Spatial-Temporal Analysis of Urban Heat Island Phenomenon 2014-2021 in Cianjur District

Cianjur District is one of the areas in Cianjur Regency, West Java, Indonesia, which is the center of community activities. These community activities allow the Cianjur District area to experience massive urbanization and development to meet the growing demand for land increase. This research aims to analyze the spatial distribution of urban phenomena Heat Island (UHI) in Cianjur District and the association between LST and NDVI in Cianjur District. This research uses Landsat 8 data to analyze UHI with LST values. Correlation analysis was also performed to see the association between LST and vegetation density. Based on LST data processing, the temperature in 2014 and 2021 showed that 2014 10 Villages had a surface temperature above 30°C, so they experienced the UHI Phenomenon. The village comprises Bojongherang, Muka, Pamoyanan, Solokpandan, Sayang, Sawahgede, Mekarsari, Limbangansari, Nagrak, and Sukamaju. In 2021, the UHI phenomenon will expand to all villages. LST in Cianjur Regency has a robust negative correlation with NDVI, both in 2014 and 2021. So, the more dense vegetation, the lower is LST, and vice versa.


Introduction
Urban regions are already home to more than 50% of the world's population, which will continue to rise, particularly in developing countries [1].This rising population is marked by the widespread change in natural land cover to urban land cover, which causes higher temperatures in urban areas than in surrounding rural areas [2].Numerous studies on climate have therefore shown that the thermal, optical, and surface geometrical qualities influence the heat and radiation absorption properties and result in the so-called UHI effect [3].
A phenomenon known as an "Urban Heat Island" (UHI) occurs when the air temperature in highly populated cities is more significant than it is in nearby suburban and rural areas [4,5].The notable disparities in ambient air temperature between cities and their surrounding rural areas, which collectively indicate the impact of UHI, have been the subject of climate studies for a long time [6].Anthropogenic heat produced by transportation, industry, and residential structures worsens the UHI effect by altering energy exchange and conductivity levels in the city's dense building mass [7].Between 2 and 8°C can be added to the air temperature in metropolitan areas due to the UHI effect, and the more precise range, according to recent studies, is between 5 and 15°C [8].Increased urbanization and tree cover are two strategies suggested for reducing the heat island effect [9,5].Cianjur Regency is the outermost part of the Jabodetabekpunjur metropolitan area [10].As the urban population increases and the limited land available around Jakarta, the metropolitan area will increasingly expand to areas closest to the city center [11].This expanded area allows the Cianjur Regency to experience massive urbanization and development to meet the ever-increasing demand for land [12].One of the areas in Cianjur Regency, which is the center of community activities, is Cianjur District.Due to surveying collected data, the existing land cover in Cianjur District is dominated by built-up areas (Data Processing, 2022).Besides the land cover, the population factor also influences temperature increases [13].According to BPS, in 2021, the highest population was in Cianjur District, with 173,265 people [14].Therefore, Cianjur District will likely experience the urban heat island phenomenon.This study aims to analyze the spatial distribution of UHI phenomena in Cianjur District and the association between NDVI and LST in Cianjur District.

Research Area
The area observed in this study is in the Cianjur District, Cianjur Regency, West Java.Cianjur District is one of the Districts in Cianjur Regency, located at coordinates -6.822, 107.137.Cianjur District consists of 11 villages, namely Nagrak, Sukamaju, Sayang, Solokpandan, Muka, Bojongherang, Pamoyanan, Sawahgede, Limbangansari, Mekarsari, and Babakankaret villages.Cianjur District has an area of 1705,641 ha, with the broadest village, Babakankaret Village, with an area of 500,270 ha (BPS, 2021).Cianjur District is bordered by Mande District in the north, Karangtengah District in the east, Cilaku District in the south, and Cugenang District in the west (Figure 1).

Research Flow
This research was conducted in the urban area of Cianjur District using Landsat 8 data to analyze the Urban Heat Island phenomenon using Land Surface Temperature (LST) data.In addition, correlation analysis was also carried out to see the association between LST and vegetation density.The following is a research flow that can be seen in Figure 2.

Research Data
Primary and secondary data are the two data sources used in this study.Primary data are those that researchers directly gather.At the same time, Secondary data is information gleaned from or gathered by researchers from pre-existing sources.

Primary Data
The primary data used in this study is air temperature data which will be measured directly using a temperature and humidity measuring device at several sample points in the field during map validation.In addition, documentation is also primary data in this study which is used to strengthen the analysis.

Secondary Data
Secondary data in this study were obtained indirectly through literature studies, previous research, and the Google Earth Engine (GEE).The secondary data needed is:

Data Collection
Primary data collection in the form of air temperature data for validation is measured through direct measurements using temperature and humidity gauges.Measurements were made at sample points that have been divided based on distance from the urban center.In addition, secondary data in the Cianjur Administrative vector dataset was obtained through tanahair.indonesia.go.id and processed in ArcGIS Pro 2.9.2.Meanwhile, Landsat 8 Surface Reflectance Tier 1 data for 2014 and 2021 were obtained through the Google Earth Engine.

Data Processing
The data processing used in this study is cloud computing-based data processing using the Google Earth Engine.Landsat 8 Surface Reflectance Tier 1 satellite imagery data in 2014 and 2021 are processed into Land Surface Temperature (LST) data using a script.After that, to see the association between LST and NDVI, Landsat data is also processed using the NDVI algorithm in GEE to produce a vegetation index value.The data is exported to be later processed in ArcGIS Pro 2.9.2 to make analysis easier.The following is the script code used to obtain LST and NDVI data. https://code.earthengine.google.com/1dcfc0a1e8136649e57f92893e8ca5c0 In the ArcGIS Pro 2.9.2 application, data is processed using the Reclassify tool to determine the temperature class.In addition, NDVI data are also reclassified to facilitate analysis.After that, the raster data is cut based on administration using Extract By Mask.In addition to data processing for UHI analysis, LST data is also processed using SPSS software with Pearson Correlation analysis to determine the correlation between the UHI phenomenon and vegetation density in Cianjur.

Data Analysis
Analysis conducted for this study is overlay analysis to see the pattern of LST distribution and the UHI phenomenon on vegetation density.The LST map that has been made is overlaid with the vegetation density index map and the pattern of each temperature class against NDVI.The expansion of the area experiencing the UHI phenomenon was also seen using calculate geometry in ArcGIS Pro 2.9.2.In addition, this study also uses a quantitative descriptive analysis method with the statistical formula Pearson Correlation (Table 1).Pearson Correlation statistical analysis was used to analyze the relationship between vegetation density (NDVI) and land surface temperature (LST).

Distribution of Land Surface Temperatures
Considering the outcomes of Landsat 8 image processing into an LST map (Figure 3) in 2014, it is evident that the surface temperature that dominates in the Cianjur District area is surface temperature in the class less than 27.5 °C.Subdistricts dominated by land surface temperatures below 27.5 degrees are Babakankaret, Mekarsari, Limbangansari, Nagrak, and Sukamaju Districts.Meanwhile, in the eastern part of the Cianjur District, the temperature is dominated by classes of 28.9 -30°C and 30 -32°C.The area comprises the Districts of Bojongherang, Muka, Pamoyanan, Solokpandan, Sayang, and Sawahgede villages.Whereas in 2021, it can be seen that the temperature classes of 28.9 -30°C and 30 -32°C have expanded significantly so that they dominate the Cianjur District area, as shown in Figure 4.In addition, in 2021, it had a higher temperature class hotter than in 2014, namely the temperature class above 32°C.In 2021 data, high temperatures dominate the eastern region of the Cianjur District (Figure 5).At the same time, the 2022 data obtained from field validation shows that high temperatures dominate the western part of the Cianjur District (Figure 6).This temperature difference in a pattern can be caused by differences in the time of data collection and variables not used in this study.2).The increase in the area of UHI occurred in several villages that had not previously experienced the UHI phenomenon (Figure 9).Spatially, areas that experience the Urban Heat Island phenomenon are areas with low vegetation density.In other words, areas with more built-up land tend to have higher temperatures.Therefore, there is a difference in temperature between urban areas, dominated by built-up land, and rural areas, which tend to have very dense vegetation still.In the distribution of the UHI phenomenon, it is evident that the built-up area in Cianjur District dominates the eastern part.In addition, based on NDVI data processing, it was discovered that there was a decrease in vegetation density in several areas in Figure 10    Field validation of the UHI phenomenon was carried out by measuring the air temperature at 24 sample points (Figure 12).In addition, the 2022 UHI map was also produced from processing validation data in the field in the form of air temperature data (Figure 13).The validation air temperature data, which is point data, is interpolated using the spline method so that it can interpret the surrounding area.The UHI area generated in AST data processing shows a different pattern from the UHI map produced using    3).Pearson correlation calculations were carried out using 24 random sample points and made tabular data for analysis (Table 4).In conducting the analysis, the significance level used was 95%.Based on Table 4, the correlation value in the 2014 data produced is -0.829, so the association between LST and NDVI in 2014 is very strong and has a negative or inverse relationship.In the 2021 data, the correlation value is -0.846, which means the association between LST and NDVI in 2021 is very strong and has a negative relationship.Therefore, it can be concluded that the smaller the NDVI value, the higher the land surface temperature.

Figure 3 .
Figure 3. LST Map of Cianjur District in 2014

Figure 4 .
Figure 4. LST Map of Cianjur District in 2021

Figure 5 . 8 Figure 6 .
Figure 5. Spatial Model Map of LST in Cianjur District Based on Landsat 8

Figure 7 .Figure 8 .
Figure 7. Urban Heat Island Distribution Map of Cianjur District in 2014

Figure 9 .Figure 10 .
Figure 9. Map of UHI Changes in Cianjur District in 2014 -2021 and Figure 11.This vegetation density is IOP Publishing doi:10.1088/1755-1315/1291/1/0120229in line with the increased built-up land, which causes higher temperature changes.Changes in vegetation density significantly affect the increase in the area that experiences the Urban Heat Island phenomenon yearly.

Figure 13 .
Figure 13.Urban Heat Island Map of Cianjur District Based on AST Measurement 1. Cianjur District Administrative Boundary Shapefile Administrative boundary shapefile data was obtained from the Inageoportal page and processed using ArcGIS Pro 2.9.2. 2. Landsat 8 satellite image of 2014 Landsat 8 Surface Reflectance Tier 1 data is taken from image search results on Google Earth Engine.The Landsat 8 satellite image data was Landsat 8 data recorded on June 9, 2014.3. Landsat 8 satellite image of 2021 Landsat 8 Surface Reflectance Tier 1 data is taken from image search results on Google Earth Engine.Landsat 8 satellite image data used is Landsat 8 data recorded on May 11, 2021.

Table 2 .
Area of the UHI phenomenon in the Cianjur sub-district in 2014 and 2021

Table 3 .
LST and NDVI Value Correlation Analysis of Land Surface Temperature with NDVI Correlation analysis is done by making a sample point to see the LST and NDVI values simultaneously.The correlation analysis used is Pearson Correlation on IBM SPSS Statistics 23 software.The following results are from calculating the correlation between LST and NDVI (Table

Table 4 .
Pearson Correlation Calculation Results Based on the processing of Land Surface Temperature data in 2014 and 2021, it was found that 2014 10 villages had surface temperatures above 30°C, so they experienced the Urban Heat Island phenomenon.The sub-districts consist of Bojongherang, Muka, Pamoyanan, Solokpandan, Sayang, Sawahgede, Mekarsari, Limbangansari, Nagrak, and Sukamaju Villages.In 2021, the Urban Heat Island phenomenon will expand to all urban villages.Land Surface Temperature in Cianjur District negatively correlates with NDVI in both 2014 and 2021.The Pearson Correlation value generated in data processing was -0.829 in 2014 and -0.846 in 2021.Those values show a solid inverse relationship between LST and NDVI.In other words, the denser the vegetation, the lower the land surface temperature, and vice versa.