MSc Applied Geographical Information Systems and Remote
Sensing
GEOG6087: Practical Skills in Remote Sensing
Practical: Land Surface Temperature, Urban Greenness and Land Use/Cover
in Glasgow, UK
Assignment 2: “The relationship between LST, urban greenness and land use/cover
types in Glasgow, UK.”
Due Date: 12 noon, 27th May 2022, Word Count 2500
The overall aim of this practical is to evaluate the relationship between land surface temperature
urban greenness, and land use/cover in Glasgow UK. The specific objectives include:
• Generate land surface temperature data from a Landsat image and use this to identify and
map the presence of urban heat island in Glasgow area, UK
• Generate NDVI index and land cover/use map from a Landsat image and relate this to land
surface temperature data to identify any trends
• Write up the report from the practical on “The relationship between LST, urban greenness
and land use/cover types in Glasgow, UK”
Before beginning the practical download, the following data:
• Go to EarthExplorer (https://earthexplorer.usgs.gov/) and download Landsat 8 image
acquired on the 28th of June 2019 (path: 205, Row: 021)
• Download both Landsat 8 Collection 2 Level 1 and Landsat 8 Collection 2 Level 2 data for
the same date. Create a Folder called Level 1 and unzip the level 1 data here. Create another
folder called Level 2 and unzip the level 2 data here.
• You will use Landsat 8 Collection 2 level 1 data when generating LST and Landsat 8
Collection 2 level 2 data for other analysis (e.g., calculating NDVI and Land cover/use map)
• Glasgow Shapefile from Blackboard site
Part 1 Generating Land Surface Temperature from Landsat imagery
N/B: Some of the procedure described here follows that provided in Sobrino et al., 2008; Avdan &
Jovanovska, 2016 paper (provided in references).
The first step when generating LST is to calcite the Top of Atmosphere Spectral Radiance. We will
use the Thermal Infrared band to achieve this (Band 10) from Level 1 data
A. Calculating Top of Atmosphere Spectral radiance
1. Open ENVI 5.x software
2. Load Band 10.TIF from the Collection 2 Landsat 8 Level 1 data folder (make sure you use the
Level 1 data)
3. The formula for calculating the TOA spectral radiance is as follows:

1. TOA(R) = ML*Qcal + AL ……………Equation 1

o Where ML is the bands specific radiance multiplicative rescaling factor, which is
provided in the Metadata file (i.e., MTL.txt file in the folder) To find this value
you can go to Go to the Level 1 folder and open the MTL.txt file- In the MTL.txt
file go to GROUP = LEVEL1_RADIOMETRIC_RESCALING >
RADIANCE_MULT_BAND_10 and find the value.
o AL is band specific additive rescaling factor, which is provided in the Metadata
file (i.e., MTL.txt file in the folder) To find this value you can go to Go to the
Level 1 folder and open the MTL.txt file- In the MTL.txt file go to GROUP =
LEVEL1_RADIOMETRIC_RESCALING > RADIANCE_ADD_BAND_10 and find the
value.
o
N/B: We have slightly varied the TOA formula from Avdan and Jovanovska’s 2016 paper
(i.e., TOA(R)= ML∗ Qcal + AL – Oi), where they used Oi to correct for stray light effects in
band 10. We do not need this correction as this has been implemented in Landsat 8 level
1 data already. Therefore, we just use the first part of their equation (Equation 1 above)
to generate TOA in this practical.

2. After finding the various values of Equation 1 from the metadata (or Table 1 in Avdan
and Jovanovska’s 2016 paper) implement Equation 1 using Band Math tool in ENVI (type
Band Math here to find the tool)



3. Double click on the band math and then type the following in the Enter Expression
section: float (0.000342*b10 + 0.1). After typing the equation Click Add to List and then
OK.

4. In the available bands list > click on band10 image to set it as the variable used in
expression



5. Press Choose and select a folder where you want to save the output of the band math
calculation. Give the file appropriate name (e.g., TOA_radiance) and press OK.
The next step in calculating LST involves converting the TOA radiance to Brightness temperature
B. Calculating Brightness Temperature
1. To convert the TOA_Radiance from the previous steps to brightness temperature use
the following equation ( Avdan and Jovanovskas 2016):

o =
[ (

)+] ……………….Equation 2

Where K1 and K2 are thermal conversion constants from the MTL.txt file (You can
find the values in the MTL.txt file under GROUP =
LEVEL1_THERMAL_CONSTANTS then get the value for K1 and K2 for band_10
N/B: the K1 and K2 values provided in Table 1 of the Avdan and Jovanovska’s
2016 paper are not correct, so don’t use them but use the ones from the
MTL.txt file

2. Implement Equation 2 in Band Math
3. Write the following in band math: float (1321.0789/alog((774.8853/B1) +1)) and Press
Add to List then press OK
4. Select the TOA_radiance image you produced in the Section A above as B1 in the
variable used in expression window

5. Press Choose and then select a folder to save your output. Give the output an
appropriate name (e.g., Brightness_T)
The next step is to convert the brightness temperature to temperature in degrees Celsius
C. Converting Brightness Temperature to Degrees Celsius
1. To convert the Brightness Temperature (from Section B above) to Degrees Celsius, you
need to subtract absolute zero (that is, -273.15 from the Brightness_T data)
2. Write the following in band math > float(B1-273.15), then add this to the list. Select this
equation and press OK.
3. Select the Brightness_T you produced in Section B above as B1 under the variable used
in expression
4. Choose the folder to save this product and give appropriate name e.g., Degree_C. This
data should give you Brightness temperature in Degrees Celsius
To calculate LST, we need to correct for emissivity. There are many methods of doing this, but for
this practical we will use the proportion of vegetation, which you will derive from the Normalised
Difference Vegetation Index (NDVI). Therefore, we first need to calculate NDVI.
D. Calculating NDVI
1. To calculate NDVI you will use the Collection 2 Landsat 8 Level 2 data as this has been
atmospherically corrected.
2. Load the following bands from the Landsat 8 level 2 folder into ENVI (i.e., band 4(red band)
and band 5(NIR band))
3. To calculate NDVI type the following formula in band Math: float(B5-B4)/float(B5+B4) and
press OK
4. Make sure you select the Relevant bands, that is, band 4 to B4 and band 5 to B5 in the
variable of interest window. Then choose where to save the file giving it appropriate name e.
The next step is to calculate the proportion of vegetation from the NDVI values
E. Calculating the proportion of vegetation (PV)
1. To calculate the proportion of vegetation, use the following equation (Sobrino et al., 2008;
Avdan and Jovanovska 2016):

• = (−−)2……….. Equation 3
• Where NDVIs represent the soil value (which we will take to be 0.2 in this practical)
and NDVIv represents the vegetation value (which we will take as 0.5 in this practical

2. Impalement Equation 3 in band math in ENVI as follows
• float((B6-0.2)/ (0.3)) ^2
3. When implementing the equation select NDVI from Section D above as B6 in the variable
used in the expression. Then choose an appropriate folder and name (e.g., PV) and save the
output
The next step is to calculate Land surface emissivity

F. Calculating land surface emissivity
1. We will use a simplified version of emissivity equation from Sobrino et al 2008 and Avdan
and Jovanovska’s 2016 paper to calculate land surface emissivity
2. The simplified equation is as follows:
• = ∗ + ………….Equation 4
• Where Ɛv represents vegetation emissivity and C represents the proposed surface
roughness. We will use constants derived/proposed in the above papers that is, C =
0.005 and Ɛv = 0.985
3. Implement Equation 4 in Band Math in ENVI as follows:
• float (0.005*B7)+0.985
4. When implementing the above equation choose PV from Section E above as B7
5. Save the output in appropriate folder, giving appropriate name (e.g., LSE)
After calculating the land surface emissivity, the last step involves calculating emissivity corrected
Land Surface Temperature

G. Calculating Land Surface temperature

1. The equation we will use in calculating the LST is a simplified version of the equation
provided in Avdan & Jovanovska, 2016 paper (N/B: I have implemented the constants in the
equation to arrive at the simplified version provided below):
• = (1+�0.00115∗
1.4388 �∗ln())) …………..Equation 5
• Where DegreesC is the data generated by transforming BT into Degrees Celsius in
Section C above and LSE is the emissivity data generated in Section F above
2. Implement the equation in BandMath as follows:

• float(B8)/float(1+((0.00115*B8)/1.4388) *(alog(B9)))

3. When implementing the equation be generous with parentheses and ensure that B8 is
DegreesC image and B9 is LSE image.
4. Save the output in an appropriate folder and give it appropriate name (e.g., LST)
5. The file you have produced is the LST data representing Land Surface Temperature over the
study site

Next step will involve you producing a land use/cover map of the study site using a supervised
classification approach.
H. Producing a supervised land cover/use classification
1. Load Band 2,3,4,5,6 from the Collection 2 Landsat 8 Level 2 folder
2. Layer stack these bands to produce a single layer stack image
3. Use the layer stack image to produce a supervised classification map consisting of the
following 4 classes: Water, Forests, Grasslands/Lawns, and Built-Up Areas.
4. Refer to your previous notes/practicals on how to do a supervised classification.

I. Sub setting/Resizing files and exporting as TIFF
• Subset the LST, NDVI and Land/cover/use map using the Glasgow Shapefile
• Load the Glasgow Shapefile/Boundary
• Type Subset in the Toolbox search window> Double Click on Subset Data from ROIs> Select
the file you want subset> Select Glasgow Boundary File > (Make sure the mask pixels outside
boundary is set to Yes and the Background Value set to NaN) > Provide appropriate file
name and Click OK.
• Export/Save the LST, NDVI and land cover/use as TIFF files (so that you can open them later
in ArcGIS). Go to File>Save As> (ENVI, NTIF, TIFF, DTED) >
•
• Choose the file you want to save>OK>Make sure the output Format is Set to TIFF

• Save the file using appropriate name

J. Comparison of LST with NDVI (the NDVI you produced in Section D) and LST with land use/land
cover map
• Start ArcGIS 10.x
• Load, the TIFF files (LST, NDVI and Land cover/use map) from Section I above and the
Glasgow shapefile
• Generate Random Points in ArcGIS
o Go to Data Management Tools in ArcToolbox > Sampling> Create Random Points>
Select where you will save the points > Give it an appropriate name> Select the
Glasgow shapefile as your Constraining Feature Class > Set Number of points to 200
and leave the rest as default and click OK.
• To extract values, from LST and NDVI using the random points above:
o Go to Spatial Analyst Tools in ArcToolbox > Extraction > Extract Values to points>
Use the Random Points as input feature> Select appropriate Raster and then give
the output feature an appropriate name (e.g., LST_points)
o Repeat the above process for NDVI
• Export the extracted values to as .dbf Table
o Right Click on the name of the values you want to export>Open Attribute Table
o Export the Table by selecting the Export> Give the file appropriate name (make sure
dBASE Table is selected as ‘Save as type’ when you are saving the file)
o
o

• Import the .dbf files into Excel for further analysis >Clean the data (e.g., removing rows with
missing values)> Then evaluate the relationship between LST and NDVI (e.g., through scatter
plot and relevant statistics).

• To generate the LST statistics for the various land cover/use types
o Use Zonal Statistics as Table function in ArcGIS to generate statistical metrices (e.g.,
mean, max, min) of the LST of various landcover types and compere the differences
in LST for the various land cover types
o The Zonal Statistics as Table can be found in >Spatial Analyst Tools>Zonal> Zonal
Statistics as Table
References
Avdan, U. and Jovanovska, G., 2016. Algorithm for automated mapping of land surface temperature
using LANDSAT 8 satellite data. Journal of sensors, 2016.
Sobrino, J.A., Jiménez-Muñoz, J.C., Sòria, G., Romaguera, M., Guanter, L., Moreno, J., Plaza, A. and
Martínez, P., 2008. Land surface emissivity retrieval from different VNIR and TIR sensors. IEEE
transactions on geoscience and remote sensing, 46(2), pp.316-327.