Impacts of Vegetation Dynamics on Land Surface Temperature and Moisture in Kendari City, Indonesia: A Remote Sensing Study

Rapid urbanisation has changed cityscapes throughout the tropics, resulting in severe ecological and microclimatic consequences. In many developing towns, unregulated urbanisation has resulted in a loss of plant cover, which naturally regulates local temperatures and humidity. The conversion of green spaces into impervious surfaces changes the surface energy balance, increases heat absorption, and raises the danger of Urban Heat Island (UHI) formation [1, 2].

Kendari City, the capital of Southeast Sulawesi Province, is one of Indonesia's fastest-growing coastal cities, with an annual population growth rate of 1.63%, which exceeds the national average of 1.49% [3]. Its importance as a regional economic hub, which supports nickel mining in neighbouring provinces, has hastened land conversion for housing and infrastructure development. The city's hilly terrain and the fact that it is located between the coast and the mountains make it especially vulnerable to environmental damage from more people moving to cities and more business units opening.

Previous research on urban heat and vegetation dynamics has been widely conducted in cities such as Manila [4], Kunming [5], and Mysuru [6], but few studies have focused on Southeast Asian mid-sized cities like Kendari, where the interactions between vegetation, surface temperature, and soil moisture are still poorly understood. Remote sensing provides an excellent way to track these characteristics throughout time [7, 8].

The study's particular objectives are to quantify spatial and temporal changes in NDVI, LST, and NDMI in Kendari City from 2014 to 2024. Analyse the statistical correlations between NDVI, NDMI, and LST. Determine spatial hotspots of thermal stress and make area-specific recommendations for sustainable urban planning.

3.1 Normalized difference vegetation index (NDVI)

The NDVI analysis of Kendari City shows the spatial and quantitative trends of plant cover during the last decade (Table 1). As illustrated in Figure 2, in 2014, areas with high greenness levels (dense vegetation) dominated suburban districts, particularly in the city's northern, southern, and southwestern regions, where the terrain ranged from steep to gently sloping. However, by 2024, several of these locations had deteriorated in vegetation quality, falling into the medium- and low-greenness classifications, particularly in the Kadia, Mandonga, and Wua-Wua districts. These center districts evolved from thickly forested zones to more developed areas, owing mostly to urban expansion, residential construction, and infrastructure development spurred by population increase and economic activity.

Table 1. NDVI class

Figure 2. Normalized difference vegetation index (NDVI) of Kendari City in 2014 and 2024

The statistics in Table 1 demonstrate that the high greenness class grew by 8.21% (1,493.73 ha) as a result of the Southeast Sulawesi Provincial Forestry Service's river basin rehabilitation initiative, particularly in Kambu and Poasia Villages. However, this good trend contrasts with a steep drop in medium and low greenness levels, which fell by 28.44% and 13.86%, respectively. These decreases point to the change of moderately vegetated zones into built-up regions, especially in the urban core and transitional districts. The Very Low Greenness class also decreased significantly (-9.23%), indicating that marginal vegetated regions were being lost to development.

Meanwhile, non-vegetated regions (no vegetation) rose by 12.42%, primarily in central Kendari and previously classed as highly vegetated. This increase in impervious surfaces suggests that urban densification and land-use transformation are ongoing, consistent with findings from Manila, Philippines [4]; Kelantan, Malaysia [13]; Kunming, China [5]; and Mysuru, India [6], where rapid urbanization resulted in vegetation degradation and the expansion of built-up land.

Overall, Kendari City's total vegetated area declined from 26,879.94 ha in 2014 to 26,674.71 ha in 2024, a net loss of 205.23 ha (-0.76%). Although the percentage reduction is slight, it represents a significant biological transition from naturally vegetated land to impermeable urban surfaces. This transition has a direct impact on thermal comfort, groundwater penetration, and the stability of urban micro-ecosystems [14].

Low NDVI values are found primarily in the city center and recently constructed areas, indicating a decreased natural ability to regulate surface temperature and humidity. This state increases the vulnerability to the urban heat island (UHI) phenomenon. In contrast, suburban areas with persistently high NDVI values [15], particularly in the northern and southern districts, are significant natural buffers that should be maintained and integrated into climate adaptation plans and long-term sustainable urban design.

The general spatial pattern is consistent with data from other tropical cities [16-18], suggesting that highly built-up regions have lower NDVI values, indicating both a loss of plant cover and a weaker ecological buffering capability in the urban landscape.

3.2 Normalized difference moisture index (NDMI)

The NDMI research shows the regional variance and temporal dynamics of land moisture and vegetation water content in Kendari City from 2014 to 2024 (Table 2 and Figure 3). The findings show a significant shift in moisture distribution patterns over the previous decade, which is linked to vegetation change and urbanization.

Figure 3. Normalized difference moisture index (NDMI) of Kendari City in 2014 and 2024

As shown in Table 2, between 2014 and 2024, wet surface areas grew by 36.28% (4,546.45 hectares), showing better soil and vegetation wetness in numerous parts of the city, particularly in northern and southern suburban zones with higher vegetation density. This increase in NDMI matches spatially to places with high NDVI values, indicating that thick vegetation plays an important role in maintaining surface moisture levels.

Table 2. NDMI class

In contrast, both the very dry and dry categories showed immense reductions of -45.80% and -30.34%, respectively, indicating a significant decline in arid zones that were widespread in 2014. These dry zones were primarily found in central and western Kendari, which have had major urbanization and infrastructure development. The change of permeable vegetated surfaces to impermeable built-up areas has most certainly impacted local hydrological processes, diminishing soil water retention capacity and increasing surface runoff.

The overall moisture distribution pattern indicates a spatial separation between the urban core and the vegetated periphery. The core districts (Kadia, Mandonga, and Wua-Wua) are significantly drier due to the prevalence of built-up areas, but the outer districts (Kambu, Poasia, and Abeli) have higher moisture levels due to increased vegetation cover. This is consistent with observations from other tropical urban environments, where plant loss and land conversion cause localized drying trends, whereas maintained green zones maintain higher moisture content and microclimatic balance [19, 20].

Although the total NDMI-based area declined by -0.78%, the increase in "Wet Level" regions suggests a partial recovery of vegetated land or the impact of improved rainfall patterns throughout the research period. However, the unequal geographical distribution of wet and dry zones demonstrates that urban expansion continues to exert significant localized pressure on surface moisture dynamics. Overall, the NDMI research highlights the strong relationship between vegetation health and density and surface moisture retention, emphasizing the necessity of protecting vegetated areas in urban development. Green infrastructure, such as urban woods, rain gardens, and permeable surfaces, may improve moisture retention and offset the detrimental consequences of urban dryness and heat building in Kendari City.

3.3 Land surface temperature (LST)

LST is a useful indicator for assessing the impact of urbanization, land-use change, and local climatic dynamics on thermal comfort [21]. The analysis reveals that Kendari City experienced significant surface warming between 2014 and 2024, which was linked to both physical land modification and spatially uneven vegetation changes.

In 2014, the city's surface temperature distribution was dominated by moderate to low temperature zones (22.0-23.5℃), especially in suburban areas with lush vegetation and undulating terrain (Figure 4). These colder zones, visually represented by green and yellow tones, corresponded to areas with dense vegetative cover and minimal development. Higher surface temperatures (red zones) were concentrated around the city center, particularly in Kadia, Mandonga, and Poasia, which were already densely populated and urbanized. This study confirms earlier results that the temperature in Kendari City rose by about 2.44℃ from 2001 to 2014 [22].

Figure 4. Land surface temperature (LST) map of Kendari City in 2014 and 2024

By 2024, high-temperature zones (> 24.5℃) had expanded dramatically, resulting in continuous heat clusters in the urban core. The quantitative classification (Table 3) shows a large decrease in low-temperature areas (-41.99%) and a corresponding increase in high- and very-high-temperature classes (+27.35% and +204.36%, respectively). The average LST increased from 22.6℃ in 2014 to 25.1℃ in 2024, representing a +2.47℃ increase over the decade.

These results reveal that surface warming in Kendari is not evenly distributed but rather concentrated in the central and western regions. The urban heat island (UHI) effect has gotten worse, with rising temperatures in densely populated areas and a gradual extension into suburban zones like Kambu, Poasia, and Puuwatu, where high-temperature areas now account for more than 90% of total surface area. This evolution is mostly due to land conversion, infrastructural expansion, and the widespread use of impermeable materials, mainly asphalt and concrete, which absorb and store solar radiation more efficiently [23, 24].

Table 3. Distribution of LST classes

Global warming exacerbates this local effect; climatic projections forecast a global average temperature rise of 1.3-1.8℃ [25], adding to the thermodynamic stress already present in tropical cities. These variables emphasize the significance of integrated climate-responsive urban design, which includes increasing green open spaces, planting urban trees, and adopting sustainable drainage systems to reduce heat accumulation and maintain hydrological balance. Integrating ESG-based spatial planning will allow local governments to manage urban demand more sustainably while achieving goals for adapting to climate change.

Despite the overall rise in surface temperature, the study discovered an 8.21% increase in "High Greenness" locations from 2014 to 2024. At first glance, the results appear contradictory: increased vegetation should theoretically reduce surface temperatures. However, a closer spatial study resolves this quandary.

The increase in NDVI values was predominantly centered in the Kambu and Poasia Districts, where government-led reforestation and watershed rehabilitation efforts took place between 2018 and 2023. These efforts were effective in restoring green cover and improving microclimate conditions in remote areas. However, these vegetated increases were spatially restricted and fragmented, unable to compensate for the significant loss of vegetation and increase in heat-emitting surfaces in the urban core districts (Kadia, Mandonga, and Wua-Wua).

In this area, land-use transformation advanced rapidly: built-up density increased, impermeable surfaces increased by approximately 12%, and evapotranspiration capacity declined drastically. As a result, the overall city thermal balance remained positive (warming), albeit some outlying areas saw local cooling effects. The results demonstrate geographical imbalance rather than contradiction: urban densification in the center surpassed ecological restoration in the periphery.

This finding is consistent with observations from other rapidly developing tropical cities, such as Manila [4], Kunming [5], and Mysuru [6], where local greening efforts coexist with citywide warming due to uneven land-use distribution, fragmented green spaces, and anthropogenic heat emissions. In each scenario, vegetation growth outside of the city center only provides local relief and has little impact on regional LST trends.

3.4 Correlation between NDVI and LST

The correlation between NDVI and LST suggests that vegetation plays an important role in controlling land surface temperature [26]. A correlation analysis between NDVI and LST yields a coefficient of -0.06610 (Table 4). This indicates that greater NDVI values result in lower LST values, and vice versa. High NDVI values suggest thick, healthy plant cover and lower surface temperatures. In contrast, locations with low NDVI, such as built-up areas (city centers), have greater temperatures due to a lack of flora capable of absorbing and decreasing heat via evaporation.

Table 4. Correlation between NDVI and LST in Kendari City

Figure 5. Correlation between NDVI and LST in Kendari City

The association graph between NDVI and LST exhibits a negative linear trend with the equation y = -7.7574x + 23.936 and a coefficient of determination R² = 0.437 (Figure 5). This suggests that vegetation cover variations account for 43.7% of LST variance, whereas 56.3% is explained by other factors such as surface material type, surface albedo, and human activities. Thus, while vegetation helps reduce land surface temperature, other things also contribute to increased surface temperature, which affects the urban heat island (UHI). As a result, to keep urban environmental temperatures stable, a comprehensive approach that combines vegetation planning and sustainable urban design is required.

3.5 Correlation between NDVI and NDMI

NDVI and NDMI are useful markers for monitoring vegetation and soil moisture levels [27]. NDVI measures the density of vegetation, whereas NDMI monitors the amount of water in the soil or vegetation. The correlation coefficient between NDVI and NDMI is 0.90817 (Table 5). This suggests that higher humidity will boost vegetation density, whereas water shortages may reduce it. The dependency of vegetation on water availability suggests a strong ecological response to local climate and hydrological conditions [28].

Table 5. Correlation between NDVI and LST in Kendari City

The graph demonstrating the association between NDVI and NDMI produced a regression equation of y = 0.7838x - 0.1047, and R² = 0.8248 (Figure 6). This reveals that NDVI accounts for 82.48% of the variation in NDMI. This strong positive linear association suggests that when the NDVI increases, so does the NDMI. In other words, a greener and healthier plant (high NDVI values) has higher moisture content (high NDMI values). This indicates that, in an ecosystem context, the local vegetation is highly dependent on moisture to maintain its density and health. As a result, continuous monitoring of these two indicators is extremely beneficial in promoting conservation and climate adaptation measures in urban settings, such as water and soil conservation and the creation of sustainable green spaces.

Figure 6. Correlation between NDVI and NDMI in Kendari City

3.6 Correlation between NDMI and LST

The correlation between NDMI and LST is negative, with a value of -0.76508 (Table 6). The linear regression equation y = -10.403x + 23.007 and R² = 0.5853 show that changes in NDMI represent 58.53% of the variation in LST (Figure 7). High NDMI tends to lower land surface temperature (LST) by natural cooling via evapotranspiration, but low NDMI tends to raise surface temperature, particularly in places with little vegetation [29]. This study suggests that the NDMI indicator could be used to monitor microclimate conditions, mitigate drought risks, and reduce the urban heat island impact.

Table 6. Correlation between LST and NDMI in Kendari City

Figure 7. Correlation of LST and NDMI in Kendari City

3.7 The impact of NDVI and LST on NDMI

The regression analysis revealed a significant correlation between NDVI and LST on NDMI. The multiple correlation coefficient (R) of 0.9343 and R² = 0.8729 showed that changes in NDVI and LST could explain 87.29% of the differences in NDMI (Table 7). This indicates that the created model has high predictive skills.

NDVI exerts a substantial positive influence on NDMI (correlation coefficient = 0.6169), signifying that regions with thick and robust vegetation possess elevated moisture levels. This illustrates the consistency of vegetation's ecological role in preserving soil moisture via shadow protection and evapotranspiration processes. Conversely, LST exhibits a substantial negative correlation with NDMI (correlation coefficient = -0.02), suggesting that rising surface temperatures lead to diminished soil moisture as a result of heightened evaporation and decreased vegetation cover.

The regression model shows that NDMI = 0.4104 + 0.6169 (NDVI) - 0.02 (LST), meaning that if NDVI goes up by one unit, NDMI increases by 0.6169 units, but if LST goes up by one degree, NDMI decreases by 0.02 units. This signifies that the reduction of plant cover and diminished soil moisture directly influence the rise in surface temperatures. In contrast, regions with elevated NDVI and NDMI values continue to have lower and stable surface temperatures. This finding affirms that plants and soil moisture significantly mitigate local warming effects and sustain urban ecological balance. Consequently, urban spatial planning focused on conserving vegetation cover, enhancing green infrastructure, and regulating impermeable surfaces is crucial for mitigating moisture loss and sustainably managing the city's microclimate, particularly in the context of urbanization and global climate change.

Table 7. The impact of NDVI, LST, and NDMI of Kendari City