Analysis of Urban Thermal Environment Effect by TIRS and GIS: A Case Study of Zhuhai, Guangdong

— The rapid development of urbanization in China is not only reflected in the tight land area and rapid population growth but also causes changes in the local urban climate, such as the increasingly obvious urban heat island effect (UHIE). This study explores the impact of urban surface cover types on the urban thermal environment. Taking Zhuhai City, Guangdong Province as an example, based on Landsat - 8 thermal infrared remote sensing (TIRS) data, the atmospheric correction method (also known as Radiation Transfer Equation, RTE) and spilt - window inversion algorithm are used to invert the land surface temperature (LST) of the study area and compare their accuracy. After applying ArcGIS to normalize the data, the standard deviation method was used to classify the LST and obtain the distribution map of surface temperature levels in the urban area. In addition, the urban heat island proportion index was used to evaluate the UHIE in the study area, and the distribution of UHIE intensity was obtained. Based on geographical and national data, a combination of mathematical and spatial statistics was used to establish a correlation between the proportion of underlying surface coverage and LST in three different types of water bodies: vegetation and impermeable water surfaces. The results of the effect of urban underlying surface layout on the thermal environment were obtained, and the overall thermal environment effect of the city was obtained


I. INTRODUCTION
Land surface temperature (LST) is of great research significance in the fields of urban thermal environment changes, landscape pattern analysis, and ecological characteristics' analysis, and is an important parameter for studying the exchange of matter and energy between land and atmosphere.In recent years, scholars have conducted research on LST based on remote sensing (RS) technology (Price, 1990;Yue et al., 2006), ranging from LST inversion to results analysis and application.One of the hotspots is the quantitative analysis of the relationship between urban surface temperature changes and the underlying surface (which is an important factor in climate formation and refers to the Earth's surface interacting with the atmosphere during heat, momentum, and water vapor exchange).
ISSN: 2456-1878 (Int.J. Environ.Agric.Biotech.) https://dx.doi.org/10.22161/ijeab.85.14 88 For example, Atsuko et al. (2009) pointed out after studying the impact of land use and land cover (LULC) on ambient temperature in Takamatsu City, Japan, that the growth of urban impermeable underlying surface area is one of the most important factors leading to temperature rise.Xiao et al. (2007) further found that there is a positive correlation between the impermeable underlying surface and LST in Beijing.Streutker (2002) and Roth et al. (1989) used RS data to invert LST and its spatial distribution in several cities along the western coast of North America.
They believe that there is a clear correlation between the thermal characteristics inside cities during the day and land use, while the correlation between nighttime heat island intensity and land use is relatively small.Chen et al. (2006) and Li et al. (2008) analyzed the correlation between NDVI (Normalized Difference Vegetation Index), MNDWI (Modified Normalized Difference Water Index), NDBI (Normalized Difference Building Index), and NDBSI (Normalized Difference Bare Soil Index) and LST, respectively.They found a clear correlation between them, and there were significant differences in LST among different LULC types.The above study obtained the correlation between underlying surface and temperature in different types of cities.
This article takes Zhuhai City, Guangdong Province, as an example and uses Landsat-8 thermal infrared (band) image data as the basis to invert its LST.Combined with the geographical and national data of the research area, a quantitative analysis is conducted on the relationship between surface temperature and underlying surface in order to provide scientific reference for the evaluation of natural resource ecological environment and urban planning in domestic cities.

Study Area
Zhuhai City has three administrative regions under its The climate of Zhuhai is pleasant, with an obvious alternation of winter and summer winds.The temperature is relatively high all year, with occasional cold showers.
The annual and daily temperature differences are small, and they belong to the transitional marine climate between the South Asian tropics and the tropics.The city is rich in solar energy and abundant in heat.It is the only city in China that has been selected as one of the "Top 40 National Tourist Attractions" for its overall urban landscape, with mountains and rivers alternating and land islands facing each other (Figure 1 and Figure 2).
with land use status maps.
(4) The climate pattern of Guangdong Province has been characterized by a continuous increase in temperature since June, with a high temperature period from July to September and a significant UHIE.Therefore, when selecting data sources, priority should be given to image data from June to September.However, due to the large amount of cloud cover in the image data during this time span, it will have a serious impact on temperature inversion.Thus, after further evaluation, February images were selected for inversion analysis.

Data Source and Preprocessing
The data sources used in this study chiefly include geographic national condition vectors and Landsat-8 satellite images.
(1) Geographic national vector data includes surface coverage data from the 2017 Geographic National Survey and the 2020 Geographic National Monitoring results.
( images were selected for analysis and comparison.When selecting images from the source database, it was found that the number of available images was relatively small.
Thus, an image mosaic method was adopted to obtain the full image of February 2016.Among them, the 10th and  The main approach is to first estimate the impact of the atmosphere on surface thermal radiation, and then subtract the atmospheric impact from the total amount of thermal radiation observed by satellite sensors to obtain the intensity of surface thermal radiation, and then convert the intensity of thermal radiation into the corresponding surface temperature.Basic steps:

Data Preprocessing
(1) Calculate radiation brightness temperature (T6): After radiation correction and atmospheric correction are In the formula (1), B (TS) is the thermal radiation brightness of the blackbody in TS derived from Planck's law, and T is the transmittance of the atmosphere in the thermal infrared band.The radiation brightness B (TS) of a blackbody at temperature T in the thermal infrared band is as formula (2): λ is the thermal radiation brightness of the blackbody at TS, which is derived from Planck's law.
T6=1321.08/log (774.89/b1+1) (2) NDVI calculation such as formula ( 4 (4) Surface emissivity (Surf) is a basic parameter of LST that mainly depends on the geological structure of the surface.This study uses the same surface emissivity calculation method as TM/ETM+6.Calculate surface emissivity using the NDVI threshold method proposed by Sobrino (2006).In band math, the formula is converted to: In the formula, b1 is vegetation coverage (VFC)

LST Inversion
(1) Calculate variables C and D, as shown in formulas (7) and ( 8): In the above equation, the conversion formula between Kelvin (K) and Celsius (℃) is: K=℃+273.and the emissivity of the pixel is εv.When the NDVI of a pixel is less than NDVIs, the vegetation cover of the pixel is 0,0; Pixel reflectance is εg .

Underlying Surface
This study is based on the results of surface coverage in geographical and national conditions.Within the study area, quantitative analysis is conducted on the relationship between LST changes and underlying surfaces, including the urban heat island proportion index and the analysis of the relationship between underlying surfaces and LST changes.Among them, the urban heat island proportion index is a new viewpoint proposed to address the difficulty of quantitative comparison of UHIE in different time periods.That is, the comparison of UHIE between different time periods cannot only consider the temperature itself but also the different temperature levels that make up the urban heat island, the proportion of the area in the urban built-up area (Li, 2020)

IV. ANALYSIS AND RESULTS
The underlying surfaces of this study are divided into three categories: vegetation, water bodies, and impermeable surfaces, where impermeable surfaces refer to surfaces covered by various impermeable building materials, such as buildings, roads, and parking lots composed of materials such as tiles, asphalt, cement concrete, etc.

Vegetation Information Extraction
The most commonly used vegetation information extraction is the normalized difference vegetation index (NDVI), which is between -1 and 1.The vegetation index is positive, and the larger the value, the more obvious the vegetation features.The calculation formula is shown in (11), and the vegetation cover map is shown in Figure 4: where RED and NIR, respectively, select the red and near-infrared bands after atmospheric correction.The calculation formula is as follows ( 13): where RNIR and RMIR are the reflected radiation values of the near-infrared and mid-infrared images, corresponding to the band 5 and band 6 OLI data.The darker the color of the NDBI image, the higher the representative value, indicating a higher proportion of building land and a higher density of buildings (Figure 6).

Analysis of UHI Ratio Index
The proportion of vegetation area in the urban area of Zhuhai remained unchanged from 2016 to 2021, while the proportion of water body area decreased slightly and the proportion of impermeable water surface area increased.
According to the geographical and national survey and monitoring data statistics of Guangdong Province (Figure 7 and Table 1).

Inverted LST Situation
The surface temperature situation of Landsat-8 remote sensing image data for temperature inversion is shown in Figure 8 and Table 2.The atmospheric correction method, also known as the RTE method, is based on the Planck equation to invert surface temperature.This method is simple and clear and has higher inversion accuracy when obtaining or simulating more accurate atmospheric  The surface temperature situation of the reverse performance (Figure 9, Table 3) shows that the daily highest and lowest temperatures in the urban area of Zhuhai in 2021 were higher than those in the same month of 2016.The standard deviation of temperature decreased, indicating a smaller dispersion of temperature distribution and gradually stabilizing.The specific steps are as follows: first, normalize the inverted LST according to formula (14), and unify the surface temperature between 0 and 1.Then, using density segmentation technology, the normalized s LST is divided into seven levels using an even distribution method, In the formula (14), LSTnorm is the normalized surface temperature value; LST that has not been normalized; and LSTmax is the maximum value of surface temperature.LSTmin is the minimum value of surface temperature; In the formula (15), URI refers to the proportion of urban heat islands; m is the number of surface temperature levels; i is the temperature level in the urban area that is higher than the moderate temperature zone; n is the number of temperature levels in urban areas that are higher than those in moderate-temperature areas; ｗ is the weight value, selecting the level value of the i-th level; and P is the percentage of level i.In this study, the natural breakpoint classification method was used to classify the LST into 7 levels, so m is 7. Areas above moderate temperatures represent the development range of urban heat islands; therefore, n is 3.
Based on the above method, the calculated values and distribution maps of Table 4

V. CONCLUSIONS
The study analysis shows that from 2016 to 2021, the proportion of impermeable water surface in the urban area of Zhuhai City has slightly increased, the proportion of water body area has slightly decreased, and the vegetation area remains unchanged.After careful comparison and analysis of the proportion of underlying surface area, it was found that the vegetation area on the underlying surface of Zhuhai City remained relatively stable at 33%.
The proportion of water body area is about 30% (27% in 2016 and 24% in 2021).From this, it can be concluded that for small and medium-sized cities, vegetation plays a greater role in alleviating the UHIE than water bodies.
That is, for small and medium-sized cities, in the process of urban development, if the high proportion of vegetation coverage in the area can be maintained, even if the proportion of impervious surface area decreases and human activity areas increase (15% in 2016 and 22% in 2021), it may not lead to a sharp intensification of the urban heat island phenomenon.
In this study, only two images of February were selected for the inversion of urban temperature.Although the months studied do not belong to the annual high temperature period, the patterns show that there are changes (increases) in urban temperature between years.
However, the temperature changes throughout the year cannot be fully demonstrated, so it is insufficient to highlight the trend of heat island changes in the city.In the

Fig. 1
Fig.1 Geographical Area Map of Zhuhai City

( 1 )
11th bands of the Landsat-8 thermal infrared sensor (TIRS) are used to estimate brightness temperature; the operational land imager (OLI) data is used to calculate the NDVI, MNDWI, and NDBI.Next, the data is subjected to radiometric calibration processing and FLAAS atmospheric correction.In addition, land use types are based on NDVI, MNDWI, and NDBI, using normalized density segmentation methods for classification, and calibrated in conjunction III.METHODOLOGY This study selected Zhuhai City as the research area, comprehensively utilizing various methods such as geographic information systems (GIS), RS technology, and spatial modeling.Based on multi-temporal Landsat image data, a supervised classification method is used to classify land use/cover.The RTE method and split-window algorithm are used to invert LST in order to study the characteristics of underlying surface changes and the spatio-temporal changes of LST during urbanization.Analysis of the relationship between underlying surface changes and LST uses the quantitative method.The specific research route is as follows (Figure 3): Analysis of the spatio-temporal dynamics of underlying surface changes: Using land use/cover change (LUCC) as the characteristic, the spatio-temporal dynamics of underlying surface changes are analyzed, as well as the characteristics of landscape pattern changes in Zhuhai's underlying surface from the perspective of patch types and landscape levels.

(
b1 represents the surface emissivity (Surf), 0.34 is the atmospheric transmittance of the day; the atmospheric transmittance is obtained by inputting the photography time and central latitude and longitude through NASA's official website.)(2) Calculate the surface temperature in degrees Celsius using the formulas (9), (10), and (11): Ts=[a * (1-C-D)+(b * (1-C-D)+C+D) * T6+D * Ta]/C (9) (Ts is the true surface temperature; a and b are constants; a = -67.355351,b = 0.458606.)C and D are intermediate variables, as shown in formulas (7) and (8).The radiant brightness temperature T6 can be obtained using the inverse function of the Planck formula (as shown in formula 3), where Ta is the average atmospheric temperature (in K).In addition, there is a linear relationship between the average atmospheric temperature Ta and the near surface temperature T0 (usually 2m) as follows: Ta=17.9769+0.91715* T0 (tropical average atmosphere) Ta=16.0110+0.92621* T0 (mid-latitude summer average atmosphere) Ta=19.2704+0.91118* T0 (mid-latitude winter average atmosphere) Among them, Ta is the average atmospheric temperature, and T0 is the local temperature at the time of remote sensing image acquisition (T0's temperature needs to be converted into Kelvin temperature).Ta=17.97669+0.91715* (273.15+19)=285.992(10) LST=Ts-273.15

2 +( 2 )( 3 )
Using the dual channel nonlinear split-window algorithm to invert surface temperature(Chen et al., 2004) (as shown in formula 12):  7 (  −   )2  (12)Among them, e and Δε Represent the mean and difference of emissivity for two channels, depending on surface classification and coverage: T I and T J are the observed brightness temperatures of two channels, b i (i=0, 1... 7) represents various coefficients that can be obtained from simulated datasets of laboratory data, atmospheric parameter data, and atmospheric radiation transfer equations.To improve inversion accuracy, coefficient b i depends on the water vapor content in the atmospheric column.Atmospheric Water Vapor Content To reduce dependence on external atmospheric conditions, a new algorithm has been developed to estimate water vapor from thermal infrared images themselves.Firstly, an empirical relationship between the atmospheric transmittance ratio T i /T J of two split window channels and the atmospheric water vapor content wv is established using MODTRAN and TIGR atmospheric profiles.Then, the transmittance ratio is estimated using the ratio of covariance to variance between the brightness temperatures of the two channels within a certain size sliding window (as shown in formula 13).. Pixel Emissivity Inversion The vegetation coverage weighting method uses Landsat-8 visible and near-infrared data to invert NDVI and vegetation coverage f to estimate pixel emissivity.  =   . +   (1 − ) + 4 <  > .(1 − ) (14) f= (NDVI-NDVIs)/ NDVIv-NDVIs) 2 (15) Among them, the emissivity data of vegetation components ε and the emissivity data of background components εg come from the spectral database.The <dε> represents the cavity effect parameter formed by multiple scattering of components within a pixel, which is determined by the red structure of the pixel canopy and surface roughness.NDVIs and NDVIv are NDVI values for bare soil and dense vegetation, respectively.To maintain consistency between different images of NDVIs and NDVIv, fixed values are taken here, namely NDVIs=0.2 and NDViv=0.86.When the NDVI of a pixel is greater than NDVIv, the plate coverage of the pixel is 10,

Fig. 6
Fig.6 Construction Land Coverage Map of Zhuhai City

Fig. 7
Fig.7 Remote Sensing Images of Land Use in Zhuhai in 2016 and 2021 (Image Source: 2022 GlobeLand30 Surface Cover Data from the Ministry of Natural Resources of China)

Fig. 8
Fig.8 Temperature Inversion Image in 2021 [Split Window Algorithm (Left), Atmospheric Correction Method (Right)] including extremely high temperature (EHT), high temperature (HT), relatively high temperature (RHT), medium temperature (MT), relatively low temperature (RLT), low temperature (LT), and extremely low temperature (ELT), with corresponding level values ranging from 7 to 1 (Pan and Han, 2011; Li and Xv, 2014).Finally, calculate the proportion of these levels in the built-up area and calculate the URI based on the formula (

Fig. 10 4 . 6
Fig.10 Temperature Inversion Normalization Processing Images [2016 (Left) and 2021 (Right)] future, we plan to make full use of geographical and national data, refine the underlying land types, and conduct quantitative analysis of the relationship between LST changes and underlying surfaces in the study area during the four seasons in order to provide a useful reference for urban development planning.Technically, this article uses the traditional atmospheric correction (RTE) method for LST inversion, ISSN: 2456-1878 (Int.J. Environ.Agric.Biotech.) https://dx.doi.org/10.22161/ijeab.85.14 99 which has room for improvement in accuracy.In the future, we plan to optimize the inverse algorithm from the perspectives of physical simulation and mathematical statistical analysis and then compare the accuracy differences with traditional calculation results.In addition, it is not possible to verify the inversion results of surface temperature due to the lack of synchronous meteorological observation data on satellite transit time and surface temperature, and the estimation of emissivity has not addressed the issue of pixel mixing.In summary, the above analysis elements are the focus of further experiments and discussions in this study in order to improve the reliability and accuracy of LST inversion.

Table 1
Proportion of Different Underlying Surface Areas in Zhuhai in 2016 and 2021

Table 2
Statistical Characteristics of Temperature Inversion