=Paper=
{{Paper
|id=Vol-2534/68_short_paper
|storemode=property
|title=Search of Changes in the Temperature of Urban Environment with Use of Satellite Data on the Example of the Krasnoyarsk
|pdfUrl=https://ceur-ws.org/Vol-2534/68_short_paper.pdf
|volume=Vol-2534
|authors=Anna A. Gosteva,Aleksandra K. Matuzko,Oleg E. Yakubailik
}}
==Search of Changes in the Temperature of Urban Environment with Use of Satellite Data on the Example of the Krasnoyarsk==
https://ceur-ws.org/Vol-2534/68_short_paper.pdf
Search of Changes in the Temperature of Urban Environment
with Use of Satellite Data on the Example of the Krasnoyarsk
Anna A. Gosteva1,2, Aleksandra K. Matuzko2,3, Oleg E. Yakubailik1,2,3
1 Federal Research Center Krasnoyarsk Science Center of the SB RAS, Krasnoyarsk, Russia
2
Siberian Federal University, Krasnoyarsk, Russia
3 Institute of Computational Modelling SB RAS, Krasnoyarsk, Russia
gissfu@gmail.com, akmatuzko@icm.krasn.ru, oleg@icm.krasn.ru
Abstract. Changes in the urban environment can be detected using satellite images of
different spatial resolutions in the visible and far infrared range. Landsat data is
currently the most accessible, complete, and open for studying these changes. Thermal
imaging is widely used for research and monitoring of man-made objects such as
pipelines, urban facilities, industrial facilities and pollution. The paper presents the
results of the assessment of the land surface temperature in the Krasnoyarsk city for the
two-year period from September 2016 to September 2018 based on the analysis of
Landsat-8 satellite images.
Keywords: thermal infrared imagery, TIR, Landsat, land surface temperature, LST,
climate of the urban environment.
1 Introduction
A change in the urban environment entails a change in the microclimate of the city, which directly affects the
change in the temperature of the earth’s surface; these changes can be estimated using field measurements and
remote methods. Far-infrared satellite imagery of the Earth is used as source data for remote methods for studying
temperature [1]. Thermal imaging is widely used for research and monitoring of anthropogenic and natural objects
[2].
Since 1984, the systematic collection of Landsat imagery has produced more 60-120 m high spatial resolution
thermal infrared (TIR) imagery of the Earth’s land surfaces than any other satellite system. Yet unlike other NASA
Earth Observation System missions, the Landsat production system does not produce derived physical parameters,
such as surface temperatures, from the calibrated at-satellite radiance data. Additional calculations, and sometimes –
additional data, are needed to determine them [3].
Launched on February 11, 2013, Landsat 8 is the most recently launched Landsat satellite. It is collecting
valuable data and imagery used in agriculture, education, business, science, and government.
The Landsat 8 satellite system consists of two major segments: the observatory and the ground system. The
observatory consists of the spacecraft bus and its payload of two Earth-observing sensors, the Operational Land
Imager (OLI) and the Thermal Infrared Sensor (TIRS) [4].
Land surface temperature (LST) is a key environmental climate variable derived from Thermal Infrared (TIR)
data that is used in surface energy balance models in studies ecology and climate research. Research in this area is
carried out for areas in different parts of the world, including the territory of Krasnoyarsk, a city in Siberia with a
population of more than a million people [5].
This paper considers examples of land surface temperature changes over two-year period from September 2016
to September 2018 based on analysis of Landsat-8 satellite TIR images, which is very important as TIR images
contain information that is virtually impossible to obtain in any other way such as using visible and near infrared
images [6]. A technique is presented for detecting temperature changes within one territory for a certain time
interval and its applicability for determining anthropogenic changes in the landscape is shown. The choice of this
time period is associated with active development of the city for the XXIX World Winter Universiade 2019 in
Krasnoyarsk [7].
2 Remote sensing data
Land Surface Temperature, in the most basic sense, is how hot the ground feels to the touch. From the point of
view of a satellite, the land surface is the first solid surface encountered, whether it is the actual ground or something
Copyright © 2019 for this paper by its authors. Use permitted under Creative Commons License Attribution 4.0
International (CC BY 4.0).
like the top of a tree canopy or building. This surface is considered to be a few millimeters in thickness, and can be
any type of terrain such as grass, forest, desert, snow, water, among others [8].
Problems with observing the Earth’s surface often occur with the presence of clouds, which prevents the satellite
from gathering accurate measurements of the land surface. The land surface temperature especially on a global scale
would be useful by itself as well as for use in obtaining other variables and properties of the Earth’s surface and
terrain. Remote sensing data provides the ability to monitor the land surface temperature over the Earth’s surface.
2.1 MODIS
The Earth Observing System (EOS) Moderate Resolution Imaging Spectroradiometer (MODIS) instrument
produces a daily land surface temperature product using multiple thermal bands. MODIS can provide global
coverage, high spectral resolution, and accurately calibrated data. MODIS utilizes multiple bands in atmospheric
windows for its LST retrieval; it implements a generalized split-window algorithm and a physics-based day/night
algorithm. With seven available thermal infrared bands, this algorithm can adjust for uncertainties in temperature
and water vapor profiles without simultaneous retrieval of surface data or atmospheric variable profiles. Emissivity
is also immediately required for an operational product, so MODIS estimates classification-based emissivities from
land-cover types using thermal infrared bidirectional reflectance distribution function (BRDF) and emissivity
modeling. Over certain land cover types in the range of 263 K to 300 K, the MODIS LST can be better than 1 K, but
can underestimate temperatures in semi-arid regions due to inaccuracies in the estimated surface emissivity [9].
MODIS does have lower spatial resolution than Landsat, which makes this product difficult to apply in certain
applications that require LST, such as field specific agriculture or irrigation studies.
2.1.1 Modis Product
The MOD11A1 V6 product provides daily per-pixel Land Surface Temperature and Emissivity with 1 kilometer
(km) spatial resolution in a 1,200 by 1,200 km grid. The pixel temperature value is derived from the MOD11L2
swath product. Above 30 degrees latitude, some pixels may have multiple observations where the criteria for clear-
sky are met. When this occurs, the pixel value is a result of the average of all qualifying observations. Provided
along with the daytime and nighttime surface temperature bands are associated quality control assessments,
observation times, view zenith angles, and clear-sky coverages along with bands 31 and 32 emissivity from land
cover types.
2.2 Landsat
Landsat, as the longest and only continuous record of the global land surface, can be applicable in work
concerning agriculture, geology, forestry, mapping, and change detection among other applications. This dataset has
been accessible to a wide community of users since all Landsat data became freely available in December 2009 [10].
Landsat sensors is designed for single thermal band imagery (Landsat 4, 5, and 7). The only modification from one
sensor to another is inserting a diff erent spectral response function.
The launch of the Landsat-8 satellite took place in February 2013. Landsat 8 satellite receives data using two
different sensors - Operational Land Imager (OLI) and Thermal Infrared Sensor (TIRS). The TIRS scanner was
created in the NASA Goddard Space Flight Center and is intended for imaging in the far infrared. GaAs-based
Quantum Well Infrared Photodetector (QWIP) photodetectors are installed in the focal plane of TIRS. Landsat-8
images consist of 11 spectral bands, where the 10th and 11th are far infrared bands with a spatial resolution of 100
m, which allows them to analyze the energy of the Earth's surface rather than the reflection of sunlight [11].
2.2.1 Landsat Product
Since 2016, all Landsat data, including Landsat-8, have been supplied with geometric and radiometric
correction. Correction photographs are stored in the Landsat Level-1 Data Processing Levels or Landsat Level-1
data product sets. Additionally, the user can only perform atmospheric correction.
3 Methodology
Given that the surface temperature changes as a result of urban development. One can estimation by changes in
temperature that changes occur in the urban environment. The paper describes a method for determining changes in
the urban environment using remote sensing data. The supporting data is satellite Google Maps. The work was done
in QGIS software.
3.1 Calculation of the atmospheric profile
Atmospheric correction using the "Radiative transfer equation" method is available due to access to all necessary
parameters in open sources and when using the Calculator of atmospheric parameters.
Removing the effects of the atmosphere in the thermal region is the essential step necessary to use the thermal
band imagery for absolute temperature studies. The emitted signal leaving a target on the ground is both attenuated
and enhanced by the atmosphere. With appropriate knowledge of the atmosphere, a radiative transfer model can be
used to estimate the transmission, and upwelling and downwelling radiance.
Traditionally, calculating the atmospheric transmission and upwelling radiance has been difficult and time
consuming. The user has to know where to get the atmospheric data, convert it to the proper format for a radiative
transfer model and integrate the results over the proper band pass. The Atmospheric Correction Parameter
Calculator facilitates this calculation [12].
To obtain atmospheric profile, one has to enter date and time, latitude and longitude, select spectral response
curve corresponding to used source of remote sensing data and optionally, surface conditions – elevation above sea
level, normalized atmospheric pressure, temperature and relative humidity. If surface conditions are not provided,
model predicted surface conditions will be used. In this work, pressure, temperature and humidity data were
obtained from weather forecast and monitoring service [https://rp5.ru/].
3.2 Calculation LST
The LST calculation is performed in QGIS software, with the atmospheric correction “Radiative transfer
equation”, using the parameters obtained in the Atmospheric Correction Parameter Calculator, using “the Land
Surface Temperature Estimation Plugin”. We make two LST calculations for 2018 and 2016.
3.3 Comparison of MODIS and Landsat-8 data
To talk about the accuracy of LST obtained because of calculations using the data of the Landsat-8 heat band, we
consider the obtained values with the data of the finished MODIS product, and with the data obtained from
Automatic Weather Station (AWS). Let's compare at several points (test sites) in the city (Fig. 1), two of them have
AWS installed. And others have a uniform distribution of the landscape on the territory of more than 1200 * 1200
meters.
Figure 1. Test sites, located in the city of Krasnoyarsk (the border of the city is marked in green).
Table 1 presents the average temperature in degrees Celsius between the time of flight of Landsat-8 (about
12 hours of the day) and MODIS (about 13 hours of the day).
Location (a) is situated near the weather station in the area of low-rise buildings in the North-Western suburbs of
Krasnoyarsk. Location (b) is located in a dense mixed forest near the Akademgorodok district. Location (c) is
located on a hill in a relatively open space near the cemetery and the racetrack. The location (d) corresponds to a
typical urban development with 5-and 9-storey buildings on the right bank of the river in the city.
Despite the ease of use of MODIS data, its spatial resolution is not sufficient for studies of changes in the urban
environment. Therefore, this paper uses Landsat-8 data. The temperature channel Landsat-8 has a resolution of
100 meters, but the temperature map is built using visible channels, which allows you to build combined images
with higher accuracy, in this case, it is possible to increase the spatial resolution up to 30 meters.
The obtained results of the comparative analysis of ready-made values of the surface temperature in the archive
of MODIS data and the calculated values of Landsat allowed us to conclude about the correctness of the
implementation methodology of the calculations. The values for several test sites are almost the same. The
discrepancy with the AWS data seems to be due to the fact that the AWS values given are the air temperature values
at a height of 2 meters, not the surface temperature.
Table 1. Comparison of temperature data for test sites.
Date 09/05/2016 09/20/2018
Data MODIS Landsat-8 AWS MODIS Landsat-8 AWS
Location
a 17,2 17,5 16,1 14,4 14,7 12,6
b 16,4 16,1 12,8 12,1
c 19,4 19,8 16,3 15,6 15 12,3
d 20,7 20,5 16,6 16,44
The result of changes in urban development, artificial changes in relief, deforestation, leads to a change in the
microclimate of the urban environment. Accordingly, there is a change in the surface temperature of the city. Thus, a
change in the surface temperature of the city is a sign of changes in the urban environment.
3.4 Methods for detection of urban changes based on Land Surface Temperature
The result of changes in urban development, artificial changes in relief, deforestation, leads to a change in the
microclimate of the urban environment. Accordingly, there is a change in the surface temperature of the city. Thus, a
change in the surface temperature of the city is a sign of changes in the urban environment.
The methodology for detecting changes in the urban environment is based on the use of temperature maps
obtained from Landsat-8 satellite data in the thermal and visible ranges [13]. During implementation, the following
stages are distinguished:
The first step is the preliminary processing of satellite data, which includes atmospheric correction of the source
layers by the Radiative transfer equation method.
The next step is calculating LST on selected dates to study, in our case this 09/20/2018 and 09/05/2016.
The study area is limited to the boundaries of the Krasnoyarsk city, therefore, further it is necessary to conduct
an overlay operation along the city border for each LST.
An important step is the normalization of data to allow comparison of the obtained temperature maps with each
other. Using the statistical characteristics of each image, namely, the average temperature value of the map, an
increment value is calculated for the normalization operation. After this, the mathematical operation of maps algebra
is carried out, allowing one map to be subtracted from another. The result is a new image containing the difference
between the temperature values in each pixel.
For the correct detection of changes in two satellite images, it is necessary to use cloudless one-season satellite
images with the most similar meteorological conditions. To test the methodology for detecting changes in the urban
environment, the authors selected satellite images to the territory of the Krasnoyarsk city with identical weather
conditions for September 2016 and September 2018.
A set of satellite images is presented on dates 09/20/2018 and 09/05/2016 from the Landsat-8 satellite. Table 2
presents the air temperature at the study date and the surface temperature of the earth obtained from satellite images.
The temperature in the image corresponds to the time of shooting Landsat-8 at 12 noon.
Table 2. Weather data for satellite images.
Date Air temperature, С Average land surface temperature on image, С
09/20/2018 day +15 night +4 15.9
09/05/2016 day +19 night +14 19.7
For temperature comparison in 2016 and 2018, we will conduct normalization by the average value of the land
surface temperature in Table 1 within the boundaries of the Krasnoyarsk city. After normalization, the average
values become equal and in the next step the mathematical operation of the map algebra is performed, subtracting
the LST2016 map from normalized LST2018.
As a result of the subtraction, a new image is obtained containing the difference between the temperature values
in each pixel for the studied dates. The obtained values are divided into 3 classes: no change, minor changes with a
temperature difference of 1-3 degrees and significant changes with a temperature difference of more than 4 degrees.
4 Results
As a result of approbation of a technique of detection of changes of an urban environment on land surface
temperature the corresponding thematic map has been created (Fig. 2). This map shows areas with significant
temperature changes, where temperatures have changed by 4 degrees or more over two years. Several characteristic
points on the map of the city, in which there is a change in the urban environment, are marked with numbers 1 - 4.
Let us consider the results in more detail. Figure 3 shows these parts of the city on a larger scale.
Figure 2. Thematic map is analysis of results.
a b
c d
Figure 3. Location with temperature changes
Figure 3a shows the change in surface temperature as a result of the construction of a sports facility. In 2016,
construction of the facility was just beginning. The red color shows areas with an increase the land surface
temperature, which coincide with the contour of the object constructed in 2018.
Figure 3b shows the territory of the former Combine Plant, where in 2017 the demolition of industrial buildings
for the construction of residential buildings began.
Figure 3c shows an example of urban development in the Soviet district, where instead of a wasteland in 2016,
warehouses appeared in 2018.
Figure 3d shows the new Preobrazhensky microdistrict, the places of changes the land surface temperature,
which correspond to new houses under construction in 2018, shown in red in the figure.
5 Conclusions
Test of the LST from remote sensing data MODIS and Landsat-8 show that they values are the same. This is
very important for further research and application of LST from Landsat-8. We can use the values LST from
Landsat-8 for search of changes in the environment of the Krasnoyarsk city. As result we created the map of the
temperature difference for two research dates (05/09/2016 and 20/09/2018).
When analyzing areas with a maximum temperature change, it is clear that this is due to a change in urban
development, for example, the construction of new microdistricts, shopping mall, and a change in the natural
landscape, for example, destruction of forests. Comparison of LST for two different dates of the same season with
similar weather conditions allows you to find changes in urban development. It is also possible to evaluate the
influence of these changes on the urban thermal outline.
References
[1] Merlin O., Duchemin B., Hagolle O., Jacob F., Coudert B., Chehbouni G., Dedieu G., Garatuza J., Kerr Y.,
Disaggregation of MODIS surface temperature over an agricultural area using a time series of Formosat-2
images. Rem. Sens. of Env., 2010, 114, 2500–12
[2] Windahl E., Beurs K., An intercomparison of Landsat land surface temperature retrieval methods under
variable atmospheric conditions using in situ skin temperature Int. J Appl. Earth Obs. Geoinf., 2016, 51, 11-27
[3] Matuzko A.K., Yakubailik O.E. Assessment of thermal conditions in Krasnoyarsk urban area with use of
different satellite data and geographic information system // Thermal Science, 2019, Vol. 23, Suppl. 2, pp.
S615-S621.
[4] James R. Irons, John L. Dwyer, and Julia A. Barsi , The next Landsat satellite: The Landsat Data Continuity
Mission// Remote Sensing of the Environment, vol. 122, 2012, p. 11-21 doi:10.1016/j.rse.2011.08.026.
[5] Matuzko A.K., Yakubailik O.E. Monitoring of Land Surface Temperature in Krasnoyarsk and its Suburban
Area Based on Landsat 8 Satellite Data // Journal of SibFU. Engineering & Technologies. 2018, Vol. 11 (8), p.
934-945.
[6] Frey, C.M.; Parlow, E. Flux measurements in Cairo. Part 2: On the determination of the spatial radiation and
energy balance using aster satellite data. Remote Sens. 2012, 4, 2635–2660.
[7] Gosteva A.A., Matuzko A.K., Yakubailik O.E. Detection of changes in urban environment based on infrared
satellite data // IOP Conference Series: Materials Science and Engineering, 2019, Vol. 537 (6), 062051.
[8] K. G. Laraby, J. R. Schott, N. Raqueno, "Developing a confidence metric for the Landsat land surface
temperature product", Proc. SPIE, vol. 9840, pp. 98400C-1-98400C-14, 2016.
[9] Wan, Z., Zhang, Y., Zhang, Q., and Li, Z. (2004). Quality assessment and validation of the MODIS global land
surface temperature. International Journal of Remote Sensing, 25(1):261–274.
[10] Irons, J. R. and Rocchio, L. (2013). From the beginning. http://landsat.gsfc.nasa.gov/about/history.html
[11] Fu P and Weng Q 2016 Consistent land surface temperature data generation from irregularly spaced Landsat
imagery Rem. Sens. of Env. 184 175-87
[12] Barsi, J.A., J.R. Schott, F.D. Palluconi, S.J. Hook. Validation of a Web-Based Atmospheric Correction Tool
for Single Thermal Band Instruments. Earth Observing Systems X, Proc. SPIE Vol. 5882, August 2005, San
Diego, CA.
[13] Effat H., Hassan O., Change detection of urban heat islands and some related parameters using multi-temporal
Landsat images; a case study for Cairo city, Egypt Urban Climate 2014,10, 171-88.