Tempospatial Pattern of Surface Wind Speed and the “Urban Stilling Island” in Beijing City

Wind is one of the most direct and exiguous climatic elements affecting the daily life of mankind. It is closely associated with air pollution, sand storms, wind power generation, sailing race, and so on (Pielke et al., 2002). Wind speed (WS), as an important indicator of wind field, is very essential to many other climatic variables. First, WS is a main factor affecting evapotranspiration, which is a key process of the climate and water cycle change at land surface (Hobbins et al., 2004). Second, WS affects the accuracy of the observation data of precipitation. Higher WS often lowers the catch-rate (under-catch) of precipitation (Yang et al., 1999; Zheng and Ren, 2017). Third, WS is also an indispensable part of urban climatology, affecting Urban Heat Island (UHI), air pollution, and surface hydrological processes (Hou et al., 2013).

Previous studies have found that, UHI is usually weaker in windy season (Unger, 1996; Alonso et al., 2007; Yang et al., 2013). As a key urban climatic indica-tor, the characteristics of UHI and its causes have been investigated by many studies (e.g., Oke, 1988; Ren et al., 2005, 2007; Yang et al., 2013; Ren and Zhou, 2014; Bian et al., 2018; Shapiro and Fedorovich, 2018). In fact, WS, precipitation, humidity, air pollution, and others, are important urban climatic indicators that are remarkably different in urban areas from those in rural areas (Ashrafi et al., 2009; Cuadrat et al., 2015; Yang et al., 2017; Chai et al., 2018).

Many studies have demonstrated that the WS shows a decreasing trend over the past decades in countries like China, Austria, Canada, Europe, and America (e.g., Klink, 1999; Pirazzoli and Tomasin, 2003; Tuller, 2004; Xu et al., 2006; Roderick et al., 2007; Azorin-Molina et al., 2016; Yao and Li, 2016; Torralba et al., 2017). In some high-latitude countries, however, opposite conclusions are drawn (Lynch et al., 2004; Aristidi, 2005; Turner et al., 2005). China is a country situated in the middle and low latitudes. Numerous investigations show that WS in China has weakened remarkably, especially in the northern parts of the country (Jiang et al., 2010, Li et al., 2011; Hou et al., 2013), and this phenomenon may be related to three factors: global warming, weakening monsoons, and rapid urbanization. The first two reasons have been already discussed by many researchers (Zhang et al., 2014; Wu and Wu, 2016).

There are some observational evidence on the relationship between WS change and urbanization in the past, but the analyses are still insufficient. One of the commonly recognized phenomenon is that the near surface WS in cities is different from that in rural areas, due to the special thermal and dynamic properties of land surface in urban areas (Bornstein and Johnson, 1977). In many cities of China, such as Beijing, Lanzhou, Nanjing, Wuhan, and Shanghai, the annual and seasonal mean WS in cities are lower than that in their rural sites (Liu et al., 2002; Miao et al., 2016; Wu and Wu, 2016). It is also recognized that higher WS is helpful to the diffusion of atmospheric pollutants (Hosler, 1961; Ashrafi et al., 2009). For example, it has been concluded that high WS is one of the dominant influences to air pollution in Washington and Oregon coastal areas (Ashrafi et al., 2009). Besides, some evidence shows that WS exceeding a threshold value could prevent the development of UHIs (Alonso et al., 2007).

Nonetheless, the previous studies of WS mostly focus on the local phenomenon on large spatial–temporal scales instead of the detailed contrast between city and rural areas, due to the insufficient high-density observations. During the past several years, many automatic weather stations (AWSs) have been built up in China. All the construction of AWSs meets the operational standard set by the China Meteorological Administration (CMA). By 2017, for example, a dense AWS network of about 200 stations has been developed in the Beijing area (Yang et al., 2011, 2013, 2017). About two-thirds of the stations of the network are able to provide the observations of hourly mean surface WS.

In this paper, the hourly mean WS data from the AWS observations in Beijing are applied to investigate the climatological tempospatial pattern of surface WS. We also make a comparison of WS in urban and rural areas by calculating the WS difference between the rural stations and urban stations, which is defined as Urban Stilling Island (USI) in this paper. In the following sections, information and basic climatological features of surface WS in Beijing during 2008–2017 will be illustrated first; then we will examine the detailed features of USI intensity (USII) in built-up urban areas, including its spatial distribution, diurnal cycle, and seasonal variation; last but not least, the reasons for the USII variation will be analyzed and discussed before the conclusions.

Beijing City, located in the north of the North China Plain and south to the Yanshan Mountains, covers an area of almost 1.6 × 10⁴ km ². The flat southeast region of Beijing occupies roughly 38%, while the rest northwest area is mostly mountainous. Beijing has a typical monsoon-driven semi-humid continental climate. It is hot in summer and cold in winter, and enjoys a seasonally high concentrated summer precipitation regime. Since the 1980s, Beijing has experienced rapid urbanization. Up to 2007, the urbanized regions have expanded and covered a much larger area than that of the 1980s. So far, a multiple Ring Road (RR) system of transportation (Fig. 1) has been developed in the urban zones (Mu et al., 2012; Yang et al., 2013), and the 4th, 5th, and 6th RRs came into service in 2001, 2003, and 2009, successively. This analysis will be focused on the urban areas (inside the 6th RR).

Figure 1.  Geographic distribution of the observation stations in Beijing. The 45 urban stations within the 6th Ring Road (RR) are denoted by black dots, whereas the 6 rural stations outside the 6th RR are denoted by triangles. The blue circles (i.e., lines A, B, and C) represent the 4th, 5th, and 6th RRs, respectively. The black curves denote boundaries of the districts of Beijing.

Hourly WS data of 160 automatic weather stations (AWSs) over Beijing for the time period 2008–2017 are obtained from the Meteorological Information Center, Beijing Meteorological Bureau (MIC/BMB). A quality-control process developed by Yang et al. (2011) is adopted in order to ensure the validity of the high-density AWS data. The stations with over 3% of missing records have been taken out directly, and the questionable records have been checked and corrected by use of the regional climate extreme thresholds (Yang et al., 2013). Finally, 45 observation stations are selected as urban stations, which are relatively evenly located inside the 6th RR, in addition to 6 rural stations outside the 6th RR (Fig. 1).

In this study, the satellite remote-sensing data are also utilized in selection of rural stations. The brightness temperature is obtained from the Moderate Resolution Imaging Spectroradiometer (MODIS) dataset first, and then the temperature isolines surrounding the stations are drawn in order to compare the WS of urban and rural sites following Ren and Ren (2011) and Yang et al. (2013, 2017). Finally, six rural stations are selected. All the selected rural stations have similar natural characteristics with those urban observational sites. In particular, their average altitude are almost the same, and their locations are quite close to the urban areas. The detailed information of the six stations is shown in Table 1.

The USII is defined herein as the WS difference between one or several urban sites and rural areas. The rural WS (WS_r) refers to the mean value of WS of the six reference stations outside the urban areas, while the ur-ban WS (WS_u) denotes the WS of any urban station, or the mean WS of certain types of urban stations. The USII (∆WS_r−u) can be thus derived as:

The significance testing is applied to the difference of mean WS between urban and rural stations, and the difference is considered statistically significant if it is at the 95% (P < 0.05) confidence level.

Figure 2 shows the spatial distributions of annual mean WS in the study area of Beijing. The contour maps with interpolated grid values used in this study are drawn with kriging methods. During the 10-yr period, the ann-ual mean WS of the urban study region is 1.11 m s⁻¹, which is 0.21 m s⁻¹ lower than that of the rural stations (annual mean WS of 6 rural stations is 1.32 m s⁻¹; Table 2 shows the annual mean WS of each rural station). The site-to-site difference between the maximum and minimum annual mean WS values is 1.58 m s⁻¹. The lowest value (0.46 m s⁻¹) is obtained at Beijing Workers’ Stadium (BWS; Fig. 2) inside the 4th RR (annual mean WS inside the 4th RR is 0.92 m s⁻¹), while the highest one (2.04 m s⁻¹) is recorded by Beijing Observatory (BO; Fig. 2) near the 5th RR. As a national station of Beijing, BO is in compliance strictly with the highest standards for observational environment, which requires quite open surroundings. Thus, the faster local airflow caused by the open space (i.e., lower surface roughness) around BO could be the main reason for the higher WS record.

Figure 2.  Spatial distribution of the annual mean wind speed (WS; contour and shading; m s⁻¹) in Beijing during 2008–2017. The black dots and curves are the same as those in Fig. 1. The two red pins indicate locations of the Beijing Workers’ Stadium (BWS) and Beijing Observatory (BO), respectively.

Figure 2 also reveals that the annual mean WSs in the east and south of the study area show generally higher values. Obviously, this large spatial difference of the annual mean WS values between urban and rural areas of Beijing is mainly indicative of the USI effect, with a significant decrease in WS over the built-up areas.

The temporal characteristics of hourly and pentad mean WS during 2008–2017 in urban and rural areas of Beijing are shown in Figs. 3a, b. Blue lines in the figure represent 1.0-m s⁻¹ isolines of mean WS, while red lines represent 2.4-m s⁻¹ isolines of mean WS. It is obvious that the temporal evolution patterns of WS are similar between urban and rural areas, and the WS is a bit stronger in rural areas. The WS experiences a smaller diurnal variation in autumn compared to other seasons. For example, the daily range of hourly mean WS in urban areas (Fig. 3a) is within 0.88 m s⁻¹ in autumn, while it reaches 1.48 m s⁻¹ in spring.

Figure 3.  Hour–pentad profiles of the mean WS (contour and shading; m s⁻¹) for (a) urban and (b) rural areas of Beijing during 2008–2017. The blue and red lines represent 1.0- and 2.4-m s⁻¹ isolines of the mean WS, respectively. Along the x-axis, Hours 0, 4, ... correspond to 0000 BT, 0400 BT, and so on. BT denotes Beijing Time.

Figure 3 also indicates that the WS shift between night and day is large in spring and small in autumn. The high values in both urban and rural areas occur at daytime in spring and winter while the low values lie in the period from midnight to early morning. Besides, considerable differences exist between the urban and rural areas. A significance test is taken to examine the WS difference between the urban and rural stations. For time series of hourly average values and pentad average values, the t values are 41.13 and −3.92 respectively, both of which pass the significance level of 0.001. In the rural areas, the maximum hourly mean WS can reach 3.7 m s⁻¹ (16 h, 23 pentad) during the strong WS stage in spring, approximately 3.0 m s⁻¹ higher than the smallest hourly mean WS of 0.8 m s⁻¹ (4 h, 44 pentad) recorded during the weak WS stage in summer. By comparison, the difference between the highest and lowest hourly mean WS values in urban areas does not exceed 2.5 m s⁻¹. Once again, the area where hourly mean WS exceeds 2.4 m s⁻¹ (red line) at daytime in rural areas is much larger than that in ur-ban areas, while the region where the WS values less than 1.0 m s⁻¹ in rural areas turns out to be smaller than that in urban areas. Obviously, the diurnal variation of WS among different seasons in rural areas is generally larger than those in urban areas.

It is known that WS has been influenced largely by urbanization (i.e., higher buildings around observational sites; e.g., Landsberg, 1956; Yu et al., 2014; Bian et al., 2018). Based on the dense observations, the apparent differences of WS can also been found in this work. Surface roughness is obviously one of the most important factors. In the urban center, because buildings increase both in density and height (Ge et al., 2016), the underlying surfaces have evolved from being heterogeneous to rough, and distribution of the airflow has changed fundamentally (Peng and Hu, 2006). Due to the increased surface roughness, the mean WS in urban areas, especially in the urban center, has been decreased significantly.

Figure 4 demonstrates the spatial pattern of annual average USII inside the urban areas of Beijing during 2008–2017. The maximum USII center (0.86 m s⁻¹) appears at BWS, which coincides with the annual mean high WS value in Fig. 2. The largest USII, 0.5 m s⁻¹ (marked by pink lines) or greater, mostly occur within the 4th RR. In fact, the annual mean USII value inside the 4th RR is 0.52 m s⁻¹. Tian’anmen (TAM; Fig. 4) is the only station that has recorded a low USII value (0.17 m s⁻¹) inside the 4th RR, which is located in the most urbanized center of Beijing City. This phenomenon can be explained by the open surroundings of the site. The most spacious urban square in the world, TAM Square, is next to the site, and several large parks are also located around TAM, such as Zhongshan Park, Beihai Park, Jingshan Park, and so on. Previous studies illustrated that the concentrated trees and grass in parks in urban areas help to weaken the UHI effect (e.g., Yang et al., 2016). Ge et al. (2016) also found that high and dense buildings weaken WS, while dispersed buildings with larger open space help to keep airflow moving. Therefore, the apparent low USII value recorded at TAM is more likely caused by the open space around the observational site.

Figure 4.  Spatial distribution of the annual mean Urban Stilling Island intensity (USII; contour and shading; m s⁻¹) in Beijing during 2008–2017. The pink line represents the 0.5-m s⁻¹ isoline of the mean USII. BWS and TAM represent Beijing Workers’ Stadium and Tian’anmen Square, respectively.

Figure 5 demonstrates the spatial patterns of seasonal average USII in Beijing’s urban and rural areas. As is shown, the area enclosed by the pink isoline of 0.5 m s⁻¹ is larger in Figs. 5a, d than in the other two panels, indicating that the USI effect is stronger in spring and winter. The USII in spring is the strongest as more than half of the sites (21 sites) have the records of over 0.5 m s⁻¹, which can be seen clearly in Fig. 5a. The only low record in the urban center is still at the TAM site, which is only 0.24 m s⁻¹. In contrast, the seasonal mean USII is a little weaker in winter as the 0.5-m s⁻¹ isoline encircles a narrower area (Fig. 5d). Although the low-value area of TAM in the urban center still exists, its USII value is a little larger (0.35 m s⁻¹) than that in spring. Overall, the USII demonstrates a more remarkable spatial difference in spring than in winter. The spatial patterns of seasonal mean USII in summer and autumn are similar. Both of their average values of USII are 0.35, and the 0.5-m s⁻¹ isoline covers a small area in or around the 4th RR. However, there is an additional area enclosed by 0.5 m s⁻¹ isoline around southwest of the 4th RR, as shown in Fig. 5b, indicating that the seasonal mean USII is slightly higher in summer than in autumn.

Figure 5.  Spatial distributions of the seasonal mean USII (contour and shading; m s⁻¹) for (a) spring, (b) summer, (c) autumn, and (d) winter in Beijing during 2008–2017. The pink line represents the 0.5-m s⁻¹ isoline of the mean USII.

Figure 6 displays the diurnal variation of annual and seasonal mean USII in Beijing City during 2008–2017. The variations of both annual and seasonal values are similar. The maximum USII value occurs in the noon while the minimum one comes at the sunset. The diurnal curve of annual mean USII demonstrates that the USII value remains low from evening [1800 Beijing Time (BT)] to the next morning (0700 BT), and it keeps high for only 4–5 h. The rest time of the day can be considered as swift-changing stages, including a surge stage (0700–1100 BT) and a sharp decline stage (1500–1800 BT). In addition, the spatial distributions of USII during periods of daytime (0700–1800 BT) and nighttime (1900–0600 BT) are both analyzed, which are similar to Fig. 4, demonstrating larger values of USII inside the 4th RR.

Figure 6.  Diurnal variations of the annual (black line) and seasonal (color lines) mean Urban Stilling Island intensity (USII) in Beijing during 2008–2017.

Although the diurnal curves are similar in the whole year, the seasonal differences still exist. First, in spite of the similar appearance time of valley values (1800 or 2000 BT), the peak time is almost different in every season, which is 1000, 1200, 1100, and 1200 BT in spring, summer, autumn, and winter, respectively (Fig. 6 and Table 3). It is evident that the USI effect remains weak from evening to next early morning (1900 to 0800 BT), and the largest seasonal hourly mean value of USII is no more than 0.45 m s⁻¹ during this time. Second, the seasonal mean USII varies in different seasons. Table 3 and Figure 6 show that the strongest USII occurs in spring, and the second strongest USII occurs in winter, while summer and autumn see relatively weak USII values. Third, the diurnal ranges of the hourly average USII values witness an inter-seasonal difference. The smallest USII diurnal range appears in winter while autumn sees a greater diurnal variation.

Month–year and hour–year profiles of the mean USII in urban areas of Beijing are demonstrated in Fig. 7. In Fig. 7a, it is clear that the whole period from 2008 to 2017 can be considered in two separate intervals: one with larger USII values (2008 to 2012) and the other with smaller USII values (2013 to 2017). This change may be attributed to the change of urbanization level around the rural observation sites leading to the weak background near-surface WS, but the details need to be further investigated. The low-layer circulation has weakened since 2013, which decreased the near-surface WS (Zhou et al., 2017). In detail, the annual value of USII peaks in 2010, while it reaches the bottom in 2016. For the whole 10-yr period, the interseasonal variability seems to be large as well. The monthly mean USII turns out to be larger in January–June and October to –December, especially in spring (March–May). On the other hand, the monthly mean USII keeps low in summer (July–September). Figure 7b demonstrates that the hourly average USII changes year by year in urban areas. The diurnal changes are obvious in the whole study period. The hourly average USII maintains high during the daytime and drops at the nighttime every year. However, the strength and duration of the high- and low-value USIIs are different each year. The high-value USIIs in 2010 and 2009 are greater, while the low-value USIIs are more remarkable in 2016 and 2017. In addition, at any time of a day, the hourly mean value of USII decreases year by year after 2010.

Figure 7.  (a) Month–year and (b) hour–year profiles of the mean USII (m s⁻¹) in the urban areas of Beijing during 2008–2017. Hour numbers on the x-axis correspond to Beijing Time (BT).

Figure 8 demonstrates the hour–pentad profiles of hourly average USII for the urban region of Beijing during the period of 2008–2017. The distribution of hourly mean USII is considerably different among the whole day and whole year. It can reach as high as 0.5 m s⁻¹ during daytime (1000 to 1400 BT) in the strong USII stage of spring and winter, approximately 0.5 m s⁻¹ larger than that in the evening in summer and autumn. In particular, the hourly mean USII during the daytime (0800 to 1600 BT) is obviously higher in spring. The maximum USII value reaches 0.7 m s⁻¹ around 1000 BT of the 17th pentad (middle spring), and the relatively strong USII appears at daytime during the late autumn and early winter.

Figure 8.  Hour–pentad profile of the mean USII for the urban areas of Beijing during 2008–2017. The blue and red lines represent 0- and 0.4-m s⁻¹ isolines, respectively. Along the x-axis, Hours 0, 4, ... correspond to 0000 BT, 0400 BT, and so on. BT denotes Beijing Time.

It is interesting to notify that there exists a negative USII stage during the evening time of late summer and early autumn, with the greatest negative values ranging from −0.2 to −0.3 m s⁻¹ during early autumn (Fig. 8). The negative USII also appears in late afternoon and early evening in other seasons, particularly in early summer and winter. This phenomenon may be related to the obviously weaker background WS field and the increased UHI intensity during the specific transitional time point from summer to autumn and during late afternoon and late evening (Yang et al., 2013); the exact reasons need to be investigated in future. The increased UHI intensity may also lead to stronger local UHI circulation and the resulting stronger near-surface WS within the urban areas.

The difference of near-surface WS between urban and rural areas has been investigated continuously. Early studies on this in Beijing City are based on comparison of the observation data obtained from two meteorologi-cal towers or two weather stations: one is within the city and the other is outside the city. The early studies on other regions have generally shown that WSs over cities tend to be lower (e.g., Landsberg, 1956; Frederick, 1964; Graham, 1968), which is mainly attributed to the greater aerodynamic surface roughness in urban areas, compared with adjacent rural regions. However, other studies indicate that UHI can make it possible to change surface WS in urban areas, which enhances the urban–rural temperature gradient and increases WS_u. For example, Chandler (1965) demonstrated the spatial difference of WS in and around London and found that the WS was greater over London than nearby rural sites.

In the last decade, especially in China, the latest observational evidence showed that the WS in cities was decreasing more significantly than that in rural areas, and in most cases, it was lower in cities than in rural regions (Liu et al., 2012; Yu et al., 2014; Wu and Wu, 2016; Bian et al., 2018). According to an analysis of the WS changes in surface and the upper air levels, about 1/3 of the decline in annual mean surface WS in the Chinese mainland on the whole is attributed to the urbanization and the change of observation environment around the current stations of the national meteorological network (Zhang et al., 2009). The simulations of flow structures in city canyon (Zhang et al., 2001) have also demonstrated that the surface WS decline would be profoundly affected by the height and density of buildings. If the urban buildings are kept away from each other with sufficient open space in between, or the buildings are not tall enough, the near-surface wind would follow the direction of prevailing wind. Conversely, if the height and density of urban buildings are increased, the near-surface wind direction would be rather different from the prevailing wind, and the mean WS would be weakened. A survey shows that the building height inside the 5th RR of Beijing generally exceeds 15 m (Mu et al., 2019; except for the TAM area on the City’s center). Based on the higher-resolution observational data used in this study, it is found that the WS in the urban areas of Beijing, especially inside the 5th RR, is obviously lower than that in the nearby rural areas. Thus, it could be inferred that, during the last decade in Beijing, changes in the urban environment and surface roughness accompanying the rapid urbanization have had a great impact on the WS_u.

By comparing Fig. 3 with Fig. 8, it is easy to find that the diurnal variations of annual mean WS and USII are similar. Both of the curves exhibit a twin-peak pattern. The maximum hourly mean value occurs at daytime in winter and spring, and the second largest value obviously appears in early morning in spring. The daily curve of hourly mean WS indicates that the WS changes remarkably during the daytime. From 0500 to 1500 BT, the hourly mean value of WS grows from 0.72 to 1.76 m s⁻¹, up more than doubled. This increase is mainly attributed to solar radiation. The strengthened solar radiation after sunrise encourages the turbulence development and the downward transmission of upper air momentum. Consequently, the WS is gradually increasing after sunrise until afternoon. On the other hand, the turbulence at night is weak and the air keeps calm, which contributes to lower near-surface WS. The diurnal curve of USII presents a similar pattern, which can be partly explained by the UHI effect. In the urban area, nighttime UHI enhancement is common in inland temperate cities like Beijing and Shijiazhuang (Yang et al., 2013; Bian et al., 2018), which makes the atmospheric stratification less stable at night in urban than in rural areas. In addition, the turbulence in the boundary layer tends to be enhanced at night due to UHI effects, and it will bring the momentum from the top downward. Thus, although the WS is weaker over the urban area than in the rural area (i.e., USII is larger) during daytime, there is little difference (i.e., USII is smaller) at night due to the UHI.

The seasonal feature of USII is also discussed here. Previous studies found that the lowest seasonal mean UHII over Beijing occurs in spring (Yang et al., 2013). The present study demonstrates that the strongest USII also occurs in spring, followed by winter (Fig. 8). Why are stronger USII concentrated in spring and winter? Previous studies indicate that urbanization affects the surface WS in two ways. One is UHI, which enhances the urban–rural temperature gradient and increases WS_u (i.e., reduces USII). The other is surface roughness, the increase of which acts to decrease WS_u (Klink, 1999). The surface roughness effect would dominate (i.e., with enhanced USII) when the large-scale wind is stronger, which is typical in spring (Born and Johnson, 1977; Lee, 1979). Thus, the roughness has a greater impact on WS in windy season like spring and winter in Beiing. Oppositely, if the large-scale wind is weaker, such as in autumn and summer in Beijing, the effect of UHI on WS would be grea-ter (i.e., a smaller USII) while the surface roughness influence is reduced (Li et al., 1982; Zhou and Yu, 1988). However, as a kind of local airflow, the UHI-induced circulation is typically weak, hence the set-off to the USII is usually small. All in all, the USII in urban areas is decreased partly by the UHI circulation. Consequently, the USII is not significant in autumn and summer when UHI dominates. Another reason for the weak wind in summer is probably attributed to varied summer precipitation, which mitigates the UHI intensity and UHI circulation. During the period of 2008–2017, for example, the accumulated precipitation in summer on average is 387 mm, almost occupying 70% of the annual total precipitation (579 mm). The weakened UHI circulation during summer may have contributed to the weak urban wind field to some extent.

The results presented in this study have clearly indicated the significant urbanization effect on the weakening surface WS in the urban areas of Beijing. The sites selected for study includes the national observation stations such as Beijing Observatory (BO), Haidian Station, and Chaoyang Station, which are all located in built-up areas. We gather that, the urbanization (Zhang et al., 2009; Liu et al., 2012), the increased roughness of land areas (Vautard et al., 2010; Bichet et al., 2012), and the decrease in thermal contrast between tropical and polar regions in the context of global warming (Zhang and Ren 2003; McVicar et al., 2012) may all influence the station observational environment. As a result, the near-surface USI facts observed from most of the stations in the Chinese mainland will also be affected as the previous studies claimed. This is important because the large-scale near-surface WS might have not declined so much as previously believed and the utilization of wind power on land will not be negatively affected by the long-term change in climate. It is very likely that the observed decline in surface WS on land has been mainly caused by some lo-cal anthropogenic effects such as urbanization.

Furthermore, the near-surface WS is also a very important factor in formation of haze or severe pollutant weather. Calm condition plays a positive role in generating and maintaining the haze weather, especially when the WS value keeps at a low level (Mao et al., 2018). Miao et al. (2016) studied the diurnal variation of PM_2.5 mass concentration and found that the diurnal curve of PM_2.5 in Beijing City has two valleys (0700 and 1600 BT), which coincide with the USII diurnal pattern as reported in this paper. In addition, PM_2.5 mass concentration keeps high from 0900 to 1600 BT (Miao et al., 2016), which is also consistent the diurnal variation pattern of USII, especially in spring. It is likely that when the pollutants emissions increase during daytime, the relatively large USII is conducive to a higher concentration of aerosols, exacerbating the urban air pollution over this period of time.

Therefore, the results obtained from this work would be important not only for understanding the WS variation between urban and rural areas and associated physical mechanism, but also for clarifying possible causes of the widely observed near-surface WS decline on land, as well as for analyzing air pollution processes and monitoring haze weather in the megacities like Beijing.

This paper examines the climatological features of the surface wind speed (WS) and the “Urban Stilling Island (USI)” phenomena in Beijing City based on an hourly station observation dataset with high quality derived from a high-density AWS network. The “Urban Stilling Island (USI)” is defined and quantified to depict the WS differences between rural and urban areas of Beijing City. The urban (rural) sites are represented by 45 (6) stations within (outside) the 6th Ring Road (RR). Main conclusions of this study are summarized as follows.

(1) The annual mean values of WS in the central ur-ban areas or nearby areas of Beijing City tend to be small. The largest monthly average WS values appear in spring and winter. The annual and seasonal average WS values in built-up urban areas are remarkably smaller than in rural areas.

(2) The largest annual mean USI intensity (USII) mostly occurs inside the 4th Ring Road (RR) of Beijing City. The seasonal mean USII is larger in spring and winter than in autumn and summer. This is because the large-scale background wind is stronger (weaker) in winter and spring (summer and autumn) in Beijing, the surface roughness effect [the Urban Heat Island effect (UHI)] dominates, leading to stronger (weaker) USII.

(3) Diurnal variations of annual and seasonal mean USII are similar. The annual mean USII keeps low from 1800 BT to the next morning (0700 BT). The diurnal variations, including the peak occurrence time, mean value, and general diurnal pattern, are different in different seasons. Hourly mean USII is higher during the daytime (0800 to 1800 BT) in spring. The maximum USII is 0.7 m s⁻¹ around 1000 BT of the 17th pentad (mid-spring). It is inferred that solar radiation plays a crucial role.

(4) Long-term variation of hourly USII data analysis shows that there are obvious two phases of change in USII during 2008–2017: the period of 2008–2012 with larger USII values and the period of 2013–2017 with smaller USII values. The variation may be attributable to the change of urbanization level around the rural observation sites.