Klang Valley, also known as Greater Kuala Lumpur, covers approximately 2,832 square kilometers, as shown in Figure 1. It is situated in the central part of West Coast Peninsular Malaysia and includes major areas like the Federal Territory of Kuala Lumpur, Putrajaya, Hulu Langat, Klang, and Petaling. The population of Klang Valley was estimated at 8.21 million in 2021. The population growth rate of the Klang Valley increased by approximately 2.4% from 2021 to 2023. Population density and growth are primarily concentrated within a 50 km radius of the city center, gradually extending outward into suburban regions, and the land surface temperatures have fluctuated between 22 and 32°C, with a mean annual temperature of 25.4°C. As one of the fastest urbanizing and economically developing regions in Southeast Asia, the Klang Valley serves as an ideal case study for this research, which focuses on the Heat Vulnerability Index (HVI). This city was chosen for the research because it has mixed urban morphology and urban environments. It is characterized by heterogeneous urban development, socio-economic diversity, and varying levels of infrastructure resilience. According to (T. et al., 2024), the landscape mosaic includes commercial, residential, and industrial areas, as well as green spaces such as agricultural zones, parks, forest reserves, and wetlands. The built-up area expanded from 20.58% in 2010 to 30.02% in 2020, driven by new building developments and influenced by infrastructure projects such as the Mass Rapid Transit (MRT) system (Man & Abd Majid, 2024). The study by (Wong et al., 2017) found that the working community in the city centre experienced the effects of UHI, including temperature rises, decreased water resources, and increased haze and air pollution.

The study employed various datasets, which included Landsat-8 OLI/TIRS imagery, a Digital Elevation Model (DEM), statistical data, and vector data. We compiled data on demographic, socio-economic, and environmental indicators relevant to heat vulnerability. These included population density, age distribution, income levels, land surface temperature (LST), vegetation cover from the Normalized Difference Vegetation Index (NDVI), and access to healthcare facilities. From the United States Geological Survey (USGS), Landsat-8 OLI/TIRS data were obtained with important spectral bands for HVI analysis. It included bands of blue to long-wave thermal infrared, which contain information concerning land cover and land surface temperature or vegetation growth. With the assistance of the visualization software of the SNAP Desktop, a DEM of Klang Valley was also generated from Sentinel-2 images, which capture terrain elevation. The Department of Statistics Malaysia (DOSM) received extensive data on population age structure and citizenship, as these factors are necessary for estimating the HVI. Lastly, the Malaysian Public Works Department (JKR) supplied a vector map, from which the road data is extracted to indicate the road density used in Klang Valley Table 1. The conceptual framework is structured around the triad of vulnerability components based on their relevance in the vulnerability literature ((Qureshi & Rachid, 2022); (Noori et al., 2023); (Tesfamariam et al., 2024)):

Prior to data processing, critical pre-processing steps were implemented to ensure that Landsat 8 OLI/TIRS and statistical data were of good quality for the research study’s usage. For instance, images with less than 30% cloud cover were acquired. Then, cloud removal was performed on the Landsat data using the Sen2Cor Processor to remove atmospheric noise, enabling accurate analysis through high-quality results. On the other hand, these were submitted to a proper cleanup, thereby making them suitable for use with the ArcGIS Pro software. These steps were essential in ensuring maintenance of data quality and enhancing its utility, which made it possible to carry out meaningful analysis and interpretation of the data collected.

The thermal emissivity data of the remote sensing imageries is important while estimating land surface temperatures LST and assessing the regional heat environmental risks. Geographically, LST retrieval from Landsat 8 has been undertaken using a number of algorithms aimed at end users, which assist in some aspects of temperature distribution comprehension in the Klang Valley.

To obtain LST, the TOA reflectance computation must be performed in particular. This procedure consists of the procedure of reflectance conversion from the raw digital numbers or radiance acquired through the space-borne sensors. This clean LST data will include algorithms that take into consideration rough estimation of spectral ratios of LST images, as well as measures of radiance for Landsat 8's band 10 (Equation 1).

(1)

Where TOA: Top of Atmosphere (TOA), ML: band-specific multiplicative rescaling factor from the metadata (RADIANCE_MULT_BAND_10), QCAL: thermal band (Band 10), AL: bandspecific additive rescaling factor from the metadata (RADIANCE_ADD_BAND_10).

The presence of vegetation can be assessed utilizing NDVI derived from the visible and nearinfrared bands of Landsat, which is a significant determinant in ascertaining factors of vulnerability and sensitivity in the region (Equation 2).

(2)

Where RED: DN values from the red band, NIR: DN values from the near-infrared band.

The physical proportion of vegetation cover, represented as PV and valued in NDVI as previously discussed, plays a crucial role in assessing other aspects of vegetation cover in the study area (Equation 3).

(3)

Where Pv: proportion of vegetation, NDVI: DN values from the NDVI image, NDVIs: minimum DN values from NDVI image, NDVIv: maximum DN values from NDVI image.

LSE is a proportionality factor used to scale blackbody radiance (Planck’s law) to estimate the emitted radiance (Jimenez-Munoz et al., 2009). LSE is the measured fraction of vegetation cover that contributes A parameter in thermal remote sensing where a value helps to form thermal attributes in Klang Valley (Equation 4).

(4)

Where ε: LSE, Pv: proportion of vegetation.

LST may be predicted using parameters like Brightness Temperature, Emissivity, and other relevant data that yield information about the distribution of surface temperature over the Klang Valley region (Equation 5).

(5)

Where LST: Land Surface Temperature, BT: top of atmosphere brightness temperature (C), W: wavelength of emitted radiance, ε: LSE.

Natural Jenks classification methods contain additional classification to classify Landsat data into LST, which allows for more information about temperature variations in the region. The classification uses the natural Jenks method, which results in five different categories, namely Very Low, Low, Moderate, High, and Very High.

Eight parameters, which include demographic distributions and population density, are picked from census data to serve as sensitivity variables. These parameters estimate the exposure of populations to heat-related impacts. The next step in the analysis of these data, after determining sensitivity variable parameters, involves a series of steps. Eight parameters that were extracted are the population density, elderly population density, median age population density, very young population density, female population density, male population density, non-citizen population density, and citizen population density. The data are first extracted from the census database, and input as the spatial data attributes according to the districts. These data are then reclassed according to the natural jenks method. The same processes were repeated to all eight spatial layers where each representing each parameter.

In the case of Klang Valley, certain parameters measuring the Adaptive Capacity of the HVI include NDVI, NDWI, Slope, Road Density, and Land Surface Albedo.

NDVI is the parameter that determines the weightage of vegetation in the heat vulnerability index (HVI) based on assessing vegetative cover and its endurance against extreme temperatures. It indicates how effective the local vegetation stands and how it might provide green urban spaces that have a cooling effect. NDVI is given by Equation 2.

Heat vulnerability assessment involves using NDWI as an important parameter. It is derived from images captured by the Landsat-8 satellite to show areas at risk on the ground surface. This can be explained as high values refer to healthy water bodies and sufficient soil moisture, promoting regions' resilience against heat. At the same time, low readings may indicate areas prone to heat-induced water stress. Equation 6 then determines NDWI readings.

(6)

Where G: DN values from the green band, NIR: DN values from the near-infrared band.

Klang Valley's road density can be computed by employing the ArcGIS Pro Line Density Tool (with a search radius of 0.037 and an output cell size of 4.48). Using data sourced from Selangor Public Works Department (Jabatan Kerja Raya Selangor), these values are autogenerated based on the extent of the input features and cell size. This is often reasonable for exploratory analyses. Because Cell size (inherits from the environment settings or source layer), while Search radius (automatically calculated based on feature distribution and extent).

Such factors as accessibility and transport infrastructure in the Klang Valley will help to assess how well the region can resist or recover from long-lasting high temperatures. High Road density can facilitate quicker emergency responses and better access to cooling resources or healthcare during heat events, hence enhancing a community's capacity to adapt.

Evaluation of the Heat Vulnerability Index's adaptive capacity for Klang Valley makes it crucial that slope analysis is conducted over the Klang Valley region. The slope data were derived from the Shuttle Radar Topography Mission (SRTM) Digital Elevation Model (DEM) at a spatial resolution of 30 meters. To ensure consistency in vulnerability assessment, slope values were standardized and reclassified into five ordinal categories (Very Low to Very High) using the Natural Jenks classification method. To assess the capacity to withstand heat waves, it is important to know about the topography of a place.

Albedo, which has been reported as an ignored factor in most studies on surface radiation and thermal properties, can be quantified through time-series analysis showing its periodicity or amplitude concerning land use and cover, thereby offering another perspective on climate modeling (Isa et al., 2013). Calculating surface albedo utilizes Landsat imagery converted to Top of Atmosphere (TOA) reflectance. The TOA is given by Equation 7.

(7)

Where Mp: Band-specific multiplicative rescaling factor from the metadata, Qcal: Quantized and calibrated standard product pixel values (DN), Ap: Band-specific additive rescaling factor from the metadata.Finally, from the computation of TOA reflectance by Smith (2010), shortwave albedo is computed using Landsat data (Equation 8).

(8)

Thus, Land Surface Albedo can be calculated using bands 1, 3, 4, and 7 for Landsat. It is important to include LSA because it represents the reflectivity of surfaces and directly influences the amount of solar energy absorbed by urban areas. Surfaces with higher albedo reflect more solar radiation and retain less heat, thus helping to moderate local temperatures.

After defining the parameters for adaptive capacity variables, these are classified according to the prevailing procedures used in previous research work. Each parameter is further divided into five classes: Very Low, Low, Moderate, High, and Very High. It is a systematic approach that assists in assessing the overall level of resilience against heat stress across a region. This systematic approach supports the evaluation of the extent of the capability of regions to withstand heat stress.

The material in this chapter encompasses the Heat Vulnerability Index (HVI) construction for Klang Valley in two ways, using the Principal Component Analysis (PCA) and the EqualWeighted Index (EWI).

The first process of PCA is a two-tiered system using the "Principal Component" Tool in ArcGIS Pro to handle parameterization of each variable's parameters. It is the design of the index that makes it possible to define the importance (inter relevant) of the given criteria for solving the Heat Index. In this case, all three variables, sensitivity, exposure, and adaptive capacity, are analyzed in a sequential manner using principal component analysis. This combines all the above obtained weights into maps of exposure, sensitivity, and adaptive capacity through the 'Weighted Overlay Analysis' Tool. In the center of gravity, PCA is used in the second stage for the iterative modification of HVI, which then undergoes another stage of Weighted Overlay Analysis. This finally leads to the HVI.

In this method, weightage is distributed equally on all layers. This tool applies a similar percentage to each feature while operating the Weighted Overlay Analysis Tool at Tier 1, where EWI was perceived as well. From these maps, those depicting exposure, sensitivity, and adaptive capacity are generated. Though on tier 2, the weighted overlay analysis tool is again utilized to generate HVI, but this time around, weights are not distributed among the three variables in any manner.

A detailed comparative analysis of both PCA and EWI has been conducted, considering various qualitative and quantitative parameters that influence the understanding of heat vulnerability about urban development of the population characteristics of Klang Valley, as well as the architectural design of the buildings in the region under study. The qualitative analysis combines the findings of the two approaches. In contrast, the quantitative analysis determines an increase in geographic spatial extent across varying degrees of vulnerability using GIS, as detailed procedures in investigating knowledge-based outcomes. The attributes, advantages, and disadvantages of all these methods were also investigated in order to find out the differences and similarities. However, quantitatively, she only compares and contrasts the numbers, and all the rest of the qualitative analysis deals with how the results are related to the information and circumstances surrounding the outcomes.

In this section, the Exposure Variables of the Heat Vulnerability Index (HVI) Specific to Klang Valley are focused on. Land Surface Temperature (LST) is one of the most important exposure determinants, and its distribution over the valley is shown in Figure 3. Temperatures range from 15°C to 33.2°C, with the middle section experiencing the highest temperatures due to urbanization, while the northwest, with less urbanization, sees lower temperatures.

Using the Natural Jenks classification technique, five groups are developed for parameters after LST calculation. It is designed to enhance both internal and external heterogeneity of data partitioning. These classes are Very Low, Low, Moderate, High, and Very High with specific temperature ranges as shown in Figure 4. This figure visualizes this classification, while Table 2 presents temperatures for each category in detail. This classification will be useful in creating an exposure map.

The section begins by examining the variation in population density across various age groups, including the elderly, 15-64-year-olds, and 0-14-year-olds. Moreover, it discusses gendered population densities, revealing differences among districts. In particular, female population density mirrors overall population density patterns, whereby higher concentration rates appear in central areas. Male population density follows a similar pattern but with some variations, especially within the southern parts. Moreover, spatial distributions are explained by contrasting the densities of Malaysian citizens against those of non-citizens. For instance, non-citizen residents are mainly concentrated in central places, while Malaysian citizenry varies from one district to another. Figure 5 and Table 3 present categorization based on population density levels; thus, they are necessary for generating the Sensitivity Map of the Klang Valley, which provides an overview of demographic landscapes and socio-economic dynamics across various regions

Slope analysis reveals differences in topography, with steeper elevations and slope angles in the northeastern part compared to other sections. It is important to understand these differences and their impact on topological factors and associated vulnerabilities. Also, local road density distribution and types of prevalence show other characteristics of the development network, highlighting greater densities within the dominant areas than in the border regions.

This is also important in the event of very high temperatures when movement and accessibility are a concern. Finally, Land Surface Albedo shows some important features of the surface that affect its ability to absorb heat. In this case, the central region has different ranges of albedo, which signifies different materials with different thermal properties, which will also alter the heat vulnerability in this region Figure 6.

The parameters employ The Natural Jenks method over NDVI, Slope, and Road Density. Classification errors will be reduced owing to the strong classification within the example. As for NDWI and Land Surface Albedo, class intervals are applied that are constructed by the known literature how to increase their precision and relevance.

Thereby, a variety of classifications is created Table 4, which will classify the Adaptive Capacity of Klang Valley as Very Low to Very High. Discussion of each parameter supported with visual aids promotes understanding of the region's overall adaptivity by classifying the outcomes of each parameter at a finer level, facilitating overall analysis (see Figure 6).

The Heat Vulnerability Index (HVI) for the Klang Valley is calculated using Principal Component Analysis (PCA). The computation has two tiers: the first-tier PCA examines exposure, sensitivity, and adaptive capacity parameters. To produce maps showing the regional exposure, sensitivity, and adaptive ability, each parameter's weightage is determined.

The weight of 100% is given to the Land Surface Temperature (LST) as the only parameter that determines exposure. Various parameters related to sensitivity, such as population density, are analyzed using PCA. Most heavily weighted among these is Population Density at 98.172% Table 5.

Factors like NDVI, NDWI, Slope, Road Density, and LSA are analyzed, with NDVI having the highest weightage at 49.607% Table 6

A complicated analysis shows how variables are weighted in the HVI, namely exposure, sensitivity, and adaptive capacity. Every variable carries a different percentage that reflects its importance in vulnerability Table 7.

The resulting HVI map is an exhaustive representation of the susceptibility of heat-related issues in Klang Valley. It indicates diverse levels of susceptibility distributed across multiple locations, which offer critical insights into spatial vulnerability patterns Figure 7.

In this method, all parameters within the Exposure, Sensitivity, and Adaptive Capacity Variables receive equal weight under EWI. This approach aims to achieve balance by considering all factors involved in heat vulnerability. Land Surface Temperature weighs and occupies 100%. This map is further partitioned into five divisions, which explicitly demonstrates the extent of heat exposure within the Klang Valley region, which serves as an excellent reference for the vulnerability assessment Figure 8.

In this uniformity of weighting of the demographic parameters, the Sensitivity Map provides a global view of the sensitivity levels prevailing in that area Table 9.

The following variable shall now be presented under the caption of Adaptive Capacity Variables. This variable coincides with the allocation of assistance known as equal-weightage and details how well it adapts to heat stressors within a particular district or subdistricts, if any exist Table 9.

HVI-EWI is arrived at through the proportionate allocation of weight among Exposure, Sensitivity, and Adaptive Capacity Variables Table 10. The HVI map resulting from the overlay classification only allows to distinguish six categories of potential vulnerabilities of the Quelimane urban area, depicting not only the focal hot spots of Kuala Lumpur Himalayan Peninsular Geosites in Petaling and Klang but other regions such as Sabak Bernam, Hulu Selangor low in vulnerability levels; this integrative assessment underlined while extenuation of climate change vulnerability is the exposure, sensitivity and adaptive capacity aspect risk towards climate change Figure 8.

4.5. Heat Vulnerability Index (HVI): Comparative Analysis of Principal Component Analysis (PCA) and Equal-Weighted Index (EWI)

Figure 9 presents a comparison between the Heat Vulnerability Index (HVI) as calculated using Principal Component Analysis (PCA) and the Equal-Weighted Index (EWI). The percentage area covered by each methodology for the five vulnerability categories is illustrated in Table 11.

The resulting HVI map is a detailed representation of Klang Valley’s heat-related vulnerability landscape. The illustration of different levels of susceptibility spread across several places in order to provide useful indicators for understanding spatial vulnerability patterns is what makes this figure important. The paper provides an overview of the process and outcomes that go into calculating the Heat Vulnerability Index for Klang Valley, which include exposure, sensitivity, adaptive capacity, among others, and finally the HVI overall.

Principal Component Analysis (PCA) and Equal-Weighted Index are two different methods used to assess heat vulnerability, with each being unique. In Table 12, we bring out the differences between PCA and EWI based on their methodological approach, impact on the estimation of heat vulnerabilities, as well as data characteristics.

In addition, it also brings out other factors when discussing how to choose between PCA or EWI, including considerations such as data characteristics influencing the choice of PCA or EWI and its use in the development of an HVI Table 13.

The dual method approach demonstrates the value of both simplicity (EWI) and statistical rigor (PCA). EWI offers ease of interpretation and policy communication, while PCA enhances the accuracy and analytical strength of the index. The application of PCA in the Klang Valley provides a novel contribution by empirically identifying weight structures aligned with regional variability. This study contributes to the literature by: a. Introducing a comparative methodological analysis in a Southeast Asian context.

a. Introducing a comparative methodological analysis in a Southeast Asian context.

b. Integrating triadic vulnerability components into a spatially explicit HVI.

c. Demonstrating the relevance of PCA for detailed and refined urban climate risk profiling.

Both PCA and EWI offer complementary strengths for HVI development. PCA is best when data quality and availability are high, allowing for detailed and refined data-driven insights. EWI excels in settings requiring simplicity, stakeholder inclusiveness, or when data limitations exist. Choosing the appropriate method depends on your goals, audience, data availability, and context of application. In addition, the PCA technique also has the advantage of managing large datasets efficiently, which is crucial in handling complex data involving the Klang Valley. Apart from this, it enables the identification and prioritization of essential indicators by assigning unique weights based on their contribution to variance, thereby ensuring that major factors are duly stressed. Hence, the PCA method is aligned with the purposes of this study and serves as a strong one, backed up by an evidence-based approach to depicting vulnerability levels, rightly comparing them across various locations.

On the contrary, the EWI Method is more straightforward compared to PCA; nonetheless, it cannot determine which indicator is more important. This will lead to an over-simplification of results, hence not capturing the varying degrees of impact each factor has on HVI. Given the uniform distribution of weights used in developing such an index, it may yield less accurate results. Therefore, this research will prefer PCA as its most suitable method.