2014 IJEDR | Volume 2, Issue 4 | ISSN: 2321-9939
Land Surface Temperature Retrival of Landsat-8 Data
Using Split Window Algorithm- A Case Study of
Ranchi District
Md Shahid Latif
M.Tech Student, Department of Remote Sensing, Birla Institute of Technology Mesra Ranchi Jharkhand-835215
________________________________________________________________________________________________________
Abstract - Land surface temperature (LST) is an important factor in global climate change, vegetation growth, and
glacier. Its impact will be more in monsoon area because of failure of monsoon and uncertainty and unpredictable in
rainfall. In this article we perform Land surface temperature estimation using SPLIT-WINDOW algorithm on Landsat 8
OLI (Operational Land Imager) Sensor and TIRS (Thermal Infrared Sensor) dataset of Ranchi District. TIRS sensor
exhibit two thermal Band 10 and 11.SPLIT- WINDOW algorithm require brightness temperature value of both band 10
and 11 as well as land surface emissivity calculated from OLI bands ( NIR AND RED) for estimation of land surface
temperature.
Keywords - Land surface temperature, Land surface emissivity, Fractional vegetation cover, Split-window, NDVI.
_______________________________________________________________________________________________________
I. INTRODUCTION
Land Surface Temperature (LST) means the temperature of the surface which we observe if directly contact or touch it with. It
is also refer as skin temperature of the surface. When it rise it causes environmental unbalance situation like melts in glacier,
vegetation, climatic condition of monsoon countries leading to unpredictable rainfall. It is measure in kelvin. Thermal infrared
remote sensing technology has become one of the important means to study the thermal characteristic of land surface. Many
numbers of algorithm are design by many researchers to calculate Land surface temperature like Split-Window algorithm (SW),
Dual Angle algorithm (DA), Single-Channel algorithm (SC) etc. In case of Landsat 8 TIRS we have two thermal bands (10 and 11).
We prefer Split-Window algorithm to estimate Land surface temperature using moderate resolution Landsat 8 bands (30m). During
estimation of LST we required OLI sensor Bands (2-5) for estimation of Land Surface Emissivity (LSE) through Fractional
Vegetation Cover (FVC). Split-Window algorithm combined Brightness temperature band 10 and 11 with Land Surface Emissivity
(LSE) to estimate LST for each ground pixels vector. In the present study we estimate LST for entire Ranchi district.
Study Area
Ranchi is the capital of Jharkhand state is located at 23.35 N latitude and 85.33 E longitude. Its elevation from sea level is
2140 feet. It is located in southern part of Chotanagpur Pleatu. Hot weather season lasting from March to mid-June, the season of
southwest monsoon rains from mid-June to October and cold-weather season from November to February. May is the hottest
month.
Data And Software Used
Landsat 8 TIRS (Thermal Infrared Sensor) Band 10 and 11 and OLI (Operational Land Imaginer) sensor Band (1-9) of Ranchi
District of date 4th April 2014. Thermal constant K1 and K2 and other image statistic are obtain from metadata of the image file.
ERDAS IMAGINE 9.1, ARC MAP 10, ENVI 4.7.
Sensor
OLI
TIRS
Band
10
11
Table-1: Metadata or statistics of Image data
No. of Bands
Path/row
9
140/44
2
140/44
Table-2: Thermal constant value
K1
774.89
480.89
K2
1321.08
1201.14
Table-3: Split-Window coefficients value
Constant
Value
C0
-0.268
C1
1.378
C2
0.183
IJEDR1404073
International Journal of Engineering Development and Research (www.ijedr.org)
3840
2014 IJEDR | Volume 2, Issue 4 | ISSN: 2321-9939
C3
54.300
C4
-2.238
C5
-129.200
C6
16.400
[ Source: Skokovic et al, 2014; Sobrino et al, 1996; 2003; Shaouhua Zhao et al, 2009]
Table-4: Emissivity values
Band 10
Band 11
0.971
0.977
s
0.987
0.989
v
[ Source: Skokovic et al, 2014; Sobrino et al, 1996; 2003; Shaouhua Zhao et al, 2009]
Emissivity
II. METHODOLOGY
Flowchart of the algorithm to be perform during LST estimation using TIRS Band 10 and 11 and OLI sensor Band 2-5 are shown
in Fig-1 given as
Fig-1: Flowchart of Algorithm
Following steps are follow to perform operation are given below:
Step1: Estimation of Top of Atmospheric Spectral Radiance of TIRS Band 10 and 11 and OLI sensor of Band 2-5 individually
using the algorithm given below. This algorithm transform raw image into spectral radiance image. Using ERDAS IMAGINE 9.1
Modeler we implement algorithm of equation-1 to perform task,
L =(
)*Band + Lmin
(1)
Where,
L Top of Atmospheric Spectral Radiance in watts/ (m2*srad* )
Lmax Maximum Spectral Radiance of respective Band
Lmin Minimum Spectral Radiance of respective Band
DNmax = Qcal max Qcal min = Difference of maximum and minimum calibration of sensor
Step2: Estimation of Brightness Temperature (TB) of Band 10 and 11. Brightness Temperature is the electromagnetic radiation
traveling upward from the top of the Earths atmosphere. Thermal calibration process done by converting thermal DN values of
raw thermal bands of TIR sensor into TOA Spectral Radiance and after using Brightness Temperature equation shown in
IJEDR1404073
International Journal of Engineering Development and Research (www.ijedr.org)
3841
2014 IJEDR | Volume 2, Issue 4 | ISSN: 2321-9939
equation-2 we got Brightness Temperature (TB) band as shown in Fig-2. Using ERDAS IMAGINE 9.1 Modeler we implement
algorithm of equation-2.
Fig-2: Flowchart of Thermal calibration process
Brightness Temperature =
TB =
273.15
(2)
Where,
K1 and K2 Thermal constant of Bands from metadata image file
L Top of Atmospheric Spectral Radiance layer
Step-3: Estimation of Normalized Difference Vegetation Index (NDVI) using OLI sensor optical Band after layer stacking of
Band 2,3,4,5 using algorithm shown in equation-3,
NDVI =
Range: -1 < NDVI < + 1
(3)
Step-4: Estimation of Fractional Vegetation Cover (FVC) for an image using NDVI image obtain from Step-3 using the equation4. Fractional Vegetation cover estimate the fraction of area under vegetation. Fig-3 show the flowchart to perform FVC. SplitWindow algorithm utilize FVC to estimate Land Surface Emissivity (LSE). Using ARC MAP 10 we reclassify the NDVI layer
into soil and vegetation and calculate NDVI for Soil and Vegetation and using ERDAS IMAGINE 9.1 Modeler we implement the
algorithm of FVC of equation-4.
Fig-3: Flowchart for FVC
FVC =
(4)
Step-5: Estimation of Land Surface Emissivity (LSE) from FVC layer obtain from step-4 using algorithm in equation-5. Land
Surface Emissivity measure the inherent characteristic of earth surface. It measure its ability to convert thermal or heat energy
into radiant energy. LSE estimation required emissivity of soil and vegetation of both Band 10 and 11 are given in Table-4. LSE
of Band 10 and 11 are individually calculated.
LSE = s*(1-FVC) + v*FVC
(5) Where,
s = Emissivity for soil
v = Emissivity for vegetation
FVC = Fractional Vegetation Cover
Step-6: Combination of LSE of Band10 and LSE of Band 11 obtain from step-5 through Mean and Difference in between them as
shown in equation-6 & 7
IJEDR1404073
International Journal of Engineering Development and Research (www.ijedr.org)
3842
2014 IJEDR | Volume 2, Issue 4 | ISSN: 2321-9939
Mean of LSE = m =
Difference of LSE =
= LSE10 LSE11
(6)
(7)
Step-7: Estimation of Land Surface Temperature (LST) using the algorithm in equation-8 implemented using ERDAS IMAGINE
9.1 Modeler.
LST = TB10+C1 (TB10-TB11) +C2 (TB10-TB11)2 + C0 + (C3+C4W) (1-
+ (C5+C6W)
(8)
Where,
TB10 and TB11 Brightness Temperature of Band 10 and 11
C0-C9 Split-Window coefficient values
m Mean LSE
Difference of LSE
W Atmospheric water-vapour content = 0.013 [Source: Meteorological Observatory of Dept. of Agricultural Meteorological,
Ranchi Agricultural College, Birsa Agricultural University, Ranchi].
III. ANALYSIS AND DISCUSSION
Fig-4 represent NDVI layer of Ranchi district of date 4th April 2014 derived from Band 5(NIR) and Band 4(RED) of OLI
sensor using ERDAS IMAGINE 9.1 Modeler .The range of NDVI is varies from 0-0.574185 (zero for negative NDVI values
using condition). Increase in NDVI range from 0 to 1 indicate for healthy and green vegetation cover area. From NDVI image we
perform reclassification of NDVI layer of Fig-4 for soil and vegetation separately and calculate NDVI for soil and vegetation
shown in Table-5. Take the value of NDVI for soil and vegetation form Table-5 we calculate Fractional vegetation cover (FVC)
using equation-4 and LSE using equation-5. We implement algorithm in ERDAS IMAGINE 9.1 Modeler as shown in Fig-5 to
calculate difference and mean LSE Layer shown in Fig-6 and Fig-7. We take TIRS band 10 and 11 to estimate Brightness
Temperature (TB) in celcius using the algorithm of equation-2 shown in Fig-8 and Fig-10. From Fig-9 and 11 of statistical graph
we observe that class 4 exhibit 55.78% of the total area at a temperature in between 35-40 from TB of Band10 and class 3 of TB
of Band 11 exhibit maximum of 62.80% of the total area at a temperature in between 30-35.Fig-12 represent the Algorithm which
we implement to estimate LST using Modeler. Fig-13 represent the final LST layer of Ranchi district on 4 th April 2014. Area
statistics graph of LST layer in Fig-14 proof that, it divide the whole area into two major surface temperature class of 4 and 5with
statistics of 51.5 % of area under an average surface temperature of 37.5 in class 4 and 41.048 % of an area under an average
surface temperature of 42.5 . Remain 5.41% of an area by class 3 exhibit an average of 32.5 . Water bodies area exhibit an
average of 27.5 . Healthy vegetation land cover area exhibit an LST in between 30-35
Table-5: NDVI for Soil and Vegetation
NDVI for Soil
NDVI for Vegetation
0.2
0.574185
Fig-4: NDVI Layer of Ranchi district on 4th April 2014
IJEDR1404073
International Journal of Engineering Development and Research (www.ijedr.org)
3843
2014 IJEDR | Volume 2, Issue 4 | ISSN: 2321-9939
Fig-5: Algorithm for Mean and Difference LSE in Modeler
Fig-6: Difference LSE layer between Band 10 and 11
Fig-7: Mean of LSE layer between band 10 and 11
IJEDR1404073
International Journal of Engineering Development and Research (www.ijedr.org)
3844
2014 IJEDR | Volume 2, Issue 4 | ISSN: 2321-9939
Fig-8: TB of Band 10 with label of Temperature intervals
Area_percetage of TB_Band 10
55.78604092
60
50
40.8140922
40
30
20
10
0.186773309 1.962862906
1.249046352 0.001184311
0
classs1
class2
class3
class4
class5
class6
Fig-9: Area statistics of classified TB layer of Band10
Fig-10: TB of Band 11 with label of Temperature intervals
IJEDR1404073
International Journal of Engineering Development and Research (www.ijedr.org)
3845
2014 IJEDR | Volume 2, Issue 4 | ISSN: 2321-9939
Area_percent _TB11
70
62.80204327
60
50
40
31.85817266
30
20
10
4.786388005
0.313695661
0.239468971 2.31E-04
0
class1
class2
class3
class4
class5
class6
Fig-11: Area statistics of classified TB layer of Band11
Fig-12: Algorithm of Land Surface Temperature using Modeler
IJEDR1404073
International Journal of Engineering Development and Research (www.ijedr.org)
3846
2014 IJEDR | Volume 2, Issue 4 | ISSN: 2321-9939
Fig-13: Land Surface Temperature Layer of Ranchi District on 4th April 2014
Area_Percent
Area Statistics of Class (%)
60
55
50
45
40
35
30
25
20
15
10
5
0
0.001
0.505
5.414
51.596
41.048
1.436
51.596
41.048
0.505
0.001
1.436
5.414
class
TEMPERATURE_CLASS
1
2
3
4
5
IJEDR1404073
Fig-14: Graph of Area occupied (%)
Table-6: Statistics of LST layer of Fig-13
TEMPERATURE_INTERVAL
AVERAGE_TEMPERATURE
20-25
22.5
25-30
27.5
30-35
32.5
35-40
37.5
40-45
42.5
International Journal of Engineering Development and Research (www.ijedr.org)
AREA (%)
0.001
0.505
5.414
51.596
41.048
3847
2014 IJEDR | Volume 2, Issue 4 | ISSN: 2321-9939
6
45-50
47.5
1.436
MID_TEMP
47.5
50
45
40
35
30
25
20
15
10
5
0
42.5
37.5
32.5
27.5
22.5
CLASS1
CLASS2
CLASS3
CLASS4
CLASS5
CLASS6
Fig-15: Graph of Mid value LST
Fig-16: Standard Deviation class of LST Layer
IV. CONCLUSIONS
Split-window algorithm a dynamic mathematical tool provide the Land surface temperature (LST) information using
brightness temperature of thermal bands of TIRS sensor and Land surface emissivity (LSE) factor derived from Fractional
vegetation cover (FVC) of optical bands of OLI sensor. 51.596% of the total area are under temperature of 35-40 and other
41.048% of the total area are under temperature of 40-45 Overall we say around 92.644% of the total area exhibit a surface
temperature of 40 in 4th April 2014. From Fig-15 we observe that average LST is 35 Thus, LST can be calculated using SW
algorithm on Landsat 8 with multiband OLI and TIR images.
REFERENCES
[1]
[2]
[3]
[4]
Akhoondzadeh.M and Saradjian.M.R, Comparison of Land Surface Temperature mapping using MODIS and ASTER
Images in Semi-Arid Area, Commission VIII, WGVIII/9.
Sobrino.J.A, Reillo.S, Cueca.J and Prata.A.J, Algorithms for Estimating Surface Temperature from ASTR-2 Data,
http://earth.esa.int/pub/ESA_ DOC/gothenburg/101sobri.pdf.
Shaohua Zhao, Qiming Qin, Yonghui Yang, Yujiu Xiong and Guoyu Qiu, Comparison og two Split-Window Mehtods
for Retrieving Land Surface Temperature from MODIS Data, Journal of Earth Syst. Science, Vol.118, No.4, pp 345353, August .
Rajeshwari A, Mani N D, Estimation of Land Surface Temperature of Dindigul district using Landsat 8 Data,
International Journal of Research in Engineering and Technology, Volume: 03 Issue: 05,pp 122-126, May.
IJEDR1404073
International Journal of Engineering Development and Research (www.ijedr.org)
3848
2014 IJEDR | Volume 2, Issue 4 | ISSN: 2321-9939
[5]
[6]
[7]
[8]
Xu Hanqiu. Image-based normalization technique used for landsat TM/ETM + imagery. Geomatics and Information
Science of Wuhan University, 2007, vol. 32(1):pp.62-66(Ch).
Xiao Rongbo, Ouyang Zhiyun, Li Weifeng. A review of the eco-environmental consequences of urban heat islands. Acta
Geographic Sinica, 2005, vol. 25(8):pp.2055-2060(ch).
S.Chen, and T. Wang, Comparison Analyses of Equal Interval Method and Mean-standard Deviation Method Used to
Delimitate Urban Heat Island, JOURNAL OF GEO-INFORMATION SCIENCE, vol.11(2),2009, pp.146-150.
Skokovic.D, Sobrino.J.A, Jimenez-Munoz.J.C, Soria.G, Julien.Y, Mattar.C and Jordi Cristobal, Calibration and
Validation of Land Surface Temperature for Landsat 8 TIRS Sensor, Land product Validation and Evolution,
ESA/ESRIN Frascati (Italy), pp 6-9, January 28-30, 2014.
IJEDR1404073
International Journal of Engineering Development and Research (www.ijedr.org)
3849
