APPLICATION OF REMOTE SENSING FOR TEMPERATURE MONITORING : THE TECHNIQUE FOR LAND SURFACE TEMPERATURE ANALYSIS

This research aimed to present the technique for land surface temperature analysis with the data from Landsat-8 Operational Land Imager (OLI) /Thermal Infrared Sensors (TIR) in Meuang Maha Sarakham District, Maha Sarakham Province, Northeast Thailand. The research was conducted as following three steps: 1) Collecting the satellite data in thermal infrared band from Landsat-8 TIR satellite to adjust the value of Top of Atmosphere (ToA) Reflectance and then analyzing the land surface temperature 2) Collecting multi-band data from Landsat-8 OLI satellite to adjust the value of Top of Atmosphere (ToA) Reflectance and then analyzing values of Normalized Difference Vegetation Index (NDVI), Fractional Vegetation Cover (FVC) and Land surface Emissivity (LSE) 3) Bringing the results of 1) and 2) to analyze the land surface temperature with split window algorithm. The research results indicated that the analysis of the data from Landsat-8 OLI/TIR satellites in 18 March 2015 indicated a mean temperature of 33.57 °C.


INTRODUCTION
Currently, urban areas have been developed as a result of economic growth.This results in changes in land use.Also, many areas have been developed, modified and changed in order to support creation of economic areas.The original areas including agricultural areas and empty areas were replaced by public infrastructures, such as buildings, streets, etc. [Wanpen, 2012].
The original areas that were surfaces with good absorption and humidity, such as plants, soils and water resources, were changed to surfaces containing concrete and asphalt.It contributed to the increase of solar heat absorption on the surfaces in urban areas which could absorb the heat during daytime more than natural surfaces.Also, they were mostly green spaces and agricultural areas.During nighttime, the surfaces in urban areas will release the accumulated heat energy during daytime into the atmosphere in a higher amount than natural surfaces.The surfaces in urban areas will accelerate moisture evaporation more than natural surfaces with a better property of moisture absorption.As a result, the temperatures in urban areas will be higher compared to surrounding rural areas [Taha, 1997;Watkins, 1999;Liang, 2004;Zhou et al., 2011].
The different temperatures cause formation of Urban Heat Island (UHI) phenomenon, which air temperatures during daytime of a big city may be higher than surrounding areas up to 1-3°C [David et al., 2011] during nighttime.The difference of temperatures can be up to 12°C in a city or a small community with a less number of population.The consequences of the Urban Heat Island phenomenon will decrease in accordance with a less number of populations.Many researches indicated wide acceptances on a relationship between temperatures and land cover by indicating that In Thailand in the year 2015, the average yearly temperature was higher than the normal temperature of 0.8°C and higher than the temperature of the previous year [0.4°C higher than the normal temperature in 2014].Many areas had the highest temperatures which were higher than the previous measured records.The average monthly temperatures were higher than every normal monthly temperature, especially November and December which had higher temperatures than 2.1°C and 1.9°C, respectively [Thai Meteorological Department, 2015].

APPLICATION OF REMOTE SENSING FOR TEMPERATURE MONITORING: THE TECHNIQUE FOR LAND SURFACE TEMPERATURE ANALYSIS
Nonetheless, the research on the influence of land cover affecting temperatures in the atmosphere of Bangkok was conducted by applying data from Landsat 5 TM satellite.The research found that if there was an increase of trees for 1 percent in an area of 0.1 square kilometers (km 2 ), the temperature will decrease at 0.028°C.Therefore, there should have a change of proportion of land cover by allocating green spaces and water resources properly [Wanpen, 2012].
The application of remote sensing technology in the study of land use and land cover by data from earth observation satellites, namely Landsat 5 TM, Landsat 7 ETM+ and Landsat 8 OLI/TIR, was conducted in order to study wide areas and to monitor changes in land use and land cover quickly [Campbell, 1996;  Generally, the data from the satellites in visible band and infrared band were mainly used.These satellites had a session of repeated record that was proper for applying the data.Also, they could almost monitor situations of land use and land cover in a real time.Moreover, they could analyze the land surface temperature by using thermal infrared band.From the importance of increasing of such the temperatures, this study aimed to present the technique for land surface temperature analysis by the data from Landsat-8 OLI/ TIR satellites in Meuang Maha Sarakham District, Maha Sarakham Province, Northeast Thailand.
For data collection from the satellites in this study, the researchers had collected the data from Landsat ing factor (Table 1) where: k 1 and k 2 = Thermal conversion constant for TIR band (Table 2)

Analysis of NDVI, FVC and LSE
Normalized Difference Vegetation Index (NDVI): The analysis of NDVI was conducted from the ratio of difference and sum total of reflection of visible light band, red band and near infrared band from objects on the earth's surfaces.The result of the calculation was an index be-tween -1 and +1.The NDVI of water surface was less than 0. The NDVI of bare ground as between 0 -0.1 and the NDVI of the ground with plants covered was over 0.1.The analyses of NDVI in this study were conducted by Equation 3 (5) where: S ε and V ε = soil and vegetative emis- sivity values of the corresponding bands (see in table 3).

Land Surface temperature analysis with Split Window Algorithm
The land surface temperature analysis with Split Window Algorithm in this study was conducted in three stages as follows: 1) Analysis of mean by Equation 6, 2) Analysis of difference by Equation 7 and 3) Analysis of land surface temperature by Equation 8and Table 4

The results of implementation in infrared band
The results of operations of adjusting the value of Top of Atmosphere (ToA) Reflectance in infrared Band 10 and Band 11 of Landsat-8 TIR to reduce errors of the energy reflected from objects on the earth's surfaces to the data recorder from surrounding environment while recording data can be shown in Figure 2 and Figure 3.The analysis results of the absolute temperature from the band radiation of data of Landsat-8 TIR can be shown in Figure 4 and Figure 5.

The results of implementation in multi-band band
The results of operations of adjusting the value of Top of Atmosphere (ToA) Reflectance in multi-band of Landsat-8 OLI to reduce errors of the energy reflected from objects on the earth's surfaces to the data recorder from surrounding environment while recording data can be shown in Figure 6 and Figure 7.The analysis results of Fractional Vegetation Cover in this study used NDVI from the analysis above to determine the NDVI for soil of 0.2 and NDVI for vegetation of 0.5.The results of the study indicated a minimum of -4.00, a maximum of 2.46, a mean of 0.61 and StdDev of 0.68.
The analysis results of LSE data were analyzed by analyzing FVC data and determining the constants of Band 10 which consisted of Emissivity for Soil of 0.971 and Emissivity for Vegetation of 0.987.The constants of the Band 11 consisted of Emissivity for Soil of 0.977 and Emissivity for Vegetation of 0.989.The study results of Band 10 indicated a minimum of 1.32, maximum of 1.59, mean of 1.46 and StdDev of 0.03.Meanwhile, the study results from Band 11 were found a minimum of 1.21, a maximum of 1.40, a mean of 1.32, and StdDev of 0.02.

The analysis results of the surface temperature with Split Window Algorithm
The analysis results of the surface temperature with Split Window Algorithm in this study were conducted in three phases: 1) Analysis of mean by Equation 6, 2) Analysis of difference by Equation 7, and 3) Analysis of land surface temperature by Equation 8.The results in sub-districts level of this study can be shown in Table 5 and the result of Meuang Maha Sarakham District was shown in Figure 8 and Table 6.Meanwhile, the land surface temperature data from Thai Meteorological Department indicated a mean temperature measured of 33.11 o C. When statistically compared the analysis of land surface temperature with the data from satellites and data measured by Thai Meteorological Department with Pair Sample T-test, it showed no statistically significant difference at a confidence level of 95%.At any rate, the researchers will keep applying the technique for surface temperature analysis with the data from the satellites in analysis of relationship between urban heat island and urban physical environment of Meuang Maha Sarakham District, Maha Sarakham Province onwards.

2 Figure 1 .
Figure 1.Meuang Maha Sarakham District -8 OLI/TIR satellites (level 1G product) in path 127 Row 49 on 18 March 2015 as following details: 1.The data of thermal infrared band for Landsat-8 TIR used Band 10 and Band 11 2. The data of multi-band for Landsat-8 OLI used Band 3, 4, 5 Implementation in Landsat-8 TIR Correcting the value of Top of Atmosphere (ToA) Reflectance in infrared Band 10 and Band 11 of Landsat-8 TIR to reduce errors of the energy reflected from objects on the earth's surfaces to the data recorder from surrounding environment while recording data, including weather, topography, temperature and angle of incidence by Equation 1 [Barsi et al., 2014; Rajeshwari and Mani, 2014].where: L λ = = Top of Atmosphere spectral radi- ance [W/(m 2 sr µm)] L M = Band specific multiplicative rescal-

2
Top of Atmosphere spectral radiance For finding the results in Celsius ( o C), the absolute temperature is revised by adding the absolute zero (approximately -273.15o C) [Xu and Chen, 2004; USGS, 2013] Implementation in Landsat-8 OLI Correcting the value of Top of Atmosphere (ToA) Reflectance in multi-band of Landsat-8 OLI to reduce errors of the energy reflected from objects on the earth's surfaces to the data recorder from surrounding environment while recording data, including weather, topography, temperature and angle of incidence by Equation 1.
[Yannawut and Teerawong, 2016].NIR = Near infrared band of Landsat 8 OLI R = Red band of Landsat 8 OLI Fractional Vegetation Cover (FVC): For the analysis of Fractional Vegetation Cover in this study, the NDVI data from the analysis above was used to determine the NDVI for soil at 0.2 and NDVI for vegetation at0.5.The analysis was conducted by Equation 4 [Shahid, 2014; Rajeshwari and Mani, 2014; Ugur and Gordana, 2016].NDVI = NDVI of mixed pixel NDVIV = NDVI of vegetation NDVIS = NDVI of soil Land Surface Emissivity (LSE): This study used the FVC data to analyze by Equation 5 [Rajeshwari and Mani, 2014].The constants of Band10 were emissivity for soil of 0.971 and emissivity for vegetation of 0.987.The constants of the Band11 were emissivity for soil of 0.977 and emissivity for vegetation of (0.989).
Difference of LSE W = Atmospheric water-vapour content

Table 6 .
The analysis results of the surface temperature in Meuang Maha Sarakham District

Table 5 .
The analysis results of the surface temperature in sub-districts level