Investigation of The Impacts of Urban Morphology On Summer-Time Urban Heat Island Using GIS And Field Measurement

This study investigates the relationships between urban morphology and summertime urban heat island in Hong Kong. A total of 33 urban design parameters describing complex high-rise high-density urban morphology are proposed and categorized into three groups, i.e., land-use intensity, built form, and space enclosure. A process combining the python script and the geoprocessing function is proposed to quickly calculate the morphological parameters for 10 sites and 160 points. Microclimate data were collected using onsite measurement equipment. Pearson correlation and multiple linear regression analysis are conducted to quantify the impact of each parameter on urban heat island (UHI) intensities at 3 pm (UHI_3pm) and 9 pm (UHI_9pm). 16 parameters are found to be statistically correlated with UHI_3pm and 9 parameters with UHI_9pm. Results show land-use intensity parameters have the highest correlation with UHI, followed by built form parameters and then space enclosure parameters. Furthermore, it is found that site-level parameters can better explain UHI variation compared to point-level parameters. This research paves the path for quickly extracting urban morphological parameters and enhances the understanding of the impact of complex high-rise high-density urban environment on summertime UHI. The results can inform policy makers with the guidelines to create more comfortable urban environment. This study investigates the relationships between urban morphology and urban heat island in Hong Kong. A total of 33 design parameters, calculated on the site and point levels, are proposed to address land-use intensity, built form, and space enclosure in the complex high-rise high-density urban environment. A process combining the python script and the geoprocessing function is proposed to quickly calculate the morphological parameters for 10 sites and 160 points. Microclimate data of 160 points are collected using onsite measurement equipment. Pearson correlation and multiple linear regression analysis are conducted to quantify the impact of each parameter on UHI intensities at 3 pm (UHI_3pm) and 9 pm (UHI_9pm). This research paves the path of quickly subtracting urban morphological parameters and enhances the understanding of the impact of complex high-rise high-density urban environment on summertime UHI. The main results can be

applicability of these parameters on the climate research in different area may also vary. Hong Kong, as a typical high-rise high-density city, exhibits a high level of complexity in the building layout. It has a population of over 7 million living within only 270 km 2 of built-up area. This extremely dense population has shaped Hong Kong into a unique high-rise high-density urban environment. The general research output from other areas may contribute to the understanding of the UHI effect in Hong Kong but cannot directly guide the urban planning and design practices. Many of the morphological parameters from previous research are more suitable for homogeneous urban form, since they cannot re ect the detailed design features of highly diversi ed building arrangement. Therefore, to better parameterize the complex high-rise high-density urban environment in Hong Kong, new parameters shall be explored. Table 1 Urban morphological parameters that in uence urban climate in previous research

Category Parameters
Land-use intensity Building density [17][18][19] / building cover ratio 20 Floor area ratio 12,19 / Rugosity 18,21 Mineralization factor 18 Built form Average building height 22 Building height variation 23,24 Height to oor area ratio 25,26 Compacity 18 Complete aspect ratio 27 / Building envelope area to site area ratio 21 Total building volume to the number of buildings ratio 21,24 Building volume to site area ratio 21 Space enclosure Canyon aspect ratio (building height / street width) 13,28−35 Street aspect ratio (building length / building height) 18,28 Sky view factor [36][37][38][39][40] Frontal area density 41 Urban porosity 18 This study aims to investigate the relationships between UHI and urban morphology of high-rise highdensity urban environments like Hong Kong. Important urban morphological parameters that in uence UHI would be identi ed by analyzing both newly proposed and existing parameters. Value of parameters are calculated from 160 points and 10 sites in real high-rise high-density urban settings where microclimate Page 4/28 measurements are conducted. Statistical analysis will be conducted to nd the correlation between UHI and different morphological parameters. Empirical models will be developed to help explain spatial UHI variation in the highly heterogeneous high-rise high-density urban environment. The research outcome would enhance the understanding of parameterizing complex high-rise high-density urban environment and serve to inform urban planners and designers with scienti c information on climate-sensitive design strategies.

Identi cation Of Urban Morphological Parameters
Urban morphology has a strong impact on UHI, which has attracted increasing research interests. Table 1 lists the urban morphological parameters proposed by previous research. It can be seen that the design features described by some of the parameters overlap with each other. This research combined the previously-proposed and newly-proposed parameters to better capture the complex morphological characteristics of high-rise high-density urban environment. The urban morphological parameters for this study are chosen and proposed based on the following principles: can be controlled and understood by urban planners and architects; can properly address the 3-D characteristics of urban morphology, which can give urban planners and architects an intuitive sense of physical structure at the conceptual design stage; To verify the effect of different scales of urban morphology on UHI, two levels of variables are proposed in the current study, including site-level and point level variables. For the site-level parameters, the data within the whole research site (400 m×400 m) were collected. For the point-speci c parameters, the data within an area with a radius of 30 m were collected 11,12,42 . Based on the criteria mentioned before, the following 3 groups of parameters are proposed, including land-use intensity, built form, and space enclosure ( Table 2). reach 100% to a height not exceeding 15 m above ground level. Under this regulation, a "tower + podium" building form has been the predominant type in Hong Kong. It has a giant podium at the bottom and highrise towers above, which is called "birthday cake building" locally. Therefore, it is necessary to distinguish between the podium and tower parts by calculating the FAR value for each respective part (FARpod and FARtow). As FAR is de ned as the total oor area divided by the site area, FARpod and FARtow represent the total podium area and tower area divided by the site area respectively.
Building density refers to the ratio of building footprint to total land area. Building density describes the twodimensional feature of intensity, while FAR the three-dimensional feature. A similar approach was adopted to calculate building density at three levels, covering the bottom, middle, and top level due to the high variation in building heights in Hong Kong. Based on the data from 10 research sites, it was found that the average building height is 40 m. Given the 15 m height limit for podiums in Hong Kong, building density was calculated at 10 m, 30 m, and 60 m height levels (Fig. 1a).
Based on the above description, FAR and building density at the site level can be calculated according to the following equations: where represents the oor area of the buildings at a certain height, represents the total area of the site which is 160,000 m 2 in this study, represents the number of oors for certain buildings.

Built form group
Building height restrictions are commonly applied in urban planning. Three parameters related to building height, i.e., average building height (H), height variation (HV), and average height to oor area ratio (H/FA) at different levels were calculated ( Table 2). HV is represented by the standard deviation of building heights. Building surface area highly in uences energy and heat exchange between the buildings and the air, which makes it a proper index in evaluating the climatic impact of building form. The ratio of building surface area to site area (SAR_St, SAR_Pt) and Compacity (COMP_St and COMP_Pt) were calculated in this study. Compacity is calculated as exterior building surface area divided by building volume, which can assess the urban compactness and capacity 18,21 . These two parameters help to evaluate building mass and its heat storage and release ability.
But it cannot provide urban designers with enough detailed suggestions on climate-sensitive design, since a certain SVF value does not represent a speci c type of urban building layout. There might be a variety of sky fraction shapes that share the same SVF value. SVF cannot precisely de ne the heterogeneous urban morphology as the relation between the sun's path and the sky fraction is not addressed. In this study, attempts were made to decompose SVF to capture the enclosure degree of surrounding buildings from four directions and also to distinguish the in uence from buildings and vegetation (Fig. 1b). Therefore, six parameters were proposed, covering SVFall_Pt, SVFbld_Pt, SVFe_Pt, SVFw_Pt, SVFs_Pt, and SVFn_Pt ( Table  2).
Street canyon aspect ratio affects the net radiation and stored solar ux of the street. It is generally de ned as building height to street width ratio in a highly simpli ed urban canyon model. In real environments, there may be large variations of building heights between adjoining buildings, while the street orientation may not be parallel to east-west or south-north directions. This poses di culty in calculating aspect ratio in the real environment since it is hard to determine the building height and spacing. In this study, a new approach was proposed to calculate the aspect ratio. To re ect the actual relation between the measurement point and surrounding buildings, four parameters were introduced to quantify the ratios in four directions. Instead of a single building height, the average height of buildings within the 90 degrees range was employed (Fig. 1c).
The width is denoted by the distance between the measurement point and the building instead of the distance between the buildings in the opposite directions.

Methodology
To quickly calculate 33 morphological parameters in 160 points and 10 sites, an automatic process combining the python script and Geographical Information System (GIS) model was introduced. These parameters were then correlated with UHI data 24,44,45 . In general, the study was conducted by the following steps: conducted eld measurements to collect microclimate data and morphological information in 10 selected sites; built GIS model of study sites by ArcGIS and calculated urban morphological parameters using python script and geoprocessing functions; conducted Pearson correlation and multilinear regression analysis between UHI and urban morphological parameters. proposed empirical models to indicate their relationships.

Site identi cation
In terms of site selection for microclimate measurements, the in uence of geographic condition, seasonal and weather variance, arti cial heat sources shall be well controlled 46 . Geographic location determines the regional climate background and the distances from cool resources, including sea surfaces and vegetated mountains. For seasonal and weather in uence, the climatic condition (temperature, humidity, solar radiation, precipitation, wind, and cloud cover) changes between different seasons and even different days, which could impose signi cant effects on UHI patterns. To reduce the geographic impact, study sites were carefully selected in the adjacent area on the northern side of Hong Kong Island. Meanwhile, to minimize the seasonal and climatic impact, the measurements were conducted on clear sunny days in the summertime. Originally there were 12 sites with typical urban form investigated, but only 10 sites are analyzed in this study (Fig. 2a). In the preliminary analysis, it was found that the data collected from Sung Hing Lane site (SHL) and Lockhart Road site (LhR) may not be representative due to the difference of measurement time at SHL and land use type at LhR. The other 10 sites were all residential blocks and measured in the months between July and early September, which were typical summer months in Hong Kong. Thus, SHL and LhR are excluded from the present analysis.
In urban climate research, it is important to de ne the site boundary according to research scales. In this study, it was decided based on a review of different kinds of research on urban morphology and environmental performance, covering main research methods (remote sensing, simulation, and eld measurement). In a study of the cooling effect of green space based on remote sensing technology, a size of 240 m × 240 m upon satellite images was employed by Kong et al. to calculate landscape metrics 47 . In a physical simulation study on urban bioclimatic design strategies, Sad de Assis and Barros Frota built a scaled 500 m × 500 m model representing an area of 25 hectares 48 . In a eld measurement research project, Krüger and Givoni measured the temperature of seven areas in Curitiba, Brazil, each with a diameter of 250m 49 . Santamouris 2 , Edussuriya et al. adopted the area size of 200m × 200m and investigated 20 different urban residential areas to study the relationship between urban morphology and air quality in Hong Kong 24 . Based on the six generic urban forms proposed by Leslie Martin and others to address land-use characteristics, Ratti et al. reassessed the environmental performance of these forms using computer analysis techniques, adopting an area size of 250m × 250m 50 . In the research on the relationships between building energy consumption and urban texture, urban areas with dimensions of 400m × 400m were adopted to extract building form data by using a computer-based image processing technique 51 . Based on the aforementioned studies, a size of 400 m × 400 m was proposed for the research sites in the present study considering the following aspects: The cooling effect of green space may extend beyond the site, as far as the park's width 52 .
According to the technical circular of Hong Kong Air Ventilation Assessment 53 the surrounding area of up to a perpendicular distance of 2H (highest building in the site) from the project boundary must be included for investigating wind performance. Since building height in the measurement area could normally reach 100m, a 400m × 400m rectangular area was proper for this study.

Microclimate measurements and eld survey
The microclimate measurements in these 10 sites were conducted during three summers from 2010 to 2012. HOBO weather stations were used to record the climate data, with all the sensors installed at the heights ranging from 1.2 to 2 m. There were 160 points measured in total, with 81 points located inside the parks and 79 points on the surrounding streets. The control point in each site was continuously measured and the other points were measured by mobile equipment (Fig. 2b). Microclimate measurements were conducted for 3 days in each site. In each day, the measurements at all points were carried out once every two hours from 1 pm to 9 pm. The microclimate data collected at 3 pm and 9 pm are analyzed in this paper, representing the hottest period during daytime and early nighttime. More details on the eld measurements and climatic data were presented in the previous paper 19 .
Urban building information and land use data were collected through on-site surveys combined with the information from GoogleMaps, and GeoInfo Map provided by Survey and Mapping O ce/ Lands Department, HKSAR. Fish-eye lens pictures at all measurement points were taken to calculate SVFs. The urban morphological information including terrain, building layout, greenery, and measurement points was consolidated into the ArcGIS models.

Parameter calculation by ArcGIS
By collecting the morphological data of all the sites, 10 GIS models with building form, area, height, and volume information were built in ArcGIS. A systematic process was established to calculate the design parameters at both site and point level: Firstly, geoprocessing functions such as "Buffer", "Intersect" etc. were used to select the buildings to be included in the calculation according to a certain site or point.
Then, Python scripts were written to calculate speci c parameters such as DENSITY and HV using either mathematical function built-in Python or Data processing tools in ArcGIS. Finally, the model function in ArcGIS was used to combine these steps to calculate the parameters proposed in Sect. 2. Figure 3 shows the schematic work ow con guration to calculate aspect ratio for each point in the "Model" interface and the lines created for getting the 1-degree resolution in aspect ratio (AR) calculation.

Data analysis
Pearson correlation and multiple linear regression analysis were conducted to quantify the impact of each parameter and to develop empirical models that explain the UHI variation. In the regression analysis, dependent variables are the UHI intensities at 3 pm and 9 pm (UHI_3pm and UHI_9pm), while independent variables are the morphological parameters identi ed in the last section. For dependent variables, UHI is de ned as the difference between the measured air temperature at each point and the data from the rural area weather station (Ta Kwu Ling weather station). Ta Kwu Ling weather station has been widely used by researchers in Hong Kong as a rural site 54,55 .
To identify important urban morphological parameters, Pearson correlation was conducted between UHI_3pm and UHI_9pm with all parameters. Only those parameters proven to be statistically correlated with UHI intensities will be considered in the multiple linear regression.
Multicollinearity problems may exist in such a large quantity of independent variables. This problem was carefully dealt with through two main steps. Firstly, Pearson correlation was carried out to examine the correlation coe cient between every two design parameters. A coe cient higher than 0.80 indicates high collinearity, which means these two parameters cannot exist in one model. Secondly, based on the rst step, models with a different combination of independent variables can be generated. Some scholars have suggested a formal detection of tolerance for these models to avoid multicollinearity 55 . In this study, a tolerance of less than 0.10 was considered as a collinearity problem. The nal models were obtained by removing the independent variables with a tolerance lower than 0.1 and P-value higher than 0.05. Figure 4 presents the UHI intensities of all sites at 3 pm and 9 pm. It can be seen that UHI intensities at 9 pm are much higher than 3 pm. The UHI at 9 pm is around 3 ℃, with a very small uctuation among all sites.

UHI at 3 pm and 9 pm
The minimum is 2.23 ℃ at the WOL site and the maximum is 3.84 ℃ at the TWS site. As for 3 pm, the UHI intensities vary substantially among these sites. Cool island effects are discovered in 5 sites, with the air temperature at these sites being lower than rural weather stations. The minimum is -1.53 ℃ at the CS site and the maximum is 2.34 ℃ at the TS site.

Calculation results of design parameters
The descriptive data of all 33 morphological parameters are presented in Table 3. Figure 5 displays 5 sitelevel and 1 point-level parameters with more details. It can be seen that density decreases from bottom to top. Densities at the bottom level (10 m above ground) are between 0.31-0.49 with an average value at 0.39, while the middle-level densities (30 m above ground) are mainly between 0.12-0.23 and the top-level densities (60 m above ground) are around 0.11. It re ects the characteristic of the typical "podium + tower" urban building form, which has a large building mass at the ground level and slender buildings above. It is worth mentioning that SAR_St shows a similar trend as FAR_St, which indicates possible collinearity among these two parameters. Alsom the measurement points are highly enclosed by surrounding buildings with average SVFall_Pt being 12.3%. The 95% con dence interval is from 11.5-13.3%.  Figure 6 shows the results of the Pearson correlation between UHI_3pm and UHI_9pm with all design parameters. It is found that 16 out of 33 parameters are correlated with UHI_3pm and 9 parameters are correlated with UHI_9pm. Among 3 groups of morphological parameters, land-use intensity parameters have the highest correlation with UHI, followed by built form parameters, while space enclosure parameters have the lowest correlation.

Pearson correlation
For the land-use intensity parameters shown in Fig. 6a, the site-level parameters exhibit a much higher correlation with UHI intensities than the point-level parameters. Strong negative relations are found between the site-level FAR and density with UHI_3pm. This can be explained by the shading effect provided by the podium and towers. In the daytime, higher construction intensity, meaning larger building mass, could reduce the air temperature in multiple ways. Firstly, it could intercept solar radiation and prevent it from reaching pedestrian levels, which is especially important for a sub-tropical city like Hong Kong. Secondly, more incoming solar radiation can be re ected by the roofs since the surface albedo of roof material is normally higher than road pavement. Lastly, building mass can absorb and store more solar heat than the rural area, where solar radiation is either re ected directly or absorbed and later emitted in a short time to heat up the air during the daytime. Smaller correlation can be found between FAR_St, FARtow_St, and DENSITYbot_St with UHI_9pm. It may be explained by the fact that more anthropogenic heat are released into the air during early nighttime in these residential blocks. Among the land use intensity parameters, FAR_St and DENSITYbot_St have the highest correlation coe cients with UHI_3pm. It suggests that in complex high-rise high-density urban environments, conventional morphological parameters such as FAR and density are most in uential for UHI variations. Park area is found to be positively related to UHI_3pm and UHI_9pm. Larger parks mean more solar radiation due to less shelter from surrounding buildings. It suggests the importance of shading in Hong Kong to ameliorate outdoor thermal stress in the afternoon and early nighttime.
For the point-level parameters in the built form group shown in Fig. 6b Compacity (Comp_St) is calculated as the total exterior building surface area divided by the total building volume within the site. Higher Comp_St means more surface area with the same building volume, resulting in more radiation absorbed by exterior walls during the daytime, and more surface to release heat at night. Theoretically, higher Comp_St would lead to higher UHI in the daytime and lower UHI in the nighttime.
However, in this study, Comp_St is found to be positively related to both UHI_3pm and UHI_9pm. The possible explanation for its positive impact on early nighttime may be either due to the delayed heat release or the excessive heat absorption during the daytime.
For the space enclosure parameters as shown in Fig. 6c, 4 out of 10 parameters are statistically correlated with UHI_3pm, while no parameter is related to UHI_9pm. It is found that SVFall_Pt (regular SVF) is positively related to UHI_3pm, which is in line with the general theory. Higher SVF means more solar radiation can reach the ground and building walls, which increases heat gain. It suggests SVF should be well controlled especially in sub-tropical cities, where solar shading is critical for reducing thermal stress. To distinguish the shading effects from the buildings and vegetation, SVFbld_Pt is calculated and analyzed as well. But the result shows that it cannot explain the UHI variation, which means vegetation plays an important role in providing shading at the point level. Also, it is found that SVFe_Pt, SVFw_Pt, and SVFs_Pt have a positive in uence on UHI, while no correlation between SVFn_Pt and UHI can be found. According to the sun path on sky view images (Fig. 1b), only the sky fraction on the east, west, and south directions would affect the incoming solar radiation into the urban canyon. It means the sky fraction related to the sun path has a stronger impact on heat gain. On the other hand, no statistical correlation can be found between SVF related parameters and UHI_9pm, and between the aspect ratio related parameters and UHI during both daytime and early nighttime. It suggests that for the high-rise high-density urban environments like Hong Kong, the proposed calculation method for aspect ratio still cannot properly capture the morphological feature.

Regression analysis
To deal with the collinearity problem, Pearson correlation was conducted among 17 independent variables that are statistically correlated with UHI_3pm or UHI_9pm to check the relation with each other (Fig. 7). Collinearity is con rmed when the correlation coe cient between two variables is higher than 0.8 (marked with black dots). It is found that the correlation among FAR_St, FARpod_St, FARtow_St, DENSITYbot_St, DENSITYtop_St, and SAR_St are quite high, indicating that these parameters cannot be included in one multiple regression model. This is because these design parameters address similar design features from different scopes. The correlation coe cients of these parameters with UHI are higher than other parameters, which suggests that they are the most important parameters in each regression model.
Stepwise multiple regression was conducted to explore empirical models for explaining UHI_3pm and UHI_9pm respectively. Regression models are selected based on the following standards: (1) signi cance level is smaller than 0.05; (2) tolerance value is higher than 0.1; (3) variance in ation factor (VIF) value is smaller than 10; (4) Pearson correlation coe cient between every two variables is smaller than 0.8; (5) the impact of all parameters agree with the Pearson correlation results in the last section; (6) the R square value of each model is larger than 0.5. The regression results for UHI_3pm and design parameters are presented in Table 4, including 5 models with the R square ranging from 0.572 to 0.711. From the above criteria, no suitable model can be established for UHI_9pm. The Pearson correlation results show that only 9 design parameters are statistically related with UHI_9pm with relatively lower coe cients. It suggests the early nighttime UHI variation in high-rise high-density residential blocks is in uenced by much more complex conditions aside from urban morphological design features.
It can be seen from Table 4

Discussion
According to Pearson correlation, the site-level parameters have a higher correlation with UHI than the pointlevel design parameters. It means the site-level urban morphology plays a more important role in in uencing urban thermal conditions. Moreover, the coe cients of FAR and density parameters with UHI_3pm are the highest among all parameters, which indicates that the land-use intensity parameters at the site level are most in uential on summertime UHI in the afternoon. By decomposing FAR and density into different levels, strong correlations are found between decomposed parameters and UHI, but the classical FAR and DENSITYbot still show the highest correlation with UHI.
In this study, building density is negatively correlated with UHI, which is contradicting to a few other studies.
Previous studies concluded that increasing building density would lead to a higher outdoor air temperature since the wind ow in urban areas was blocked 20 . On the other hand, higher density means less solar radiation can enter into the street canyon. Based on the different results in other area and Hong Kong, assumption can be made that when building density is under a certain level, ventilation plays a dominant role in governing outdoor air temperature, but when building density is higher than that certain level such as in Hong Kong, its shading effect becomes the dominant factor. This exact level of building density can be investigated in future research.
The FAR values of 10 research sites in the present study ranged from 3.98 to 7.00. The regression model shows that an increase of 1 for FAR_St can lead to a decrease of UHI_3pm by 0.759 ℃. In an empirical study conducted in Shanghai 12 , the same variation in FAR can reduce daytime UHI by only 0.124℃-0.198℃. In that study, the FAR values of most research sites are around 2-3, which is much lower than in Hong Kong. It indicates that daytime UHI is more sensitive to the FAR variation in high-rise highdensity cities in subtropical climate zone.
The positive relationship between the park area and UHI should be carefully interpreted for design implications. Suggestions cannot be made that large urban parks should be replaced by small parks. Results from the present study can only help to explain the UHI variation at 3 pm and 9 pm in a high-rise high-density urban environment. It is generally acknowledged that green area provides not only environmental bene ts but also social and psychological welfare for the community. Besides, further investigation on comparing the in uence of one large park and several small parks is needed.
The positive relationship between Comp_St and UHI suggests that buildings with compact and regular plans are favored considering outdoor thermal environment, which have lower perimeters compared to irregular shapes given the same plan area. The result is not only in line with general theory, but also the energye cient building design strategy. Concerning building energy e ciency, the least surface area would be preferred since heat is mainly lost through building envelope. Therefore, the buildings with regular and compact plans will be bene cial for both outdoor and indoor thermal environments.
Many previous studies reported the different effects of urban design on daytime and nighttime UHI 12,43 .
However, it is found in the present study that parameters con rmed to be related with both 3 pm and 9 pm UHI show consistent in uence on UHI at two different times. From a report on the difference between urban and rural climate based on the data from Hong Kong Observatory, it is demonstrated that UHI keeps rising from 3 pm to 9 pm until reaches the maximum around 6 am, but the rising rate decreases drastically after 9 pm 54 . It is different from many cities where the UHI maximum occurs several hours after sunset 56 . The authors assume it may be partially due to a large amount of anthropogenic heat release during nighttime in Hong Kong. This can also explain two results in this study. One is the consistent in uence of design parameters on both 3 pm and 9 pm UHI, while the other is the lack of effective regression models for 9 pm UHI.

Conclusion
Page 17/28 This study investigates the relationships between urban morphology and urban heat island in Hong Kong. A total of 33 design parameters, calculated on the site and point levels, are proposed to address land-use intensity, built form, and space enclosure in the complex high-rise high-density urban environment. A process combining the python script and the geoprocessing function is proposed to quickly calculate the morphological parameters for 10 sites and 160 points. Microclimate data of 160 points are collected using onsite measurement equipment. Pearson correlation and multiple linear regression analysis are conducted to quantify the impact of each parameter on UHI intensities at 3 pm (UHI_3pm) and 9 pm (UHI_9pm). This research paves the path of quickly subtracting urban morphological parameters and enhances the understanding of the impact of complex high-rise high-density urban environment on summertime UHI. The main results can be summarized as follows.
-16 out of 33 parameters are found to be correlated with UHI_3pm and 9 parameters are correlated with UHI_9pm.
-Land-use intensity parameters have the highest correlation with UHI, followed by built form parameters, while space enclosure parameters have the lowest correlation.
-The site-level parameters are better at explaining UHI variation compared to the point-level parameters, which suggests urban planning and urban design should be carefully controlled during the policy making process.
-All site-level FAR and density-related parameters are negatively related with UHI_3pm with the highest correlation coe cients, which suggests that shading effect plays a important role in regulating outdoor air temperature in high density city in the subtropical area.
-Park area is positively related with both UHI_3pm and UHI_9pm, which suggests that larger green space does not necessarily lead to lower air temperature. Delicate design of the green space shall be considered -The positive relationship between Comp_St and UHI suggests that buildings with compact and regular plans are bene cial for the outdoor thermal environment.
-SVFall_Pt, SVFe_Pt, SVFw_Pt and SVFs_Pt have positive in uence on UHI_3pm, which also con rms the importance of shading effect in in uencing summertime outdoor air temperature.
-Aspect ratio is hard to de ne the complex morphological feature in a high-rise high-density urban environment.
-Urban morphological parameters can only explain a small part of early nighttime UHI variation.

Limitation And Future Research
Due to the limitation of measurement equipment and time, the study is based on a 3-day measurement campaign on 160 points in 10 high-rise high-density sites. The selection of sample sites makes it impossible to compare the thermal performance difference between compact urban form and low-density urban form.
The empirical models developed in this study are not su cient in predicting the UHI in other urban design scenarios. For the future research, a yearlong continuous microclimate measurement will be carried out at more points. With more data to be divided into train sets and test sets, machine learning tools can be introduced to train and test the empirical models to generate more accurate models in predicting UHI at different times of the year. Also, aside from air temperature, other meteorological factors, such as wind speed, humidity, and solar radiation, should be considered systematically. To overcome the limitation of empirical study, numerical simulation shall also be adopted in the future.