Sei sulla pagina 1di 11

The Egyptian Journal of Remote Sensing and Space Sciences 21 (2018) 15–25

Contents lists available at ScienceDirect

The Egyptian Journal of Remote Sensing and Space Sciences

journal homepage: www.sciencedirect.com

Research Paper

Assessment of urban heat islands and impact of climate change on


socioeconomic over Suez Governorate using remote sensing and GIS
techniques
S. Ahmed
Physics Department, Faculty of Science, Al Arish University, El Arish, Egypt

a r t i c l e i n f o a b s t r a c t

Article history: Climate change and its effect on human life and comfort have taken increase attraction. In this paper
Received 17 April 2016 effect of UHI is analyzed using quantitative thermal, temporal remote sensing and GIS techniques.
Revised 30 June 2017 Series of Landsat TM/ETM, 8 and ASTER images of Suez Governorate acquired in winter season were used
Accepted 4 August 2017
to investigate the variation in land surface temperature (LST) from 1988 to 2014 as well as land-use/cover
Available online 10 August 2017
change. NDVI NDBI and LST have been used for classification of the study area. Vulnerable areas for UHI
have been assessed over Suez district. The effect of UHI was quantitatively described using urban thermal
Keywords:
field variance index (UTFVI). Results indicate that there are increases in the overall temperature between
Climate change
UHI
years 1988 and 2014. Suez and Al-Arbin districts which are the most populated, highly density built-up
Suez Canal and rarely vegetation zones in Suez governorate experiences a huge problem of UHI. UHI phenomenon
Suez Governorate represents total weight of 40.84% and 46.3% of their areas respectively. It is found that the mean temper-
Land surface temperature ature of UHI areas of Al-Arbin and Al-Suez districts was 7.5 above mean of District temperature. While
Land/use land/cover change the value of UHI of suburban zone e.g. Ataqa and AL-Ganayin was 2.5 degree above district mean. The
Urban sprawl results indicate that changes in UTFVI distribution can be predominantly related to the expansion of
urban area during the study time period. The hot spot of UTFVI were found mainly in the built-up areas
especially densely populated district and in heavily industrial district. It is the vulnerable areas to UHI.
Ó 2017 National Authority for Remote Sensing and Space Sciences. Production and hosting by Elsevier B.V.
This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-
nd/4.0/).

1. Introduction meanwhile, replacing natural land cover with concrete, infrastruc-


ture, and industrial activities take away the natural cooling effects
The climate is one of the most influential physical factors that (Weng and Larson, 2005). It is known that, the urban heat islands
not only impact the person psychologically and physiologically have become more and more significant and have severe impacts
but also controls the life behaviors and economic activities. The cli- on the human living environment (Anbar, 2012, Weng and
mate still affects the comfort of the person indoors and outdoors Larson, 2005, and Liu and Zhang, 2011).
(Anbar, 2012). Climate changes are accompanying by the modern Satellite remote sensing provides a straightforward way to
urbanization, especially in the case of rapid changes, where man investigate the thermal differences between urban and rural areas,
has created his habitat (Kim, 1992). The residents of the major retrieve land surface temperatures (LST) and urban heat islands.
cities in tropical and semi-arid zones often experience a regional Many authors have documented the use of remote sensing date
climate phenomenon known as the urban heat island (Kim, in retrieving LST and UHI on the spatial, spectral and temporal
1992; Streutker, 2003, and Li et al., 2005). An urban heat island scales (e.g. Kim, 1992; Streutker, 2003; Weng and Larson, 2005;
(UHI) is a metropolitan area which, significantly, warmer than its Pu et al., 2006; Giridharan and Kolokotroni, 2009; Bhatta, 2010;
surrounding rural areas. The higher the urbanization, the larger Liu and Zhang, 2011; Weng, 2011; Xiong et al., 2012; Abutaleb
temperature differences between urban and rural areas. Tall build- et al., 2015; Li et al., 2016; Zhou et al., 2014; Khandelwal et al.,
ings and narrow streets can heat the air trapped between them, 2018) as well as examining satellite sensors that are the most
appropriate for detection and assessment of UHI (e.g. Al Kuwari
et al., 2016). Other authors documented the importance of differ-
Peer review under responsibility of National Authority for Remote Sensing and
ent satellite images in detecting the relationship between land sur-
Space Sciences.
E-mail address: ahmedsamah92@yahoo.com face temperature with changes of land-use land-cover (e.g. Mundia

http://dx.doi.org/10.1016/j.ejrs.2017.08.001
1110-9823/Ó 2017 National Authority for Remote Sensing and Space Sciences. Production and hosting by Elsevier B.V.
This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).
16 S. Ahmed / Egypt. J. Remote Sensing Space Sci. 21 (2018) 15–25

and Aniya, 2005; Chen et al., 2006; Ahmed and Kaiser, 2014a; Pal
and Ziaul, 2017)
Suez Governorate urban area is experiencing rapid develop-
ment and urban sprawl according to census 2006 and land-use
build up map (The Egyptian General Survey Authority, 2012; The
State Information Service, 2014). As a result of topographic and
geologic nature of the study area where the mountain bordered
its west side (Zahran, 2010 and Sims, 2015), it experiences an
increase in temperature, and it is believable to face climate phe-
nomena of UHI.
In this research, the phenomenon of UHI was investigated over
the study area using Landsat and ASTER images through the period
from 1988 to 2014. This paper will firstly retrieve LST of Suez
Governorate and assess changes in temperature from 1988 to
2014 using the aspect of emissivity correction. Secondly, extract
UHI and ecologically evaluate the obtained UHI taking in consider-
ation the number of families may be affected by the phenomena.
Also, the changes of land-use-land-cover will be examined. For this
purpose supervised classification depending on Vegetation index
and build up index will proceed followed by gain and loss analysis.

2. Study area

Suez Governorate is one of the 26 governorates of Egypt. It is


located in the north-eastern part of Egypt, (Fig. 1) and is situated
at the northern tip of the Gulf of Suez near the southern terminus
of the Suez Canal. The Suez is one of the governorates of Suez Canal
Region’s urban governorates where the Suez Canal leads forward
into the north. It occupies a total area of 9002.21 km2. The area
is bounded by a mountain known as Gabal Ataqa to the west and
Suez Canal. To the east; it is bordered in the North by Ismailia, in
the North East by North Sinai governorate, in the South by the
Red Sea governorate, in the East by Suez Gulf, and in the West by
Cairo governorate and Helwan. (The Egyptian General Survey
Authority, 2012; The State Information Service, 2014).
The Suez Governorate formed of one city that is administra-
tively divided into five districts; Al-Suez, Al-Arbin, Faysal, Ataqa,
and Al-Ganayin, Districts. Ataqa District is considered an industrial
zone. Al-Ganayen District is a cultivated area of dense fruit culti-
vated lands. The most governmental, commercial and economic
bureaus and agencies are centralizing the Al-Suez District (The
Egyptian General Survey Authority, 2012; The State Information
Service, 2014; Zahran, 2010; Sims, 2015). The population distribu-
tion density varies over the five districts; Al-Arbaeen District is
considered the densely populated urban area with many variable
civilian activities. The districts and population distribution (num-
ber of persons) of the study area are shown in Fig. 2.
The Suez Governorate is a growing industrial and commercial
zone due to the presence of natural mineral resources, such as Fig. 1. Location of the study area.
limestone, dolomite, coal, and oil. It is also known for some eco-
nomic activities including oil distillation, processing and packing
of cement, textiles industry, glass medicine bottles, fertilizers, Also, socioeconomic data (obtained from the Central Agency for
chemicals, shipping, marine services and fishing. Worth noting that Public Mobilization and Statistics (CAPMAS) based on the Egyptian
development of northwest of Suez Gulf is one of the mega projects census operation in 2006), topographic and land use map of study
in Egypt. (The Egyptian General Survey Authority, 2012; The State area (source: The Egyptian General Survey Authority, 2012; The
Information Service, 2014). State Information Service, 2014) area available.
ENVI 5.1 and ERDAS Imagine 14 and Arc Map 10.2 were used to
carry out image processing and GIS procedures. Landsat calibration
3. Data available and methodology equations were obtained and corrected from Landsat 7 and Landsat
8 handbooks (USGS, 2013; Loyd, 2013; Chander and Markham,
In this study, series of Landsat images and ASTER data were 2003). ASTER calibration equations and required parameters were
used for the retrieving of urban heat island as well as land-cover/ derived from the ASTER user handbook (Gillespie et al., 1998 and
Land-use through classification process. These images and its date Kalinowski and Oliver, 2004). As a result, Landsat and ASTER DN
of acquisition are shown in (Table 1). All of these images are values were converted into spectral radiance in units (W m2
acquired in the early morning about 8 am under a clear sky. mm1 sr1).
S. Ahmed / Egypt. J. Remote Sensing Space Sci. 21 (2018) 15–25 17

Fig. 2. Districts and population distribution of Suez Governorate.

To retrieve UHI of the study area, the following processing was 3.2. Retrieving land surface temperature (LST)
performed:
Satellite thermal infrared sensors measure Top of the Atmo-
3.1. Estimation of at sensor brightness temperature Ti sphere (TOA) radiances, from which brightness temperature
(known as blackbody temperatures) can be derived based on
The spectral radiance of thermal infrared bands were converted Plank’s law. The TOA radiances are the result of mixing three parts
into active radiance at sensor brightness temperature (the temper- of energy. The first is the emitted radiance from earth’s surface, the
ature values of black body) using Planck’s function equation as second is the upwelling radiance from the atmosphere, and the
follows: third is the downwelling radiance from the sky (Weng and
Larson, 2005; Al Kuwari et al., 2016; Van and Bao, 2010). The dif-
C
Ti ¼  2  ference among TOA and land surface brightness temperature is
ki  ln 1 þ LCk15 subject to the influence of atmospheric conditions. Therefore, to
i i
obtain an actual land surface brightness temperature, atmospheric
C1 = 1.19104356  1016 W.m2, C2 = 1.4387685  104 mm K.where effects, including absorption upward emission and downward irra-
Ti represents sensor brightness temperature in kelvin; C1 and C2 diance reflected from the surface, should be corrected first. This
are constants; k is emitted radiance wavelength for peak response correction was conducted by calculating spectral emissivity (e),
and average limiting wavelengths (k = 11.5 mm for Landsat band 6 (Weng and Larson, 2005; Al Kuwari et al., 2016; Van and Bao,
data, k = 10.9 mm and 12 mm for Landsat 8 bands 10 and 11, respec- 2010). LSTs were obtained by recovering satellite temperature Ti
tively; k = 10.6 mm and 11.3 mm for ASTER bands 13 and 14, respec- by applying the correction for emissivity.
tively); and Li is spectral radiance, (Liu and Zhang, 2011;
Stathopoulou et al., 2009). This equation was simplified to the fol-
lowing form and used to convert Landsat data (Al Kuwari et al., 3.2.1. Determination of land surface emissivity e
2016; Chander and Markham, 2003; Van and Bao, 2010): Land surface emissivity affects the satellite measurements by
three categories:
K2
Ti ¼   ð1Þ
ln K1
Lk
þ1 i) emissivity causes a reduction of surface-emitted radiance;
ii) non-black surfaces reflect radiance; and
where K2 represents the calibration constant 2 in Kelvin; K1 is the iii) anisotropy of reflectivity and emissivity might reduce or
calibration constant 1 in W/ (m2.sr.mm); (Table 2). increase total surface radiance.
18 S. Ahmed / Egypt. J. Remote Sensing Space Sci. 21 (2018) 15–25

Table 1
Data available and specifications.

Satellite data Date of acquisition Sensor Band No. Spectral range (Wavelength mm) Spatial resolution, m
Landsat 5 11 Jan 1988 TM 1 0.45–0.52 30, band 6: 120 resampled to 30
2 0.52–0.60
3 0.63–0.69
4 0.76–0.90
5 1.55–1.75
6 10.40–12.50
7 2.08–2.35
Landsat 7 18 Jan 2008 ETM+ 1 0.45–0.52 30, band 6: 60 resampled to 30, band 8 15
2 0.52–0.60
3 0.63–0.69
4 0.76–0.90
5 1.55–1.75
6 10.40–12.50
7 2.08–2.35
8 0.52–0.90
ASTER 23 Nov 2011 VNIR 1 0.52–0.60 15
2 0.78–0.86
3N 0.63–0.69
3B 0.78–0.86
SWIR 4 1.60–1.70 30
5 2.145–2.185
6 2.185–2.225
7 2.235–2.285
8 2.295–2.365
9 2.360–2.430
TIR 10 8.125–8.475 90
11 8.475–8.825
12 8.925–9.275
13 10.25–10.95
14 10.95–11.65
Landsat 8 28 Dec 2014 OLI 1 0.43–0.45 30, band 8:15
2 0.45–0.51
3 0.64–0.67
4 0.53–0.59
5 0.85–0.88
6 1.57–1.65
7 2.11–2.29
8 1.36–1.38
9 0.50–0.68
TIRS 10 10.60–11.19 100 resampled to 30
11 11.50–12.51

Table 2 Table 3
Landsat thermal band calibration constant. NDVI values and its corresponding values of Land surface spectral
emissivity.
Constant: Units K1:W/(m2.sr.mm) K2: Kelvin
NDVI Land surface emissivity e
Landsat5 TM 607.76 1260.56
Landsat7 ETM+ 666.09 1282.71 NDVI < 0.185 0.995
Landsat8 TIRS (band 10) 774.8853 1321.0789 -0.185  NDVI<0.157 0.970
Landsat8 TIRS (band 11) 480.8883 1201.1442 0.157  NDVI0.727 1.0094 + 0.047 ln (NDVI)
NDVI > 0.727 0.990

Emissivity as a function of wavelength is controlled by several


environmental factors such as surface water content, chemical where Ts represents land surface temperature, Ti indicates sensor
composition, structure, and roughness. For vegetated areas, emis- brightness temperature in Kelvin, k is the wavelength of the emit-
sivity varies significantly with plant species, areal densities, and ted radiance (for peak response and average limiting wavelengths),
growth rates. In fact, land surface emissivity is closely related to e represents land surface spectral emissivity, q ¼ hcr = 1.438  10
2

34
NDVI (Weng and Larson, 2005; Van and Bao, 2010). Therefore, mk, where h is Plank’s constant (6.626  10 Js), c indicates light
the emissivity can be estimated from NDVI used Table 3 (Liu and velocity (2.998  108 m/s), andr is the Boltzmann constant
Zhang, 2011). (5.67  108 Wm2 k4=1.38  1023 J/k).

3.2.2. LST (Ts) estimates


The emissivity-corrected land surface temperature was 3.3. LST normalizing and obtaining urban heat island (UHI)
obtained using the following equation (Xiong et al., 2012; Yue
et al., 2012): As a result of interannual variation and different atmospheric
Ti conditions within the same period among years; it was not
Ts ¼  appropriate to compare multiple data images from different years.
1 þ ðk  Tqi Þ lne
Therefore, to compare LST from different dates, a normalization
S. Ahmed / Egypt. J. Remote Sensing Space Sci. 21 (2018) 15–25 19

method was performed using the following equation (Abutaleb bMIR  bNIR
NDBI ¼
et al., 2015): bMIR þ bNIR
Ts  Tm where bMIR, bNIR are the pixel values of middle infrared and near
UHI ¼
SD infrared bands.
where: Ts is the land surface temperature, Tm is the mean of the The idea is that the NDVI could be taken as an indicator of veg-
land surface temperature of the study area, and SD is the standard etation coverage whereas the NDBI has been used in extraction and
deviation. mapping of the built-up area. Both NDVI and NDBI were consid-
The urban heat island in the study area is defined as the areas ered as closely relating to LST (Xiong et al., 2012). Vegetated areas
that have normalized value greater than 1.5 °C above the mean have a negative correlation with LST while the buildup areas
of land surface temperature. At this level, the study area is classi- (NDBI) shows high correlation with LST, it is assumed that the
fied into two classes. First: areas with UHI (given value 1) and sec- built-up area could be taken as urbanized area on the classification
ond: places have no UHI effect (given value 0). results. Water bodies have both low NDVI and low LST. From the
Finally, the effect of UHI, at district level taking into considera- obtained feature space between NDVI- LST, and NDBI taking in
tion socioeconomic parameter, was quantitatively described using consideration the objective of this study, we can precisely discrim-
urban thermal field variance index (UTFVI). UTFVI could be calcu- inate four main classes. These classes are water, urban, vegetated
lated using the following equation (Liu and Zhang, 2011; Zhang, areas, and rural areas. Landsat (5), date 1988 and Landsat 8
2006): (2014) have been prepared and used to find out the overall LULC
classification of the study area. Correlation between LST, NDVI,
Ts  Tm NDBI for the mentioned date is carried out, correlation coefficient
UTFVI ¼
Ts is generated (Table 5). Classified images were compared and veri-
UTFVI was divided into six levels by six different ecological fied, visually, with the land use map of Suez governorate (The
evaluation indices (Liu and Zhang, 2011; Zhang, 2006). Thresholds Egyptian General Survey Authority, 2012).
in the six UTFVI levels are shown in Table 4:
3.5. Change detection of both LST and Land-use/Land-cover
3.4. Classification and retrieving of Land-use/Land-cover (LULC)
Documenting changes of both LULC and LST change is one of the
The intent of the image classification process is to categorize all most important goals in this research. Change detection can be
pixels in a set of digital images into one of several land cover defined as the process of identifying the difference in the state of
classes, or ‘‘themes”. This categorized data may then be used to an object or phenomenon by observing it at different times. To
produce thematic maps of the land cover present in an image retrieve changes of both LULC and LST, time series analysis of clas-
(Eastman, 1997). Normally, multispectral data are used to perform sified and LSTs images from years 1988 to 2014 were compared
image classification, and the spectral pattern present within the using cross tabulation through gain and loss analysis. Matrix of
data for each pixel is used as the numerical basis for categorization change is obtained for each category.
(Lillesand et al., 2004 and Jensen, 1986). Several techniques of clas-
sification process including methods of supervised and unsuper- 4. Results and discussion
vised classification have been reported in the literature e.g.
(Rawat and Kumar, 2015; Seto et al., 2002). Some authors report A set of Landsat and ASTER images have been prepared and
the classification independence of NDVI, e.g. (Van De Griend and used to retrieve LST and LULC of the study area. The generated
Owe, 1993). Some other report the NDVI and NDBI (normalized dif- LST have been analyzed to observe the temporal variation of sur-
ference built-up index) to characterize the LULC types, e.g. (Xiong face temperature, its impact on socioeconomic parameters as well
et al., 2012) and to quantitatively study the relationship between as the association of LULC change. The objective was to obtain typ-
the thermal environment and urban expansion. In another litera- ical variation by measuring the change of these parameters during
ture, the feature space between LST and NDVI has been used to cat- study period from 1988 to 2014.
egorize the pixels in an area, e.g. (Ahmed and Kaiser, 2014b; Hu,
2007). In this study, Land-use/ land-cover classification of the 4.1. LST results and analysis
study area was obtained through segmentation processing of the
satellite images. Image segmentation is the process of partitioning The calculated LST values are segmented into six levels of tem-
a digital image into multiple segments based on the pixel values. perature zone. LST patterns from year 1988–2014 are displayed
To segment an image we need to enter a range of pixel values, clearly in Fig. 3. In this Fig. the red, yellow, and orange tones high-
thresholds, to use in the segmentation for each class. In this study, light higher temperature, while the green, blue and cyan tones
threshold values obtained from the feature space- 2 D reflect low temperature zones. From these results, it can be empha-
scatterplot- between LST and NDVI and verified with NDBI index. sised that, there are rapid increase in the overall temperature
NDBI have been calculated using the equation: between years 1988 and 2014. The dark green color which given
to low temperature range (0–15) is approximately transformed
to another color with higher temperature ranges. While the blue
colors are given to temperature range from 15 to 18 degree mostly
Table 4
Threshold values of urban thermal field variance index.

Urban heat island phenomenon Urban thermal field variance index Table 5
none <0 Correlation coefficient between LST, NDVI, and NDBI of the study area.
weak 0.000–0.005
LST NDVI NDBI
middle 0.005–0.010
strong 0.015–0.015 LST 1
stronger 0.015–0.020 NDVI 0.85 1
strongest >0.020 NDBI 0.69 -0.41 1
20 S. Ahmed / Egypt. J. Remote Sensing Space Sci. 21 (2018) 15–25

Fig. 3. LST results obtained for study area over time period 1988–2014 dates, (a): Landsat 5 1988, (b): Landsat 7 2008, (c): ASTER 2011 and (d): Landsat 8 2014.

changed to red (range 18–21 c°) expect that areas of the gulf which 2014. Areas in km2 of these changes are documented and shown in
experience no change. The red color which was rare in date 1988 Tables 6–8, respectively. At the same time, the overall changes
becomes dominant in 2014. from 1988 to 2014 are represented in Table 9. This type of tabula-
Gain and loss analysis between pairs of LST images have been tion offers total areas under each category and its temporal varia-
carried out in order to quantitate these changes. This type of anal- tion and provides the magnitude of LST variation of the entire area
ysis is carried out between dates 1988–2008, 2008–2011 and 2011, as well as the spatial extent of that change.
S. Ahmed / Egypt. J. Remote Sensing Space Sci. 21 (2018) 15–25 21

Table 6
Cross tabulation summary of gain and loss of LST from 1988 to 2008, (areas in km2 and temperature range in degree centigrade).

Final State 2008 Initial state 1988 ranges


0–15 15–18 18–21 21–24 24–27 27–31 Total area
0–15 211.56 106.85 8.21 0.2 0.0 0 326.82
15–18 494.63 2166.06 67.63 2.15 0.07 0 2730.55
18–21 243.08 1568.47 279.75 4.44 0.06 0.02 2095.81
21–24 22.23 605.02 360.1 14.05 0.2 0 1001.60
24–27 0.38 29.00 33.25 10.8 0.65 0 74.07
27–31 0.09 0.64 2.60 4.05 0.21 0 7.0
Total area 971.99 4476.09 751.57 35.68 1.19 0.03
changes 760.43 2310.02 471.81 21.63 0.54 0.03

Table 7
Cross tabulation summary of gain and loss of LST from 2008 to 2011, (areas in km2 and temperature range in degree centigrade).

Final State 2011 Initial state 2008 ranges


0–15 15–18 18–21 21–24 24–27 27–31 Total area
0–15 58.15 98.21 8.93 5.88 0.94 0.39 172.5
15–18 136.07 1338.88 56.04 17.17 1.71 0.97 1550.84
18–21 14.21 514.98 740.22 179.49 6.44 1.54 1456.88
21–24 1.35 14.03 249.06 297.26 7.25 1.34 570.29
24–27 0 0.01 0.24 30.32 15.57 0.18 46.32
27–31 0 0 0 0 4.06 2.78 6.84
Total area 209.78 1966.11 1055.12 526.12 35.41 7.21
changes 151.63 627.23 314.9 228.86 19.84 4.43

Table 8
Cross tabulation summary of gain and loss of LST from 2011 to 2014, (areas in km2 and temperature range in degree centigrade).

Final State 2014 Initial state 2011 ranges


0–15 15–18 18–21 21–24 24–27 27–31 Total area
0–15 59.54 48.8 15.2 7.88 0.63 0 132.05
15–18 35.05 1271.81 59.83 12.64 1.66 0 1380.99
18–21 0.87 219.1 657.06 435.84 7.32 0 1320.19
21–24 0.37 0.82 361.01 443.72 23.84 1.66 831.42
24–27 0 0 0.39 0.21 30.62 0.16 31.38
27–31 0 0 0 0 9.14 3.45 12.59
Total area 95.83 1540.53 1104.5 900 73.21 5.27
changes 36.29 268.72 447.44 456.28 49.37 1.82

Table 9
Cross tabulation summary of gain and loss of LST from 1988 to 2014, (areas in km2, percentage and temperature range in degree centigrade).

Final State 2014 Initial state 1988 ranges


0–15 15–18 18–21 21–24 24–27 27–31 Total area
Km2 % Km2 % Km2 % Km2 % Km2 % Km2 % Km2 %
0–15 189.83 18.92 10.46 0.24 1.26 0.17 0.01 0.03 0 0 0 0 201.56 100
15–18 260.78 26.00 1507.37 33.95 16.52 2.19 0.5 1.31 0 0 0 0 1785.18 100
18–21 543.63 54.20 1927.83 43.42 235.46 31.14 5.02 13.1 0.02 1.86 0.02 64.29 2711.99 100
21–24 8.1 0.81 972.37 21.9 437.63 57.88 24.16 63.03 0.17 14.21 0.01 35.71 1442.45 100
24–27 0.61 0.06 19.6 0.44 60.73 8.03 7.46 19.47 0.99 82.14 0 0 89.4 100
27–31 0 0 2.27 0.05 4.48 0.59 1.18 3.07 0.02 1.89 0 0 7.95 100
Total area 1002.96 100 4439.91 100 756.09 100 38.33 100 1.21 100 0.03 100 201.56
changes 813.13 81.07 2932.53 66.05 520.63 68.86 14.17 36.97 0.22 17.86 0.03 100 1785.18

Table 10
Variation of UHI of Suez districts and its weight.

Suez Districts Total area (km2) Number of families UHI area (km2) Weight of UHI
1988 2011 2014 1988 2011 2014
Al-Ganayin 1220.042355 19962 80.18 192.81 242.02 6.57% 15.8% 19.84%
Faysal 1170.109075 31178 12.18 68.72 74.31 1.04% 5.87% 6.35%
Al-Suez 70.938479 19962 19.35 20.13 28.97 27.28% 28.38% 40.84%
Al-Arbin 44.834302 51829 19.55 20.55 20.76 43.61% 45.84% 46.30%
Ataqa 6940.675667 6298 411.94 687.79 762.34 5.94% 9.91% 10.98%
22 S. Ahmed / Egypt. J. Remote Sensing Space Sci. 21 (2018) 15–25

As shown in these tables, data in column state the changes of


each temperature range (zone) in its initial date while the rows
give the gain or loss of the final corresponding date. The term total
area express the overall area in km2 for each region and the term
changes express the total change in km2 of that zone. From these
tables we can discuss the temporal variation of LST. Temperature
zone 0–15 experience a great shift to higher temperature from
1988 to 2014. About 81%, (see Table 9), have been changed. That
change was mostly the raise of LST to zone 18–21 which indicate
that the minimum temperature has been raised. Through the
study period (26 years) the minimum LST has been raised with
about 3 c °, this increase is mostly occurred through time period
1988–2008 (see Table 6). The variation of temperature zone 18–
21 is considered the most effective LST increase on the socioeco-
nomic parameter since this variation occurred mostly on the res-
idential areas (cyan tone transformed to reddish, see Fig. 3a, d).
LST pattern (15–18) experience an increase with about 66% from
its total area in the past. From Table 9 we can see the transforma-
tion from this zone to zones 18–21 and 21–24. This gives the same
impression of minimum temperature increase. Most of this
increase obtained through time period 1988–2011, (see Tables 6
and 7).

4.2. Effect of LST variation on socioeconomic parameter

Several studies have focused primarily on the impact of bio-


physical and meteorological factors on land surface temperatures
(LSTs). Less attention has been paid to study of the influence of
socioeconomic factors on UHIs within a city (Chen et al., 2012).
The integration of remote sensing (RS), geographic information
system (GIS), and ecological factors was used to investigate the
relationships between UHIs and socioeconomic factors (e.g. Chen
et al., 2012 and Anderson et al., 2010). To investigate the variation
of UHI and its effect on socioeconomic data, the LST presented in
Fig. 3 are normalized to two levels. The first level classifies the
areas of each of Suez districts into places that experience UHI it
given the value one and another with no UHI gave the value zero.
Data distributed over dates 1988, 2011, and 2014 have been
selected for further analysis. This data set is used for evaluation
of UHI changes over years. Aster data is used to take the advantage
that the high resolution allows the study of land-cover tempera-
ture interaction at the district scale. The areas in km2 of places
experience UHI are calculated using overlay module; it is shown
in the Table 10. The weights of these values are calculated by divid-
ing the area of UHI over the total district area. From this table it can
be seen that the Suez and Al-Arbin districts which are the most
populated zones in Suez governorate. They are very high density
built-up with rare vegetation zone experiences a huge problem
of UHI. It experiences a UHI phenomenon represents total weight
of 40.84% and 46.3% respectively. That give an indication about
how the population may affect by this phenomena. At the same
time Al-Ganayin district which is the densely vegetated with med-
Fig. 4. Classification of UHFVI over study area, (a): date 2014 and (b) date 1988.
ium density built up zone experience relatively moderate phenom-
ena. As a mean of vegetation has cool effect on UHI and

Table 11
Distribution of UTFVI areas over Suez districts - year 2014.

District Area in Km2 Number of families Urban thermal field variance index (area in km2)
None Weak middle strong stronger strongest
Al-Ganayin 1220.042355 19962 188.9 583.85 232.2 155.69 44.45 14.95
Faysal 1170.109075 31178 369.71 498.48 227.45 59.18 15.2 0.09
Al-Suez 70.938479 19962 22.33 3.7 7.6 12.33 12.48 12.5
Al-Arbin 44.834302 51829 9.57 11.49 10.41 7.56 3.99 1.81
Ataqa 6940.675667 6298 3562.7 1404.59 1211.04 508.32 244.59 9.43
S. Ahmed / Egypt. J. Remote Sensing Space Sci. 21 (2018) 15–25 23

Table 12
Distribution of UTFVI areas over Suez districts- date 1988.

District Area in Km2 Number of families Urban thermal field variance index
None Weak middle strong stronger strongest
Al-Ganayin 1220.042355 19962 732.14 227.98 200.49 50.54 8.85 0.04
Faysal 1170.109075 31178 801.66 290.69 73.8 3.82 0.14 0
Al-Suez 70.938479 19962 10.68 18.53 27.73 12.05 1.87 0.08
Al-Arbin 44.834302 51829 1.002 4.88 25.72 12.15 1.01 0.09
Ataqa 6940.675667 6298 4181.91 1789.56 797.92 102.97 52.6 15.72

temperature rise. Ataqa district experience relatively low UHI phe-


nomena where the district area is widely dispersed, it is the biggest
district of Suez governorate. The district area in considered rural
area, it is an industrial zone with low-density built-up area; most
of the district area are quarries.
With the aid of image processing software, the mean tempera-
ture of UHI areas of each district has been calculated. It is found
that the mean temperature of UHI areas of Al-Arbin and Al-Suez
districts was 7.5 above mean of District temperature. While the
value of UHI of suburban zone e.g. Ataqa and AL-Ganayin was 2.5
degree above district mean.
The second level of LST normalization is the ecological evalua-
tion of UHI by calculation of UTFVI at district level. Fig. 4 displays
the resulted UHFVI of time 1988 and 2014. From the fig, it is clearly
shown that area had yellow color and classified as moderate UHI
mostly transformed into dark green, red and cyan colors which
represent strong to strongest UHI. Areas with blue color which rep-
resents areas with no UHI transformed into another higher UHI. It
is a good indication of temperature increase and the areas that
have UHI more than moderate considered as areas vulnerable to
temperature rise.
Areas in km2 of the distribution of UTFVI over each of Suez dis-
trict is calculated and shown in Table 11 for the date 2014 and
Table 12 for year 1988. We can realize that the total areas of strong,
stronger and strongest are relatively equal to the areas segmented
as UHI (Table 10). Comparing Fig. 4a and Fig. 5b, it can be seen that
the strongest distribution of UTFVI in 2014 were concentrated in
the urban area and partly in developed areas.

4.3. LULC change and its association with LST variation

Urbanization is one of the most active forces that derive Land-


use/land cover changes and impact climate warming (Lal et al.,
2017; Mosammam et al., 2017; Xiong et al., 2012; Gong et al.,
2011; Hu, 2007; Kalnay and Cai, 2003). It is possible to use remote
sensing techniques to reveal the degree of urbanization and LULC
changes of the study area. Taking into consideration the objective
of this study, 6 classes have been generated. Fig. 5 displays the
retrieved LULC of the study area. From this fig, it can be seen that
there is a large area of reclaimed lands appears the year 2014
which was rural areas in 1988. Most of the reclaimed areas are
transformed to urban or vegetated areas. There is increase of urban
areas (yellow color) and vegetated areas (red color) as well.
Changes as well as gains and losses in each class between years
1988 and 2014 are documented in the Table 13.
From this table it can be determined the LULC pattern substan-
tially changed from 1988 to 2014 and the urban area expands
rapidly. The results also indicate the growth of urban areas which
was 258.87 km2 in 1988 and become 642.25 in 2014. The extent of
the urban area increased 383.38 km2 in 26 years. During this per-
iod, the urban area is duplicated. This is enhancing the urbaniza-
tion trend of the study area. Rural lands present areas under
change. It shows significant decrease from 1988 to 2014. Most of
these area are reclaimed (1393 km2), vegetated (78 km2), and a Fig. 5. Classified image of the study area, (a): 2014, (b): 1988.
24 S. Ahmed / Egypt. J. Remote Sensing Space Sci. 21 (2018) 15–25

Table 13
Cross tabulation summary of gain and loss of LULC from 1988 to 2014, (areas in km2, and percentage).

Final case 2014 Initial case 1988


Water Urban Vegetation Reclaimed area Rural area (1)&(2) Total
Km2 % Km2 % Km2 % Km2 % Km2 % Km2 %
Water 1365.88 99.55 0.02 0.01 0.15 0.41 11.27 1.03 2.8 0.08 1380.12 100
Urban 5.32 0.39 256.99 98.89 3.31 9.13 220.78 20.07 155.85 4.43 642.25 100
Vegetation 0.18 0.01 0.45 0.17 32.42 89.43 49.54 4.52 77.56 2.21 160.15 100
Reclaimed area 0.03 0.002 0.71 0.27 0.0 0 811.17 73.99 1393.1 39.63 2205.01 100
Rural area (1) & (2) 0.72 0.05 0.71 0.27 0.38 1.05 3.61 0.33 1885.97 53.66 1891.39 100
Total class 1372 100 258.87 100 36.25 100 1096.37 100 3514.87 100
Class change 6.25 0.452 1.88 0.72 3.83 10.59 285.2 25.95 1628.9 46.35

Table 14
Main statistical values of LST for each LULC types for selected years 1988 and 2014.

LULC Type Temperature range, Mean, SD and autocorrelation C°/year


1988 2014
Range Mean SD Auto-corre Range Mean SD Auto-corre
Water 16.2–18 17 6.87 0.99 14.5–18 16.35 6.89 0.99
Urban 15–24 19.7 3.73 0.84 18–29.39 24.2 4.45 0.61
Vegetation 16.5–21.76 19.13 1.36 0.85 17–25 21 3 0.71
Reclaimed areas 11–16 13.5 5.15 0.76 15–24 21 8.8 0.77
Rural area (1) 15–19 17 7.31 0.81 12–29.38 20.5 8.2 0.71
Rural area (2) 15–21 16.5 7.37 0.77 18–24 21 7.81 0.73

significant part (about 156 km2) are transformed to urban areas. affecting LST of each LULC are different. An overall increase of
comparing Fig. 3 and Fig. 5 it could be indicate that the spatial pat- LST is indicated as well. Development
tern of LST increase (reddish tone, Fig. 3) actually highlighted in
areas that reclaimed to urban (yellow tone, Fig. 5). All percentage 5. Conclusion
of areas under change of each category is shown in Table 13.
Statistical analysis has been carried out in order to quantita- Studies of temperature and land cover interaction provide
tively document the association between LST and different LULC. valuable insights for the analysis of urban environment as well
Many authors have documented this association using various sta- as assistance in various decision-making processes for the city
tistical methods (e.g. Li et al., 2014 and Pal and Ziaul, 2017). In this planning and development. Remote sensing is a valuable data
research the relation between different LULC wise LST variations as source to accomplish this objective, and a multi-sensor approach
well as changes of LST with the association of LULC types from date meets the temporal and spatial requirements of frequent and
1988 to 2014 were documented. The following analysis was used highly detailed observations from visible to thermal infrared part
to test the variation of LST and summarize the main statistical of the spectrum. In this study, satellite data and socio-economic
characteristics of LST among each of LULC type. In the beginning, data have been analyzed to investigate the land cover and tem-
each of the 6 obtained LULC types was separated in an individual perature interactions, which begun increase due to emerging
layer for the selected dates. Then, with the aid of ArcGIS modules, economic and climate change, and the effect of human of Suez
each of these individual layers was associated with LST layer. Governorate for the years 1988–2014. The results indicate that
Finally, basic statistical characteristics (minimum, maximum, changes in UTFVI distribution can be predominantly related to
mean, standard deviations) are calculated as well as spatial auto- the expansion of urban area during the study time period.
correlation (auto-corr)) by using global Moran’s I. Table 14 summa- Strong, stronger and strongest UTFVI shown in dark green, red
rizes the main statistical values of LST of each LULC types for and cyan respectively (Fig. 4), were mainly distributed in the
selected years 1988 and 2014. urban area (yellow color, Fig. 5). The hot spot of UTFVI were
From this table it can be illustrating that for selected years all found mainly in the built-up areas especially densely populated
LULC types except water gain higher LST. Urban area experienced district and in heavily industrial district. It is the vulnerable
increase of LST mean by 4.5 °C, while vegetated areas gain 2 °C areas to UHI. The mean temperature varies from 7.5 degree
above the LST mean from 1988 to 2014. Reclaimed areas exhibits above mean in the densely populated zone to 2.5 degree in sub-
highest LST increase, 7.5 °C above mean with increase of SD, min- urban/ rural zone. The LST, NDVI, NDBI together can be used to
imum and maximum ranges which gives an indication of variation describe the spatial distribution of LULC depending on the corre-
of land natures, i.e. mixed LULC types as a results of development lation among them. Variation of UHI of Suez districts form years
of lands into built up, industrial, roads and vegetated areas. This 1988 to 2011 and 2014, indicate that it can predict the uncon-
implies that the newly development land has gained more temper- trollable gigantic increase of the phenomenon through the
ature. Rural areas (1), it is observed that areas accumulated maxi- incoming years. There are dramatic needs of controlling the rea-
mum LST range variations it reached range from 12 to 29.38 °C but son of temperature rise such as greenhouse gases, urban sprawl
the mean increase with 3 °C. Rural areas (1) experiences homoge- and construction of industries inside cities, looking for human
neous LST increase about 3 °C above mean this due to land nature. comfort. Highlighting more, we are mercilessly hurting the envi-
From this table it can be implies that the mean values of LST vary ronment and then, screams, ‘‘Why are not we living in a healthy
among different land use types and suggesting that the factors surrounding?”
S. Ahmed / Egypt. J. Remote Sensing Space Sci. 21 (2018) 15–25 25

References Li, X., Li, W., Middel, A., Harlan, S.L., Brazel, A.J., Turner II, B.L., 2016. Remote sensing
of the surface urban heat island and land architecture in Phoenix, Arizona:
Combined effects of land composition and configuration and cadastral–
Abutaleb, K., Ngie, A., Darwish, A., Ahmed, M., Arafat, S., Ahmed, F., 2015.
demographic–economic factors. Remote Sens. Environ. 174, 233–243.
Assessment of Urban Heat Island using remote sensed imagery over greater
Lillesand, T.M., Kiefer, R.W., Chipman, J.W., 2004. Remote Sensing and Image
Cairo, Egypt. Adv. Remote Sens. 4, 35–47.
Interpretation. John Wiley & Sons Inc, p. 763.
Ahmed, S., Kaiser, M.F., 2014. Geo-Environmental Assessment of the Suez Canal
Liu, L., Zhang, Y., 2011. Urban heat island analysis using the Landsat TM data and
Area, using remote sensing and GIS Techniques. IEEE International Geoscience
ASTER data: A case study in Hong Kong. Remote Sens. 3, 1535–1552.
and Remote Sensing Symposium, pp: 1662–1665.
Loyd, C., 2013. Putting Landsat 8’s Bands to Work, published on June 14, 2013.
Ahmed, S., Kaiser, M.F., 2014b. Monitoring water pollution of Lake Maryout on the
Mosammam, H.M., Nia, J.T., Khani, H., Teymouri, A., Kazemi, M., 2017. Monitoring
Mediterranean coast of Egypt. Int. J. Remote Sens. Appl. 4 (1), 36–40.
land use change and measuring urban sprawl based on its spatial forms: the
Al Kuwari, N.Y., Ahmed, S., Kaiser, M.F., 2016. Optimal satellite sensor selection
case of Qom city. Egypt. J. Remote Sensing Space Sci. 20 (1), 103–116.
utilized to monitor the impact of urban sprawl on the thermal environment in
Mundia, C.N., Aniya, M., 2005. Analysis of land use/cover changes and urban
Doha City, Qatar. J. Earth Sci. Clim. Change 7, 1.
expansion of Nairobi city using remote sensing and GIS. Int. J. Remote Sens. 26
Anbar, M., 2012. The Climate impact on human comfort in the eastern Nile Delta.
(13), 2831–2849.
Journal of Faculty of Art, Cairo University, Vol. (72) no. 7, pp. 267–319. (In
Pal, S., Ziaul, S., 2017. Detection of land use and land cover change and land surface
Arabic).
temperature in English Bazar urban centre. Egypt. J. Remote Sensing Space Sci.
Anderson, S.J., Tuttle, B.T., Powell, R.L., Sutton, P.C., 2010. Characterizing
20 (1), 125–145.
relationships between population density and nighttime imagery for Denver,
Pu, R., Gong, P., Michishita, R., Sasagawa, T., 2006. Assessment of multi-resolution
Colorado: issues of scale and representation. Int. J. Remote Sens. 31, 5733–5746.
and multi-sensor data for urban surface temperature retrieval. Remote Sens.
Bhatta, B., 2010. Analysis of Urban Growth and Sprawl From Remote Sensing Data.
Environ. 104, 211–225.
Springer.
Rawat, J.S., Kumar, M., 2015. Monitoring land use/cover change using remote
Chander, G., Markham, B., 2003. Revised Landsat-5 TM radiometric calibration
sensing and GIS techniques: a case study of Hawalbagh block, district Almora,
procedures and post calibration dynamic ranges. IEEE Trans. Geosci. Remote
Uttarakhand, India. Egypt. J. Remote Sens. Space Sci. 18, 77–84.
Sens. 41 (11), 2674–2677.
Seto, K.C., Woodcock, C.E., Song, C., Huang, X., Lu, J., Kaufmann, R.K., 2002.
Chen, Z., Gong, C., Wu, J., Yu, S., 2012. The influence of socioeconomic and
Monitoring land-use change in the Pearl River delta using Landsat TM. Int. J.
topographic factors on nocturnal urban heat islands: a case study in Shenzhen,
Remote Sens. 23, 1985–2004.
China. Int. J. Remote Sens. 33 (12), 3834–3849.
Sims, D., 2015. Egypt’s desert dreams development or disaster. Published by The
Chen, X.L., Zhao, H.M., Li, P.X., Yin, Z.Y., 2006. Remote sensing image-based analysis
American University in Cairo press. Ver. (1).
of the relationship between urban heat island and land use/cover changes.
Stathopoulou, M., synnefa, A., Cartalis, C., Santamouris, M., Karlessi, T., Akbari, H.,
Remote Sens. Environ. 104, 133–146.
2009. A surface heat island study of Athens using high resolution satellite
Eastman, J.R., 1997. Supervised Classification in IDRISI for Windows Version 2,
imagery and measurements of the optical and thermal properties of commonly
Tutorial Exercises. Clark University, Worcester, Massachusetts, pp. 86–94.
used building and paving materials. Int. J. Sustain. Energy 28 (1–3), 59–76.
Gillespie, A., Rokugawa, S., Matsunaga, T., Cothern, J.S., Hook, S., Kahle, A.B., 1998. A
Streutker, D.R., 2003. A remote sensing study of the urban heat island of Houston,
temperature and emissivity separation algorithm for advanced spaceborne
Texas. Int. J. Remote Sens. 23, 2595–2608.
thermal emission and reflection radiometer (ASTER) images. IEEE Trans. Geosci.
The Egyptian General Survey Authority, 2012. www.esa.gov.eg.
Remote Sens. 36 (4), 1113–1126.
The State Information Service, 2014. www.sis.gov.eg.
Giridharan, R., Kolokotroni, M., 2009. Urban heat island characteristics in London
USGS, Landsat 7 Thermal Band Calibration Update, 2013. USGS/EROS, available on
during winter. Sol. Energy 83, 1668–1682.
the USGS web site.
Gong, C., Chen, J., Yu, S., 2011. Spatiotemporal dynamics of urban forest conversion
Van De Griend, A.A., Owe, M., 1993. On the relationship between thermal emissivity
through model urbanization in Shenzhen, China. Int. J. Remote Sens. 32 (24),
and the normalized difference vegetation index for natural surfaces. Int. J. Rem.
9071–9092.
Sens. 14, 1119–1131.
Hu, Z.Y., 2007. Using NDVI differencing and temporal logic to enhance ISODATA
Van, T.T., Bao, H.D.X., 2010. Study of the impact of urban development on surface
classification in urban environments. GISci. Remote Sens. 44, 48–67.
temperature using remote sensing in Ho Chi Minh city, north Vietnam.
Jensen, J.R., 1986. Introductory Digital Image Processing. Prentice-Hall, New Jersey,
Geographical Res. 48 (1), 86–96.
p. 379.
Weng Q., and Larson R.C., 2005. 10 satellite remote sensing of urban heat islands:
Kalinowski, A., Oliver, S., 2004. ASTER Mineral Index Processing Manual Remote
current practice and prospects.
Sensing Applications Geoscience Australia October 2004.
Weng, Q., 2011. Remote sensing of impervious surfaces in the urban areas:
Kalnay, E., Cai, M., 2003. Impact of urbanization and land-use change on climate.
requirements, methods and trends. Remote Sens. Environ. 117, 34–49.
Nature 423, 528–531.
Xiong, Y., Huang, S., Chen, F., Ye, H., Wang, C., Zhu, C., 2012. The impacts of rapid
Khandelwal, S., Goyal, R., Kaul, N., Mathew, A., 2018. Assessment of land surface
urbanization on the thermal environment: A remote sensing study of
temperature variation due to change in elevation of area surrounding Jaipur,
Guangzhou, South China. Remote Sens. 4, 2033–2056.
India. Egypt. J. Remote Sensing Space Sci. 21 (1), 87–94.
Yue, W., Liu, y., Fan, P., Ye, x., Wu, c., 2012. Assessing spatial pattern of urban thermal
Kim, H.H., 1992. Urban heat island. Int. J. Remote Sens. 13, 2319–2336.
environment in Shanghai, China. Stoch. Env. Res. Risk Assess. 26, 899–911.
Lal, K., Kumar, D., Kuma, A., 2017. Spatio-temporal landscape modeling of urban
Zahran, M.A., 2010. Climate-Vegetation: Afro-Asian Mediterranean and Red Sea
growth patterns in Dhanbad Urban Agglomeration, India using geoinformatics
Coastal Lands. Springer.
techniques. Egypt. J. Remote Sensing Space Sci. 20 (1), 91–102.
Zhang, Y., 2006. Land surface temperature retrieval from CBERS-02 IRMSS thermal
Li, W., Wang, Y., Peng, J., Li, G., 2005. Landscape spatial changes associated with
infrared data and its applications in quantitative analysis of urban heat island
rapid urbanization in Shenzhen, China. Int. J. Sustainable Dev. World Ecol. 12
effect. J. Remote Sens. 10, 789–797.
(3), 314–325.
Zhou, D., Zhao, S., Liu S., Zhang L., and Zhu C., 2014. Surface urban heat island in
Li, W., Bai, Y., Chen, Q., He, K., Ji, X., Han, C., 2014. Discrepant impacts of land use and
China’s 32 major cities: Spatial patterns and drivers. Remote.
land cover on urban heat islands: A case study of Shanghai, China. Ecol. Ind. 47,
171–178.

Potrebbero piacerti anche