Land Surface Temperature Estimation

Using Remote Sensing Data

Vijay Solanky, Sangeeta Singh and S. K. Katiyar

Abstract India is facing crucial problem as the other world, by the impact of
Global warming which is resulting from climate change. Global warming is an
increase in Earth’s average surface temperature, land surface temperature
(LST) plays a critical role in Global climate change and is used in a range of
hydrological, meteorological, and climatological applications. As needed for most
modelling and climate analysis applications. Remote sensing (RS) and geographic
information system (GIS) have found wide application areas in climate change
analyses also this can be used for LST calculation. There are various sensor whose
data is useful in generating LST such as: advanced very high resolution radiometer
(AVHRR), moderate resolution imaging spectroradiometer (MODIS), Landsat-8,
and many more. LST can be estimated by using many algorithms i.e. split-window
(SW), dual-angle (DA), single-channel (SC). In this paper we have used SW
algorithm to estimate LST for Bhopal region using LANDSAT-8 Thermal Infrared
bands. SW algorithm uses spectral radiance and emissivity of TIR bands as input
for deriving LST. The spectral radiance was estimated using TIR bands and
emissivity was calculated using NDVI threshold technique. Finally Land surface
temperature for Bhopal region have been estimated and statistics are presented.

Keywords Remote sensing (RS) Land surface temperature (LST)



Split window algorithm NDVI threshold Climate change

V. Solanky (&)
S. Singh
S. K. Katiyar

Centre for Remote Sensing, GIS & GPS, Department of Civil Engineering,
M.A.N.I.T., Bhopal 462003, MP, India
e-mail: makusolanky@gmail.com
S. Singh
e-mail: sangeeta211.aec.it@gmail.com
S. K. Katiyar
e-mail: sunil64k@gmail.com

© Springer Nature Singapore Pte Ltd. 2018 343

V. P. Singh et al. (eds.), Hydrologic Modeling, Water Science
and Technology Library 81, https://doi.org/10.1007/978-981-10-5801-1_24
344 V. Solanky et al.

Introduction

Over the past several decades in Remote Sensing community, land surface tem-
perature (LST) is found to be one of the most important parameters in the physical
processes of surface energy and water balance at local through global scales
(Anderson et al. 2008; Brunsell and Gillies 2003; Karnieli et al. 2010; Kustas and
Anderson 2009; Zhang et al. 2008). In more general terms LST can be defined as
hotness of surface of earth, from RS satellite’s perspective, the surface is whatever
it sees when it looks through the atmosphere to the ground. LST estimation pro-
vides information about temporal and spatial variations of the surface equilibrium
state and is of fundamental importance in many applications.
LST is being used in a variety of areas such as evapotranspiration, climate
change, hydrological cycle, vegetation monitoring, urban climate and environ-
mental studies, among others (Arnfield 2003; Bastiaanssen et al. 1998; Hansen et al.
2010; Kalma et al. 2008; Kogan 2001; Su 2002; Voogt and Oke 2003; Weng 2009;
Weng et al. 2004). Since the launch of the Landsat-8 satellite (also known as
Landsat Data Continuity Mission, LDCM) in February 2013 continuity of remote
sensing data at high spatial resolution acquired by instruments on board previous
Landsat satellites such as the multispectral scanner system (MSS), the thematic
mapper (TM), and the enhanced thematic mapper plus (ETM+).
Landsat-8 carries two sensors, i.e., the operational land imager (OLI) and the
thermal infrared sensor (TIRS). OLI collects data at a 30 m spatial resolution with
eight bands located in the visible and near-infrared and in the shortwave infrared
regions of the electromagnetic spectrum, plus an additional panchromatic band at
15 m spatial resolution. TIRS sensor measures the TIR radiance at 100 m spatial
resolution using two bands located in the atmospheric window between 10 and
12 lm (Irons et al. 2012). Since Satellite-based thermal infrared (TIR) data is
directly linked to the LST through the radiative transfer equation.
The retrieval of the LST from remotely sensed TIR data has attracted much
attention, specially from Landsat-8 TIR bands. Direct estimation of LST from the
radiation emitted in the TIR spectral region is difficult to perform with that accu-
racy, because radiances measured by the radiometers onboard satellites depend not
only on surface parameters (temperature and emissivity) but also on atmospheric
effects. Therefore, besides radiometric calibration and cloud screening, the deter-
mination of LSTs from space-based TIR measurements requires atmospheric cor-
rections (Vidal 1991). In past many studies have been carried out to estimate LST
from satellite derived TIR data, by using different approaches and methods
(Jiménez-Muñoz and Sobrino 2003).
Before the invention of earth observation satellites (EOS), it was hard to estimate
the LST of an area. Generally, it was calculated for a particular set of sample points
and interpolated into isotherms to generalize the point data into area data. But now a
day’s remotely sensed data is being used for LST estimation by using thermal data.
LST greatly affected by the increasing green house gases in the atmosphere. As it
rises, it melts the glaciers and ices sheets in the polar region, it also increases flood
Land Surface Temperature Estimation Using Remote Sensing … 345

and sea level, Increase in LST also affects the climatic condition of the monsoon
countries leading to unpredictable rainfall hence vegetation in the entire Earth
surface will be affected by this (Rajeshwari and Mani 2014).
Land use/Land cover (LU/LC) of an area can be used for estimating the amount
of LST. The natural and anthropogenic activities change the LU/LC of an area. It
also influences LST of that area. As its value changes the local climate of the area
also changes, this draws attention of many researchers to calculate the amount of
LST for any particular region. In literature it has been found that many methods of
LST estimation are developed and adopted some of the most widely used methods
are as follow: split-window (SW), SC algorithm, Dual-Angle, Sobrino, Mao, and
Submao. Most of the studies were done for urban areas and arid and semi-arid
regions and in many of the studies, single thermal band was used. LST products are
routinely generated from moderate-resolution satellite data (e.g., moderate resolu-
tion imaging spectroradiometer (MODIS) on the National Aeronautics and Space
Administration’s (NASA) Terra and Aqua platforms, Advanced Along-Track
Scanning Radiometer on the European Space Agency’s Environmental Satellite)
because they indicate the surface energy that is available at the land–atmosphere
interface (Norman and Becker 1995).
In this paper, investigations have been carried out for LST estimation using
Landsat-8 TIR bands, and SW algorithm which uses NDVI Threshold technique.
The rest of paper is structured as follow:
Section 2 describes LST retrieval algorithms which also covers theoretical
aspects of these algorithm, details of study area, data resources and satellite data
used in this study are given in Sects. 3 and 4 addresses methodology adopted for
carrying out this study, Results are discussed in Sect. 5. In Section 6 concluding
remarks are presented which is followed by some future recommendations.

LST Retrieval Algorithms

Previous Landsat series has single thermal band in the atmospheric window
between 10 and 12 lm for TM/ETM+sensors. Landsat-8 TIRS has some
advancement over previous sensors and split single thermal band into two TIR
bands. Previous LST algorithms use singe channel (SC) algorithm for the retrieval
of surface temperature, which is fully based on single thermal band. But in TIRS
sensor SC algorithm may cause some problem because the bandwidths of TIRS
bands are narrower than the previous TM/ETM+TIR band. To overcome this
problem a new mathematical structure of split-window (SW) algorithm was pro-
posed, in which two TIR bands were used instead of single TIR band. In this paper
a SW algorithm has been used for LST (Land surface temperature) estimation.
346 V. Solanky et al.

Single-Channel Algorithm

The SC algorithm developed by Jiménez-Muñoz et al. (2014) for the estimation of

LST. The mathematical structure of SC algorithm is:

Ts ¼ c½1=eðw1 Lsen + w2 Þ + w3  + d ð1Þ

where e is the surface emissivity and (c, d) are two parameters given as:

c  Tsen
2
=bc Lsen ; d  Tsen  Tsen
2
=bc ð2Þ

where Tsen is the at-sensor brightness temperature and w1, w2, and w3 are atmo-
spheric functions, given as:

w1 ¼ 1=s; w2 ¼ Ld  Lu /s; w3 ¼ Ld ð3Þ

Split-Window Algorithm

The split-window algorithm proposed by based on two TIR bands typically located
in the atmospheric window between 10 and 12 lm. The mathematical structure of
SW algorithm is given as:

Ts ¼ Ti þ c1 ðTi Tj Þ þ c2 ðTi Tj Þ2 þ c0 þ ðc3 þ c4 wÞð1  eÞ þ ðc5 þ c6 wÞDe ð4Þ

where Ti and Tj are the at-sensor brightness temperatures of bands i and j, e is the
mean emissivity, and given by e ¼ 0:5ðei þ ej Þ, De is the emissivity difference,
De ¼ ðei ej Þ, w is the total atmospheric water vapour content (in g cm−2), and
c0–c6 are the SW coefficients. The basic inputs to SW algorithm are brightness
temperature and Land Surface Emissivity LSE.

Study Area and Data Resources

Study area selected for this study is the capital city of Indian state of Madhya
Pradesh (MP) having area central point coordinates 23° 55′N Latitude and 76° 57′E
Longitude. Investigations for this study have been carried out for the LANDSAT-8
satellite images and details of each test image is presented in Table 1. Study area is
shown in Fig. 1 and LST retrieval algorithm is implemented in ArcGIS 10.3
software environment and ERDAS imagine version 2011 software has been used
for other image analysis task.
Land Surface Temperature Estimation Using Remote Sensing … 347

Table 1 Details of data resources

S. No. Sensor Satellite Area Date of acquisition
1 OLI LANDSAT-8 BHOPAL (INDIA) 23-04-2014
2 TIRS LANDSAT-8 BHOPAL (INDIA) 23-04-2014
3 OLI LANDSAT-8 BHOPAL (INDIA) 26-04-2015
4 TIRS LANDSAT-8 BHOPAL (INDIA) 26-04-2015

Fig. 1 Study area (Bhopal District) on map

Detailed description of satellite data used in this paper is formulated in tabular

form and is shown in Table 1.

Methodology Adopted

To carry out this research work, we have used split-window (SW) algorithm for
LST retrieval, procedure for each and every step involved in this process is shown
in Fig. 1. LANDSAT-8 satellite images are freely available for research purpose,
and can be easily downloaded by accessing the USGS (United States geological
survey) website. USSG offers these images in a bundled form which have 12
different band images. For this study we have only used 4 MS (Multi-spectral)
bands i.e. red, green, blue and near-infrared and 2 and TIR bands out of 12 bands.
Initially OLI images are pre-processed to generate a stacked image and then a false
colour composite (FCC) image is generated to clearly distinguish between different
land cover features. After that sub-setting is applied to extract admin boundaries of
348 V. Solanky et al.

Bhopal district by using the generated FCC image. Then pre-processing steps are
applied on TIRS bands.
After pre-processing of both data sets, NDVI map is generated using 2, 3, 4, 5
bands of LANDSAT-8 i.e. 4 MS bands by using Eq. 1. and this NDVI map is
reclassified to calculate the LSE. Brightness temperature is calculated using thermal
band i.e. 10 and 11, images which aids in the process of LST retrieval (Fig. 2).
NDVI (Normalized difference vegetation index) threshold technique is used to
obtain land surface emissivity (LSE), from reclassified NDVI map and fractional
vegetation cover (FVC) given in Eq. 6.

NDVI  Red
NDVI = ð5Þ
NDVI + Red

LANDSAT-8 OLI LANDSAT-8 TIRS

images images

Pre-processing

NDVI calculation Pre-processing

Reclassification

LSE Determination Brightness Temperature

Calculation

Apply SW Algorithm

Retrieved LST

Fig. 2 Flowchart for LST retrieval using LANDSAT-8

Land Surface Temperature Estimation Using Remote Sensing … 349

 2
NDVI  NDVImin
FVC ¼ ð6Þ
NDVImax  NDVImin

where
NDVI NDVI value
NDVImin minimum value of NDVI
NDVImax maximum value of NDVI
To calculate the LST, we have given the two inputs to SW algorithm i.e. LSE
and brightness surface as described in Eq. 5. And all other parameters used n
formula are determined in previous steps, which have led to determination of LST
for Bhopal district.

Results and Discussion

LST retrieval process is completed using SW algorithm, which is best suited algo-
rithm for LANDSAT-8 TIR bands as found in literature. LST maps are generated for
Bhopal district as shown in Fig. 3a, b, for the months April 2014 and 2015
respectively. Degree Celsius (°C) is considered as basic unit for temperature for all
the analysis, since original LST calculated was in Kelvin hence a Kelvin to °C
conversation is applied and results are presented in °C.

Fig. 3 a LST map of Bhopal district for April 2014 b LST map of Bhopal district for April 2015
350 V. Solanky et al.

Fig. 4 Maximum and LST Analysis For Bhopal District

minimum temperature for 40

Temperature in 0C
Bhopal district 35
30
25
20
15
10
5
0
Max. Temperature Min. Temperature
2014 April 34.5298 17.5204
2015 April 35.6746 19.5409

As per the results obtained in this paper it has been found that Land surface
temperature is significantly increased within one year, i.e. for April 2014 maximum
temperature was found 34.52 °C and it was found 35.67 °C that of April 2015,
which shows an increase in value by 1.15 °C and shown in Fig. 4.
As shown in Fig. 4, there is a rise in maximum and minimum land surface
temperature of 1.15, 2.02 °C which is not desirable. and clearly indicates that
region is affecting from environmental changes.

Conclusions

LST retrieval using Split window algorithm has been investigated in this research
paper and conclusions are summarised as follows:
I. Since the availability of thermal bands (TIR) of LANDSAT-8, many algo-
rithms have been adopted for LST estimation and it has been found that Split
window outperform others.
II. Estimated LST for two consecutive years i.e. 2014 and 2015 shows a sig-
nificant rise in minimum and maximum temperature from which this can be
inferred that region of study area i.e. Bhopal (India) is affecting from global
environmental changes and need to be taken care of.
III. LST was found high in barren regions and low in agricultural, vegetated
areas, also lowest value was found in water bodies, North, south region of
district is having high LST as compare to east and west and central region of
district is having moderate temperature values.
IV. A hybrid approach based upon SC and SW algorithms can be adopted in
order to refine the results for LST retrieval.
Land Surface Temperature Estimation Using Remote Sensing … 351

References

Anderson MC et al (2008) A thermal-based remote sensing technique for routine mapping of

land-surface carbon, water and energy fluxes from field to regional scales. Remote Sens
Environ 112:4227–4241
Arnfield AJ (2003) Two decades of urban climate research: a review of turbulence, exchanges of
energy and water, and the urban heat island. Int J Climatol 23:1–26
Bastiaanssen WGM et al (1998) A remote sensing surface energy balance algorithm for land
(SEBAL) 1. Formulation J Hydrol 212:198–212
Brunsell NA, Gillies RR (2003) Length scale analysis of surface energy fluxes derived from
remote sensing. J Hydrometeorol 4:1212–1219
Hansen J et al (2010) Global surface temperature change. Rev Geophys 48:RG4004
Irons JR et al (2012) The next landsat satellite: the landsat data continuity mission. Remote Sens
Environ 122:11–21, doi:10.1016/j.rse.2011.08.026
Jiménez-Muñoz JC, Sobrino JA (2003) Split-window coefficients for land surface temperature
retrieval from low-resolution thermal infrared sensors. IEEE Geo sci Remote Sens Lett 5
(4):806–809
Jiménez-Muñoz JC et al (2014) Land surface temperature retrieval methods from landsat-8 thermal
infrared sensor data. IEEE Geosc Remote Sens Lett 11(10):1840–1843
Kalma JD et al (2008) Estimating land surface evaporation: a review of methods using remotely
sensed surface temperature data. Surv Geophys 29:421–469
Karnieli A et al (2010) Use of NDVI and land surface temperature for drought assessment: merits
and limitations. J Clim 23:618–633
Kogan FN (2001) Operational space technology for global vegetation assessment. Bull Am Meteor
Soc 82:1949–1964
Kustas W, Anderson M (2009) Advances in thermal infrared remote sensing for land surface
modeling. Agric For Meteorol 149:2071–2081
Norman JM, Becker F (1995) Terminology in thermal infrared remote sensing of natural surfaces.
Agric For Meteorol 77(3):153–166
Rajeshwari A, Mani ND (2014) Estimation of land surface temperature of Dindigul district using
Landsat 8 data. Int J Res Eng Technol 03(05). (eISSN: 2319-1163, pISSN: 2321-7308)
Su Z (2002) The surface energy balance system (SEBS) for estimation of turbulent heat fluxes.
Hydrol Earth Syst Sci 6:85–100
Vidal A (1991) Atmospheric and emissivity correction of land surface temperature measured from
satellite using ground measurements or satellite data. Int J Remote Sens 12:2449–2460
Voogt JA, Oke TR (2003) Thermal remote sensing of urban climates. Remote Sens Environ
86:370–384
Weng Q et al (2004) Estimation of land surface temperature–Vegetation abundance relationship
for urban heat island studies. Remote Sens Environ 89:467–483
Weng Q (2009) Thermal infrared remote sensing for urban climate and environmental studies:
methods, applications, and trends. ISPRS J Photogram Remote Sens 64:335–344
Zhang R et al (2008) Two improvements of an operational two-layer model for terrestrial surface
heat flux retrieval. Sensors 8:6165–6187