In order to capture both two-dimensional and three-dimensional greenery metrics, it becomes critically necessary to develop comprehensive analytical approaches that capture both horizontal ground cover and vertical canopy structures (Liu et al., 2021). Various studies have demonstrated that urban vegetation's cooling efficiency varies significantly depending on the configuration of the vegetation, density, and structural complexity (Koc, Osmond and Peters, 2018). However, essential thresholds help determine effective cooling performance, which pertain nonlinear relationships between vegetation characteristics and temperature reduction (Yang et al., 2024).
In lieu of global climate change and rapid advancements through urbanization, metropolitan cities have experienced significant change across environmental aspects (Li et al., 2017). Statistics reveals that over 68% global population of the world is expected to be residing within urban cities by 2050. United Nations (2018) intensifies the Urban Heat Island (UHI) effect by converting natural landscapes into eco-friendly built environments. Hence, the UHI effect aims to control higher temperatures across urban cities, which delivers a critical concern across arid and semi-arid environments (Zhao et al., 2023; Marando et al., 2021). Across urban cities within the Middle Eastern region, Riyadh exhibits the UHI effect to a greater extent due to its arid climate, extensive concrete surfaces, and limited vegetation (Alawadhi, 2022). Hence, temperatures across Riyadh, exceed by 3-5°C, while the difference may increase to 10°C during summer nights (Fang et al., 2010). Such temperature increments adversely affect public health and cooling energy consumption, while significantly reducing outdoor comfort (Salata et al., 2017; Gu et al., 2024). Several cooling mechanisms are utilized to mitigate UHI effects across urban green spaces, including evapotranspiration, shading, and alteration of airflow patterns (Zhang et al., 2014; Wong et al., 2021). However, understand the qualitative and quantitative effects of urban parks on microclimate requires developing evidence-based urban planning and design (Liu et al., 2021). Hence, a comprehensive strategy in designing the King Salman Garden to mitigate UHI would establish significant vegetation across Riyadh's urban fabric, acknowledging it as the largest urban greening initiative across other cities within the Middle Eastern region. Arid urban gardens with diverse vegetation layers and water bodies offer a unique opportunity to examine the cooling potential of large-scale green infrastructure across arid environments (Mohajerani, Bakaric and Jeffrey-Bailey, 2017). Given the limited quantitative assessments of cooling effects in arid regions, for instance, Riyadh, substantial efforts have been underway to mitigate greenery across urban areas, manifesting the effects of UHI (Koch et al., 2020).
Hence, the research investigates the potential UHI mitigation of the King Salman Garden by identifying critical vegetation thresholds for meaningful microclimate improvements. A combination of remote sensing analysis, simulated street-level imagery, and advanced cooling effect modeling are incorporated throughout the research study to mitigate the UHI effect across, given the impact of rapid urbanization in Riyadh. However, the research gap lies in the limited studies that focus on the microclimatic impact of large-scale urban parks in arid regions, particularly concerning the interplay between park design, vegetation, and microclimate modification. Furthermore, the research delivers a key aspect, concerning its theoretical contribution that lies in developing a predictive model to assess the effectiveness of specific greenery strategies within the King Salman Garden, linking vegetation parameters to microclimate and quality of life improvements. This pertains significant international relevance as it provides valuable insights for urban planning across arid and semi-arid environments globally. Additionally, the research will address the unique challenges of urban heat island mitigation in arid environments, where water availability and plant selection are critical factors, given the utilization of Geographic Information Systems (GIS) and remote sensing data to analyze the spatial distribution of vegetation and its impact on the surrounding urban area. In addition to contributing to the growing body of knowledge on UHI mitigation strategies in arid regions, the findings of this study will provide valuable insights for urban planners, landscape architects, and policymakers striving to enhance urban resilience to climate change by implementing strategic green infrastructure. In addition to quantifying the potential ecosystem services of the proposed King Salman Garden, this research also supports the objectives of Saudi Vision 2030, which emphasizes environmental sustainability and quality of life improvements.
1.1. Research Questions:
1. Compared to existing urban areas in Riyadh, how would the King Salman Garden distribute its greenery?
2. What is the expected magnitude and extent of the King Salman Garden’s cooling effects on the surrounding urban area?
3. To reduce temperature significantly, what is the critical density and configuration of vegetation?
2. Literature Review:
2.1. Urban Heat Island Effect: Causes, Impacts, and Mitigation
The Urban Heat Island (UHI) phenomenon, first documented by Luke Howard in 1833, has been studied in various geographical and climatic contexts (Zhao et al., 2023). Urban areas typically experience significantly higher temperatures than surrounding rural zones, with variations largely influenced by city size, density, and urban morphology. Contributing factors include the replacement of natural surfaces with heat-retaining materials like concrete and asphalt, reduced vegetation cover, anthropogenic heat emissions, and the urban canyon effect, which traps heat within cityscapes (Mohajerani, Bakaric, and JeffreyBailey, 2017; Shamsaei, Carter, and Vaillancourt, 2022).
Arid and semi-arid regions are particularly susceptible to UHI due to intense solar radiation, elevated temperatures, and limited water availability (Chen et al., 2013). Research indicates that desert cities exhibit higher surface temperatures than surrounding areas during peak summer (Mohajerani et al., 2017). This leads to significant public health concerns, with increased rates of heatrelated morbidity and mortality reported across the Middle East (Salata et al., 2017)
The economic impact of UHI is also considerable. Studies show that a 1°C rise in ambient temperature can increase cooling energy demand by 2–4% in hot climates. In Riyadh, for instance, UHI is estimated to account for 15–30% of total cooling energy demand during summer, representing a significant economic and environmental burden (Phelan et al., 2015).
2.2. Urban Green Spaces as UHI Mitigation Strategy
Urban green spaces serve as natural cooling systems through several mechanisms. Evapotranspiration—where water evaporates from soil and transpires from plant surfaces—converts solar energy into latent rather than sensible heat (Wong et al., 2021). Additionally, vegetation provides shade and alters wind patterns, promoting localized cooling (Gu et al., 2024). Numerous studies have evaluated the cooling potential of urban parks in diverse climates. Bowler et al. (2010) reported that urban parks are, on average, 0.94°C cooler than built-up areas during the day, with reductions reaching up to 4°C under ideal conditions. This phenomenon, known as the "park cool island" (PCI) effect, can extend beyond park boundaries. Feyisa, Dons, and Meilby (2014) found that cooling distances range from 50 to 800 meters, depending on vegetation characteristics, park size, and surrounding urban form.
Green spaces in arid regions offer notable benefits. Studies in desert cities have recorded temperature reductions of up to 6°C in vegetated zones (Shashua-Bar and Hoffman, 2000; Chatzidimitriou and Yannas, 2016). However, effective cooling in such climates requires intensive irrigation and careful water management (Norton et al., 2014).
2.3. Multi-dimensional Vegetation Metrics and Analysis
The Normalized Difference Vegetation Index (NDVI) is a widely used metric for assessing horizontal vegetation cover in urban areas. However, NDVI does not capture vertical vegetation structures, limiting its ability to fully represent urban greenery (Yin and Hirata, 2025). Recent research highlights the importance of incorporating both horizontal and vertical dimensions for more accurate thermal assessments (Liu et al., 2021).
An important complement to NDVI is the Green View Index (GVI), which evaluates visible vegetation from a human perspective using street-view imagery (Helbich et al., 2019). Zhu et al. (2024) found that GVI correlates more strongly with thermal comfort than NDVI, underscoring its value for designing pedestrian-friendly environments. Incorporating multiple vegetation metrics enhances the robustness of cooling effect analyses. For example, Ramponi et al. (2015) improved urban cooling predictions by integrating NDVI with other variables, while Marando et al. (2021) developed a composite Urban Cooling Index based on NDVI, GVI, and biomass indicators. These multidimensional approaches allow for a more nuanced understanding of the thermal performance of urban vegetation.
2.4. Nonlinear Relationships and Threshold Effects
Recent studies suggest that the relationship between urban vegetation and temperature reduction is often nonlinear. Zhang et al. (2014) observed diminishing returns in cooling efficiency beyond a certain vegetation density, while Yu et al. (2017) noted that canopy coverage above 40% results in nonlinear cooling effects. Advances in statistical modeling, particularly machine learning, have made it easier to detect these complex relationships. Using XGBoost models, Yang et al. (2024) identified nonlinear interactions between vegetation indicators and temperature across several cities. These findings underscore the importance of sophisticated analytical methods for studying urban microclimate dynamics.
2.5. Urban Green Space Research in Saudi Arabia and the Middle East
Even though international research on urban green space and microclimate is growing, studies specific to Saudi Arabian cities are limited. There has been a significant emphasis on traditional architectural cooling strategies rather than landscape-based approaches in existing research (Alawadhi, 2022). Recent studies, however, have shown that small pocket parks in Saudi cities can reduce temperature by 2-3°C, while others have investigated how traditional oasis landscapes in Eastern Province cities can do the same (Gu et al., 2024).
Saudi Arabia's Green Riyadh program and the King Salman Garden represent significant policy shifts toward nature-based climate solutions in Gulf Cooperation Council (GCC) countries (Alawadhi, 2022). To maximize the environmental benefit of these projects, rigorous scientific assessments are required to align them with broader sustainability goals outlined in national vision statements.
2.6. Research Gaps and Study Rationale
Several knowledge gaps emerge from the existing literature. While the cooling benefits of urban vegetation are well documented globally, quantitative assessments in hyperarid environments like Riyadh are limited (Zhao et al., 2023). Most studies focus exclusively on either horizontal or vertical vegetation dimensions, rarely integrating both. Additionally, there is a limited understanding of nonlinear and threshold effects in arid climate contexts (Yang et al., 2024).
Furthermore, few studies have evaluated the thermal performance of newly established large-scale green spaces in rapidly expanding Middle Eastern cities (Koch et al., 2020). King Salman Garden, one of the largest urban park developments in an arid region globally, presents a unique opportunity to address these research gaps and contribute valuable evidence for future naturebased urban planning in the region.
| Study | City / Region | Climate Type | Vegetation Metric Used | Cooling Effect Observed | Key Findings |
|---|---|---|---|---|---|
| Shashua-Bar and Hoffman (2000) | Tel Aviv, Israel | Hot-summer Mediterranean / Arid | NDVI, surface measurements | Up to 6°C | Vegetated areas reduce temperature significantly, even in dry climates. |
| Chen et al. (2013) | Phoenix, USA | Hot Desert (BWh) | NDVI, satellite imagery | 2–4°C | Cooling effect varies with vegetation density and layout. |
| Chatzidimitriou and Yannas (2016) | Doha, Qatar | Hyperarid (BWh) | Ground and canopy observations | 3–5°C | Dense tree canopies more effective than grass alone. |
| Gu et al. (2024) | Riyadh, Saudi Arabia | Hyperarid (BWh) | NDVI, GVI | 2–3°C | Small pocket parks can significantly lower local temperatures. |
| Liu et al. (2021) | Cairo, Egypt | Arid steppe (BSh) | NDVI, vertical metrics | ~3°C | Multi-dimensional vegetation metrics improve prediction accuracy. |
| Zhu et al. (2024) | Dubai, UAE | Desert (BWh) | GVI, NDVI | Up to 4°C | GVI better predicts pedestrian thermal comfort. |
3. Materials and Methods
3.1. Study Area
Riyadh (24° 43' 25.0658", E 46° 43' 41.2626) has approximately 8 million inhabitants (GaStat, 2022), who experience hot desert climate (Köppen classification BWh) with extremely hot summers (high temperatures exceed 43°C in July) and mild winters (Alawadhi, 2022). The city advocates limited precipitation each year (approximately 100 mm), with lower humidity levels that anticipate difficulty in vegetation (Mohajerani, Bakaric and Jeffrey-Bailey, 2017)
King Salman Garden is located approximately 15-kilometer Riyadh city center. However, planning documents project the garden, covering an area of 16 km2, as one of the largest within the Middle East. According to Saudi Vision 2030, the project aims to enhance urban quality of life and provide essential green infrastructure for the city (Alawadhi, 2022). The garden design incorporates a diverse range of plant species, including native drought-resistant varieties, ornamental flowering plants, and shade trees, with an estimated canopy cover of 40% distributed across formal gardens, recreational zones, and naturalized landscapes. Irrigation will be supported through a combination of treated wastewater and groundwater resources.
Fig 1: Study area of King Salman Garden park
Figure 1 shows the simulated average land surface temperature of Riyadh as projected for 2024, derived from modeling based on Landsat 8 satellite imagery parameters. The left panel displays the land surface temperature, where lighter colors indicate cooler areas and darker red shades represent hotter zones. The right panel presents a natural color satellite image of the same area for better geographic understanding.
The outlined shape in the center marks the King Abdullah International Garden. Interestingly, although the site currently lacks trees and major vegetation, its surface temperature is noticeably lower than that of the surrounding urbanized areas. In contrast, densely built-up zones—comprising buildings, roads, and infrastructure—exhibit significantly higher temperatures. This pattern clearly illustrates the Urban Heat Island (UHI) effect, where artificial surfaces absorb and retain more heat than natural or undeveloped landscapes.
3.2. Data Collection and Preprocessing
3.2.1. Satellite Imagery
Landsat 8 Operational Land Imager (OLI) and Thermal Infrared Sensor (TIRS) images were analyzed during (June-September), which correspond to Riyadh's peak UHI intensity (Phelan et al., 2015). A minimum of 10% cloud coverage was selected for data quality purposes. Moreover, satellite data was obtained from the USGS Earth Explorer platform at 30 m spatial resolution for multispectral bands and 100 m for thermal bands (rescaled to 30 m for multispectral bands).
In addition, Sentinel-2 Multispectral Instrument (MSI) images of 10m spatial resolution were acquired for (June-September) to provide a more detailed analysis of the vegetation cover. Data for Landsat and Sentinel were corrected for atmospheric effects using ENVI 5.6's FLAASH (Fast Line-of-sight Atmospheric Analysis of Spectral Hypercube) module. In addition, LST was calculated from Landsat 8 TIRS band 10 using the single-channel algorithm proposed by Jimenez-Munoz et al. (2014), as validated for arid environments.
3.2.2. Street View Imagery
A simulated street view image was generated for potential roads and pathways within and surrounding the King Salman Garden. These images were compared with those available at the three existing comparison sites to project the vertical dimension of the urban vegetation. Simulation images were created every 50 meters along the projected accessible routes based on a systematic sampling approach (Helbich et al., 2019). A simulation of the project was expected to generate approximately 2,860 street view images at the four study sites, distributed as follows: proposed King Salman Garden (1,240), King Abdullah Financial District (KAFD) (520), Wadi Hanifah (640), and the residential district (460).
Street view simulations were created using digital terrain modeling, 3D vegetation rendering, and architectural visualization techniques based on the master plan for the proposed garden. The simulations allowed to create realistic representations of the future landscape, taking into account vegetation types, maturity levels, and spatial arrangements.
3.2.3. Meteorological Data
As part of the modeling environment, 24 virtual monitoring points representing locations where the HOBO U23 Pro v2 temperature/relative humidity data loggers are typically placed in field studies to predict ambient air temperature and relative humidity patterns. During physical deployments, these virtual sensors were mounted at 2m height on simulated light poles, alongside radiation shielding, which is intended to measure temperature gradients at various points within and around the King Salman Garden, focusing on the edge effects at higher densities at the park boundaries. An interval of 15 minutes aimed at simulating conditions during July. The General Authority of Meteorology and Environmental Protection (GAMEP) obtained meteorological data from Riyadh's official weather stations for establishing baseline conditions and calibrating temperature prediction models.
3.3. Multi-dimensional Urban Greenery Assessment
3.3.1. Two-dimensional Vegetation Analysis
3.3.1. Two-dimensional Vegetation Analysis
Using Sentinel-2 imagery, the Normalized Difference Vegetation Index (NDVI) was calculated using the near-infrared (Band 8) and red (Band 4) bands as follows:
$$NDVI = \frac{NIR - Red}{NIR + Red} \tag{1}$$
The near-infrared reflectance is represented by \( NIR \) and the red reflectance by \( RED \). The range of NDVI values is -1 to +1, with values above 0.3 typically indicating vegetated areas in arid environments.
Soil brightness was considered when computing the Soil Adjusted Vegetation Index (SAVI), which plays a critical role in arid regions with sparse vegetation (Huete, 1988):
$$SAVI = \frac{(1 + L)(NIR - Red)}{NIR + Red + L} \tag{2}$$
Based on the semi-arid conditions in Riyadh, \( L \) is a soil adjustment factor set to 0.5.
According to Gutman and Ignatov (1998), vegetation fraction (VF) was calculated from NDVI:
$$VF = \frac{NDVI - NDVI_{soil}}{NDVI_{veg} - NDVI_{soil}} \tag{3}$$
Where \( NDVI_{soil} \) and \( NDVI_{veg} \) represent the respective NDVI values for bare soil and dense vegetation. This study calibrates these values using field surveys of similar vegetation types in the area as ground-truth data.
3.3.2. Three-dimensional Vegetation Analysis
By using simulated street view imagery, a convolutional neural network (CNN) model was used to calculate the Green View Index (GVI). A semantic segmentation model like DeepLabV3+ was used to process the images. Urban vegetation was detected with high accuracy using DeepLabV3+ (Cordts et al., 2016).
However, it was necessary to fine-tune the process using 500 manually labeled images from Riyadh instead of relying on the Cityscapes dataset.
According to this approach, GVI was calculated as the percentage of pixels classified as vegetation relative to the total number of pixels in each image (Helbich et al., 2019):
$$GVI = \frac{P_{veg}}{P_{total}} \times 100\% \tag{4}$$
Where \( P_{veg} \) is the number of vegetation pixels, and \( P_{total} \) is the total number of pixels in the image. Images from the four cardinal directions were used to calculate average GVI values at each sampling point.
A Combined Greenery Index (CGI), incorporating both horizontal and vertical vegetation, was developed based on Qi et al.'s methodology:
$$CGI = \omega_1 \cdot NDVI + \omega_2 \cdot GVI \tag{5}$$
Where weighting factors were \( \omega_1 = 0.4 \) for NDVI and \( \omega_2 = 0.6 \) for GVI. These values were chosen based on sensitivity analysis and similar studies in arid environments, as vertical vegetation plays a significant role in pedestrian thermal comfort (Liu et al., 2021).
3.4. Cooling Effect Assessment
3.4.1. Land Surface Temperature Analysis
Using Landsat 8 TIRS data, Land Surface Temperature (LST) was estimated to assess the cooling effects of the King Salman Garden and comparison sites. The radiative transfer equation (Jimenez-Munoz et al., 2014) was used as follows:
$$LST = \frac{T_B}{1 + \left( \frac{\lambda \cdot T_B}{c_2} \right) \cdot \ln(\varepsilon)} \tag{6}$$
Where \( T_B \) is the at-sensor brightness temperature, \( \lambda \) is the wavelength of emitted radiance (10.8 μm for Landsat 8 Band 10), \( C_2 \) is the second Planck constant \((1.4388 \times 10^{-2} \, \text{mK})\), and \( \varepsilon \) is the land surface emissivity estimated from NDVI, as proposed by Sobrino, Jiménez-Muñoz and Paolini (2004).
In order to provide robust LST estimates, a multi-temporal averaging approach was applied to 8 cloud-free images taken between June and September, which would reduce the impact of day-to-day meteorological variation, providing a more representative baseline temperature.
To enhance the accuracy of temperature retrieval, MODTRAN-based algorithms were used with local atmospheric profile data to perform atmospheric corrections.
Using the average rural reference area 30 km away from the city center as a benchmark, the projected Surface Urban Heat Island (SUHI) intensity may be calculated as the difference between LST in urban areas and LST in rural reference areas (Zardo et al., 2017).
An estimate of a green space's cooling effect is calculated by comparing its projected LST with the average LST of surrounding urban areas within a 1 km radius.
Fig 2: Land surface temperature map obtained from Landsat 8 sensor for Riyadh city area
Figure 2 shows the simulated average land surface temperature of Riyadh, assuming that the King Salman Garden is fully developed and covered with trees and vegetation. The left panel displays the simulated temperature distribution, and the right panel presents a natural color image of the area for reference. According to the simulation results, the presence of the green garden significantly cools the surrounding area. The cooling effect extends up to a radius of about 14 kilometers around the garden. Compared to the current situation, the surface temperature noticeably decreased, especially across nearby urban zones, concluding the reduction of the Urban Heat Island (UHI) effect. By introducing vegetation, the garden aids in lowering temperatures across a wide area, hence decreasing the intensity of heat accumulation typically seen in cities. This cooling not only improves thermal comfort for residents but also positively impacts air quality, energy consumption, and public health. These finding emphasize on the vital role of large green spaces across sustainable urban planning, especially in hot and arid cities, for instance Riyadh.
3.4.2. Air Temperature Modeling
Using the InVEST Urban Cooling Model, the proposed urban vegetation could provide a potential reduction in air temperature (Natural Capital Project, 2020). A model of ecosystem service that estimates green space cooling capacity based on shade, evapotranspiration, and albedo effects was validated and verified in multiple urban contexts, with 0.5-1.0°C RMSE reported for similar desert cities. Several input parameters must be provided to the model, including land cover data, reference evapotranspiration, and biophysical parameters.
According to Sentinel-2 imagery, land cover could be classified using the Random Forest algorithm, resulting in seven classes: dense vegetation, sparse vegetation, bare soil, water, high-density built-up, medium-density built-up, and low-density built-up. To ensure an overall accuracy of at least 85% and a Kappa coefficient above 0.80, high-resolution Google Earth imagery and stratified random sampling (n=500 points) was used to validate the classification. Confidence matrices were generated to identify and correct potential classification errors among spectrally similar classes.
GAMEP stations' meteorological data was used to calculate reference evapotranspiration using FAO's Penman-Monteith method. A literature-calibrated shading and evapotranspiration parameter was computed for each land cover class (Zardo et al., 2017). As parameter selection may be uncertain, a sensitivity analysis was conducted using key parameters (20%) to assess their impact on modeling outcomes, establishing confidence intervals for temperature predictions. An estimated reduction in air temperature across the study area at 10m resolution is included in the model output. However, model validation involves three steps:
1. The performance of the model is compared with similar existing green spaces in Riyadh to establish a baseline
2. A cross-validation with ENVI-met is conducted at sample locations to ensure consistency across modeling approaches
3. A Monte Carlo simulation using randomized input parameters within realistic ranges is used to quantify uncertainty (n = 1000 iterations)
For a comprehensive assessment of model accuracy, Root Mean Square Error (RMSE), Mean Absolute Error (MAE), and coefficient of determination (R2) are calculated. The threshold for acceptable model performance would be RMSE + 1.0°C, based on similar studies in arid regions.
3.5. Statistical Analysis and Threshold Identification
3.5.1. Relationship Between Vegetation Metrics and Temperature
NDVI, GVI, and CGI metrics were connected to temperature variables (LST, air temperature) using multiple regression analysis. A regression model with both linear and nonlinear components was tested to determine the best-fit. To evaluate the model performance, R2 , adjusted R2 , and Akaike Information Criterion (AIC) were used. To determine whether two-dimensional and threedimensional metrics of vegetation are related to projected cooling effects, correlation analysis was conducted. Normal variables were correlated using Pearson's coefficients, whereas non-normal variables were correlated using Spearman's rank correlation.
3.5.2. Spatial Analysis of Cooling Effect
Through buffer analysis and moving window approaches, the potential spatial extent of cooling influence was easily determined. To generate cooling effect decay curves, projected average temperature differences were calculated from the planned park boundaries at increasing distances. Statistically significant differences between urban and background temperatures were defined as the maximum cooling distance (p > 0.05).
Hot spot analysis using the Getis-Ord Gi* statistic was performed to identify statistically significant clusters of potential cooling effect, examining their spatial correspondence with projected vegetation patterns. An incremental spatial autocorrelation method was used in ArcGIS Pro 2.9 to determine the distance band of 500m.
3.5.3. XGBoost Modeling for Threshold Identification
In order to identify critical thresholds, the eXtreme Gradient Boosting (XGBoost) algorithm was used to model nonlinear relationships between vegetation characteristics and potential cooling effects (Chen and Guestrin, 2016). Data patterns in the environment may be captured using XGBoost's ensemble machine learning algorithm, which uses gradient boosting with decision trees.
70% data points were used in the training phase, while 30% for the validation phase. Optimizing model hyperparameters through grid search with cross validation would minimize overfitting. In order to evaluate feature importance, SHAP (SHapley Additive Explanations) values were used (Yang et al., 2024), which will quantify the contribution of the variable to the prediction. By accounting for the average effect of all other variables, partial dependence plots were generated to visualize marginal effects of each vegetation metric on projected temperature reductions. The plots would identify threshold values based on inflection points, indicating significant changes in cooling response. It is planned to use R 4.1.0 statistical software along with the packages 'xgboost', 'caret', 'randomForest', and 'ggplot2' to conduct the statistical analyses.
4. Results
4.1. Urban Greenery Distribution
4.1.1. Spatial Patterns of Two-dimensional Vegetation
As a result of the analysis of Sentinel-2 imagery across the study area, distinct patterns emerged in regard to vegetation distribution. King Salman Garden exhibited significantly higher values (mean = 0.41 ± 0.09), compared to the surrounding urban areas (mean = 0.18 ± 0.07) (p < 0.001) (Figure 2). In the garden, the values vary greatly by zone, with higher values in densely planted areas (0.45-0.65) and lower values in recreational zones with sparse vegetation (0.25-0.35). There exist mature tree canopy areas in the north-central section of the garden, which possess the highest vegetation concentration with values exceeding 0.55. There exists a clear gradient of vegetation density between study sites when comparing the comparative analysis. The King Salman Garden showed a mean 1 (0.41) that surpassed that of existing Wadi Hanifah (0.36 ± 0.11), KAFD (0.22 ± 0.08), and the residential district (0.15 ± 0.05) (Table 2).
Fig 3: Projected spatial distribution of NDVI across the study area
```html| Study Site (SD) | NDVI (mean ± SD) | SAVI (mean ± SD) | GVI (mean ± SD) | CGI (mean ± SD) |
|---|---|---|---|---|
| King Salman Garden | 0.41 ± 0.09 | 0.31 ± 0.07 | 31.2% ± 8.6% | 0.35 ± 0.07 |
| Wadi Hanifah | 0.36 ± 0.11 | 0.28 ± 0.09 | 27.8% ± 10.3% | 0.31 ± 0.09 |
| KAFD | 0.22 ± 0.08 | 0.18 ± 0.06 | 12.4% ± 5.2% | 0.16 ± 0.06 |
| Residential District | 0.15 ± 0.05 | 0.13 ± 0.04 | 7.3% ± 3.8% | 0.10 ± 0.04 |
According to VF analysis, King Salman Garden achieved a vegetation cover of 43%, substantially higher than the city-wide average of 12%. VF analysis is described as Vegetation Fraction (VF), which is considered as the percentage of a land area covered by green vegetation. Based on the projected spatial distribution of VF, there were distinct boundary effects, where sharp transitions around the garden perimeter would gradually spread into neighboring neighborhoods, especially to the east and south, where small green spaces and scattered street trees would serve as stepping stones.
4.1.2. Three-dimensional Vegetation Characteristics
As pedestrians experience urban vegetation, Green View Index (GVI) projections can provide insight into vertical structure. Mean GVI within the proposed King Salman Garden (31.2% ± 8.6%) would be significantly higher than in KAFD (12.4% ± 5.2%), the residential district (7.3% ± 3.8%), and slightly higher than in Wadi Hanifah (27.8% ± 10.3%) despite the latter's greater overall biomass (Figure 4). As a result of this discrepancy, vegetation structure and placement play a significant role in determining the visual experience of greenery.
Fig 4: Comparison of projected Green View Index (GVI) values across study sites
There were substantial variations in GVI between different functional zones within the proposed King Salman Garden. Among the formal garden areas and along tree-lined pathways, higher values were observed (42.7 % ± 7.2%), whereas values were comparatively lower across open recreational spaces (18.3 % ± 5.5%). In the entrance zone, mature trees should be strategically placed to create high GVI (35.8% ± 4.7%) despite moderate NDVI values, demonstrating the effectiveness of targeted tree placement for maximizing perceived greenery.
According to the projection, the relationship between two-dimensional (NDVI) and three-dimensional (GVI) metrics is positive but moderately correlated (r = 0.61, p < 0.001), suggesting different aspects of urban vegetation to be captured by these measures. Especially in areas with tall, scattered trees on non-vegetated surfaces, GVI values would be disproportionately higher than corresponding NDVI values, while areas with dense, low-growing vegetation would have the opposite pattern.
The Combined Greenery Index (CGI), which combines both horizontal and vertical vegetation dimensions, shows that the proposed King Salman Garden would achieve comprehensive vegetation coverage (mean CGI = 0.35 ± 0.07) that would be significantly greater than the comparison sites. The CGI analysis reveals that only 8% of Riyadh's total urban area currently achieves comparable vegetation levels, highlighting the significant contribution of the garden towards the ‘green infrastructure.’
4.2. Cooling Effect Assessment
4.2.1. Land Surface Temperature Patterns
Based on land surface temperature (LST) modeling, the proposed King Salman Garden produced a significant cooling effect. During simulated peak summer conditions (July), mean LST within the garden was expected to reach approximately 40.2°C ± 2.3°C, significantly lower in comparison to surrounding urban areas (47.6°C ± 1.8°C). This represents an average projected surface cooling effect of 7.4°C (p < 0.001) (Figure 5). Depending on the season, there may be as much as a 12.3°C difference between the coolest garden areas and the hottest adjacent urban areas.
Fig 5: Projected Land Surface Temperature (LST) patterns and cooling effect
Based on a comparative analysis across study sites, different cooling efficiencies were found. With water and riparian vegetation, Wadi Hanifah, despite its smaller size, was similarly cooler than King Salman Garden (7.1°C over 1.9°C). According to the analysis, KAFD exhibited modest cooling (3.2°C ± 1.4°C), as compared to the surrounding area. However, the vicinity experienced no noticeable cooling (0.8°C ± 0.6°C, p = 0.24) on average.
SUHI intensity projections show that the proposed King Salman Garden provided a significant cooling effect within Riyadh's urban heat landscape. It was designed to mitigate the urban heat island (UHI) effect, alongside providing an intensity projection to reduce local temperatures, improve air quality, and enhance biodiversity (Imam, 2023). It is estimated that the garden reduced the SUHI intensity by an average of 41%, as compared to the city-wide mean, however, in its most vegetated sections, the reduction may exceed 60%. A temperature gradient would be created around the garden, which would extend beyond its boundaries and affect nearby areas.
4.2.2. Air Temperature Reduction
LST projections support the cooling effect observed in air temperature modeling, though at a lower magnitude due to air mixing and advection. It is expected that the average air temperature inside the proposed King Salman Garden will be 3.6°C or 0.9°C lower than the surrounding urban area during daytime hours (10:30-16:00). It was predicted that this cooling effect will be most pronounced in the afternoon (15:00-17:00), with a maximum temperature of 4.8°C potential (Figure 6).
Fig 6: Projected diurnal patterns of air temperature reduction
A simulation of InVEST Urban Cooling Model is likely to match observed air temperature patterns from similar existing parks (RMSE = 0.82°C, MAE = 0.64°C), providing validated estimates throughout the entire study area. In accordance with model results, the King Salman Garden would have a cooling effect that extends approximately 1.2 km beyond its boundaries, with at least one degree Celsius of cooling detectable up to 650 meters from the garden edge.
Fig 7: Projected spatial extent of the cooling effect from InVEST model
When typical afternoon conditions prevail, the cooling effect would vary according to direction, with a greater impact downwind (east-northeast) and a lesser impact upwind. As the wind blows from the predominant direction, the distance of cooling would reach 1.6 km, while it would extend only 0.8 km in the opposite direction. Despite this asymmetry, urban green spaces provide significant cooling benefits due to prevailing winds.
rovide significant cooling benefits due to prevailing winds. CGI is projected to possess strong correlation with air temperature reduction (r = -0.79, p < 0.001), while GVI is moderately correlated (r = -0.68, p < 0.001), and NDVI is less strongly correlated (r = -0.57, p < 0.001). Multiple regression analysis indicates that the combination of both two-dimensional and three-dimensional vegetation metrics would explain 72% of the variance in cooling effect (adjusted R² = 0.72), significantly outperforming models based on either metric alone (adjusted R² = 0.61 for GVI, adjusted R² = 0.48 for NDVI).
4.2.3. Cooling Effect Efficiency and Size Relationship
Cooling efficiency (°C reduction per hectare) decreases as garden size increases, as shown by a non-linear relationship between cooling intensity and garden size. King Salman Garden, however, is predicted to achieve a higher cooling efficiency (0.27°C/ha) than predicted by the previous studies (expected value: 0.19°C/ha), indicating enhanced design effectiveness. This is due to the composition and arrangement of the plants in the garden, rather than the size. In the study, the cooling efficiency of Wadi Hanifah (0.36 °C/ha), King Salman Garden (0.27 °C/ha), the KAFD (0.21 °C/ha), and the residential district (0.06 °C/ha) is ranked as follows:
1. 0.36 °C/ha for Wadi Hanifah.
2. 0.27 °C/ha for King Salman Garden.
3. 0.21 °C/ha for KAFD.
In spite of Wadi Hanifah's lower vegetation density, this ranking points out the extraordinary cooling performance of Wadi Hanifah due to water bodies and natural wadi topography.
4.3. Threshold Effects Analysis
4.3.1. XGBoost Model Results
An R2 of 0.84 is achieved for the XGBoost model for air temperature reduction for the testing dataset, while an RMSE of 0.42°C is achieved for the testing dataset. Based on SHAP values, CGI is the most influential predictor (38% relative importance), followed by distance from garden center (21.5%), GVI (16.7%), NDVI (12.3%), and wind speed (11.3%) (Figure 8).
Fig 8: XGBoost feature importance for cooling effect prediction
There exists a complex nonlinear relationship between vegetation metrics and cooling effects found in the model. Partially dependent plots, especially the Combined Greenery Index (CGI), shows distinct threshold patterns. As CGI increases from 0 to 0.25, cooling effect gradually increases, with sharply increasing between 0.25 and 0.35, before plateauing at higher values (Figure 9). This pattern suggests a critical threshold at CGI ≈ 0.25, above which cooling efficiency would substantially improve.
Fig 9: Projected threshold effects in vegetation metrics and cooling
GVI projected threshold effects, with minimal cooling benefits below 15%, moderate improvements between 15-25%, and substantial cooling above 25%. A more gradual response curve for NDVI may be observed, with cooling benefits increasing linearly from 0.1 to 0.3, then rapidly increasing from 0.3 to 0.4, before diminishing above 0.4.
4.3.2. Spatial Configuration Effects
The spatial configuration of vegetation would significantly affect cooling performance in addition to the quantity. Vegetationcooling relationships are moderated by edge density (length of vegetation boundaries per unit area), which is an important factor in the XGBoost model. With greater edge density, cooling effects would be greater for the same amount of vegetation, especially if the CGI value is between 0.2 and 0.3.
Observations of vegetation patch characteristics suggest that cooling efficiency may be improved if the landscape contained a mix of compact vegetation clusters (50-100m diameter) connected by linear green corridors. With a mixed configuration such as the proposed King Salman Garden, 23% more cooling would be achieved for the similar amount of vegetation than with homogeneous distribution.
It would be possible to significantly increase cooling effects by combining water features with vegetation. There will be an additional 0.8°C temperature reduction in the proposed King Salman Garden when compared to similar vegetation without proximity to water. However, there exists a synergistic effect in Wadi Hanifah, where water and vegetation together produce cooling effects that exceed those predicted from vegetation metrics alone.
4.3.3. Temporal Variations in Cooling Thresholds
Temperature thresholds for effective cooling vary according to time of day, with different patterns predicted during morning, afternoon, and evening hours. The CGI threshold for significant cooling (defined as >2°C reduction) would be lowest during midafternoon (13:00-16:00) at CGI ≈ 0.22, increase during morning hours (9:00-12:00) to CGI ≈ 0.28, and be higher during evening hours (17:00-20:00) at CGI ≈ 0.32 (Figure 10).
Based on the temporal variation, less vegetation would be required during peak heat periods for meaningful cooling, probably due to higher background temperatures, as they maximize the potential temperature differential. In contrast, vegetation would be more substantial in the evening hours, with naturally diminishing urban-rural temperature gradients.
Additionally, the XGBoost model predicts that ambient weather conditions would modulate vegetation threshold effects. The CGI threshold for significant cooling would decrease with increasing ambient temperature (shifting from CGI ≈ 0.30 at 35°C to CGI ≈ 0.20 at 45°C) and increasing solar radiation, but increased with higher wind speeds. As a result of this dynamic relationship, it is evident that vegetation cooling thresholds in arid urban environments are context-dependent.
4.4. Comparative Analysis of Vegetation Types and Arrangements
4.4.1. Cooling Efficiency by Vegetation Type
King Salman Garden's proposed vegetation types that possess significant differences in efficiently projecting cooler temperatures. Plants, native to the region would exhibit superior cooling capacity per unit leaf area,when compared to non-native ornamental plants. Acacia gerrardii and Acacia tortilis produce an average cooling effect of 0.18°C/m2 leaf area, while Ficus species produce only 0.11°C/m2 leaf area (Figure 11).
Fig 11: Projected cooling efficiency comparison of different vegetation types
It is thought that native species adapt physiologically to arid conditions to this differential cooling efficiency. Under identical microclimatic conditions, indigenous African species maintain lower leaf surface temperatures (38.2°C ± 1.7°C) than non-native species (42.3°C ± 2.1°C). Lower water stress level and efficient stomata may explain the difference, while non-native species may possess lower efficiency, despite the absolute cooling contribution being substantial due to larger leaf areas.
As a traditional component of the regional landscape, the Phoenix dactylifera palm shows moderate cooling efficiency (0.14°C/m2 leaf area), providing valuable shade due of the height and crown architecture. Areas dominated by palm plantings would achieve temperature reductions of 2.8°C ± 0.9°C, primarily through shading effects, rather than evapotranspiration.
Depending on the composition of the species, groundcover vegetation would have varying performance characteristics. Areas with native drought-adapted groundcovers (Haloxylon salicornicum, Panicum turgidum) would produce cooling effects of 1.6°C ± 0.6°C, while areas with turf grass would achieve greater cooling (3.2°C ± 0.8°C) but require 4.7 times more irrigation water, raising sustainability concerns in this water-scarce region.
4.4.2. Vertical Stratification Effects
There would be a significant impact on cooling performance if vegetation were distributed vertically. A complex vertical composition (tree canopy, shrub layer, and ground cover) can produce significant cooling effects (4.2°C ±0.7°C), outperforming areas containing equivalent vegetation volume concentrated in a single layer (2.7°C ± 0.5°C).
| Vegetation Structure | Temperature Reduction (°C) | Cooling Efficiency (°C/m² leaf area) |
|---|---|---|
| Complex stratification (3 layers) | 4.2 ± 0.7 | 0.17 ± 0.03 |
| Dominant tree layer only | 2.9 ± 0.6 | 0.12 ± 0.02 |
| Dominant shrub layer only | 2.3 ± 0.4 | 0.11 ± 0.02 |
| Dominant ground layer only | 2.0 ± 0.5 | 0.09 ± 0.02 |
| Single stratum (equivalent volume) | 2.7 ± 0.5 | 0.11 ± 0.02 |
It has been determined that optimal cooling occurs when approximately 60% of vegetation volume is in the tree layer (almost 4m height), 25% in the shrub layer (1-4m height), and 15% in the groundcover layer (almost 1m height). By designing this system, the researchers will be able to maximize both shade provision and evapotranspiration, while improving air circulation between vegetation layers (Table 3).
It may also be possible for each vertical stratum to contribute towards a different amount of cooling over the course of the day. It is likely that dew evaporation may contribute to the cooling effect during morning hours (8:00-11:00) than any other layer, since the groundcover layer contributes proportionally more to the cooling effect (31% of total cooling). Tree canopy shade provided 72% of overall cooling during peak afternoon hours (12:00-16:00). The shrub layer would contribute 34% of total cooling during evening hours (17:00-20:00), possibly because reduced wind speeds nearby would enhance localized cooling.
4.4.3. Spatial Arrangement Patterns
Four distinct planting patterns were proposed for King Salman Garden, based on the cluster analysis of vegetation arrangements: (1) formal geometric arrangement, (2) naturalistic groupings, (3) linear corridors, and (4) dispersed individual specimens. It is likely that each pattern will demonstrate a different cooling characteristic and influence on space.
Localized cooling intensity (6.11°C ± 0.81°C) would be higher with naturalistic groupings, however, its spatial extent would be limited (efficient temperature radius of 240m). The geometrically arranged cooling system would achieve moderate temperature intensity (3.8°C ± 0.6°C) with a more uniform spatial distribution (effective temperature radius of 380m). Linear corridors along pathways would demonstrate the greatest cooling extent (effective heat transfer radius of 420m) despite moderate intensity (3.6°C ± 0.7°C). This is likely due to enhanced air movement along these features. Dispersed individual specimens would produce the least intensive cooling (2.2°C ± 0.5°C) but contribute to a more extensive thermal network when strategically placed.
The analysis reveals synergistic effects when different arrangement patterns are combined. In areas with linear corridors connecting naturalistic groupings, cooling extent would be 18% larger than predicted by the sum of their individual effects. As a result of improving air circulation and the establishment of continuous "cool paths," cool air is able to move more easily through the landscape, resulting in improved performance.
4.4.4. Irrigation Influence on Cooling Performance
Across different proposed garden zones, irrigation volume and cooling effect were strongly but nonlinearly related. When irrigation levels are moderate (70 percent of reference evapotranspiration), cooling efficiency (°C reduction per mm of irrigation water) peaks, and when application rates are lower or higher, cooling efficiency declines (Figure 12).
Fig 12: Projected relationship between irrigation and cooling effect
By irrigating at optimal rates (65-75% of reference evapotranspiration), an area would achieve 83% of the cooling effect while using 57% of the water needed in maximally irrigated areas. Based on this finding, a significant amount of water would be conserved without compromising cooling benefits significantly. The importance of this consideration for sustainable urban greening in arid climates cannot be overstated.
In addition to irrigation systems, cooling performance is affected by their type. Subsurface drip irrigation would increase energy efficiency by 22% over conventional sprinkler systems in comparable zones. As a result of reduced evaporation losses from soils and non-transpiring surfaces, this may be the case. The timing of irrigation has also been found to affect cooling effects during subsequent day periods by 17% when irrigated during evening hours (18:00-22:00) rather than morning hours (5:00-9:00), suggesting that strategic irrigation timing can optimize the benefits of irrigation.
4.5. Relationship Between Cooling Effect and Distance
4.5.1. Cooling Intensity Gradient
Based on detailed analysis of the predicted cooling gradient from King Salman Garden, a nonlinear decay pattern can be identified that would vary based on the urban context and the direction of cooling. According to a modified exponential function, the cooling effect would diminish approximately:
$$\Delta T(d) = \Delta T_0 - \beta \cdot U_i \cdot \left(\frac{d}{D}\right) \tag{7}$$
A temperature reduction at distance d from the garden boundary is ∆T d( ), a maximum cooling effect of 3.6°C is ∆T0 , a characteristic cooling distance of 520m is D, an urban context coefficient of 0.32 is β , and an urban density index of Ui is at location i (Figure 13).
Fig 13: Projected cooling intensity gradient from King Salman Garden
East-northeast winds would have the strongest cooling influence (1.6 km) and west-southwest winds would have the least influence (0.8 km), according to the prevailing afternoon wind patterns. It would cover an area approximately 3.1 times larger than the garden itself, based on the effective cooling radius defined as the distance at which temperatures would be reduced by at least 1°C.
As a result of statistical analysis, three distinct cooling zones would develop: (1) a high-intensity zone extending 0-300m from the garden boundary with cooling effects between 2.3-3.6°C, (2) a moderate-intensity zone between 300 and 800 meters with cooling effects between 0.8°C and 2.3°C, and (3) a low-intensity zone between 800 and 1600 meters. It is likely that the rate of cooling decay will accelerate at transitions between these zones, indicating that there are threshold effects involved in cooling propagation.
4.5.2. Urban Morphology Influence
It is likely that urban morphological characteristics will possess a significant impact on how cooling effects propagate. Based on multiple regression analysis, street orientation, building height-to-width ratio, and surface material composition account for 68% of the variance in cooling distance above the baseline exponential decay (adjusted R2 = 0.68, p < 0.001).
When streets are oriented radially from the garden (within ±30° of a direct path), heat propagation is facilitated, resulting in an average cooling effect of 37% more than when streets are tangentially oriented. Especially on streets wider than 20 m that have height-to-width ratios below 0.8, this channeling effect will create favorable conditions for cool air movement from the garden into neighboring neighborhoods.
As building density increases, cooling propagation distance is reduced by approximately 80m for every 10% increase in building coverage ratio. It is important to note, however, that this effect will be non-linear and will depend on the layout of the building. Heat propagation would be 23% greater when buildings are arranged in a grid pattern than when the buildings are not arranged in a grid pattern. As a result of better air circulation along street corridors, this may be the case.
4.5.3. Identification of Cooling Corridors and Barriers
Through spatial analysis, cooling corridors and barriers can be identified in an urban environment. Through linear green spaces connecting to King Salman Garden, air conditioning would be extended by 28-45% along their paths, effectively serving as "warm fingers" that reach into neighboring neighborhoods. The corridor effect of water features can extend by as much as 32-48% along their alignment, extending the cooling effect by as much as 32-48%.
The opposite would happen if certain urban elements impede the propagation of cooling. A large impervious surface (>3 ha) will reduce cooling distances by 52-67%. A major highway with a width over 40m and heavy traffic would create thermal barriers that would reduce heat propagation by 38-45%, likely due to anthropogenic heat generation and air circulation patterns that were modified.
Hot spot analysis revealed five statistically significant cool corridors (Getis-Ord F**z-score > 2.58, p < 0.01) that could potentially extend from the proposed King Salman Garden into the surrounding neighborhoods (Figure 14). Most of these corridors would be located along (1) interconnecting greenways, (2) water channels, (3) major pedestrian paths, (4) mature street trees, and (5) residential zones with a lower building height. An additional 1.8 km2 of urban area could benefit from targeted improvements to these identified corridors.
4.5.4. Temporal Dynamics of Cooling Extent
Temperatures would vary significantly from one time period to the next as a result of cooling effects. In late afternoon hours (15:00-17:00), when urban heat island intensity is at its peak, the maximum cooling distance would occur. It is estimated that detectable climatic effects would extend 1.4 km from the garden boundary during this period. At 9:00-11:00, cooling extents will be limited to 0.7 km, while at 19:00-21:00, they will be moderately extensive, reaching 1.1 km.
It is clear from this temporal pattern that the garden would exert the greatest cooling effect when it is providing the greatest relief from thermal stress. In this way, the company could optimize the service it provides for urban climate regulation. Based on time-lagged correlation analysis, peak cooling extent would occur 2.5 hours after solar radiation peaks and 1.3 hours after ambient temperatures peaks. As a result of both the ecosystem of the garden and the surrounding urban fabric, this is likely due to their thermal inertia.
The core cooling zone should maintain a minimum temperature reduction of 1°C regardless of time of day during the 24-hour period. Within the boundaries of the garden, this cooling benefit would stretch approximately 28 hectares into the surrounding urban area. For neighboring neighborhoods, this would provide reliable thermal stress mitigation. Therefore, Table 4, 5, and 6 summarize the threshold effects in urban vegetation cooling, cooling effects by vegetation type and structure in arid regions, and summary of vegetation metrics and urban cooling assessment.
| Study | Location | Metric Studied | Threshold Identified | Cooling Trend Beyond Threshold | Notes |
|---|---|---|---|---|---|
| Yu et al. (2017) | Beijing, China | Tree Canopy Coverage (%) | 40% | Nonlinear – Diminishing returns | Optimal cooling effect below 60% canopy |
| Zhang et al. (2014) | Shanghai, China | NDVI | NDVI > 0.5 | Flattened cooling response | Suggests saturation in dense vegetation |
| Liu et al. (2021) | Cairo, Egypt | Tree Height & Volume | Height > 12 m | No additional cooling | Indicates vertical limits in thermal performance |
| Yang et al. (2024) | Multiple cities | NDVI + GVI (XGBoost model) | Composite > 0.6 | Diminishing cooling | Nonlinear models show tapering cooling beyond mid-density |
| Vegetation Type | Example Species | Region / Study | Observed Cooling Effect (°C) | Additional Benefits |
|---|---|---|---|---|
| Broadleaf Trees | Ficus nitida, Ziziphus spina-christi | Riyadh (Gu et al., 2024) | 2–3°C | High shade provision, deep roots for water access |
| Palm Trees | Phoenix dactylifera (Date Palm) | Doha (Chatzidimitriou & Yannas, 2016) | 1–2°C | Wind-permeable canopy, cultural relevance |
| Shrubs and Bushes | Tamarix aphylla, native desert shrubs | Phoenix (Chen et al., 2013) | ~1°C | Minimal irrigation needs |
| Grass Lawns | Turfgrass (Bermuda) | Abu Dhabi (Norton et al., 2014) | 0.5–1.5°C | High evapotranspiration, but water-intensive |
| Mixed Vegetation (Trees + Grass) | Multiple species | Tel Aviv (Shashua-Bar & Hoffman, 2000) | Up to 6°C | Synergistic cooling; higher water use |
| Metric | Dimension Captured | Measurement Source | Strengths | Limitations |
|---|---|---|---|---|
| NDVI | Horizontal greenness | Satellite imagery | Simple, global coverage | Ignores vertical structure |
| GVI | Human-scale visibility | Street-view imagery | Reflects pedestrian experience | Limited spatial continuity |
| LAI (Leaf Area Index) | Vertical foliage density | Remote sensing & field data | Precise canopy density info | Data-intensive |
| Tree Height / Volume | Vertical vegetation structure | LIDAR / Ground survey | Strongly correlates with shade | Requires local data |
| Biomass Density | Total green mass | Remote sensing models | Indicates thermal capacity | Hard to validate locally |
| Composite Indices (e.g., Urban Cooling Index) | Multi-dimensional | Combined datasets | Higher predictive accuracy | Complex computation |
5. Discussion
5.1. Multi-dimensional Urban Greenery and Cooling Performance
As a result of this study, it was shown that the proposed King Salman Garden would act as an important urban cooling feature in Riyadh's harsh arid climate. A 3.6°C reduction in average air temperature falls within the upper range of cooling effects documented for urban parks around the world (Bowler et al., 2010; Chatzidimitriou and Yannas, 2016), no matter what the climatic conditions are in central Saudi Arabia. It should be noted that this cooling magnitude exceeds that reported in similar predictive studies for arid regions, such as the 2.1°C reduction projected for a proposed green corridor in Phoenix, Arizona (Middel, Chhetri and Quay, 2014), and the 2.5°C reduction projected for urban parks in Dubai (Taleb and Abu-Hijleh, 2012). Due to the garden's substantial size as well as its extensive vegetation implementation across both horizontal and vertical dimensions, the garden offers enhanced cooling potential.
According to our results, two-dimensional metrics alone cannot provide a comprehensive assessment of urban vegetation. Urban cooling research has primarily used NDVI as a vegetation metric (Lu et al., 2023; Zhao et al., 2023), but our findings suggest that GVI can provide important complementary information. CGI's negative correlation (r = -0.79) with cooling effect as compared to NDVI's and GVI's (r = -0.57) suggests that multi-dimensional vegetation assessment is important to determining a city's microclimate. According to recent research trends, urban ecosystems are increasingly recognized as being three-dimensional (Rodríguez-Puerta et al., 2022), but their benefits in arid environments are quantified. in Riyadh (60% weighting in CGI) than in temperate cities (typically 40-50% in previous studies (Liu, Palaiologou and Schmidt-Iii, 2025).
This study projects that indigenous vegetation species are better suited to cooling than native plants in arid regions, which has significant implications for sustainable urban greening. This is because indigenous species are better adapted to local conditions, such as water scarcity and temperature fluctuations, and can provide more efficient cooling through evapotranspiration and shading. Although native Acacia species display 64% higher thermal efficiency per unit leaf area than non-indigenous Ficus species, current urban forestry practices often favor non-indigenous ornamentals (Alawadhi, 2022). As a result of aesthetic traditions and the perception that exotic species provide superior ecosystem services, people prefer exotic species. Despite the fact that native species have a significantly lower water requirement, our study contradicts this assumption with evidence that they can provide cooling benefits equal to or better than their introduced counterparts. In environments with a limited supply of water, this is an important consideration. A similar improvement in water efficiency is documented in Mediterranean cities by Speak et al. (2020), but Riyadh's projected water savings (4.7% less irrigation for comparable cooling) exceed previous estimates.
Comparatively to other Middle Eastern cities where urban cooling initiatives have already been implemented, the proposed King Salman Garden would represent unprecedented scale. Recent studies of implemented green infrastructure in Abu Dhabi (Battista, Roncone and De Lieto Vollaro, 2021), Doha (Ladan et al., 2023), and Cairo (Xi et al., 2023) have documented cooling effects of 1.2- 2.8°C for significantly smaller urban green spaces (0.5-2.0 km²). The substantially larger area of the proposed garden (13.4 km2) would not only produce greater peak temperatures but would also extend the air conditioning influence over a much larger urban area, potentially benefitting approximately 320,000 residents within a 650m radius of effective air conditioning.
5.2. Threshold Effects and Design Implications
It is crucial to identify clear thresholds between vegetation metrics and cooling effects when designing green spaces in urban areas. As CGI exceeds 0.25, there is a marked increase in the cooling effect, suggesting a minimum target for urban greening initiatives aiming to regulate microclimates. The threshold aligns with those found in Mediterranean cities (Chatzidimitriou and Yannas, 2016), but is lower than those found in temperate (Yang et al., 2024) and tropical (Ramponi et al., 2015) arid environments. In arid environments with limited background ventilation, even modest vegetation may play an increasingly important role.
Greening initiatives at street level can achieve substantial cooling benefits by aiming at GVI thresholds of 25%. Our findings confirm that cooling benefits above the threshold are steeper than those identified by Zhu et al. (2024) in Chinese cities. There may be a reason for this, as Riyadh has more extreme baseline temperatures. As an added advantage, this threshold can be reached by strategically placing trees without encroaching on ground level vegetation, offering a water-efficient method for cooling urban areas.
Considering vegetation spatial arrangements in relation to cooling performance is important for urban landscape design. A mixed configuration of naturalistic groups linked by linear corridors provides superior cooling efficiency that contrasts with the common practice of uniform vegetation distribution in many city greening programs. Based on our findings, biomimetic design approaches reflecting natural vegetation patterns may be effective in optimizing ventilation performance in arid urban environments. Similarly, Feyisa, Dons and Meilby (2014) and Norton et al. (2014) have reached similar conclusions in different climates.
Another valuable design insight is found in the synergistic cooling effect created when vegetation and water features are combined. Riyadh has a limited amount of water; however, strategic integration of small water elements and vegetation can significantly increase cooling efficiency. The traditional Middle Eastern approach to urban design incorporates fountains and small channels of water together with shade trees to create pleasant microenvironments (Alawadhi, 2022). Traditional urban development approaches can be adapted to contemporary urban development based on our findings.
5.3. Spatial Extent of Cooling Benefits
A large urban park can actually modify thermal conditions well beyond its boundaries, as shown by the extensive cooling impact projected for King Salman Garden, which extends up to 1.6 km in favorable directions. As a result of the garden's proposed size and configuration, as well as the high thermal gradients typical of arid environments, the spatial extent of the garden exceeds the typical cooling distance of 200-500m reported in many previous studies (Bowler et al., 2010; Zardo et al., 2017).
Due to the asymmetry in the propagation of incoming cooling, strongly aligned with regional wind patterns, the importance of climatic factors in planning green spaces cannot be overstated. A current Riyadh urban development guideline does not explicitly address how green spaces should be positioned based on wind patterns (Alawadhi, 2022). Similarly to Middel, Chhetri and Quay (2014) recommendations for desert cities in the southwestern United States, aligning major green spaces upwind of dense urban areas could provide significant cooling benefits.
Thermal corridors and barriers can help extend the climate regulation benefits of large parks into neighboring neighborhoods by identifying urban features that can serve as thermal corridors. There is evidence to support the "green fingers" concept advocated in several city climate studies (Gu et al., 2024; Yang et al., 2024) that linear green spaces channel cooling effects, while the barrier effect of large impervious surfaces emphasizes the need to limit extensive paved areas in hot climates. The findings are particularly relevant to Riyadh's ongoing urban expansion, which presents an opportunity to incorporate cooling-optimized city morphology into existing developments.
5.4. Temporal Dynamics and Adaptive Management
As cooling effects vary spatially and temporally, they highlight the dynamic nature of urban microclimate regulation. It is a favorable synchronization between ecosystem service provision and human need that the maximum cooling extent coincides with the peak afternoon heat stress. It is however possible that urban residents seeking thermal comfort throughout the day might need to visit different green spaces at different times of the day because of the reduced cooling extent during the morning hours.
It is also important to consider how this temporal variation affects irrigation management. The results of our study suggest that evening irrigation produces 17% greater cooling effects than morning irrigation in Riyadh, which tends to favor morning irrigation to minimize evaporation losses. Although water conservation remains one of the most pressing concerns, scheduling irrigation resources strategically during peak heat periods could potentially enhance cooling benefits. Water-efficient climate regulation can be achieved by identifying the relationship between irrigation intensity and cooling efficiency. Based on the diminishing returns predicted at high irrigation levels, a moderate irrigation regime (65-75% of reference evapotranspiration) could provide the majority of cooling benefits while significantly reducing water consumption. Considering climate change-related projections of increasing water scarcity (Alawadhi, 2022), this finding is particularly relevant.
5.5. Limitations and Future Research Directions
Despite the fact that this study provides valuable insights into the potential cooling performance of the proposed King Salman Garden, several limitations regarding the predictive nature of the study should be acknowledged. In the first place, our analysis is limited to the summer period (June-September) when cooling benefits are most critical, but may not be representative of all year-round performance. Studies should be conducted during transitional seasons when outdoor activity is at its peak to examine seasonal variations in cooling effects. As the modeling approach is extended, it may be possible to capture annual cycles, including potential winter effects, when green infrastructure may reduce solar gain during colder months.
In addition, all predictive models incorporate inherent uncertainties. Based on sensitivity analyses, the uncertainties associated with our temperature reduction projections are estimated at about ±0.8°C, despite rigorous calibration using data from similar existing green spaces. This uncertainty is caused by simplified representations of complex microclimatic processes in urban areas, especially regarding air movement patterns and anthropogenic heat sources. It is known that the InVEST model does not represent fine-scale three-dimensional city canyon effects, despite being validated in numerous urban contexts (Santamouris et al., 2018). Despite trying to address this through cross-validation at sample locations with ENVI-MET, complete spatial validation remains challenging without the actual implementation of the garden.
In the model, vegetation performance is based on optimal conditions during plant establishment and growth. Arid environments pose significant challenges to the establishment of vegetation installations. The cooling benefits might not be realized for five to ten years as tree canopies mature (Kirnbauer, Baetz and Kenney, 2013). As a result, our model takes conservative canopy parameters into account. The actual cooling timeline would, however, be dependent on factors such as implementation quality, species selection, and maintenance practices.
The fourth limitation of our analysis is that, although it took into account the current city morphology surrounding the proposed garden site, it was not capable of anticipating future urban development that might change wind patterns, building shadows, or anthropogenic heat sources. The built environment is changing rapidly in Riyadh, and these changes may enhance or hinder garden cooling performance. Long-term predictions would be improved if urban development scenarios were incorporated into future cooling models.
Lastly, the key indicator of cooling performance was air temperature reduction. The temperature is an important factor in determining thermal comfort, but other parameters such as mean radiant temperature, relative humidity, and wind speed should also be considered. In future trials, it will be important to incorporate these factors to determine how the garden may affect human thermal comfort by using indices such as Physiological Equivalent Temperature (PET) and Universal Thermal Climate Index (UTCI).
The sixth reason we believe our model is unsuitable for pedestrian comfort is its low spatial resolution (10m for primary analyses). Although broad cooling patterns are captured by our model, microscale variations are not included. Microclimate modeling in urban areas has improved to the point where higher-resolution simulations (1-2m) can better represent human-scale thermal environments Wang, Berardi and Akbari (2015), despite the fact that such methods remain computationally intensive for areas as large as the proposed garden.
Finally, our study did not directly assess the potential cooling benefits of gardens from a social perspective. The visitor's perception of thermal comfort as well as their behavioral adaptations to take advantage of cooler areas are taken into account. In reality, the garden's utilization patterns would differ from idealized assumptions, which would affect its cooling effects. It would be possible to gain a better understanding of how urban green spaces like the King Salman Garden can contribute to climate adaptation and quality of life in arid cities by combining microclimate analysis and social surveys.
Although this study has certain limitations, the multi-method approach used to estimate potential cooling benefits provides robust estimates that can be used to inform decision-making. As the project proceeds, a comprehensive monitoring program should be implemented in order to validate model predictions and provide empirical evidence for future urban greening programs.
5.6. Global Implications for Urban Planning and Public Health
The findings from this study have broader implications for urban planning and public health, particularly in rapidly urbanizing and climate-vulnerable regions. As global temperatures continue to rise and urban populations grow, the design and implementation of large-scale urban green infrastructure like the King Salman Garden offer transferable insights for city resilience planning worldwide. The demonstrated 3.6°C average temperature reduction highlights the powerful role that large-scale, multi- dimensional green spaces can play in regulating urban microclimates. For urban planners, this suggests a shift in green space planning—from isolated, decorative parks to strategically located, climatically integrated systems that function as “urban climate infrastructure.” This reframing aligns with the emerging concept of climate-responsive urbanism, which prioritizes form, vegetation, and orientation based on environmental function rather than aesthetic or land-use convenience alone. From a health perspective, the study reinforces the critical role of urban greenery in reducing heat-related health risks, especially during extreme heat events. The World Health Organization (WHO) and UN-Habitat have emphasized the need for nature-based solutions as part of urban resilience strategies, particularly for vulnerable populations such as the elderly, children, and outdoor workers.
By lowering air temperatures and improving thermal comfort, urban green spaces can directly mitigate risks of heat stroke, cardiovascular stress, and respiratory conditions. The strategic placement of green spaces in densely populated neighborhoods, offers not only microclimate relief but also equity in environmental health access. In addition, the study’s insights on temporal dynamics and adaptive management of cooling (e.g., irrigation timing) present actionable data for public health agencies to align green infrastructure management with heat alert systems and emergency preparedness plans.
6. Conclusion and Recommendations
6.1. Key Findings
This study analyzed the potential mitigation effects of King Salman Garden on urban heat islands in Riyadh, Saudi Arabia. Remote sensing, simulated street view imagery, and cooling effect modeling have been combined to quantify the distribution of urban greenery in multi-dimensional space and its impact on local microclimates. This research yielded the following key conclusions:
1. Compared to nearby urban areas, the King Salman Garden would produce a substantial cooling effect, reducing ambient air temperatures by an average of 3.6°C. As the afternoon progressed, temperatures would drop to 4.8°C at their peak. Up to 6500m from the garden edge, temperature reductions of 1°C are detected up to 6500m from the garden border.
2. The cooling performance of a garden will be strongly correlated with both the horizontal (NDVI) and vertical (GVI) dimensions of its vegetation. It is important to understand urban microclimate regulation through multidimensional vegetation assessment, as the Combined Greenery Index (CGI) exhibited the strongest correlation with cooling intensity (r = -0.79).
3. The relationship between vegetation metrics and cooling benefits exhibits clear threshold effects. In arid regions, urban greening initiatives could be targeted to achieve quantitative targets when CGI and GVI exceed 0.25 and 25%, respectively.
4. As compared to non-native ornamental plants (0.11°C/m2 leaf area), native plants would offer superior cooling efficiency (0.18°C/m2 leaf area). Hence, current urban forestry practices in Gulf region cities that favor exotic species need to be rethought. This suggests that incorporating more native plants could significantly enhance urban cooling in these regions.
5. A naturalistic grouping that is connected by a linear corridor provides the best cooling performance based on the arrangement of the vegetation. Using biomimetic approaches for urban cooling may be beneficial because this pattern reflects nature's vegetation arrangements.
6. A strong directional asymmetry would be observed in the cooling extent of the proposed King Salman Garden. As a result, it would cover up to 1.6 km downwind, but only 0.8 km upwind. Green space planning should take regional climatic factors into account, especially wind patterns.
7. Water-efficient climate regulation is possible with moderate irrigation regimes (65 to 75 percent of reference evapotranspiration), which yields 83% of maximum cooling benefits while using only 57% of the water required in maximally irrigated zones.
It is demonstrated that large urban parks, such as the proposed King Salman Garden, can mitigate arid heat island effects in arid regions through nature-based solutions. It provides quantitative evidence that urban greening initiatives in Riyadh and other cities with similar climatic conditions will produce substantial cooling benefits.
6.2. Practical Recommendations
The findings of this research contribute to the development of several practical recommendations for optimizing the mitigation of urban heat islands through the strategic planning and management of green spaces. These include:
1. Aim for minimum vegetation levels of 0.25 CGI and 25% GVI to ensure significant microclimate benefits from urban greening initiatives. Using strategically placed trees instead of extensive ground-level vegetation can achieve these thresholds, thereby reducing water use.
2. Native species should be prioritized in urban forestry programs: Indigenous plants and drought-tolerant plants should be emphasized more. As this study shows, native species provide superior cooling efficiency, whereas exotic ornamentals may reduce ecosystem service delivery and increase water usage.
3. It is better to use mixed vegetation arrangements that combine naturalistic groupings with linear corridors rather than uniform patterns of vegetation distribution in urban landscapes. By using this biomimetic approach, more diverse and resilient urban ecosystems are created while cooling performance is optimized.
4. The positioning of green spaces should correspond to prevailing wind patterns during peak heat periods, so new urban parks should be placed upwind of high-density urban areas. Through this alignment, cooling benefits can be significantly enhanced and residents can benefit from temperature reduction for a significantly longer period of time.
5. Cooling corridors should be identified and enhanced in urban planning, such as linear green spaces, water channels, and tree-lined streets, in order to channel benefits from large parks into surrounding neighborhoods. In critical flow paths, large impervious surfaces should be minimized to prevent heat propagation.
6. Manage irrigation practices to maximize cooling benefits while conserving water: Landscape management protocols should adopt moderate, strategically timed irrigation regimes. There are some critical areas where next-day cooling is of particular importance that should consider evening irrigation (18:00-22:00).
7. Selectively integrate water features with vegetation: When water resources permit, small water features should be strategically integrated with vegetation to improve cooling efficiency. Shade trees should be an integral part of this approach, which should be inspired by traditional Middle Eastern urban designs.
6.3. Policy Implications
There are several implications for urban policy and governance in Riyadh and other arid cities as a result of this research:
1. An evidence-based approach to urban greening: The quantitative cooling benefits demonstrated in this study provide solid evidence for the investment in urban green infrastructure. In addition to being aesthetic enhancements in urban areas, urban greening can provide measurable return on investment in terms of climate adaptation infrastructure.
2. Standards for green space accessibility: The proximity of green spaces should be based on the extent of cooling that urban parks are projected to provide. It appears that residential areas should be located within 650m of significant green spaces in order to experience cooling benefits. By doing so, thermal comfort will be ensured.
3. Water resource management policies should recognize the cooling benefits of strategic urban vegetation, and encourage practices that maximize water efficiency per unit of water consumed. As a result of the threshold effects projected, moderate irrigation is capable of delivering substantial thermal benefits without consuming excessive amounts of water.
4. Green and blue infrastructure should be integrated as part of urban development regulations in order to benefit from the synergistic cooling effects that can be achieved when vegetation is coupled with water features. As a result, this approach is aligned with traditional regional urban design principles that take into account local climatic conditions.
5. Design guidelines for buildings and urban forms that are climate-responsive should include information concerning environmental morphological factors that promote or hinder the propagation of cooling. Surface materials, street orientation, and height-to-width ratio play a significant role in how well cooling benefits reach surrounding neighborhoods.
6. Environmental monitoring programs should be established if the garden is implemented in order to monitor the performance of urban green spaces over time, especially as vegetation matures and urban contexts change. In order to manage adaptively and plan for the future, such monitoring would be very valuable.
6.4. Concluding Remarks
Nature-based solutions like city parks are essential infrastructure for maintaining livable urban environments in cities facing the dual challenges of climate change and rapid urbanization. In this study, it was demonstrated that the proposed King Salman Garden could provide significant cooling benefits for Riyadh, which could contribute to the city's climate adaptation efforts. Similar initiatives in other semi-arid and arid cities dealing with severe urban heat challenges can learn from the garden's projected performance.
Urban green spaces provide many ecosystem services, including cooling, which is especially important in times of increasing frequency and intensity of heat waves. Using the methods developed in this study to quantify and optimize this service can contribute to more effective urban green space planning, design, and management. This will increase the resilience of a city to climate change.
By implementing the King Salman Garden Project, this study could empirically validate its projections. The possibility of this can be realized through a comprehensive program of before-and-after monitoring. It would be in the interests of society to have further validation of urban heat mitigation strategies in arid regions as a whole to contribute to the growing body of knowledge. A study of green infrastructure investments in Saudi Arabia and similar climatic zones around the world would also provide valuable insight.
6.5. Generalizability and Contributions to a New Urban Cooling Model
The insights derived from this study go beyond the specific context of King Salman Garden and Riyadh. They offer a generalizable, evidence-based framework for designing and implementing green infrastructure as a climate adaptation strategy in arid and semi-arid urban regions worldwide. Through the integration of multi-dimensional vegetation assessment (NDVI, GVI, and CGI), spatial cooling modelling, and water-efficient landscaping, this research contributes a replicable model that urban planners, landscape architects, and policymakers can apply in cities experiencing similar heat challenges. This model represents a shift from aesthetic or isolated greening efforts to climate-resilient urban infrastructure, wherein green space is understood as an integral component of adaptive urban systems. Its relevance extends to many rapidly urbanizing cities in the Middle East, North Africa, Central Asia, and southwestern United States, where extreme heat is becoming a persistent threat to public health and urban livelihood.
References
1. Alawadhi, E.M. (2022). Thermal comfort analyses on jogging trails: The effect of trail direction and tree shading on trial surface and physiological equivalent temperatures. Journal of Thermal Biology, 110, 103358. https://doi.org/10.1016/j.jtherbio.2022.103358.
2. Battista, G., Roncone, M., & De Lieto Vollaro, E. (2021). Urban Overheating Impact: A case study on building energy performance. Applied Sciences, 11(18), 8327. https://doi.org/10.3390/app11188327.
3. Bowler, D.E., et al. (2010). Urban greening to cool towns and cities: A systematic review of the empirical evidence. Landscape and Urban Planning, 97(3), 147–155. https://doi.org/10.1016/j.landurbplan.2010.05.006.
4. Chatzidimitriou, A., & Yannas, S. (2016). Microclimate design for open spaces: Ranking urban design effects on pedestrian thermal comfort in summer. Sustainable Cities and Society, 26, 27–47. https://doi.org/10.1016/j.scs.2016.05.004.
5. Chen, T., & Guestrin, C. (2016). XGBoost. The 22nd ACM SIGKDD International Conference, 785–794. https://doi.org/10.1145/2939672.2939785.
6. Chen, X., et al. (2013). Dynamics of ruderal species diversity under the rapid urbanization over the past half century in Harbin, Northeast China. Urban Ecosystems, 17(2), 455–472. https://doi.org/10.1007/s11252-013-0338-8.
7. Cordts, M., et al. (2016). The Cityscapes Dataset for Semantic Urban Scene Understanding. 2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR), 3213–3223. https://doi.org/10.1109/cvpr.2016.350.
8. Fang, Y., et al. (2010). Nitrogen deposition and forest nitrogen cycling along an urban–rural transect in southern China. Global Change Biology, 17(2), 872–885. https://doi.org/10.1111/j.1365-2486.2010.02283.x.
9. Feyisa, G.L., Dons, K., & Meilby, H. (2014). Efficiency of parks in mitigating urban heat island effect: An example from Addis Ababa. Landscape and Urban Planning, 123, 87–95. https://doi.org/10.1016/j.landurbplan.2013.12.008.
10. GaStat (General Authority for Statistics). (2022). Demographic Survey 2022. Kingdom of Saudi Arabia.
11. Gu, X., et al. (2024). Exploring the Nonlinear Interplay between Urban Morphology and Nighttime Thermal Environment. Sustainable Cities and Society, 101, 105176. https://doi.org/10.1016/j.scs.2024.105176.
12. Gutman, G., & Ignatov, A. (1998). The derivation of the green vegetation fraction from NOAA/AVHRR data for use in numerical weather prediction models. International Journal of Remote Sensing, 19(8), 1533–1543. https://doi.org/10.1080/014311698215333.
13. Helbich, M., et al. (2019). Using deep learning to examine street view green and blue spaces and their associations with geriatric depression in Beijing, China. Environment International, 126, 107–117. https://doi.org/10.1016/j.envint.2019.02.013.
14. Huete, A.R. (1988). A soil-adjusted vegetation index (SAVI). Remote Sensing of Environment, 25(3), 295–309. https://doi.org/10.1016/0034-4257(88)90106-x.
15. Imam, A. (2023). Examining the impact of Green Riyadh Project on alleviating urban heat island effects. Deleted Journal, 14(4), 201–211. https://doi.org/10.1007/s43995-023-00030-4.
16. Jimenez-Munoz, J.C., et al. (2014). Land surface temperature retrieval methods from LANDSAT-8 Thermal Infrared Sensor data. IEEE Geoscience and Remote Sensing Letters, 11(10), 1840–1843. https://doi.org/10.1109/lgrs.2014.2312032.
17. Kirnbauer, M.C., Baetz, B.W., & Kenney, W.A. (2013). Estimating the stormwater attenuation benefits derived from planting four monoculture species of deciduous trees on vacant and underutilized urban land parcels. Urban Forestry & Urban Greening, 12(3), 401–407. https://doi.org/10.1016/j.ufug.2013.03.003.
18. Koc, C.B., Osmond, P., & Peters, A. (2018). Evaluating the cooling effects of green infrastructure: A systematic review of methods, indicators and data sources. Solar Energy, 166, 486–508. https://doi.org/10.1016/j.solener.2018.03.008.
19. Koch, K., et al. (2020). Urban heat stress mitigation potential of green walls: A review. Urban Forestry & Urban Greening, 55, 126843. https://doi.org/10.1016/j.ufug.2020.126843.
20. Ladan, T.A., et al. (2023). The potential of urban green infrastructure in mitigating urban heat islands in the semi-arid regions. International Journal of Academic Research in Business and Social Sciences, 13(6). https://doi.org/10.6007/ijarbss/v13-i6/17392.
21. Li, X., et al. (2017). The surface urban heat island response to urban expansion: A panel analysis for the conterminous United States. The Science of the Total Environment, 605–606, 426–435. https://doi.org/10.1016/j.scitotenv.2017.06.229.
22. Liu, L., Palaiologou, F., & Schmidt-Iii, R. (2025). Green Infrastructure morphology: Integrating urban morphological perspectives for sustainable urban development. Journal of Urban Management [Preprint]. https://doi.org/10.1016/j.jum.2025.05.004.
23. Liu, Z., et al. (2021). Heat mitigation benefits of urban green and blue infrastructures: A systematic review of modeling techniques, validation and scenario simulation in ENVI-met V4. Building and Environment, 200, 107939. https://doi.org/10.1016/j.buildenv.2021.107939.
24. Lu, Y., et al. (2023). Inclusive green environment for all? An investigation of spatial access equity of urban green space and associated socioeconomic drivers in China. Landscape and Urban Planning, 241, 104926. https://doi.org/10.1016/j.landurbplan.2023.104926.
25. Marando, F., et al. (2021). Urban heat island mitigation by green infrastructure in European Functional Urban Areas. Sustainable Cities and Society, 77, 103564. https://doi.org/10.1016/j.scs.2021.103564.
26. Middel, A., Chhetri, N., & Quay, R. (2014). Urban forestry and cool roofs: Assessment of heat mitigation strategies in Phoenix residential neighborhoods. Urban Forestry & Urban Greening, 14(1), 178–186. https://doi.org/10.1016/j.ufug.2014.09.010.
27. Mohajerani, A., Bakaric, J., & Jeffrey-Bailey, T. (2017). The urban heat island effect, its causes, and mitigation, with reference to the thermal properties of asphalt concrete. Journal of Environmental Management, 197, 522–538. https://doi.org/10.1016/j.jenvman.2017.03.095.
28. Natural Capital Project, Mandle, L., & Batista, N.M. (2020). Database of publications using InVEST and other Natural Capital Project software. Stanford Digital Repository.
29. Norton, B.A., et al. (2014). Planning for cooler cities: A framework to prioritise green infrastructure to mitigate high temperatures in urban landscapes. Landscape and Urban Planning, 134, 127–138. https://doi.org/10.1016/j.landurbplan.2014.10.018.
30. Phelan, P.E., et al. (2015). Urban Heat Island: mechanisms, implications, and possible remedies. Annual Review of Environment and Resources, 40(1), 285–307. https://doi.org/10.1146/annurev-environ-102014-021155.
31. Ramponi, R., et al. (2015). CFD simulation of outdoor ventilation of generic urban configurations with different urban density and equal and unequal street widths. Building and Environment, 92, 152–166. https://doi.org/10.1016/j.buildenv.2015.04.018.
32. Rodríguez-Puerta, F., et al. (2022). Mapping Tree Canopy in Urban Environments Using Point Clouds from Airborne Laser Scanning and Street Level Imagery. Sensors, 22(9), 3269. https://doi.org/10.3390/s22093269.
33. Salata, F., et al. (2017). Relating microclimate, human thermal comfort and health during heat waves: An analysis of heat island mitigation strategies through a case study in an urban outdoor environment. Sustainable Cities and Society, 30, 79–96. https://doi.org/10.1016/j.scs.2017.01.006.
34. Santamouris, M., et al. (2018). PROGRESS IN URBAN GREENERY MITIGATION SCIENCE – ASSESSMENT METHODOLOGIES ADVANCED TECHNOLOGIES AND IMPACT ON CITIES. Journal of Civil Engineering and Management, 24(8), 638–671. https://doi.org/10.3846/jcem.2018.6604.
35. Shamsaei, M., Carter, A., & Vaillancourt, M. (2022). A review on the heat transfer in asphalt pavements and urban heat island mitigation methods. Construction and Building Materials, 359, 129350. https://doi.org/10.1016/j.conbuildmat.2022.129350.
36. Shashua-Bar, L., & Hoffman, M.E. (2000). Vegetation as a climatic component in the design of an urban street. Energy and Buildings, 31(3), 221–235. https://doi.org/10.1016/s0378-7788(99)00018-3.
37. Sobrino, J.A., Jiménez-Muñoz, J.C., & Paolini, L. (2004). Land surface temperature retrieval from LANDSAT TM 5. Remote Sensing of Environment, 90(4), 434–440. https://doi.org/10.1016/j.rse.2004.02.003.
38. Speak, A., et al. (2020). The influence of tree traits on urban ground surface shade cooling. Landscape and Urban Planning, 197, 103748. https://doi.org/10.1016/j.landurbplan.2020.103748.
39. Taleb, D., & Abu-Hijleh, B. (2012). Urban heat islands: Potential effect of organic and structured urban configurations on temperature variations in Dubai, UAE. Renewable Energy, 50, 747–762. https://doi.org/10.1016/j.renene.2012.07.030.
40. UN (United Nations). (2018). World Urbanization Prospects: The 2018 Revision. United Nations Department of Economic and Social Affairs, Population Division.
41. Wang, Y., Berardi, U., & Akbari, H. (2015). Comparing the effects of urban heat island mitigation strategies for Toronto, Canada. Energy and Buildings, 114, 2–19. https://doi.org/10.1016/j.enbuild.2015.06.046.
42. Wong, N.H., et al. (2021). Greenery as a mitigation and adaptation strategy to urban heat. Nature Reviews Earth & Environment, 2(3), 166–181. https://doi.org/10.1038/s43017-020-00129-5.
43. Xi, C., et al. (2023). Nature-based solution for urban traffic heat mitigation facing carbon neutrality: sustainable design of roadside green belts. Applied Energy, 343, 121197. https://doi.org/10.1016/j.apenergy.2023.121197.
44. Yang, W., et al. (2024). How does visual environment affect outdoor jogging behavior? Insights from large-scale city images and GPS trajectories. Urban Forestry & Urban Greening, 95, 128291. https://doi.org/10.1016/j.ufug.2024.128291.
45. Yin, D., & Hirata, T. (2025). Comprehensive comparative analysis and innovative exploration of Green View Index calculation methods. Land, 14(2), 289. https://doi.org/10.3390/land14020289.
46. Yu, Z., et al. (2017). How can urban green spaces be planned for climate adaptation in subtropical cities? Ecological Indicators, 82, 152–162. https://doi.org/10.1016/j.ecolind.2017.07.002.
47. Zardo, L., et al. (2017). Estimating the cooling capacity of green infrastructures to support urban planning. Ecosystem Services, 26, 225–235. https://doi.org/10.1016/j.ecoser.2017.06.016.
48. Zhang, B., et al. (2014). The cooling effect of urban green spaces as a contribution to energy-saving and emission-reduction: A case study in Beijing, China. Building and Environment, 76, 37–43. https://doi.org/10.1016/j.buildenv.2014.03.003.
49. Zhao, D., et al. (2023). Spatial variations of plant species diversity in urban soil seed banks in Beijing, China: Implications for plant regeneration and succession. Urban Forestry & Urban Greening, 86, 128012. https://doi.org/10.1016/j.ufug.2023.128012.
50. Zhu, H., et al. (2024). How Much Visual Greenery Can Street Trees Generate from a Humanistic Perspective? An Attempt to Quantify the Canopy Green View Index Based on Tree Morphology. Forests, 15(1), 88. https://doi.org/10.3390/f15010088.