Effect of urbanization on gust wind speed in summer over an urban area in Eastern China

Previous studies have extensively examined effects of urbanization on mean wind speed, but few studies were aimed at gust wind speed, while large wind gust could cause safety and economic hazard to a variety of activities. In this study, the effect of urbanization on the gust wind speed in Nanjing, China is assessed using the Weather Research and Forecasting (WRF) model with a parameterization of the gust wind speed. The WRF simulations are run for the summer period of 2013 with the underlying surface before and after the urbanization. The results indicate that although the mean wind speed is reduced, the gust wind speed in the urban areas is increased significantly due to the enhanced friction velocity and less atmospheric stability induced by the urbanization, while the contribution of deep convection is relatively small. The gust wind speed increases more in the nighttime (0.6–0.9 m s−1) than in the daytime (less than 0.3 m s−1), since the turbulence is enhanced more in the nighttime than in the daytime after the urbanization. The probability distribution shows that the increase of gust wind speed is mainly between 0.0–0.5 m s−1 in the urban areas. In different urban land categories, the increase of the gust wind speed is larger in the commercial or industrial areas than in high-intensity and low-intensity residential urban areas. Averagely, the gust wind speed in the entire city after the urbanization increases by 0.02, 0.36 and 0.19 m s−1 for the daytime, nighttime and daily mean, respectively.


Introduction
The wind gust is a short-duration maximal wind speed occurring in a long-duration time series of wind speed. According to the standard definition of the World Meteorological Organization (WMO 2014), the gust wind speed is the maximal 3-second moving-average wind speed in a 10 minute wind speed time series. In practice, the maximal instantaneous wind speed in an hour or day is usually taken as the hourly or daily gust wind speed (Patlakas et al 2017, ECMWF 2021. Wind gust often represents safety and economic hazard to a variety of activities, since it can cause damage to wind turbines, buildings and other structures (Beljaars 1987, Hawbecker et al 2017, destroy vegetation and lead to large socio-economic impacts (Fink et al 2009), endanger aviation and induce train derailments and truck rollovers (Wu et al 2019, Montenegro et al 2020, cause and spread forest fire and dust storms (Suomi et al 2013, Forzieri et al 2020, and so forth. Therefore, it is important to study on the wind gust over a certain area as it may be helpful in accurate severe weather warnings and valuable for societal, scientific and economic developments (e.g. Blackmore and Tsokri 2004, Suzuki 2009, Liu et al 2023. During recent decades, many developing countries such as China have been undergoing accelerated urbanization. It has already been well known that urbanization can modify local weather/climate and atmospheric environment by changing dynamical and thermodynamical properties of underlying surface and emitting anthropogenic heat, aerosols and pollutants etc in the urban areas (e.g. Ryu et al 2013, Wang et al 2014, Tao et al 2017, Huszar et al 2018, Li et al 2019. Some key climatic effects of urbanization have been widely investigated, such as the urban heat islands (UHI) effect (Oke 1982, Zhao et al 2013, the slow-down of near-surface wind by enhanced drag force and greater roughness length (Liao et al 2015, Wang et al 2019, the thermally-driven local atmospheric circulation (Lin et al 2008, Chen et al 2009, and the possible precipitation changes (Miao et al 2011, Georgescu et al 2014, Holst et al 2016. Urbanization can also modify the characteristics/ structures of the planetary boundary layer (PBL) including the boundary layer height in urban areas (Ryu et al 2013), decrease the surface evaporation (Wienert and Kuttler 2005) and relative humidity (Zhao et al 2013), and increase the friction velocity because of the enhanced surface frictional dynamic effect , Ao et al 2022.
Although so far there have already been a large amount of literatures on effect of urbanization on weather/climate and atmospheric environment, in these researches, there have been rarely studies on its effect on the wind gust. This may be because the wind gust is not directly simulated due to the physical mechanism for the generation of the wind gust, which is the transfer/transportation of momentum caused by the deep convection and atmospheric turbulence, of which both cannot be directly predicted in present numerical models. Nevertheless, some schemes have been developed for the parameterization of the wind gust, especially in the last decade (Brasseur 2001, Suomi 2017, Gutierrez and Fovell 2018, Kurbatova et al 2018, Sheridan 2018. These schemes are based on common methods that wind gusts result from mean wind, atmospheric turbulence and deep convection and are thus influenced by the variables related to turbulence and convection. These approaches thus enable the study on the effect of urbanization on wind gusts by numerical models. In this study, the effect of urbanization on the wind gust in the Nanjing megacity is analyzed. As an important central city in eastern China, Nanjing has experienced significant urbanization (Zhao et al 2013), which has led to obvious expansion of the urban areas and changed the urban climate. In this study, we use the simulation results from the Weather Research and Forecasting (WRF) model as the input for a parameterization scheme of the wind gust to estimate the gust wind speed. Then the temporal and spatial change of the wind gust are examined through comparison of the gust wind speed between before and after the urbanization in Nanjing to reveal the effect of urbanization on the wind gust.

Model configuration
In this study, the WRF model is employed to assess the impact of urbanization on the wind gust. We select the summer (June, July and August) of 2013 as the simulation time. Two sensitivity simulations are conducted, named as 'CTL' and 'Urban' , respectively. In the CTL (i.e. control; before the urbanization) case the U.S. Geological Survey (USGS) 1 km land use map in the early 1990s is used, while in the urban case the urban land use type is updated by the 2012 Landsat map. The urban land cover is further divided into three types by the definition of the WRF Single-Layer Urban Canopy Model, including low-(LIR) and high-intensity residential (HIR), and commercial or industrial (COI) urban areas. Both the simulations use the same initial and boundary conditions and thus the effect of climate change is not taken into account in this study. The details of the model configuration can be found in supplementary information.

Parameterization of the wind gust
The parameterization scheme employed in this study to calculate the gust wind speed is the scheme used by the European Center for Medium-Range Weather Forecasts (ECMWF 2021, see supplementary information).

Validation of the WRF simulation
The evaluations in this study and the validations in our previous studies, which can be found in supplementary information, suggest that WRF simulations are in good agreement with the observations and are reliable in simulating the impact of urbanization on urban meteorological fields, thus can be employed to investigate the effect of urbanization on the wind gust.

Spatial variation
The spatial differences of the mean and gust wind speeds between before and after the urbanization during daytime and nighttime as well as in the daily mean are shown in figure 1. The corresponding percentage changes are shown in figure S4. The daytime is denoted as 07:00-18:00 LST and the nighttime as the rest of a day (00:00-06:00 LST and 19:00-23:00 LST).
The results indicate that the mean wind speed reduces significantly over the urban areas, while the change is small in the surrounding countryside areas. The daytime mean wind speed in LIR urban areas reduces by approximately 0.1-0.2 m s −1 (8%-12%), and the reduction of the mean wind speed increases with the increase of building density, as the largest reduction is in COI urban areas where it reduces by approximately 0.5 m s −1 (20%), implying that stronger drag force and greater roughness length make larger wind speed reduction, which is in agreement with prior studies (Liao et al 2015, Wang et al 2019. The nighttime wind speed reduction is less than that at daytime, being between 0.1-0.3 m s −1 (4%-8%) over most urban areas, which could be because the PBL is more stable and the wind speed is generally smaller in the nighttime, thus leading to less change of the nighttime wind speed (Wang et al 2020). The daily mean wind speed reduces between 0.2-0.4 m s −1 (8%-16%) in most urban areas, which is consistent with prior studies simulating the effects of urbanization on wind speed (Wang et al 2014, Yang et al 2015, Huszar et al 2018. In contrast, the gust wind speed generally increases in urban areas with larger increment in COI urban areas, and the increment is greater in the nighttime than that at daytime. As shown in figures 1(d)-(f) and S4(d)-(f), the daytime gust wind speed in most urban areas increases less than 0.3 m s −1 (10%) and slightly decreases in some nature land pixels. At night, the gust wind speed increases in all urban pixels, and the increment could exceed 0.9 m s −1 (30%) in COI urban areas. On average, the daily mean gust wind speed increases by 0.3-0.6 m s −1 (10%-20%) over most urban areas.
The effect of the urbanization on the gust wind speed can be viewed in another way. Figure 2 shows the probability distribution of the change of gust wind speed between CTL and Urban case in the urban areas. Whether averaged over the daytime or nighttime or the daily mean, the positive differences have larger probability than the negative ones, indicating the gust wind speed has increased in the urban areas after the urbanization. The positive differences have larger probability in the nighttime than in the daytime and in the daily mean, indicating the gust wind speed is more increased in the nighttime than in the daytime. Moreover, the probability distribution of the gust wind speed increasing by 0.0-0.5 m s −1 is the largest, reaching 7.0, 6.0 and 6.4% at night, during the day and for the daily mean, respectively, indicating that the change of gust wind speed is mainly between 0.0-0.5 m s −1 in the urban areas.
As the gust wind speed increases while the mean wind speed decreases, it is necessary to look into the effects of the second and third terms at the right hand of equation (S1) in order to investigate how they can overcome the decrease in the mean wind speed and make the gust wind speed increase. Shown in figure 3 is the differences between before and after the urbanization of the friction velocity, surfacelayer atmospheric stability z/L (hereafter abbreviated as the atmospheric stability, where z is the height of 10 m above ground level and L is the Obukhov length calculated in equations (S3)-(S5) in supplementary information), and mean wind speed difference between 850 hPa and 950 hPa (i.e. wind shear) over the daytime, nighttime and daily mean in the summertime of 2013. The corresponding percentage changes are shown in figure S5. The friction velocity is increased after the urbanization, especially the increase is greater in the nighttime than that in the daytime. The daytime friction velocity increases approximately 0.05-0.10 m s −1 (10%-20%) over most urban areas. At night, the friction velocity increases by 0.10-0.15 m s −1 (25%-40%) in most urban areas, and could exceed 0.20 m s −1 (50%) in some COI urban areas. On average, the daily mean friction velocity increases by 0.05-0.15 m s −1 (15%-30%) over the urban areas. The increase of friction velocity illustrates that the surface frictional  and dragging effects are obviously increased with the increase of building density and building heights , Ao et al 2022. Since the friction velocity is proportional to the standard deviation of the wind speed fluctuations in the surface layer similarity relation, the increase of the friction velocity also represents the increase of the turbulence intensity characterizing the strength of turbulent motion (see supplementary information).
The surface-layer atmospheric stability is reduced (i.e. the surface-layer atmosphere becomes more unstable) after the urbanization, and similarly the reduction in the atmospheric stability is larger (i.e. further towards unstable) in the nighttime than in the daytime. The atmospheric stability decreases less than 0.10 (20%) in most urban areas in the daytime, whereas it decreases by 0.30-0.40 (40%-80%) during nighttime in vast urban areas, and could exceed 0.80 (100%) in some urban areas where the land type changed from nature to urban surfaces. Some researchers showed that the difference of the turbulence intensity between the urban areas and the surrounding countryside areas is more distinct in the nighttime than in the daytime , Wang et al 2019. This is because usually in the daytime the sensible heat flux is positive while the atmospheric stability is negative both in the urban areas and in the surrounding countryside areas, but in the nighttime the positive sensible heat flux and negative atmospheric stability can still be maintained in the urban areas while a near-neutral to stable boundarylayer is often formed in the surrounding countryside areas. Thus in the urban areas, the change of the atmospheric stability after the urbanization is greater in the nighttime than that in the daytime.
The wind shear in the urban areas is also increased after the urbanization, which is because the mean wind speed in the lower atmosphere is reduced in the urban areas after the urbanization while it is less affected in the upper atmosphere (see figures S6 and S7 in supplementary information). While the urbanization also reduces the atmospheric stability and thus tends to decrease the wind shear, the dragging effect reducing the wind speed is stronger, leading to weak increase of the wind shear. Nevertheless, compared to the friction velocity and atmospheric stability, the increase in the wind shear is not apparently different between the daytime and nighttime, which could be because the mean wind speed in the upper atmosphere does not behave with as strong diurnal variation as in the lower atmosphere. Generally, the daily mean wind shear increases 0.20-0.25 m s −1 (10%-25%) over most urban areas.

Diurnal variation
Although it is shown in figures 1 and 2 that the gust wind speed in the urban areas is increased after the urbanization, the increase is apparently inhomogeneous in terms of space and time. To look into the inhomogeneity as well as to further understand how the urbanization affects the gust wind speed, figure 4 compares the diurnal cycles of the differences of the mean wind speed, gust wind speed, friction velocity, atmospheric stability and wind shear averaged over the grids in all urban pixels (i.e. mean) and in each of the three urban types (i.e. COI, HIR, LIR) between before and after urbanization (see figure S8 for CTL and Urban results in supplementary information).
The mean wind speed ( figure 4(a)) is reduced in the urban areas, and the decrease is largest in the daytime. Overall, the mean wind speed could decrease over 0.30 m s −1 during morning and noon time (07:00-14:00 LST) in all urban categories. The mean wind speed over the entire city (i.e. mean) decreases by 0.29, 0.08 and 0.18 m s −1 for the daytime, nighttime and daily mean, respectively. For different urban land types in the urban areas, relatively the least reduction happens in LIR urban areas, although the distinction between the different categories is not significant. This is reasonable since there is relatively weaker drag effect and smaller roughness length in LIR urban areas, thus the mean wind speed is less reduced in the areas.
The diurnal trends of the gust wind speed changes (figure 4(b)) are associated with the changes of mean wind speed. The gust wind speed increases significantly during nighttime. However, there are slight reductions in the late morning and noon. For the three urban land types, the increase is largest in COI urban areas, and is least in LIR urban areas. The maximum increase of gust wind speed for all the three urban land types occurs at 18:00-19:00 LST, with the daily peaks of 0.90, 0.85 and 0.75 m s −1 in COI, HIR and LIR urban areas, respectively. On the other hand, the gust wind speed slightly reduces by 0.10 m s −1 in COI urban areas during 08:00-09:00 LST, and the reduction is relatively larger in HIR and LIR urban areas, which reduces by 0.20 m s −1 in the late morning. The reduction of the gust wind speed in the morning and noon time is mainly caused by the significant decrease of the mean wind speed. Averagely, the gust wind speed over the entire city (i.e. mean) after the urbanization increases by 0.02, 0.36 and 0.19 m s −1 for the daytime, nighttime and daily mean, respectively.
The diurnal variations of the friction velocity changes (figure 4(c)) are also similar to the changes of mean wind speed, as the friction velocity is increased throughout the day in the urban areas after the urbanization and the increase is greater in the nighttime than that in the daytime. The friction velocity is outputted by the WRF model with a surface-layer parameterization scheme that employs the Monin-Obukhov Similarity Theory. Based on the theory, the friction velocity is formulated as a function of the vertical difference of mean wind speed in the surface layer between a certain height and the ground, while the function is regulated by the atmospheric stability conditions (e.g. Stull 1988, Liu et al 2013. The vertical difference of mean wind speed is actually itself since the wind speed at the ground is zero. Thus in the urban areas, the diurnal variation of the friction velocity also partly reflects and is similar to that of the mean wind speed. In addition, the averaged increase in the friction velocity after the urbanization is largest in COI urban areas while least in LIR urban areas, suggesting the turbulent motion is stronger in COI urban areas with greater surface roughness , Ao et al 2022, which also contributes to stronger wind gust enhancement in COI urban areas. In general, the minimum increase of friction velocity in COI, HIR and LIR urban areas is approximately 0.020, 0.010 and 0.005 m s −1 at 08:00 LST, while the maximum increase of friction velocity in the three urban land categories is approximately 0.100, 0.095 and 0.080 m s −1 at 19:00 LST, respectively. Although not showing a clear trend of diurnal variation, the atmospheric stability ( figure 4(d)) consistently decreases over all the urban land categories after the urbanization, especially the decreases are relatively larger in the nighttime with the maximum exceeding 0.5. This is in agreement with existing simulations and observations that the surfacelayer atmosphere is more unstable in the urban areas than in the surrounding countryside areas (Chen and Zhang 2018, Huszar et al 2018, Wang et al 2020, especially in the nighttime when there often exists a nearneutral to stable boundary-layer in the surrounding countryside areas (Barlow et al 2011). More unstable atmosphere is favorable to inducing and maintaining stronger turbulent motion and convection (Barlow 2014), which thus enhances the wind gust in the urban areas.
The wind shear between the upper and lower atmosphere represented by 850 hPa and 950 hPa pressure level (figure 4(e)) consistently increases in the urban areas after the urbanization, which also contributes to the enhancement of the wind gust. Figure 5. The city-wide hourly-averaged differences of the gust wind speed and its contributing components between before and after the urbanization.
Although the differences do not show a strong diurnal variation, it has two peaks around 07:00 LST (0.30 m s −1 ) and 18:00 LST (0.30 m s −1 ) corresponding to the time of the development and collapses of the PBL height. This is an interesting phenomenon that warranties further study. In addition, the distinction of the wind shear between different urban land categories is not significant, which is similar to the case for the mean wind speed.

Cause for gust wind speed change
According to equation (S1), the gust wind speed is the mean wind speed added with the effects of the turbulent motion and deep convection reflected in the wind shear between 850 hPa and 950 hPa pressure level. Since the mean wind speed mainly decreases while the gust wind speed mainly increases in the urban areas as shown in figure 1, it reveals that the increment in the second term (i.e. the atmospheric turbulence) and third term (i.e. deep convection) at the right-hand side of equation (S1) is larger than the decrement in the first term (i.e. the mean wind speed). To further investigate the contribution of each term to the gust wind speed, figure 5 shows the city-wide (i.e. mean) changes for the gust wind speed and the three items between before and after the urbanization. The percentage contributions (%) of each term is shown in figure S9, and the changes and percentage contributions (%) of each term in COI, HIR and LIR urban areas are shown in figures S10 and S11 in supplementary information. Figures 5 and S9 show that although mean wind speed reduces throughout the day for the entire city, the increase of turbulent motion is relatively greater due to larger friction velocity and less atmospheric stability, thus the gust wind speed still increases for most of the day in the urban areas. On the other hand, the increment of deep convection is not significant throughout the day, suggesting a relatively small contribution to the change of gust wind speed. The diurnal change of gust wind speed is also mainly affected by the mean wind speed and turbulence. In the daytime, although the turbulence increases in the urban areas, the decrease of mean wind speed is relatively greater, especially during the late morning to noon time, leading to a small reduction of gust wind speed in the late morning. During the nighttime, the decrease of mean wind speed is small, while the increase of turbulence is greater than that in the daytime since the turbulence is still stronger in the nighttime when the positive (i.e. upward) sensible heat flux and negative (i.e. unstable) atmospheric stability can still be maintained in the urban areas while the contrary is prevalent in the surrounding rural areas (e.g. Zhang et al 2017, Huszar et al 2018, Wang et al 2019, thus leading to a significant increase of gust wind speed at night. For different urban land types (see figures S10 and S11 in supplementary information), there is small difference for the change of mean wind speed, while the difference due to the change of turbulence is greater in COI urban areas, resulting in greater increment of gust wind speed in COI urban areas than in HIR and LIR urban areas.
In general, the changes of gust wind speed induced by the urbanization are mainly affected by the reduced mean wind speed, enhanced friction velocity and more unstable surface-layer atmosphere in the urban areas, while the contribution of deep convection is relatively small.

Conclusions and discussion
This study employs a parameterization of the gust wind speed and the WRF model to investigate the effect of urbanization on the gust wind speed in Nanjing, Eastern China. After the urbanization, the daily mean 10 m wind speed reduces in the urban areas between 0.2-0.4 m s −1 , while the daily mean gust wind speed increases in the urban areas by 0.3-0.6 m s −1 . Temporally, the mean wind speed reduction is larger in the daytime than in the nighttime, while the increases of gust wind speed are larger in the nighttime than in the daytime. In terms of different urban land categories, the relatively largest reduction of mean wind speed and the relatively largest increase of gust wind speed take place in the COI areas.
The overall increased gust wind speed is mainly contributed to by the increased friction velocity and less atmospheric stability in the urban areas after the urbanization, while the contribution of deep convection is relatively small. The diurnal change of gust wind speed is affected by the diurnal variations of the changes of mean wind speed and friction velocity. Larger gust wind speed in COI areas is contributed to by stronger turbulent motion and convection and more unstable surface-layer atmosphere in the areas.
At present there is a lack of adequate and detailed studies on the effect of urbanization on the wind gust, this study may serve as a preliminary investigation on the mechanism of generating and maintaining of wind gusts and thus be potentially beneficial to forecasting/warning of wind gusts in urban areas.

Data availability statement
The data that support the findings of this study are openly available at the following URL/DOI: https:// doi.org/10.6084/m9.figshare.22065284.