Introduction
A particular problem in urban areas is the increase in surface temperature due
to the
conversion of vegetated surfaces into asphalt roads, as well as residential, commercial
and industrial areas. Presently, climate change in the cities is occurring
through
anthropogenic activities and land use changes (Weng et al., 2004). The
atmospheric conditions of the urban areas, land surface temperature, warming,
evaporation and absorption of solar radiation may be changed through anthropogenic
changes. The study of surface temperature in the cities located in the arid
and semi-arid areas is necessary (Mallick et al., 2008), because high
temperature leads to energy consumption for cooling buildings, which is
economically very costly especially in the warm months of year (Santamouris
et al., 2001).
The study of soils requires an interdisciplinary approach involving
geologists, biologists, physicists, chemists, anthropologists, economists,
engineers, medical professionals, military professionals, sociologists and
even artists. The most important soil-related critical challenges for
research are biodiversity, energy security, climate change, ecosystem
services, food security, human health, land degradation and water security
(Brevik et al., 2015). Soils provide fundamental ecosystem services, and
management to change a soil process in support of one ecosystem service can
provide co-benefits to other services. Fundamental research is needed to
better understand the relationships between soils and the array of ecosystem
services. More knowledge about soil and agricultural systems is required to
utilize ecosystem services in order to protect and enhance soils in the long
term (Smith et al., 2015). Soil scientists believe that soil knowledge has
a key role in major global issues like food production, the loss of
biodiversity and the availability of water resources. Agronomic studies
show that land management strongly influences pollutant behavior in soils
because it influences the filter and buffer functions of soil (Mol and
Keesstra, 2012). Using modern soil information can improve the results of
studies related to food security, water scarcity, climate change,
biodiversity loss and health threats (Keesstra et al., 2016). Management
can help to improve the social, economic and environmental sustainability of
agricultural and food systems (Decock et al., 2015). Loss of plant species
diversity has important effects on the erosion resistance of slopes. The
protection and restoration of vegetated covers are necessary to minimize
soil erosion, which will not only contribute to greater safety in the most
densely populated areas of the world, but will also help maintain soil
fertility on pasture lands (Berendse et al., 2015).
Urban planning is often hindered by a lack of knowledge on how land use changes. In wet periods, saturation overland flows occur on urban and
agricultural soils. Hydrophilic urban and agricultural soils are
characterized by increased infiltration capacity during dry periods. In
contrast, urban soils remain mostly hydrophilic, and have relatively high
infiltration capacities, whereas on the agricultural sites a rise in soil
moisture leads to a decline in infiltration capacity, with soil saturation
in areas of shallow soils (Ferreira et al., 2015). The spatial and temporal
variability of overland flows and infiltration affect flow connectivity
depending on land use and soils. Understanding the mentioned concepts is
necessary for land management in order to improve urban planning to minimize
flood hazards in territory (Ferreira et al., 2015). Parameters such as water
repellence, soil moisture content, vegetation and litter surface soil cover
are important for overland flow generation. Urban land use changes affect
soil properties and processes. The complexity in hydrological properties and
processes of the land use changes as a result of the urban sprawl increases
the unpredictability. Urban sprawl has the potential to alter the
hydrological response and erosion processes in small catchments. Vegetation
induces soil water repellence, promoting discontinuous overland flow
processes, which affect soil moisture content. These have major implications
for sustainable urban planning (Ferreira et al., 2012). Urban agriculture is
a desirable land use for these spaces, but degraded soils are common. Urban
agriculture can be productive in vacant urban land and also can improve soil
quality at previously degraded sites (Beniston et al., 2016).
Remote sensing can be applied to detect and study various phenomena in Earth
sciences and natural resources. For example, forest fires can be simulated
using thermal remote sensing and normalized difference vegetation index (NDVI) indices derived from satellite
imagery. NDVI data are a suitable substitute for ground-based measurements
in post-fire runoff predictions (Van Eck et al., 2016). Vegetation indices
derived from the satellite data and remote sensing methods can be applied to
investigate dynamic performance of plantations in terms of multitemporal dry
biomass production (Zucca et al., 2015). Remote sensing instruments are key
players to study and map land surface temperature (LST) at temporal and
spatial scales (André et al., 2015). The LST indicator shows effects of
different types of phenomena and features in the electromagnetic energy
dispatch (Bingwei et al., 2015). Remote sensing methodology requires
less time and lower cost than field methods to investigate various phenomena
on the land surface (Niu et al., 2015). The advantages of using remote
sensing methodology are the repetitive and consistent coverage, high
resolution and evaluation of land surface characteristics (Owen et al.,
1998). Thermal infrared (TIR) data in remote sensing can help us obtain
quantitative information of surface temperature. Landsat imagery can be
applied for monitoring different types of land use in arid and semi-arid
regions (Baojuan et al., 2015). The Landsat Thematic Mapper (TM) and ETM+ (Enhanced Thematic Mapper Plus) sensor images can
be used to study relationship between surface temperature and land use types
using thermal quantitative indicators (Weng, 2003; Streutker, 2003). Land
surface temperature can be retrieved using data from NOAA-11 (National Oceanic and Atmospheric Administration) AVHRR (Advanced Very High Resolution Radiometer) channels 4 and 5 by emissivity calculation (France and Cracknell, 1994).
The using of LST values, which vary according to the surface characteristics,
is a new method for investigating the effects of different land surface
features on the surface temperature especially in urban areas (Guanhua et
al., 2015). Derivation of LST from medium to high
spatial resolution data of remote sensing is very important to study climate
change and environment (Juan et al., 2014). In the several studies the
relative warmth of cities was estimated by knowing air temperature and land
use changes. The LST index provides important information about climate and
the surface physical characteristics. Land use changes and anthropogenic
activities affect the environment and land surface temperature (Dehua
et al., 2012; Weng and Schubring, 2004). The estimation of the LST from the
radiative transfer equation, the mono-window and single-channel algorithms
can be used to retrieve the LST from thermal
infrared data of the TM sensor (José et al., 2004).
Yazd city location in Iran.
![]()
The NDVI is a good indicator for
identifying long-term changes in the vegetation covers and their status
(Baihua and Isabela, 2015). The NDVI calculation method using surface
emissivity can be applied to areas with different soil and vegetation types
and where the vegetation cover changes (Valor and Caselles, 1996). Therefore,
analysis of spatial variability of NDVI, surface temperature and the relationship
between these parameters is essential in environmental studies. Combined
study of NDVI, surface temperature and the temporal relation these two
parameters with land use changes can be used to investigate climate
change and global warming (Schultz and Halpert, 1993). Vegetation cover
change is the main factor that causes surface temperature changes. It
should be noted that increasing surface temperature may increase vegetation
cover density in the area, but only in areas where there are sufficient
water resources (Weixin et al., 2011). Different types of vegetation cover
have different spatial responses to climate changes (Dehua et al., 2012). In
the environmental studies, researchers have investigated land surface
temperature using vegetation indices (Wei et al., 2015). Analysis of NDVI
and LST of the different times (days, months, seasons and years) can be used
to detect land use changes, which were formed because of deforestation,
forest fires, mining activities, urban expansion and grassland regeneration
(Sandra et al., 2015). Changes in land use and land cover can be evaluated
by analysis of the vegetation cover and NDVI trends. The vegetation
phenology was detected using Terra MODIS NDVI data by Gong et al. (2015).
Land use changes, vegetation cover and soil moisture have strong effects on
the land surface temperature; therefore, surface temperature can be applied
to study land use changes, urbanization and desertification. Surface
emissivity calculation is important to estimate surface temperature. In the
several studies, laboratory measurements of the emissivity data were used
for estimation of land surface temperature (Salisbury and D'Aria, 1992, 1994).
In the present study, heterogeneous surface temperature and NDVI of Yazd in
Iran were calculated using Landsat TM sensor data. Surface temperature
variation over different land use types in the Yazd are investigated and
the relationship between NDVI and land surface temperature are analyzed. The
aims of this study are to apply temperature–emissivity separation (TES) algorithm for LST retrieving from Landsat TM thermal data and to analyze the
NDVI, different land use types and their roles in the surface
temperature change. The main advantages of this study are TES algorithm
(calculating emissivity) for retrieving LST, full statistical analysis for
results validation and simultaneous analysis of NDVI, LST and land use. The
main difference between this study and previous studies is to investigate
the relationship between NDVI, LST and land use changes, and analysis of the
effect of the land use changes on land surface temperature.
Material and methods
Study area
Yazd has been chosen as the study area, since there is a combination of
different land use categories and it is susceptible to dust storms. Yazd is
located in 31∘47′37′′–31∘57′56′′ N latitude and
54∘13′28′′–54∘27′10′′ E longitude in the Iran. This
city has an altitude of 1230 m and covers an area of
2491 km2
(Fig. 1). The study area is located in the arid and semi-arid belt in the
Northern Hemisphere. In the arid and semi-arid areas, vegetation covers are
affected by high diurnal and seasonal variations of temperature, low amounts
of precipitation and high evaporation (Dehghan, 2011). The predominant
features of the territory are residential areas, waste land and bare soil.
Land surface temperature in Yazd is affected by a warm, arid and semi-arid
climate, low precipitation and the remoteness of major water resources, such as
the Caspian Sea, Persian Gulf and Oman Sea.
Satellite data processing and methodology
The Landsat TM sensor data of 8 August 1998 and 6 August 2009 (daytime) were used in the present study. A surface emissivity calculation is the
first step of land surface temperature retrieving by the TES algorithm. The emissivity per pixel was obtained directly
from Landsat TM sensor data. Natural surfaces at the resolution of 30 m
are heterogeneous and they differ from each other in their emissivity. In
addition, the surface emissivity is affected by surface roughness,
vegetation cover and different land use types. In the present study, surface
emissivity was evaluated by analysis of NDVI and the fraction of
vegetation cover per pixel. Emissivity is a quantification of the intrinsic
ability of a surface in converting heat energy into above surface radiation
and depends on the physical properties of the surface and on observation
conditions (Sobrino et al., 2001). Surface emissivity can be extracted using
NDVI values of the bare soil, full vegetation and a mixture of bare soil and
vegetation (Sobrino et al., 2004). In this study, the following equation was
used to extract land surface emissivity for each pixel:
ε=εv⋅Pv+εs⋅1-Pv+dε,
where ε is the LSE (Land Surface Emissivity), Pv is the proportion of vegetation,
εv is vegetation emissivity (0.99) and εs is soil
emissivity (0.97). The term dε shows the geometric distribution
effect of natural surfaces and their internal reflection. This term
(dε) for our study area was not considered because it is
negligible for surfaces with little height difference. The proportion of
vegetation (Pv) is calculated with the following equation:
Pv=NDVI-NDVIminNDVImax-NDVImin2.
NDVI is obtained from spectral reflectance measurements in the visible (RED) and near-infrared regions (NIR) in the ArcGIS environment by
NDVI=NIR-REDNIR+RED.
The NDVI was calculated for two different time periods (1998 and 2009) to
investigate the status of vegetation cover for the study area. High values of
NDVI indicate dense and healthy vegetation. This method needs elementary
knowledge of emissivity and NDVI for the different features and land use
types.
The at-sensor spectral radiance is the amount of energy received by the
satellite sensor. Calculation of spectral radiance is the fundamental step
in converting satellite image data into a physically radiometric scale.
Radiometric calibration of the Landsat TM sensor involves rescaling the raw
digital numbers of the satellite image to calibrated digital numbers. The
pixel values of unprocessed image data were converted to spectral radiance
by radiometric calibration.
TM spectral range and post-calibration dynamic ranges.
Band
Spectral
Center
LMIN
LMAX
Grescale
Brescale
number
range (µm)
wavelength
(w/(m2 ⋅ sr ⋅ µm))
(w/(m2 ⋅ sr ⋅ µm))
((w/(m2 ⋅ sr ⋅ µm))/DN)
(w/(m2 ⋅ sr ⋅ µm))
1
0.452–0.518
0.485
-1.52
169
0.671339
-2.19
-1.52
193
0.765827
-2.29
2
0.528–0.609
0.569
-2.84
333
1.322205
-4.16
-2.84
365
1.448189
-4.29
3
0.626–0.693
0.660
-1.17
264
1.043976
-2.21
4
0.776–0.904
0.840
-1.51
221
0.876024
-2.39
5
1.567–1.784
1.676
-0.37
30.2
0.120354
-0.49
6
10.45–12.42
11.435
1.2378
15.3032
0.055376
1.18
7
2.097–2.349
2.223
-0.15
16.5
0.065551
-0.22
Spectral radiance (Lλ) at the sensor's aperture in watts/(m2 ⋅ ster ⋅ µm) is provided with the following equation:
Lλ+Grescale⋅QCAL+Brescale
Spectral radiance is also expressed as
Lλ=LMAX-LMINQCALMAX-QCALMIN⋅QCAL-QCALMIN+LMIN,
where QCAL is the quantized calibrated pixel value in DN (Digital Number), Grescale is
band-specific rescaling gain factor in (watts/(m2 ⋅ ster ⋅ µm))/DN, Brescale is band-specific rescaling bias factor in watts/(m2 ⋅ ster ⋅ µm)), LMIN is the spectral radiance that is scaled to
QCALMIN in watts/(m2 ⋅ ster ⋅ µm), LMAX is the spectral
radiance that is scaled to QCALMAX in watts/(m2 ⋅ ster ⋅ µm),
QCALMIN is the minimum quantized calibrated pixel value (corresponding to
LMIN) in DN and QCALMAX is the maximum quantized calibrated pixel value
(corresponding to LMAX) in DN.
Table 1 summarize the TM spectral range, post-calibration dynamic ranges
(LMIN and LMAX scaling parameters, the corresponding rescaling gain
(Grescale) and rescaling bias (Brescale) values). QCALMIN = 0 for data
processed using NLAPS (National Landsat Archive Production System) and QCALMIN = 1 for data processed using LPGS (Level 1 Product Generation System).
Thermal band data (band 6 on TM) can be converted from spectral radiance to
an
effective brightness temperature. The brightness temperature assumes that
the Earth's surface is a black body (spectral emissivity of the black body
is 1). Thermal radiance values were converted from spectral radiance to
brightness temperature using the thermal constants with the following equation:
T=K2lnK1Lλ+1,
where T= satellite brightness temperature (Kelvin), Lλ= TOA (Top of Atmosphere) spectral radiance, K1= calibration constant 1 from the
metadata and
K2= calibration constant 2 from the metadata (Table 2).
TM thermal band calibration constants.
Constant 1 – K1
Constant 2 – K2 (Kelvin)
(w/(m2 ⋅ ster ⋅ µm))
Landsat 5
607.76
1260.56
Since brightness temperature (T) is a blackbody temperature, the final step
is the spectral emissivity according to the nature of the surface by
temperature correction (Weng et al., 2004):
ST=TB1+λ⋅TBρ⋅Lnε,
where TB= satellite brightness temperature (Kelvin), λ= wavelength of emitted radiance (11.5 µm), ε= land
surface emissivity, ρ=h⋅cσ= 1.438 × 10-2 mK (σ= Boltzmann constant = 1.38 × 10-23 J K-1, h= Planck's
constant = 6.626 × 10-34 Js and
c= velocity of light = 2.998 × 108 m s-1).
Finally, derived land surface temperature in Kelvin was converted to Celsius
by subtracting from 273.15.
For processing satellite data and database building, the Gauss–Krueger
coordinate system was selected. The false color composite imagery of Landsat
TM data (8 August 1998 and 6 August 2009) covering the study area
was produced in the ArcGIS environment (Fig. 3). Land use classes were
selected in the false color composite imagery for supervised classification
of image. Different land use types were classified using maximum likelihood
classification in ArcGIS environment (Fig. 4). Following land
classification was accepted using the Landsat TM data: asphalt road, park
and green spaces, waste land and bare soil, fallow land, residential
(urban), commercial and industrial (Mallick et al., 2008). Classification
was performed on Landsat TM for spectral reparability of the land use
classes existing in the study area. Finally, the relationship between land
use classes, surface temperature and NDVI in Yazd was analyzed in detail.
Calibration and validation of Landsat TM LST and land use classification
The validation of LST by satellite products is
performed using ground-based measurements of surface temperature. In the
present study, ground-based data of land surface temperature received from
the Yazd Meteorological Bureau were used to validate the LST of the Landsat TM sensors.
Measuring land surface temperature was performed using thermometers with
SMT160 temperature sensor. The thermometers have a temperature range of -45
to 130 ∘C. The most important characteristics of the SMT160
temperature sensor include absolute accuracy ∓0.7 ∘C and
measurement range 175 ∘C. Geographical characteristics of the
measurement points can be seen in Table 3 and Fig. 2.
Coordinates of the land surface temperature measurement points and
their ground-based and satellite temperatures.
8 August 1998
6 August 2009
Measurement
Latitude
Longitude
Ground-based
Landsat TM
Ground-based
Landsat TM
point
measured
LST
measured
LST
temperature (∘C)
temperature (∘C)
1 (Yazd synoptic station)
31∘53′59.9994′′ N
54∘16′59.9982′′ E
31.1
32.06
33.8
34.5
2
31∘53′51.792′′ N
54∘14′20.833′′ E
38.95
39.93
40.56
41.45
3
31∘53′42.415′′ N
54∘15′52.937′′ E
38.97
39.55
41.09
42.21
4
31∘54′7.576′′ N
54∘15′53.826′′ E
40.56
41.07
41.89
42.59
5
31∘54′53.781′′ N
54∘15′55.217′′ E
40.15
41.45
41.5
42.21
6
31∘54′51.239′′ N
54∘17′33.628′′ E
41.02
41.83
42.09
42.96
7
31∘54′48.834′′ N
54∘19′5.99′′ E
36.75
37.22
39.8
40.69
8
31∘55′49.672′′ N
54∘19′8.177′′ E
36.24
37.61
39.18
39.93
9
31∘54′46.463′′ N
54∘20′44.16′′ E
34.72
35.26
35.2
36.05
10
31∘53′40.635′′ N
54∘23′49.896′′ E
38.49
39.93
38.87
39.16
11
31∘52′50.098′′ N
54∘22′15.626′′ E
34.37
34.46
35.88
37.22
Location of the land surface temperature measurement points.
![]()
False color composite (bands 2, 3 and 4) image of Landsat TM data
of Yazd city: (a) 8 August 1998, (b) 6 August 2009.
![]()
Land use classified image of Yazd city by maximum likelihood
classification: (a) 8 August 1998, (b) 6 August 2009.
![]()
Land use distribution of Yazd city using maximum likelihood
classification.
Land use
Image of the 8 August 1998
Image of the 6 August 2009
Difference between
images of 1998 and 2009
Study area
Study area
Study area
Study area
km2
%
(km2)
(%)
(km2)
(%)
Asphalt roads
13.586029
6.9
18.358887
9.33
4.772858
2.43
Parks and green spaces
6.558943
3.33
4.935898
2.51
-1.623045
-0.82
Waste land and bare soil
77.454838
39.34
42.078257
21.37
-35.376581
-17.97
Fallow land
38.03299
19.32
28.8658
14.66
-9.16719
-4.66
Residential (urban)
59.359028
30.15
99.858542
50.72
40.499514
20.57
Commercial and industrial
1.896892
0.96
2.779129
1.41
0.882237
0.45
Statistical analysis
Validation of the obtained temperatures from 8 August 1998 and
6 August 2009 images was performed in the Excel software using the
root mean square error (RMSE) and coefficient of determination (R2)
between the measured (ground-based) and predicted (by satellite data) land
surface temperatures (Xiaolei et al., 2014). For the investigated area, accuracy
assessment of land surface temperatures retrieved from the TES algorithm was
performed by RMSE and R2 parameters (Fig. 9 and Table 8). The
statistical analysis of the relationship between NDVI, surface
temperature and land use type changes was performed (Fig. 10 and Tables 9,
10).
The Kappa coefficient and overall accuracy were used to validate the land
use classification. The error matrix is used as an analytical statistical
technique. The simplest descriptive statistical indicator is overall
accuracy, which can be calculated using error matrix. Producer's accuracy is
the probability that a pixel in the classified image is placed in the same
class on land. In the calculation, the correct pixel total in each class is
divided by the pixel total of that class as derived from the reference
data. The producer's accuracy indicates the probability of a reference pixel
being correctly classified. In the user's accuracy the correct pixels total
in a land use class is divided by the total number of pixels that were
classified in that class. The user's accuracy is indicative of the
probability that a pixel classified on the image actually represents that
category on the ground (Story and Congalton, 1986). User's accuracy is the
probability that a specific class of land was classified in the same class
on the classified image.
Results
Analysis of land use
By classifying land uses of the study area, six land use types for the study area
were considered: asphalt roads, parks and green spaces, waste land and bare
soil, fallow land, residential, commercial and industrial areas. The land
use distribution in Yazd is described in Table 4. By comparing area
percentage values of different land use classes between images, it can be
concluded that land use types of the study area were significantly converted in
the 11-year period. Table 4 shows that the asphalt roads, commercial,
industrial and residential (urban) classes of land use were increased.
However, parks, green spaces, fallow lands, waste lands and bare soils were
decreased during the time. The study results show that residential areas
increased most compared to the other classes of land use. The main changes
include conversion of parks, green spaces, bare soils and fallow lands to
residential areas.
For example in the 1998 image, by comparing the reference and classified data in
the Table 5, it was observed that although 100 % of the asphalt roads were
being correctly identified as asphalt roads, only 90.9 % of the areas
called asphalt roads are actually asphalt roads.
And in the 2009 image by comparing the reference and classified data, it was
observed that 95.24 % of the asphalt roads were being correctly identified
as asphalt roads, while 86.96 % of the areas called asphalt roads are
actually asphalt roads (Table 6). The Kappa coefficient values of the 1998
and 2009 classified images were 0.96 and 0.95, respectively. The error
matrixes indicate an overall accuracy of 0.97 and 0.95 for the 1998 and 2009
classified images, respectively. These values were obtained from the analysis of the
referenced data (ground-based data) and the classification output (Tables 5 and 6 and Figs. 5 and 6).
Error matrix used to assess the accuracy of a classification of the
8 August 1998 Landsat TM image.
Reference data
Land use types
Asphalt
Park and
Waste land
Fallow
Residential
Commercial
Total
User's
road
green space
and bare soil
land
(urban)
and industrial
accuracy (%)
Classified data
Asphalt road
40
0
0
0
0
4
44
90.9
Park and green space
0
43
0
1
0
0
44
97.72
Waste land and bare soil
0
0
51
0
3
0
54
94.44
Fallow land
0
0
0
42
0
0
42
100
Residential (urban)
0
0
2
0
84
0
86
97.67
Commercial and industrial
0
0
0
0
0
13
13
100
Total
40
43
53
43
87
17
283
–
Producer's accuracy (%)
100
100
96.23
97.67
96.55
76.47
–
–
Error matrix used to assess the accuracy of a classification of the
6 August 2009 Landsat TM image.
Reference data
Land use types
Asphalt
Park and
Waste land
Fallow
Residential
Commercial
Total
User's
road
green space
and bare soil
land
(urban)
and industrial
accuracy (%)
Classified data
Asphalt road
20
0
0
0
2
1
23
86.96
Park and green space
0
21
0
0
0
0
21
100
Waste land and bare soil
0
0
26
0
0
0
26
100
Fallow land
0
2
0
22
0
0
24
91.67
Residential (urban)
1
0
2
0
42
0
45
93.33
Commercial and industrial
0
0
0
0
0
16
16
100
Total
21
23
28
22
44
17
155
–
Producer's accuracy (%)
95.24
91.3
92.86
100
95.45
94.12
–
–
Land use classes User's and producer's accuracy of the 8 August 1998 Landsat TM image.
![]()
Land use classes User's and producer's accuracy of the 6 August 2009 Landsat TM image.
![]()
Spatial distribution of NDVI obtained from Landsat TM data:
(a) 8 August 1998, (b) 6 August 2009.
![]()
Land surface temperature retrieved from Landsat TM TIRS data at
urban area of Yazd city: (a) 8 August 1998, (b) 6 August 2009.
![]()
Analysis of vegetation situation and NDVI
The spatial distribution of NDVI values from the Landsat TM image can be
seen in Fig. 7. The 1998 NDVI values are in the range of -0.2 to 0.59,
having a mean value of 0.195, and the 2009 NDVI values are in the range of
-0.17 to 0.41, having a mean value of 0.12 (Fig. 7). In the figure it is
shown that low values of NDVI (light green area) correspond to waste land,
bare soil, commercial, industrial and residential areas on the different
parts mainly in the southern and western sections of the study area. High
values of NDVI (dark green) that were observed in the central, north and
southwest parts of the images correspond to parks and green spaces. The
medium NDVI values were observed over fallow land and asphalt roads, in the
central, north and southwest parts of the study area. By comparing NDVI of
two different time periods (1998 and 2009) we concluded that NDVI values decreased
over the studied period of time. The maximum values of derived emissivity
are observed over parks and green spaces. The emissivity values of parks and
green spaces for the year 1998 image are from 0.987 to 0.99. In addition,
emissivity values of the fallow land class are found in the range of 0.986
to 0.987. Asphalt roads, commercial and industrial, residential, waste land
and bare soil have very similar emissivity values from 0.986 to 0.9863. The
emissivity values of parks and green spaces in the year 2009 image are from
0.982 to 0.99, emissivity values of the fallow land class are from 0.979 to
0.982 and emissivity values of the asphalt roads, commercial and industrial,
residential, waste land and bare soil are from 0.97 to 0.979. Wheat and
barley are mainly grown in the agricultural lands of the study area (seasonal
plants), which in the performed classification are placed on the fallow land
class.
Linear relationship between ground-based measured temperature and
land surface temperature (LST) of Landsat TM sensor: (a) 8 August
1998, (b) 6 August 2009.
![]()
Land surface temperature of different land use categories of Yazd
city.
Min. temperature ( ∘C)
Max. temperature ( ∘C)
Mean ( ∘C)
Land use and land cover
8 August 1998
6 August 2009
8 August 1998
6 August 2009
8 August 1998
6 August 2009
Asphalt roads
29.6
31.24
44.08
45.57
36.84
38.41
Parks and green spaces
27.1
30.42
41.83
42.96
34.47
36.69
Waste land and bare soil
31.65
33.66
45.57
44.46
38.61
39.06
Fallow land
27.94
30.42
44.83
44.46
36.39
37.44
Residential (urban)
31.24
30.01
45.2
45.2
38.22
37.61
Commercial and industrial
31.24
34.46
42.21
45.2
36.73
39.83
Accuracy assessment of land surface temperature derived from
Landsat TM sensor data using calculation RMSE and R2.
LST Landsat TM sensor images
Root mean square error (RMSE)
Coefficient of determination (R2)
8 August 1998
0.9
0.98
6 August 2009
0.87
0.99
Land surface temperature analysis
In the present study, land surface temperature was retrieved by the TES algorithm from TIRS (Thermal Infrared Sensor) data of the Landsat TM. The spatial distribution of surface temperature of the
1998 and 2009 images is shown in the Fig. 8. Surface temperature of the
1998 LST image ranged from 27.1 to 45.57 ∘C (mean of
36.34 ∘C), and surface temperature of the 2009 LST image
ranged from 30.01 to 45.57 ∘C (mean of 37.79 ∘C). In the
Fig. 8 high surface temperatures are shown by the dark red areas, this
means that outlying parts of the city have a temperature higher than the central
part. The results indicated that the average temperature of Yazd increased from
36.34 to 37.79 ∘C. Developments in asphalt roads, residential and
commercial areas increased dramatically between 1998 and 2009, and
vegetation cover has been reduced. The combination of mentioned factors
caused an increase the overall temperature of the city.
In the 1998 classified image, bare soil and waste land (mean value
38.61 ∘C) and residential (mean value 38.22 ∘C) classes
have maximum values of surface temperature, and parks and green spaces have
minimum values (mean value 34.47 ∘C). In the 2009 images, bare soil
and waste land (mean value 39.06 ∘C), commercial and industrial
(mean value 39.83 ∘C) classes have maximum values of surface
temperature, and parks and green spaces have minimum values (mean value
36.69 ∘C) (Fig. 8 and Table 7).
Relationship between surface temperatures with NDVI of the different
land use categories.
Mean temperature ( ∘C)
NDVI value (mean)
Land use and land cover
8 August 1998
6 August 2009
8 August 1998
6 August 2009
Asphalt roads
36.84
38.41
0.085
0.023
Parks and green spaces
34.47
36.69
0.278
0.182
Waste land and bare soil
38.61
39.06
-0.058
-0.051
Fallow land
36.39
37.44
0.235
0.155
Residential (urban)
38.22
37.61
0.072
0.006
Commercial and industrial
36.73
39.83
-0.063
-0.086
For the investigated area, results show that the land surface temperatures
retrieved from TES algorithm using Landsat TM sensor data have a high
accuracy with a RMSE of 0.9 ∘C and 0.87 ∘C for 1998 and 2009
images, respectively. The R2 between the
measured and predicted temperatures is 0.98 and 0.99 for 1998 and 2009,
respectively (Fig. 9 and Table 8).
Linear relationship between NDVI values and land surface
temperature (LST) of Landsat TM sensor: (a) 8 August 1998,
(b) 6 August 2009.
![]()
Variance analysis of relationship between land surface temperature
and normalized difference vegetation index (NDVI) retrieved from Landsat TM
data.
Landsat TM data date
Source
Sum of squares
df
Mean square
P level
8 August 1998
Between groups
0.510
5
0.102
0.000
Within groups
0.000
24
0.000
Total
0.510
29
6 August 2009
Between groups
0.295
5
0.059
0.000
Within groups
0.000
24
0.000
Total
0.295
29
Statistical analysis of the relationship between land surface
temperature, normalized difference vegetation index (NDVI) and land use
changes
In the present study, a strong relation was observed between NDVI,
surface temperature and land use types changes (Fig. 10 and Tables 9 and 10).
The spatial variations of surface temperature are affected by the conversion
of land for human-dominated use (land use change) and vegetation cover.
There is a linear regression between surface temperature and NDVI (Fig. 10).
The significant differences less than 0.01 between NDVI values and
temperatures of land use types for the 1998 and 2009 cases were obtained through
statistical analysis of NDVI values and surface temperatures. According to the
mentioned significant difference can be concluded that with a probability of 99 %
a decrease in the NDVI values caused to increase temperature (Table 10).
Discussions
This paper proposed the TES algorithm to obtain LST from Landsat TM and
change detection in land surface temperature in relation to land use change.
The spatial scales of TIRS pixel (100 m), land use and NDVI products (30 m)
can reduce accuracy of the results. The classified images show that the
agricultural lands were classified as fallow land, and it was also observed
that in some places of study fallow lands have been combined with waste
land and bare soil because of the similar values between their spectral reflectance.
As a result, the main factor of the conversion of land use types of study
area in the 11-year period is the human activities for urban growth. Based
on the results, the classified images have significantly good accuracy over
different land use types. Wheat and barley are mainly grown in the
agricultural lands of the study area (seasonal plants) that in the performed
classification are placed on the fallow land class because of similar values
of spectral reflectance of these products and fallow lands. It should be
noted that these agricultural products were harvested in May. As
mentioned in Sect. 3.2, the NDVI values decreased over the studied period
of time (1998–2009) because of plant-covered surface reduction. It should
be noted that the central part of the study area had a slight decrease in the
surface temperature. It was probably due to the design and implementation of
artificial green spaces and parks by managers in the city. Also, many
cooling devices, which are used in residential areas (especially in the
summer), were effective in creating this micro-climate. However, reduction of
vegetation cover and NDVI values and the consequent rise in temperature were
observed in the entire study area. A higher temperature of outlying parts
than the central part of city was due to keeping and increase parks and green
spaces in the central part and the destruction of vegetation in the outlying
parts of the city. Meanwhile, an increase in the overall temperature of Yazd
is due to different reasons, including vegetation loss and land use change in the
area.
It is generally observed that surface temperature has been increased in
all types of land use, but the greatest increase is registered in the
commercial and industrial sites. It should be noted that residential areas
had a slight decrease in the surface temperature. This is probably due to
many cooling devices, which are used in residential areas of the Yazd,
especially in the summer. The high R2
and low RMSE of satellite LST indicate that the study results have a high
accuracy. Values of the NDVI have an inverse relation to land surface
temperature. This means that a decrease in NDVI corresponds to an increase in
the temperature of land use types and vice versa. The strongest inverse
relationship between surface temperature and NDVI values was observed in the
fallow lands, parks and green spaces. By reducing vegetation density (in
this study waste land and bare soil), the inverse relationship between surface
temperature and the NDVI value was weaker. By increasing vegetation cover, the NDVI
values increased and surface temperature decreased. Also, an increase
in temperature and a decrease in NDVI values were observed by increasing
waste and barren lands in the area. Land surface temperature can be estimated
using a linear regression between surface temperature and NDVI, if NDVI
values are known with reasonable accuracy. According to the received
results, NDVI values were decreased during the study time. Land use type
changes were causing surface temperature increasing, and temperature
increasing subsequently was causing changes in NDVI and vegetation covers in
the study area. The decrease in vegetation cover and an increase in
residential areas are the main reasons of decreasing NDVI values of
Yazd. Land use gradual changes during the time are causing the temperature
change. In Yazd, surface temperature has increased as a result of
increases in asphalt roads, commercial, industrial and residential areas and
a decrease in parks, green spaces and fallow land classes in 2009 compared
to 1998. However, waste land and bare soil decreased, whereas
asphalt roads, commercial, industrial and residential areas increased, and
these changes caused a rise in temperature. Other studies also confirm that
the LST and NDVI changes are due to changes of the vegetation cover and
residential areas. Studies by Gong et al. (2015) and Valor and Caselles (1996)
showed that NDVI and LST values show trends in vegetation cover and
phenology changes. A study by Sandra et al. (2015) showed that land
degradation and regeneration areas can be investigated using NDVI satellite
data. NDVI correlates highly with tree canopy cover and other types of land
uses (Wei et al., 2015). Mallick et al. (2008) investigated LST index using
the vegetation abundance, and in their study, a strong correlation was
observed between surface temperatures with NDVI over different land use
types.