Numerical Studies on the Influence of Building Morphology on Urban Canopy Wind Speed

Buildings increase the urban surface roughness and reduce near‐surface wind speeds in the urban canopy due to the drag effect. Urban heat storage and other effects cause urban warming as well, which decreases the urban boundary layer stability and enhances the turbulence exchange between upper and lower layer. As upper momentum is transported downward, the wind speed of urban canopy increases. Quantitative descriptions of these mechanisms are still lacking currently. This paper presents high‐resolution numerical simulation results of a mega city, Shanghai, China from 2016 to 2020 using the building effect parameterization in WRF (WRF‐BEP) with urban morphological parameters. The dynamic and thermal effects of building morphology on urban canopy wind speed were separated and their quantitative expression functions were given. The results indicate that the influence of building morphology on urban canopy wind speed is mainly dynamic resulting in a wind speed attenuation of approximately 50% and nearly constant. The thermal effect of building morphology on urban canopy wind speed increases with the urban heat island intensity, and the thermal effect could increase urban canopy wind speed by about 30% under the condition of strong urban heat island. The relative contributions of the dynamic and thermal effects of building morphology to urban canopy wind speed change with the wind speed. As wind speed increases, the contribution of the thermal effect of building morphology to urban canopy wind speed gradually decreases. This paper provides a quantitative relationship between the urban canopy wind speed variation and urban morphology, as well as urban heat island intensity.


Introduction
The world has undergone rapid urbanization in the past few decades.The urban population is projected to reach 6.7 billion by 2050, when more than two-thirds of the world's population will live in cities (United Nations et al., 2019).Urbanization alters the properties of the urban underlying surface, emits a significant amount of anthropogenic heat and pollutants, and has a substantial impact on the local climate (Christen & Vogt, 2004;Grimmond, 2006;Grimmond & Oke, 1999;Wang et al., 2020).A series of meteorological and environmental problems have occurred, including the typical urban heat island (UHI) and air pollution problems, as well as the increasingly frequent extreme weather (Miao et al., 2009;Wang et al., 2016;Zhang et al., 2009).
Cities expand both horizontally and vertically in the Urbanization processes and alter the surface properties of urban areas significantly.The vertical extension of the cities creates a new canopy (the urban canopy), which increases the roughness of the urban underlying surface and changes the dynamic and thermal properties of the canopy (Arnfield, 2003;T. R Oke, 1982).Urbanization significantly alters the energy balance process and creates new dynamic and thermal processes in urban areas.The wind obstruction caused by buildings and the turbulence they generate have a considerable impact on the urban wind field's characteristics (Chen et al., 2011;Wang et al., 2009;Zhang et al., 2010).Previous studies have demonstrated that urbanization causes a decline in the average urban wind speed and changes the surface dynamical characteristic.Liu et al. (2018) found that the mean wind speed presents an evident decreasing trend during 1991-2011 in Beijing based on observational data collected at 15 levels on a 325 m meteorological tower.Li et al. (2018) evaluated the impact of urbanization on urban surface wind speed based on the wind speed observation data of 506 meteorological observation stations in eastern China from 1991 to 2015.The results showed that the observed surface wind speed over East China is distinctly weakening with a rate of 0.16 m s 1 deca 1 during 1991-2015 and the wind declining intensity is closely related to the urbanization rhythms.Zhang and Wang (2021) evaluated the impact of urbanization on surface wind speed over China.The results showed that urbanization led to a decrease of 11% in surface wind speed over China from 1985 to 2017 and the attenuation of the surface wind speed due to urbanization was different in different regions.Urbanization leads to an increase in urban surface roughness, which in turn enhances the drag effect on the wind.Buildings are the most important surface elements of the city and vary in height, shape, and orientation, resulting in strong non-uniformity characteristics of the urban underlying surface.The morphological characteristics of buildings are the most significant factor contributing to this increase in roughness and have a profound impact on the urban surface wind speed.
In addition to increasing the urban surface roughness, urbanization also indirectly affects the urban surface wind speed (Zhang et al., 2016).The urban warming caused by urbanization enhances the turbulent exchange in the vertical direction of the urban area, and the upper atmospheric momentum is transported down to the surface, increasing the urban surface wind speed.The effects of these two mechanisms of urbanization on urban surface wind speed are opposite, and are closely related to building morphology.The characteristics of building morphology determine the overall effect of urbanization on urban surface wind speed.Several studies suggest a critical threshold value of the regional wind speed: above that threshold, roughness effects and deceleration dominate, while below it, thermal effects and acceleration are significant or even dominant (Bornstein & Johnson, 1977;Lee, 1979;Oke et al., 2017;Wong, 1976;Wong & Dirks, 1978).It is likely that the threshold value is related to physical properties of the urban surface, such as its fabric, surface cover, structure and metabolism, but that has yet to be shown.In the past, most studies on the impact of urbanization on urban surface wind speed were based on observational analysis.Although the conclusion that urbanization leads to the decrease of urban surface wind speed was obtained, the quantitative research on the relevant physical mechanism was relatively scarce.And due to the limited temporal and spatial resolution of observation data, many studies have overlooked the indirect impact of urban heat island on urban surface wind speed.
The first urban canopy model was developed by Nunez and Oke (1977) by representing the built environment using urban canyon.With the development of urban canopy model over the past few decades, it has advanced the research on urbanization-related issues.The WRF model provides various urban canopy parameterization schemes to describe the impact of buildings on the energy balance of the urban surface, but they have varying degrees of simplification in characterizing building morphology.The commonly used multi-layer urban canopy model, BEP (Building Effect Parameterization Scheme) provides a relatively complete description of the complex physical processes of cities.However, it heavily relies on the high-quality urban morphological parameters (Martilli, 2002).These parameters effectively represent the morphological characteristics of buildings at the grid scale and improve the description of the non-uniform characteristics of the urban underlying surface.It is difficult to characterize the urban underlying surface using a unified description method.To address this issue, Sun et al. (2021) developed a high-resolution urban morphological parameter data set based on the fine 3D building data of 60 major cities in China, and evaluated the representativeness of WRF default morphological parameters in China.The results show that the current WRF default morphological parameters cannot accurately represent the morphological characteristics of major cities in China.Replacing the default morphological parameters with high-resolution urban morphological parameters improves the simulation of the WRF model in urban areas by better reflecting the fine distribution of building morphology (Ching, 2007).
By coupling the WRF model with high-resolution urban morphological parameters, we can quantitatively study the dynamic and thermal effects of building morphology on urban canopy wind speed, and gain an in-depth understanding of the physical mechanism involved.As the central city of the Yangtze River Delta, Shanghai is a typical megacity in China.Conducting research on the impact of building morphology on urban canopy wind speed in Shanghai is crucial in understanding the climate effect of urbanization.Moreover, the findings will provide important reference values for future urban planning and policy formulation in other cities.

Data and Methods
In this paper, two sets of numerical simulations of Shanghai were conducted based on WRF, referred to as CTL and UCP cases.In the CTL case, building height was set to zero in urban areas, treating the city as a slab.In contrast, the UCP case applied the high-resolution urban morphological parameters to represent the actual distribution of buildings (Sun et al., 2021).The simulations covered the winter (from December of the previous year to February) and summer (from June to August) seasons from 2016 to 2020.All the simulations have three two-way nested domains (Figure 1) centered at 32°N, 118.8°E with horizontal grid spacing (grid points) of 9 km (160 × 160), 3 km (162 × 162) and 1 km (135 × 135), respectively.The vertical coordinate contains 43 sigma levels from the surface to 50 hPa.Simulations were conducted with the initial and boundary conditions provided by data from the National Center for Environment Prediction operational Global Final Analyses at a horizontal resolution of 0.25°× 0.25°.The spin-up time of each group of experiments is 7 days.We also updated the urban cover fraction of the underlying surface of Shanghai based on the impervious surface coverage data from FROM-GLC10 (Gong et al., 2019).The specific parameterization scheme settings are shown in Table 1.The distribution of Building Plan Area Fraction, Building Surface to Plan Area Ratio and building height probability of Shanghai are shown in Figure 2. The high-resolution UCPs can describe the refine distribution of buildings and improve the expression of the heterogeneous characteristics of the urban underlying surface.
The simulated 2 m temperature and 10 m wind speed of UCP cases were compared with the observations of Shanghai station (31.4°N, 121.467°E) from 2016 to 2020.The Mean Bias (MB), Root Mean Square Error (RMSE) and Correlation Coefficient (R) are taken to evaluate the performance of WRF model.Figure 3 shows the comparison of the average simulated and observed 2 m temperature and 10 m wind speed in summer and winter of Shanghai from 2016 to 2020.In can be seen that the WRF model can well reproduce the variation of 2 m temperature and 10 m wind speed.The statistical results presented in Table 2 show that the WRF model has good performance in simulating the 2 m temperature and 10 m wind speed in Shanghai.The correlation coefficients of the 2 m temperature and 10 m wind speed in summer (winter) are 0.89 (0.93) and 0.57 (0.66) and their RMSE are 2.3°C (2.6°C) and 0.9 m/s (1.1 m/s), respectively.For long-term simulation, such errors are acceptable and can meet the accuracy requirements of this study.To further evaluate the simulation of UCP case using high-resolution urban morphological parameters, we validated the modeled 2 m  Longwave radiation RRTM scheme (Mlawer et al., 1997) Shortwave radiation Dudhia scheme (Dudhia, 1988) Land surface Unified Noah land surface model (Chen & Dudhia, 2001) Boundary layer BouLac scheme (Bougeault & Lacarrere, 1989) Urban canopy BEP scheme (Martilli et al., 2002) temperature and 10 m wind speed against multiple automated weather stations (AWSs) in Shanghai (Figure 4).Due to the data sharing policy and quality of observation data of AWSs, we only compared hourly data from 43 AWSs in July 2017.The result shows that UCP case could reproduce the observed variability of the 2 m temperature and 10 m wind speed.

The Dynamic and Thermal Effects of Building Morphology on Urban Canopy Wind Speed
The influence of building morphology on urban canopy wind speed in Shanghai was shown in Figure 5.It can be seen that the overall effect of building morphology on urban canopy wind speed is to reduce the wind speed, but the influence of building morphology on urban canopy wind speed varies seasonally and diurnally.The attenuation of urban canopy wind speed is stronger at night than during the day, and greater in winter during the day compared to summer.However, there is no significant difference in the attenuation of canopy wind speed at night between winter and summer.The average reduction in canopy wind speed caused by building morphology in Shanghai is 1.5 and 1.1 m/s in winter daytime and summer daytime, respectively, and approximately 1.8 m/s in winter nighttime and summer nighttime.Converted to percentage, the wind speed attenuation caused by building morphology is 35% and 30% in winter daytime and summer daytime respectively, which is consistent with the  37% wind speed attenuation caused by urbanization in a previous study in the Yangtze River Delta (Zhang et al., 2010).
Buildings generally have an attenuation effect on near-surface wind speed, but there is an area outside the city where wind speed increases during the day.This phenomenon of increased near-surface wind speed may be related to the urban heat island effect.Buildings increase the intensity of the urban heat island and the momentum of the upper layer is transported downward to the surface, resulting in an increase in the urban near-surface wind speed.The urban warming effect caused by buildings is stronger in summer than in winter, and the downward transportation of momentum in summer is also stronger, resulting in weaker attenuation of wind speed.This is the reason why the attenuation effect of buildings on urban near-surface wind speed is weaker in summer than in winter.The urban heat island circulation triggered by the urban heat island makes the wind field converge in the city, increasing the wind speed in the peripheral area of the city.Buildings also enhance the sea-land breeze in Shanghai during the day, resulting in an increase in the near-surface wind speed in the northeast of Shanghai.
To account for the influence of urban morphology on the wind flow, the WRF model utilizes urban morphology parameters in the urban canopy parameterization schemes.These parameters represent the real morphological characteristics of buildings in urban areas.There are currently seven urban morphological parameters in the WRF model, each of which represents different characteristics of building morphology.The surface urban heat island  (SUHI) intensity is significantly affected by urban morphology and the impact assessed using different urban morphological parameters is consistent on the SUHI intensity (Hou et al., 2023).However, the impact of urban morphological parameters on urban flows are less studied.To investigate the impact of urban morphology on urban flows, we calculated the correlation between the mean urban canopy wind speed attenuation caused by buildings from 2016 to 2020 and various morphological parameters (including Mean Building Height, Standard Deviation of Building Height, Plan Area Fraction, Building Surface to Plan Area Ratio, Frontal Area Index and Area Weighted Mean Building Height).The result showed that Building Surface to Plan Area Ratio has the best correlation with canopy wind speed attenuation (Figure S1).This parameter can best reflect the influence of building morphology on urban canopy wind speed.Although the Building Surface to Plan Area Ratio is a single indicator, it contains information of both horizontal and vertical building surfaces.Therefore, it can be used to represent the urban morphology.If we select other single morphological parameter (such as Mean Building Height or Building Plan Area Fraction), these parameters can only represent the characteristics of one dimension of the building, which may bring certain uncertainty to the conclusion.
To describe the impact of buildings on the wind flow, BEP adds the attenuation term of the building surface area to the momentum in the momentum equation.BEP calculates the attenuation of wind caused by building surface at different heights from the ground to the height of the tallest building.The attenuation increases with the building surface area.The presence of horizontal surface such as roofs or canyon floors induces a frictional force with consequent loss of momentum, while the vertical surfaces (walls) induce pressure and viscous drag forces on the flow.Due to the different attenuation mechanisms of the building horizontal surface and vertical surface to the flow, BEP makes different parameterizations for the two surfaces.Building Surface to Plan Area Ratio contains information of both the horizontal and vertical surface of the building, making it the typical morphological parameter used to describe the attenuation of buildings to the urban canopy wind speed.
As a whole, urban areas typically have larger roughness elements (mostly buildings) than the surrounding rural areas (Bottema, 1997).Due to this effect, wind speed tends to be reduced as the air moves toward urban areas (Hou et al., 2013;Zhu et al., 2016).This effect can be considered as the dynamic effect of building morphology on urban canopy wind speed.However, in most cases, the dynamic and thermal effects of buildings on wind speed coexist.
To separate the dynamic effects, it is necessary to exclude the influence of thermal effects of buildings.The thermal effect of buildings on urban canopy wind speed is mainly induced by the urban heat island, and its intensity weakens as the wind speed increases.Theoretically, there is a threshold wind speed, and when the wind speed reaches this threshold, the heating effect of urban will be offset (Bornstein & Johnson, 1977;Wong, 1976;Wong & Dirks, 1978).At that time, the impact of buildings on urban canopy wind speed will be only the dynamic effect.To determine this threshold wind speed, we analyzed the wind speed in summer and winter in Shanghai from 2016 to 2020. Figure 6 shows the statistical results of the mean wind speed in summer and winter.Based on the statistical results, the urban wind speed was simply divided into three categories: wind speeds less than 3 m/s were classified as weak wind, those between 3 and 5 m/s were classified as medium wind, and those greater than 5 m/s were classified as strong wind.The critical threshold wind speed obtained in previous studies was 3.5-5 m/s, which is consistent with the wind speed here (Oke et al., 2017;Wong, 1976;Wong & Dirks, 1978).The distribution of buildings in urban areas is highly variable, which results in spatial variation in the attenuation of urban canopy wind speed.The Building Surface to Plan Area Ratio is an important parameter that reflects the morphology of buildings and can be used to model the attenuation of urban canopy wind speed.Figure 7 shows the relationship between the change of urban canopy wind speed with Building Surface to Plan Area Ratio during the daytime and nighttime in Shanghai from 2016 to 2020 (14:00 is taken as the typical daytime and 02:00 is taken as the typical nighttime).The results show that the attenuation of urban canopy wind speed caused by buildings varies with wind speed, and the attenuation of wind speed also shows different characteristics during the day and night.As wind speed increases, the attenuation of urban canopy wind speed also increases.Under strong wind conditions, the attenuation of wind speed can reach up to 80% due to the presence of buildings.Even in weak wind conditions, the attenuation of wind speed can reach up to 50%-60%.
The attenuation of urban canopy wind speed caused by buildings has obvious diurnal variation characteristics.
During the day, the attenuation of urban canopy wind speed increases as the mean wind speed increases.This is due to the fact that the urban surface heats up during the day and destabilizes the atmospheric stratification, which weakens the attenuation of wind speed caused by buildings.The thermal effect of buildings on the wind speed is mainly caused by urban heating.The heating caused by buildings varies depending on the wind speed, leading to different characteristics of wind speed attenuation during the day.However, the general trend is that the attenuation ratio increases as the wind speed increases.At night, the urban environment lacks heat sources, and the atmospheric stability tends to be neutral or even stable.This means that the thermal effect of buildings on wind speed can be ignored, and only the dynamic effect of buildings on the urban canopy wind speed is considered.The presence of buildings increases the roughness of the urban underlying surface and enhances the drag effect on the wind field.During the night, the dynamic effect dominates, and the attenuation of wind speed caused by the dynamic effect if fixed, remaining consistent under different wind speed conditions.
Besides the differences of daytime and nighttime, the attenuation of urban canopy wind speed varies depending on the wind speed conditions.To investigate this further, we conducted a statistical analysis of the wind speed attenuation with the Building Surface to Plan Area Ratio under different wind speed conditions.Figure 8 shows the long-term statistical characteristics of wind speed attenuation under different wind speed conditions.It can be seen that the attenuation of canopy wind speed is higher at night than during the day.However, as wind speed increases, the difference in attenuation between day and night gradually decreases.The diurnal difference of canopy wind speed attenuation is due to the competition between the thermal and dynamic effects of buildings on the wind speed.Under weak wind or even static wind conditions, the increase in the roughness of the urban underlying surface caused by buildings has little effect on the urban canopy wind speed as there is no relative movement.Thus, the main contribution of building morphology to the wind speed is purely thermal.However, as wind speed increases, the dynamic effect of building morphology on the urban canopy wind speed begins to manifest.Additionally, due to the advection of heat by the wind, the intensity of urban heat island is weakened to some extent and the dynamic effect begins to dominate.When the wind speed exceeds a certain threshold, the attenuation of wind speed becomes purely dynamic.Under strong wind conditions, the attenuation of urban canopy wind speed remains consistent during both day and night.The main influence of buildings on the wind speed is dynamic and the contribution of thermal effects can be ignored.From Figures 7 and 8, it is apparent that urban grids experience attenuation under medium wind and strong wind conditions.However, under weak wind conditions, the canopy wind speed increases in some urban grids, and this phenomenon is only observed during the day.At night, the canopy wind speed of urban grids always experiences net attenuation under all conditions.From the perspective of roughness, buildings increase urban roughness, which enhances the drag effect of the urban underlying surface on the flow, thereby reducing the wind speed.The dynamic effect of buildings on the wind only leads to a reduction in wind speed.The phenomenon of increased wind speed in some urban grids is likely due to the thermal effects of buildings.These urban grids with increased wind speeds have a small Building Surface to Plan Area Ratio and are mostly located in the outskirts of the city.There are two main reasons for the increased wind speed during the daytime in these grids: (a) the enhanced turbulent activities in urban areas result in enhanced momentum exchange between the upper and lower atmospheric layers, which compensates for the loss of momentum near the ground; (b) under weak wind conditions, the probability of typical urban heat island circulation increases, and there are downdrafts and convergence flows of the heat island circulation at the edge of the city, which cause an increase in wind speed.
After analyzing the impact of wind speed on the attenuation of urban canopy wind speed, we then studied the vertical distribution of TKE in urban areas under different wind speed conditions.Figure 9 demonstrates that the TKE during daytime is significantly higher than that at nighttime under weak wind and medium wind conditions, but under strong wind conditions, the TKE distribution is basically the same during the day and night.Thermal turbulence and mechanical turbulence are the two types of excitation mechanisms for urban turbulence.The difference in turbulence between day and night under unchanged wind speed mainly arises from the excitation of thermal turbulence, which is closely related to the intensity of urban heat island.To further investigate the influence of buildings on UHI intensity under different wind speed conditions, we found a significant negative correlation between the increase of UHI intensity caused by buildings and wind speed.As the wind speed increases, the change of UHI intensity becomes smaller.Under weak wind, medium wind, and strong wind conditions, the ∆UHI (UHI_UCP-UHI_CTL) values are 5.68°C, 3.73°C, and 0.46°C, respectively.Under strong wind conditions, the UHI intensity is too weak to generate thermal turbulence, which explains why the vertical distribution of TKE of daytime and nighttime is almost the same under strong wind conditions.During weak wind and medium wind conditions, the thermal effect of the city is significant, resulting in a significant difference in TKE between day and night.More thermal turbulence is generated during the day, resulting in the difference in TKE in the vertical direction.
In addition to diurnal differences, the thermal characteristics also vary between seasons.Figure 10 displays the variation of urban canopy wind speed attenuation with Building Surface to Plan Area Ratio under different wind speed conditions during the winter in Shanghai.It can be observed that the attenuation of canopy wind speed in winter is comparable to that in summer, with more substantial attenuation at night than during the day.But in winter, the thermal effect of buildings can be ignored under the condition of medium wind and the attenuation of wind speed during the day and night is the same under medium wind and strong wind conditions.Although the thermal effect of buildings in winter is relatively weaker than that in summer, buildings still have a thermal effect in winter.As can be seen from Figure 10a, are still some urban grids on the outskirts of the city where the wind speed increases.This phenomenon is similar to that observed in summer and is caused by the influence of the urban heat island circulation superimposed with the momentum transportation from the upper atmosphere.
To summarize, the attenuation of urban canopy wind speed is influenced by both the dynamic effect (related to roughness) and the thermal effect (related to atmospheric stability).The roughness effect can be quantified by the Building Surface to Plan Area Ratio, while the thermal effect is related to the heating effect of buildings on the surrounding atmosphere.It is challenging to quantitatively describe the impact of buildings on atmospheric stability, but the change in urban heat island intensity (∆UHI) caused by buildings can be used as an indicator of their thermal effect.Overall, both dynamic and thermal effects should be considered when analyzing the wind field in urban areas.
Buildings play a significant role in reducing urban canopy wind speed, and the degree of wind speed attenuation is closely related to Building Surface to Plan Area Ratio.Therefore, we attempt to quantify the effect of building morphology on the attenuation of wind speed using Building Surface to Plan Area Ratio as a metric.To isolate the dynamic effect of buildings on the wind speed attenuation, we need to minimize the influence of thermal effect.Thus, we separate the dynamic effect under strong wind conditions.Figure 11 presents the fitting curves of wind speed attenuation with Building Surface to Plan Area Ratio under different wind speed conditions, as well as the corresponding change in UHI intensity.Figure 11a shows the pure dynamic effect of building morphology on wind speed under strong wind conditions, where the thermal effect can be neglected.Meanwhile, Figures 11c and  11e display the fitting curves of the combined dynamic and thermal effects of buildings on wind speed attenuation under medium wind and weak wind conditions, respectively.The dynamic effect of building morphology on urban canopy wind speed can be described as: where Y represents the ratio of the attenuation wind speed to the unattenuated wind speed, and λ b is Building Surface to Plan Area Ratio.When there are no building, λ b = 0, Y = 1, meaning that the urban canopy wind speed does not attenuate.As the number of buildings increases, λ b increases and Y gradually decreases until λ b = 4.9, which denotes that the wind speed attenuates to 0. Therefore, without considering the thermal effect, the value of λ b must reach 4.9 before the urban canopy wind speed attenuates to 0. As the typical megacity in China, the maximum of Building Surface to Plan Area Ratio of Shanghai is about 2.3, which is much lower than the threshold of 4.9.Hence, it is nearly impossible for the urban canopy wind speed to decrease to 0 by relying solely on the dynamic effect of building morphology.
In theory, the attenuation of wind speed caused by the dynamic effect is constant.Therefore, we can isolate the influence of the thermal effect by subtracting the dynamic effect from the total effect of the building morphology on urban canopy wind speed.From the fitting curves obtained under different ∆UHI (i.e., different atmospheric stability conditions), it can be seen that the thermal effect of building morphology on the urban canopy wind speed is reflected in the residual term of Equation 1 and it leads to an increase in urban canopy wind speed.Under medium wind speed conditions, the thermal effect results in an increase in urban canopy wind speed of approximately 8%, while it is 20% under weak wind speed conditions.
The relationship between urban morphology and wind speed is complex, as it is influenced by both the dynamic and thermal effects of buildings.By separating the dynamic effect from the total effect, the thermal effect can be analyzed and quantified.This effect is mainly caused by the impact of building morphology on atmospheric stability, which in turn affects the exchange of momentum between the upper and lower layers, leading to an increase in surface wind speed.UHI intensity can effectively characterize the thermal properties of the urban underlying surface, and ∆UHI can reflect the variations in atmospheric stability.As shown in Figure 11, ∆UHI varies with wind speed and the contribution of thermal effects on wind speed also varies under different ∆UHI.
To obtain quantitative conclusions about the relationship between urban canopy wind speed and ∆UHI, the wind speed was classified more precisely.The average urban canopy wind speed was categorized into 5 groups with an interval of 1 m/s, and each group was analyzed separately.Figure 12 shows the relationship between ∆UHI and Building Surface to Plan Area Ratio under different wind speed conditions.It can be seen that ∆UHI decreases as the wind speed increases, and the influence of building morphology on ∆UHI can be disregarded when the wind speed is greater than 5 m/s. Figure 13 shows the relationship between the attenuation of urban canopy wind speed and Building Surface to Plan Area Ratio under different ∆UHI.The fitted curves under different ∆UHI reveal that the thermal effect of buildings on the canopy wind speed only affects the residual term of Equation 1.In other words, the thermal effect of building morphology does not impact the dynamic effect, and the effects of both on the urban canopy wind speed are superimposed.Based on this, we revised the Equation 1: which could also be expressed as: In Equation 3, 1 Y represents the attenuation of urban canopy wind speed caused by the building morphology, 0.75 * ̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅ ̅ (ln(λ b + 1)) √ represents the contribution of the dynamic effects, and f(∆ UHI ) represents the contribution of the thermal effects.The dynamic effect of building morphology is known to cause attenuation of the wind speed, while the thermal effect tends to counteract this attenuation (since strong stable atmospheric stratification is uncommon in urban areas).These two effects coexist and can compete with each other, leading to complex patterns of airflow in urban environments.
After separating the dynamic and thermal effects of building morphology on urban canopy wind speed, it is possible to compare the relative contributions of the two effects.Figure 14 shows the contribution of building morphology's dynamic effect to the urban canopy wind speed and the corresponding contribution of thermal effect in suppressing wind speed attenuation under different wind speed conditions.We can observe that the contribution of dynamic effect is nearly constant under different win speed conditions, which only depends on the roughness of the urban underlying surface.On the other hand, the contribution of building morphology's thermal effect in suppressing wind speed attenuation decreases as the average wind speed increases.When the average wind speed ranges from 1 to 2 m/s, the contribution of the thermal effect can reach up to 26%, but when the average win speed exceeds 5 m/s, the contribution of thermal effect tends to be negligible.In all conditions, the contribution of building's dynamic effect to urban canopy wind speed is greater than that of the thermal effect.According to the statistical results of the average wind speed in Shanghai from 2016 to 2020, the wind speed is mainly dominated by weak wind speed.Under such wind speed conditions, although the thermal effect is not as significant as the dynamic effect, it can still increase the wind speed by about 20%, which can't be ignored.
The impact of building's thermal effect on wind speed is mainly achieved through its influence on the UHI intensity.To quantify the relationship between f(∆ UHI ) and ∆ UHI , a fitting is performed and the result is shown in Figure 15.The thermal effect of building morphology can be described as: f (∆ UHI ) = 0.005 * ∆ UHI 2.13 .(4) Combining Equations 2 and 4, the impact of building morphology on urban canopy wind speed can be described by Building Surface to Plan Area Ratio as: The thermal effect of buildings on the urban canopy wind speed can be summarized into two aspects: (a) The presence of buildings lead to an enhancement of turbulence excitation, which includes both the increase of thermal turbulence due to urban warming and the mechanical turbulence caused by the increased urban surface roughness.The enhancement of turbulence in urban areas leads to downward transportation of momentum from the upper layer, which in turn suppresses the attenuation of urban canopy wind speed.(b) The presence of buildings enhances the ability of urban surface to absorb radiation.The increased UHI intensity favors the excitation of urban heat island circulation, which will affect the structure and intensity of the urban canopy wind speed.When the wind speed is large, the ideal "dome" structure of circulation will be replaced by the urban heat plume.At this time, the influence of the urban heat island circulation on the urban canopy wind speed will be transported to the downwind suburbs.Therefore, the impact of urban heat island circulation on the urban canopy wind speed will only manifest itself when the wind speed is very weak.

Summary and Discussion
This study aimed to investigate the impact of building morphology on urban canopy wind speed using the WRF-BEP model with high-resolution urban morphological parameters (UCPs).The dynamic and thermal effects of building morphology on urban canopy wind speed were separated and their quantitative expression functions were given.The results showed that the effects of the two on urban canopy wind speed are opposite.The impact of building morphology on urban canopy wind speed is mainly dominated by the dynamic effect resulting in a wind speed attenuation of approximately 50%, and the dynamic effect is relatively constant.The thermal effect of building morphology on urban canopy wind speed suppressed the decrease in wind speed and the thermal effect decreased (increased) with increasing wind speed (UHI).The relative contributions of the dynamic and thermal effects varied with wind speed, and with increasing wind speed, the contribution of the thermal effect gradually decreases.It is noteworthy that under weak wind speed conditions, the thermal effect can increase the wind speed by about 20%, which can't be ignored.This study provided a quantitative relationship between the urban canopy wind speed variation and urban morphology, as well as urban heat island intensity.Based on the quantitative description, the dynamic and thermal effects could be separated.The urban morphology can be obtained from the urban morphological parameter data set, but the change of UHI intensity caused by buildings is difficult to obtained directly.Further research is needed to develop more accurate and reliable quantitative descriptions of the dynamic and thermal effects on urban canopy wind speed, using more intuitive variables such as UHI intensity and wind seed.This may require the use of complex mathematical models and computational methods, as well as extensive field observations and data collection to validate and optimize these methods.

Figure 1 .
Figure 1.The configuration of two-way nested domains for WRF simulations and geographical locations of the AWSs in Shanghai (red dot: Shanghai station, blue dots: AWSs).

Figure 3 .
Figure 3. Average simulated and observational 2 m temperature and 10 m wind speed in Shanghai of (a) summer and (b) winter from 2016 to 2020 (black solid line represents simulated values and red dotted line represents observed values).

Figure 4 .
Figure 4. Time series of averaged observed and modeled hourly 2 m temperature and 10 m wind speed over the AWS sites in Shanghai and the corresponding model grids in July 2017.

Figure 5 .
Figure 5.The influence of building morphology on urban canopy wind speed in Shanghai (UCP-CTL, m/s), (a) daytime in summer; (b) daytime in winter; (c) nighttime in summer; (d) nighttime in winter.

Figure 6 .
Figure 6.Frequency of urban average wind speed of (a) summer and (b) winter in Shanghai from 2016 to 2020.

Figure 7 .
Figure 7.The attenuation of wind speed caused by buildings varies with the Building Surface to Plan Area Ratio of summer in Shanghai, (a, b) represent the attenuation of daytime, while (c, d) represent the attenuation of nighttime.

Figure 8 .
Figure 8.The variation of urban canopy wind speed attenuation with Building Surface to Plan Area Ratio under different wind speed conditions, (a, b) represent the attenuation under weak wind condition, (c, d) represent the attenuation under medium wind condition, (e, f) represent the attenuation under strong wind condition.

Figure 9 .
Figure 9.The vertical distribution of (a) TKE and (b) ∆TKE (TKE_UCP-TKE_CTL) in urban areas under weak wind, medium wind and strong wind conditions during daytime and nighttime.

Figure 10 .
Figure 10.The variation of urban canopy wind speed attenuation with Building Surface to Plan Area Ratio in winter under (a) weak wind, (b) medium wind and (c) strong wind conditions.

Figure 11 .
Figure 11.The left column shows the fitted curves of the wind speed attenuation with Building Surface to Plan Area Ratio and the right column shows the corresponding changes in UHI intensity (UHI_UCP-UHI_CTL) under different wind speed conditions, (a, b) represent strong wind speed condition, (c, d) represent medium wind speed condition, (e, f) represent weak wind speed condition.

Figure 12 .
Figure 12.The variation of ∆UHI (UHI_UCP-UHI_UCP) with Building Surface to Plan Area Ratio under different wind speed conditions.

Figure 13 .
Figure 13.The relationship between wind speed attenuation and Building Surface to Plan Area Ratio for different ∆UHI (the black solid line indicates the fitting curve between wind speed attenuation and Building Surface to Plan Area Ratio).

Figure 14 .
Figure 14.The dynamic contribution of building morphology to the attenuation of urban canopy wind speed and the thermal contribution to the suppression of wind speed under different wind speed conditions.

Figure 15 .
Figure15.The fitting curve of building's thermal effect on the urban canopy wind speed.

Table 2
Evaluations of the Simulated 2 m Temperature (°C) and 10 m Wind Speed (m/s) in Shanghai From 2016 to 2020 Note. "*" represents passing the t-test with a significance level of 95%.