one hand and the huge
changes of land classification types brought about by the construction of
According to the three concentric circles set in Section 3.3, the LST in
airports on the other hand.
the study area (after normalization) was counted in Table 4 and Fig. 9. In
Subtracts the proportion of the land classification type in 2014 from
2014, the mean temperatures within the ten radii showed little differ­
the proportion of the land classification type in 2019, the largest changes
ence, and the values were all close to the normalized value of 0, indi­
in URI4 and URI5 regions are also high reflectivity buildings and bright
cating that the overall temperature in the study area was low. Since the
soil, with the change proportion of high reflectivity buildings and bright
airport was started to be built in 2017, the normalized average tem­
soil being 8.7% and 1.9%, indicating that compared with 2017, The
perature of the ten concentric circles is relatively large, indicating that
increase in bright soil was mitigated by the continued increase in the
the overall temperature is relatively high. The maximum temperature is
found in the circle with the minimum radius, and the average temper­
ature decreases gradually with the increase of radius. When the airport is
completed in 2019, compared with 2017, when the circle radius is the
same, the degree of influence is lower. This phenomenon is due to the
heat generated during the construction in 2017, which increases the
temperature in the research area. The temperature drops obviously at
7–7.5 km, and the influence tends to be gentle after 7.5 km. After the
operation of the airport in 2021, the mean temperature difference in the
circle area with a radius of 5 km and 7.5 km is 0.0581 units, and the
mean temperature difference in the circle area with a radius of 7.5 km
and 9.5 km is 0.0152 units, indicating that the influence radius of the
airport on the surrounding environment is about 7.5 km, which is more
serious. The influence area is gentle for the area beyond 7.5 km.

5. Discussion

Large-scale infrastructure construction changes the surface thermal

radiation and aggravates the severity of the urban thermal environment.
Therefore, it is very necessary to accurately and rapidly analyze the
impact of large-scale infrastructure construction on the urban thermal
environment. Unlike previous studies that focused on comparing SUHI
changes in a large number of cities (Meng et al., 2018; Schwarz et al.,
2011; Wan and Zhu, 2021; Zhou et al., 2018) and SUHI changes within
individual cities (Feyisa et al., 2014a, 2014b; Lin et al., 2018; Imhoff
et al., 2010; Seto and Christensen, 2013; Wan and Zhu, 2021), this study
focuses on the degree and scope of influence on the surrounding thermal
environment in each process from birth, growth to work of a specific
functional area. Taking Beijing Daxing International Airport as an
example, this paper discusses the influence of large-scale infrastructure
construction on the surrounding thermal environment, such as replacing
the natural landscape, changing the surface radiation and thermal
characteristics of urban areas.
Compared with before the construction of the airport, URI increased
by 0.065 at the beginning of the construction of the airport, 0.086 after
the operation of the airport, and 0.021 after the construction of the
airport, which not only indicates that land use will affect surface urban
heat island (Kato and Yamaguchi, 2005), but also that different pro­
duction modes will further affect surface urban heat island. As far as we
know, there is a lack of research to explore the effects of production
modes on the thermal environment. In the study area, the main con­
Fig. 8. Percentage difference of Land Classification types-URI for each data. tributors to the thermal environment prior to airport construction are

8
J. Wan et al. Ecotoxicology and Environmental Safety 237 (2022) 113521

Table 4
The influence area of airport heat generation (temperature is normalized and summed).
R (5 km) R (5.5 km) R (6 km) R (6.5 km) R (7 km) R (7.5 km) R (8 km) R (8.5 km) R (9 km) R (9.5 km)

2014 0.2904 0.29 0.2897 0.2885 0.2867 0.2849 0.2947 0.2948 0.2943 0.2844
2017 0.4645 0.4642 0.4632 0.4596 0.4588 0.4555 0.4532 0.4521 0.4518 0.4501
2019 0.4429 0.4418 0.4401 0.4318 0.4287 0.4058 0.4012 0.3988 0.3989 0.397
2021 0.4918 0.491 0.4891 0.4802 0.4788 0.4337 0.421 0.4209 0.4198 0.4185

combined true color data from Landsat-8 with historical data from
Google Earth, random forest classification in Google Earth Engine (GEE)
is based on a single target rather than scenario-based. The accuracy of
the current method is only 80%, so we will consider using a more ac­
curate algorithm for scene classification in the future (since the thermal
environment is a continuous and dynamic process, only the temperature
value of a single pixel is considered in this study). In the analysis of the
results, we only analyzed the thermal pollution brought by airport
construction and operation to the surrounding environment and the
types of thermal pollution caused by the changes. Next, we plan to
explore in depth how measures can be taken to improve heat pollution.

6. Conclusions

Large-scale infrastructure construction has an enhanced influence on

the heat island effect of surface cities. Taking Beijing Daxing Interna­
tional Airport as an example, this study uses Landsat data to analyze the
impact of airport construction and operation on the thermal environ­
ment. To compare land surface temperatures from remote sensing data
over different periods, we use the urban heat island ratio index (URI).
We then performed a spatio-temporal analysis of SUHI to detect changes
Fig. 9. The normalized mean values of circles with different radii. in thermal pollution during construction and operation. Finally, the
relationship between SUHI and land cover type, URI, and before and
dark buildings (towns and industrial areas). After the construction of the after construction and operation of the airport was studied to determine
airport, the pattern changed and the high reflectivity buildings in the the cause of heat pollution intensification after construction and oper­
airport area became the main contributor to the thermal environment. In ation of Daxing Airport. The main conclusions are as follows:
addition, the construction of the airport has brought some economic
development to the study area, which has changed the residential type (1) After the completion of Daxing Airport, the URI value in this area
from a low-density residential area with much vegetation to a increased from 0.12 in 2014–0.185 in 2017, indicating that the
high-density residential area with little vegetation (For example, the urban surface heat island effect in this area greatly increased.
buildings in the lower right corner of the airport were relocated after the After the operation of Daxing Airport, the URI value in this area
completion of the airport). Although some studies have shown that increased from 0.153 in 2019-0.206 in 2021, indicating that the
low-density residential development mode contributes more to the for­ surface urban heat island effect in this area increased again. This
mation of surface heat islands than high-density residential develop­ impact is related to airport construction and operation.
ment mode (Renard et al., 2019), the overall impact of large-area (2) Random forest algorithm was used to classify land types based on
high-density residential development on the thermal environment is pixels, and the relationship between URI and Land classification
the most severe in a specific research area. In addition, with the opening type was discussed. The results show that dark buildings and bare
of airports, anthropogenic activities (such as aircraft fuel burning, land (bright soil and dark soil) have a significant impact on the
exhaust emissions and airport crowd activities) have increased, which thermal environment of the airport before construction. After the
also contributes to the heat island effect to some extent. completion of the airport, this contribution pattern was trans­
According to the statistical results of the ten concentric circles, the formed into the airport’s high reflectivity buildings and bright
mean temperature in the concentric circles gradually decreases with the soil.
increase of the radius. The influence is more serious at the radius of (3) The thermal pollution generated by the airport has a greater
7.5 km, and less at the radius of 9.5 km, which means that we can set impact on the ground objects within the range of 7.5 km, and a
measures in a fixed range to deal with the negative effects of large-scale smaller impact on the ground objects outside the range of 9.5 km.
infrastructure. For example, vegetation and water was added at 5 km Our results can provide a reference for the study of the thermal
and 7.5 km in this study area. environment.
Although this study has made some progress in understanding the
thermal environment problems caused by the construction and opera­ The conclusion above can provide a reference for urban planners and
tion of Beijing Daxing International Airport, the following problems stakeholders when constructing new large-scale transportation infra­
remain to be further studied. In terms of data, due to cloud pollution and structure in the future. To minimize heat pollution from similar climates.
problems surrounding satellite operation, we only obtained data of the That is, the rational development of land use, in the process of devel­
4-scene research area, which had certain limitations. Thermal environ­ opment on the active use of governance means, economic development
ment is a dynamic process, which requires more spatial and temporal and environmental protection go hand in hand.
resolution data. We can consider multi-source data fusion and spatial
data interpolation. In terms of research methods, the accuracy of land
classification needs to be improved. Although in sample training, we

9
J. Wan et al. Ecotoxicology and Environmental Safety 237 (2022) 113521

CRediT authorship contribution statement Kardinal Jusuf, S., Wong, N.H., Hagen, E., Anggoro, R., Hong, Y., 2007. The influence of
land use on the urban heat island in Singapore. Habitat Int. 31 (1), 232–242.
Kato, S., Yamaguchi, Y., 2005. Analysis of urban heat-island effect using ASTER and ETM
Jikang Wan: Conceptualization, Methodology, Software, Formal + data: separation of anthropogenic heat discharge and natural heat radiation from
analysis, Writing. Bin Yong: Conceptualization, Writing – review & sensible heat flux. Remote Sens. Environ. 99 (1), 44–54.
editing, Project administration, Funding acquisition. Xiaofeng Zhou: Kim, M., Lee, K., Cho, G.-H., 2017. Temporal and spatial variability of urban heat island
by geographical location: a case study of Ulsan, Korea. Build. Environ. 126 (1),
Writing – review & editing. 471–482.
Li, J., Zhan, W., Hong, F., Lai, J., Dong, P., Liu, Z., Wang, C., Huang, F., Li, L., Wang, C.,
Fu, Y., Miao, S., 2021a. Similarities and disparities in urban local heat islands
Declaration of Competing Interest responsive to regular-, stable-, and counter-urbanization: a case study of Guangzhou,
China. Build. Environ. 199 (3).
Li, T., Rong, L.L., 2021. Vulnerability analysis of high-speed rail network in China from
The authors declare that they have no known competing financial spatial-temporal perspective. J. Railw. Sci. Eng. 1–9.
interests or personal relationships that could have appeared to influence Li, X., Li, W., Middel, A., Harlan, S.L., Brazel, A.J., Turner, B.L., 2016. Remote sensing of
the work reported in this paper. the surface urban heat island and land architecture in Phoenix, Arizona: combined
effects of land composition and configuration and cadastral–demographic–economic
factors. Remote Sens. Environ. 174 (1), 233–243.
Data Availability Li, X., Zhou, W., Ouyang, Z., 2013. Relationship between land surface temperature and
spatial pattern of greenspace: what are the effects of spatial resolution? Landsc.
Urban Plan. 114 (1), 1–8.
Some or all data, models, or codes that support the findings of this
Li, Y., Gao, G., Cheng, C., 2021b. Hazard intensity of rainstorm affecting high-speed
study are available from the corresponding author upon reasonable railway operation in China. Sci. Geogr. Sin. 41 (10), 1843–1851.
request. Lin, P., Lau, S.S.Y., Qin, H., Gou, Z., 2017. Effects of urban planning indicators on urban
heat island: a case study of pocket parks in high-rise high-density environment.
Landsc. Urban Plan. 168 (1), 48–60.
Acknowledgements Lin, Y., Jim, C.Y., Deng, J., Wang, Z., 2018. Urbanization effect on spatiotemporal
thermal patterns and changes in Hangzhou (China). Build. Environ. 145 (1),
166–176.
This work was supported by the National Key Research and Devel­ Manoli, G., Fatichi, S., Schlapfer, M., Yu, K., Crowther, T.W., Meili, N., Burlando, P.,
opment Project of China (2021YFB3900601) and National Natural Sci­ Katul, G.G., Bou-Zeid, E., 2019. Magnitude of urban heat islands largely explained by
ence Foundation of China (92047301). Thanks to the editor and two climate and population. Nature 573 (1), 55–60.
Meng, Q., Zhang, L., Sun, Z., Meng, F., Wang, L., Sun, Y., 2018. Characterizing spatial
anonymous reviewers for their suggestions.
and temporal trends of surface urban heat island effect in an urban main built-up
area: A 12-year case study in Beijing, China. Remote Sens. Environ. 204 (1),
References 826–837.
Mohajerani, A., Bakaric, J., Jeffrey-Bailey, T., 2017. The urban heat island effect, its
causes, and mitigation, with reference to the thermal properties of asphalt concrete.
Aboelata, A., 2020. Vegetation in different street orientations of aspect ratio to mitigate
J. Environ. Manag. 197 (1), 522–538.
UHI and reduce buildings’ energy in arid climate. Build. Environ. 172.
Natanian, J., Aleksandrowicz, O., Auer, T., 2019. A parametric approach to optimizing
Aflaki, A., Mirnezhad, M., Ghaffarianhoseini, A., Ghaffarianhoseini, A., Omrany, H.,
urban form, energy balance and environmental quality: The case of Mediterranean
Wang, Z.H., Akbari, H., 2017. Urban heat island mitigation strategies: a state-of-the-
districts. Appl. Energy 254 (1).
art review on Kuala Lumpur, Singapore and Hong Kong. Cities 62 (1), 131–145.
Oke, T.R., Stewart, I.D., 2012. Local climate zones for urban temperature studies. Bull.
Boehme, P., Berger, M., Massier, T., 2015. Estimating the building based energy
Am. Meteorol. Soc. 93 (1), 1879–1900.
consumption as an anthropogenic contribution to urban heat islands. Sustain. Cities
Parker, D.E., 2010. Urban heat island effects on estimates of observed climate change.
Soc. 19 (2), 373–384.
Wires Clim. Change 1 (1), 123–133.
Bouyer, J., Inard, C., Musy, M., 2011. Microclimatic coupling as a solution to improve
Peng, J., Jia, J., Liu, Y., Li, H., Wu, J., 2018. Seasonal contrast of the dominant factors for
building energy simulation in an urban context. Energy Build. 43 (3), 1549–1559.
spatial distribution of land surface temperature in urban areas. Remote Sens.
Brousse, O., Georganos, S., Demuzere, M., Vanhuysse, S., Wouters, H., Wolff, E.,
Environ. 215 (1), 255–267.
Linard, C., van Lipzig, N.P.M., Dujardin, S., 2019. Using local climate zones in Sub-
Peng, S., Piao, S., Ciais, P., Friedlingstein, P., Ottle, C., Breon, F.M., Nan, H., Zhou, L.,
Saharan Africa to tackle urban health issues. Urban Clim. 27 (1), 227–242.
Myneni, R.B., 2012. Surface urban heat island across 419 global big cities. Environ.
Cai, Z., Han, G., Chen, M., 2018. Do water bodies play an important role in the
Sci. Technol. 46 (1), 696–703.
relationship between urban form and land surface temperature? Sustain. Cities Soc.
Renard, F., Alonso, L., Fitts, Y., Hadjiosif, A., Comby, J., 2019. Evaluation of the effect of
39 (2), 487–498.
urban redevelopment on surface urban heat islands. Remote Sens. 11 (1).
Cai, Z., Tang, Y., Chen, K., Han, G., 2019. Assessing the heat vulnerability of different
Salvati, A., Coch Roura, H., Cecere, C., 2017. Assessing the urban heat island and its
local climate zones in the old areas of a Chinese megacity. Sustainability 11 (2).
energy impact on residential buildings in Mediterranean climate: Barcelona case
Cai, Z., Tang, Y., Zhan, Q., 2021. A cooled city? Comparing human activity changes on
study. Energy Build. 146 (1), 38–54.
the impact of urban thermal environment before and after city-wide lockdown.
Schwarz, N., Lautenbach, S., Seppelt, R., 2011. Exploring indicators for quantifying
Build. Environ. 195, 3.
surface urban heat islands of European cities with MODIS land surface temperatures.
Chibuike, E.M., Ibukun, A.O., Abbas, A., Kunda, J.J., 2018. Assessment of green parks
Remote Sens. Environ. 115 (1), 3175–3186.
cooling effect on Abuja urban microclimate using geospatial techniques. Remote
Seto, K.C., Christensen, P., 2013. Remote sensing science to inform urban climate change
Sens. Appl. Soc. Environ. 11 (2), 11–21.
mitigation strategies. Urban Clim. 3 (1), 1–6.
Clinton, N., Gong, P., 2013. MODIS detected surface urban heat islands and sinks: global
Shen, H., Huang, L., Zhang, L., Wu, P., Zeng, C., 2016. Long-term and fine-scale satellite
locations and controls. Remote Sens. Environ. 134 (1), 294–304.
monitoring of the urban heat island effect by the fusion of multi-temporal and multi-
Connors, J.P., Galletti, C.S., Chow, W.T.L., 2012. Landscape configuration and urban
sensor remote sensed data: A 26-year case study of the city of Wuhan in China.
heat island effects: assessing the relationship between landscape characteristics and
Remote Sens. Environ. 172 (2), 109–125.
land surface temperature in Phoenix, Arizona. Landsc. Ecol. 28 (1), 271–283.
*Shima Taslima, D.M.P., Shafaghata*, Arezou, 2015. Urban design guidelines to mitigate
Du, H., Zhan, W., Liu, Z., Li, J., Li, L., Lai, J., Miao, S., Huang, F., Wang, C., Wang, C.,
urban heat island (UHI) effects in hotdry cities. J. Teknol. (Sci. Eng. ) 74 (2),
Fu, H., Jiang, L., Hong, F., Jiang, S., 2021. Simultaneous investigation of surface and
119–124.
canopy urban heat islands over global cities. ISPRS J. Photogramm. Remote Sens.
Sun, R., Lü, Y., Yang, X., Chen, L., 2019. Understanding the variability of urban heat
181 (1), 67–83.
islands from local background climate and urbanization. J. Clean. Prod. 208 (1),
Feng, Y., Du, S., Myint, S.W., Shu, M., 2019. Do urban functional zones affect land
743–752.
surface temperature differently? a case study of Beijing, China. Remote Sens. 11, 2.
Velazquez-Lozada, A., Gonzalez, J.E., Winter, A., 2006. Urban heat island effect analysis
Feyisa, G.L., Dons, K., Meilby, H., 2014a. Efficiency of parks in mitigating urban heat
for San Juan, Puerto Rico. Atmos. Environ. 40 (1), 1731–1741.
island effect: an example from Addis Ababa. Landsc. Urban Plan. 123 (1), 87–95.
Wan, J., Zhu, M., 2021. Contribution degree of different surface factors in urban interior
Feyisa, G.L., Meilby, H., Fensholt, R., Proud, S.R., 2014b. Automated Water Extraction
to urban thermal environment. Adv. Meteorol. 2021 (2), 1843053.
Index: a new technique for surface water mapping using Landsat imagery. Remote
Wan, J., Zhu, M., Ding, W., Bonafoni, S., 2021. Accuracy evaluation and parameter
Sens. Environ. 140, 23–35.
analysis of land surface temperature inversion algorithm for landsat-8 data. Adv.
Giridharan, R., Emmanuel, R., 2018. The impact of urban compactness, comfort
Meteorol. 2021 (2), 1–16.
strategies and energy consumption on tropical urban heat island intensity: a review.
Weng, Q., 2012. Remote sensing of impervious surfaces in the urban areas: requirements,
Sustain. Cities Soc. 40 (2), 677–687.
methods, and trends. Remote Sens. Environ. 117 (3), 34–49.
Imhoff, M.L., Zhang, P., Wolfe, R.E., Bounoua, L., 2010. Remote sensing of the urban heat
Yao, R., Wang, L., Huang, X., Gong, W., Xia, X., 2019. Greening in rural areas increases
island effect across biomes in the continental USA. Remote Sens. Environ. 114 (1),
the surface urban heat island intensity. Geophys. Res. Lett. 46 (3), 2204–2212.
504–513.
Yao, R., Wang, L., Huang, X., Sun, L., Chen, R., Wu, X., Zhang, W., Niu, Z., 2021. A robust
Jiang, L., Zhan, W., Hu, L., Huang, F., Hong, F., Liu, Z., Lai, J., Wang, C., 2021.
method for filling the gaps in MODIS and VIIRS land surface temperature data. IEEE
Assessment of different kernel-driven models for daytime urban thermal radiation
Trans. Geosci. Remote Sens. 1–15.
directionality simulation. Remote Sens. Environ. 263, 1.

10
J. Wan et al. Ecotoxicology and Environmental Safety 237 (2022) 113521

Ye, H., Wang, K., Zhao, X., Chen, F., Li, X., Pan, L., 2011. Relationship between Zhou, D., Xiao, J., Bonafoni, S., Berger, C., Deilami, K., Zhou, Y., Frolking, S., Yao, R.,
construction characteristics and carbon emissions from urban household operational Qiao, Z., Sobrino, J., 2018. Satellite remote sensing of surface urban heat islands:
energy usage. Energy Build. 43 (1), 147–152. progress, challenges, and perspectives. Remote Sens. 11 (2), 1–13.
Yihan Wu, A.S., Yang, Perry, Borjigin, Habura, Murakami, Daisuke, Yamagata, Yoshiki, Zhu, Z., Zhou, Y., Seto, K.C., Stokes, E.C., Deng, C., Pickett, S.T.A., Taubenböck, H., 2019.
2018. Mapping building carbon emissions within local climate zones in Shanghai. Understanding an urbanizing planet: strategic directions for remote sensing. Remote
ScienceDirec 152 (1), 815–822. Sens. Environ. 228 (2), 164–182.
Yu, Z., Yang, G., Zuo, S., Jørgensen, G., Koga, M., Vejre, H., 2020. Critical review on the
cooling effect of urban blue-green space: a threshold-size perspective. Urban For.
Urban Green. 49 (1), 1–14. Further reading
Yusuf, Y.A., Pradhan, B., Idrees, M.O., 2014. Spatio-temporal assessment of urban heat
island effects in Kuala Lumpur metropolitan city using landsat images. J. Indian Soc. Demuzere, M., Kittner, J., Bechtel, B., 2021. LCZ generator: a web application to create
Remote Sens. 42 (2), 829–837. local climate zone maps. Front. Environ. Sci. 9 (2).
Zhang, L., Meng, Q., Sun, Z., Sun, Y., 2017. Spatial and temporal analysis of the Lai, D., Liu, W., Gan, T., Liu, K., Chen, Q., 2019. A review of mitigating strategies to
mitigating effects of industrial relocation on the surface urban heat island over improve the thermal environment and thermal comfort in urban outdoor spaces. Sci.
China. ISPRS Int. J. Geo-Inf. 6 (2), 1–14. Total Environ. 661 (1), 337–353.
Zhao, L., Lee, X., Smith, R.B., Oleson, K., 2014. Strong contributions of local background Lai, J., Zhan, W., Quan, J., Liu, Z., Li, L., Huang, F., Hong, F., Liao, W., 2021. Reconciling
climate to urban heat islands. Nature 511 (1), 216–229. debates on the controls on surface urban heat island intensity: effects of scale and
sampling. Geophys. Res. Lett. 39 (1), 1–11.

11