J. Meteor. Res.  2016, Vol. 30 Issue (6): 961-982   PDF    
http://dx.doi.org/10.1007/s13351-016-6012-3
The Chinese Meteorological Society
0

Article Information

JIN Lili, LI Zhenjie, HE Qing, MIAO Qilong, ZHANG Huqiang, YANG Xinghua . 2016.
Observation and Simulation of Near-Surface Wind and Its Variation with Topography in Urumqi, West China. 2016.
J. Meteor. Res., 30(6): 961-982
http://dx.doi.org/10.1007/s13351-016-6012-3

Article History

Received February 25, 2016
in final form July 7, 2016
Observation and Simulation of Near-Surface Wind and Its Variation with Topography in Urumqi, West China
JIN Lili(金莉莉)1,2, LI Zhenjie(李振杰)3, HE Qing(何清)1,2, MIAO Qilong(缪启龙)4, ZHANG Huqiang(张虎强)5, YANG Xinghua(杨兴华)1,2     
1. (Institute of Desert Meteorology, China Meteorological Administration, Urumqi 830002, China);
2. (Desert Atmosphere and Environment Observation Station at Taklimakan, Tazhong 841000, China);
3. (Lincang Meteorological Bureau of Yunnan Province, Lincang 677099, China);
4. (College of Applied Meteorology, Nanjing University of Information Science & Technology, Nanjing 210044, China);
5. (Bureau of Meteorology, Melbourne, VIC 3001, Australia);
ABSTRACT: Near-surface wind measurements obtained with five 100-m meteorology towers, 39 regional automatic stations, and simulations by the Weather Research and Forecasting (WRF) model were used to investigate the spatial structure of topography-driven flows in the complex urban terrain of Urumqi, China. The results showed that the wind directions were mainly northerly and southerly within the reach of 100 m above ground in the southern suburbs, urban area, and northern suburbs, which were consistent with the form of the Urumqi gorge. Strong winds were observed in southern suburbs, whereas the winds in the urban, northern suburbs, and northern rural areas were weak. Static wind occurred more frequently in the urban and northern rural areas than in the southern suburbs. In the southern suburbs, wind speed was relatively high throughout the year and did not show significant seasonal variations. The average annual wind speed -1 in this region varied among 1.9-5.5, 1.1-3.6, 1.2-4.3, 1.2-4.3, and 1.1-3.5 m s within the reach of 100 m above ground at Yannanlijiao, Shuitashan, Liyushan, Hongguangshan, and Midong, respectively. The flow characteristics comprised more airflows around the mountain, where the convergence and divergence were dominated by the terrain in eastern and southwestern Urumqi. Further analysis showed that there was a significant mountain-valley wind in spring, summer, and autumn, which occurred more frequently in spring and summer for 10-11 h in urban and northern suburbs. During daytime, there was a northerly valley wind, whereas at night there was a southerly mountain wind. The conversion time from the mountain wind to the valley wind was during 0800-1000 LST (Local Standard Time), while the conversion from the valley wind to the mountain wind was during 1900-2100 LST. The influence of the mountain-valley wind in Urumqi City was most obvious at 850 hPa, according to the WRF model.
Key words: mountain-valley wind     near-surface wind     simulation     topography     Urumqi    
1Introduction

Measured wind comprises the large-scale background wind field (system wind) and the local wind, whereas the local prevailing wind direction is related mainly to geographical position and terrain (Zhang et al., 2012). Thus, measured wind is influenced by the large-scale background of the wind field, terrain, heat, and other factors. The annual variation, seasonal variation, and vertical profile of the wind field are determined mainly by the wind field of associated largescale system, local terrain, background wind field, and their dynamic and thermodynamic interactions with the surface (Wang and Jiang, 2006).

There is a special category of local or tertiary winds directly related to the topography, e.g., sea-land breezes, mountain and valley winds, airflow over mountains, and urban heat island circulation. These local winds have been observed in, and simulated for, numerous cities, mountains, and coastal areas worldwide. For example, Miao et al. (2013) simulated urban flow in Beijing by coupling a computational fluid dynamics model with the Weather Research and Forecasting (WRF) model. Seto and Clements (2011) observed the evolution of a fire whirl during a valley wind-sea breeze reversal. Laiti et al. (2013) used airborne and surface measurements to study the atmospheric boundary layer structures associated with a lake-breeze-valley-wind in the Alps. De Wekker et al. (2012) used airborne Doppler lidar measurements to study the valley flows in complex coastal terrain. Li et al. (2005) studied the mechanism of urban boundary layer structure formation during winter in Beijing by using the MM5 model, revealing that the topographic effect was the primary forcing of the main airflow characteristics. Li et al. (2008) simulated the effect of terrain on the wind field over the Qingdao Olympic Sailing Regatta waters by using the WRF model, demonstrating that the model could compensate for the lack of accurate observations by obtaining better simulations of the backwater effect as well as the function of the detouring flow and narrow pipe effect due to the piedmont and a blocking effect. Miao and Chen (2008) and Miao et al. (2009) simulated the impact of urban heat islands, mountain-valley circulation, and urbanization on meteorological conditions, indicating that the WRF model can simulate the intensity characteristics and spatial distribution of the urban heat island, as well as the diurnal variation of the wind field, the horizontal convection vortex, the convection bubble, and the low-level jet of the boundary layer at night in Beijing. Li et al. (2008) used the WRF model to study the impact of terrain on the wind field over the Olympic Sailing Competition areas of Beijing, while Li et al. (2010) determined the impacts of boundary layer flows and terrain on afternoon sea-breezes at the Qingdao Olympic Sailing venue. Acharya et al. (2014) used the WRF model to study early monsoon local flow characteristics over the Hetauda valley and their implications. And lastly, Schmidli and Rotunno (2012) performed numerical experiments to study the influence of valley surroundings on the evolution of valley winds, having previously (Schmidli et al., 2011) employed the WRF model to study a daytime valley-wind system.

Many studies have shown that mountain-valley winds are a key component of the atmospheric boundary layer over complex terrain. Mountain-valley winds are caused by altitudinal differences in solar radiation, and thus temperature, which at higher terrain results in a heat source during the day and a heat sink at night. According to pressure gradient force, the wind direction will change with the temperature variations during the day and at night. The wind circulation is upslope during the day and downslope at night (Jü et al., 2003). In mountain areas, the diurnal variation of mountain-valley winds has an important effect on their climatology, which is highly consistent with the long-term climate. Mountain-valley winds occur mainly on sunny days when the background wind is weak (Sturman et al., 1999; Whiteman, 2000; Martínez et al., 2008). The wind speed, distance, and duration, as well as the start time of mountain-valley winds, differ among areas according to various factors, including topography, land cover, soil humidity, solar radiation, partial occlusion, and the surface energy balance (Zängl, 2004). Many of these factors demonstrate clear seasonal variations; indeed, the amplitude of the horizontal pressure gradient has been shown to drive seasonal variations in valley winds (Cogliati and Mazzeo, 2006).

Mountain-valley winds, mountain air diffusion, and the urban ecological environment have important effects on cities (Qin, 2001; Jü et al., 2003; Zhang et al., 2003; Wu and Zhang, 2010). At present, the forecasting of air pollution in urban areas is focusing closely on mountain-valley wind circulation, urban heat island circulation, and local circulation caused by disturbances in fluctuating terrain. Many studies have shown that local circulation is important for mesoscale circulation because of the dynamics of the inhomogeneous underlying surface and thermodynamic forcing, which control diurnal variation and the complex vertical distribution of pollution (Shou, 1993).

The fluxes induced by valley winds can be much greater than near-surface turbulent fluxes (Weigel et al., 2007). Mountain-valley winds can influence the characteristics of local weather and climate, such as surface temperature, wind speed, cloudiness, and precipitation (Egger et al., 2000). Pollutant levels in urban areas can be exacerbated by mountain-valley winds through repeated transport and accumulation. For example, Hu et al. (2005) showed that, in summer, the unique topography of the Beijing area results in southern anabatic wind during the day and northern katabatic wind at night, while this phenomenon disappears in winter due to the effect of strong synoptic systems. Laiti et al. (2013) employed airborne measurements to study atmospheric boundary layer thermal structures in a mountain-valley area. Liu (2014) suggested that the formation of mountain-valley wind circulation is influenced by the topography in the northwest valley of the Beijing area, where changes in valley winds occur at 0800 and 2000 BT (Beijing Time). The wind is affected mainly by a combination of the mountains northwest of the Beijing region and the plains, where there are hundreds of kilometers of plain area to the southeast. Bossert (1997) indicated that external winds from less polluted regions appear to invade the Mexico Basin and are effective at displacing the distinct and highly polluted basin air mass. Chen et al. (2009) showed that mountain-valley wind circulation affects the diffusion of urban pollutants in Beijing City; they demonstrated a “chimney effect” of the mountain-valley wind circulation, in which emissions around the mountain are transported to the atmospheric boundary layer in the daytime above the free atmosphere.

Urumqi (42°4503200-44°0800000N, 86°3703300-88°5802400E) is the largest city in the western interior of China, with a total area of 1.42×104 km2 and a semi-arid climate. The urban area (365.88 km2) lies in the Urumqi valley, which is at the northern foot of the Tianshan Mountains and at the southern edge of Jungger Basin (Fig. 1). The city’s surrounding mountains and hills range from 1397 to 5445 m above sea level (a.s.l.) on three sides (Boda Mountain, Kalatage Mountain, and Dongshan Mountain to the east; Karaza Mountain and Xishan Mountain to the west; and Tianger Mountain to the south), while the av erage altitude of the urban area is 800 m a.s.l. (Li et al., 2012). In general, the urban terrain (on an alluvial plain to the north of the central Tianshan Mountains) comprises a bell-mouthed area, where the larger end faces the northern Jungger Basin and the smaller southeastern end connects with the central Tianshan Mountains pass (Fig. 1) (Li et al., 2012). The elevation of the city’s terrain decreases from the southeast to the northwest, where the total distance from north to south is about 154 km, and that east to west about 190 km (Wang et al., 2010). Urumqi is the largest and most industrialized modern city in Xinjiang, with a population of approximately three million. Previous studies have investigated the wind and its influence on air pollution in Urumqi City, where southeasterly gales frequently occur in spring and autumn (Li et al., 2012; Li, 2013). In winter, these gales are usually lifted up by the cold pool present in the up-valley region; they are unable to penetrate the deep and stable inversion layer and thus cannot flush away the cold air pool over Urumqi (Zhang et al., 1986). Strong near-surface winds are often observed in the southern suburbs, while the urban area remains almost windless. In a recent study, Li et al. (2015) observed the critical role of the sandwich foehn in the formation of heavy air pollution events in Urumqi. However, information on the spatial structure of topography-driven flow in the Urumqi area is limited. In this study, based on meteorological tower and automatic weather station observations and WRF model simulations, we investigated the local wind field in Urumqi, including the effects of the mountain-valley wind, its characteristics, and range of impacts. The aim in conducting this study was to provide a valuable reference for planning short-and long-term pollution control strategies for the city.

Figure 1 (a) The three nested WRF model domains. (b) Actual terrain in the innermost domain. (c) As in (b) but the terrain height around the urumqi city is set as sea level. Locations of the (d) five 100-m tower sites and (e) automatic weather stations in Urumqi City.
2Methods 2.1Climate setting

In the urban area of Urumqi, the mean annual temperature is 7.3℃, precipitation is 298.6 mm, evaporation is 2004.9 mm, and the mean annual wind speed is 2.3 m s-1, according to the records of Urumqi me-teorological station between 1981 and 2010. July is the hottest month (mean temperature of 23.8℃) and January is the coldest (mean temperature of -12.1℃), and the maximum (minimum) temperature is 40.6 (-30)℃.

2.2Site description

Five 100-m-high towers are deployed in the Urumqi urban area from south to north, e.g., Yannanlijiao (YNLJ; located in the southern suburbs, surrounded by buildings, a reservoir, and trees), Shuitashan (STS; located in the urban area, surrounded by 5-60-m high buildings), Liyushan (LYS; located in the urban area, surrounded by 5-80-m high buildings), Hongguangshan (HGS; located in the northern rural area, with 5-60-m high buildings to the north), and Midong (MD; located in the northern suburbs, surrounded by the Gobi and a farmland transitional zone) (Fig. 1). The spatial coverage of the five towers from north to south (YNLJ to MD) is about 37.6 km, and from east to west about 6.2 km; the strike direction of every tower is in the north-south direction. The tower foundations of YNLJ, STS, LYS, HGS, and MD are at altitudes of 1009, 890, 821, 717, and 568 m, respectively.

The 39 regional automatic monitoring sites are distributed mainly in the urban area and piedmont of Urumqi City (Fig. 1). Topographical studies based on the wind can be conducted by using regional automatic weather station observations, but there are fewer observation stations in the surrounding mountain area and in Dabancheng, which limited our study in terms of the topographic effects (such as the change in wind speed close to the eastern Tianshan Mountains in the main city).

2.3Data

The five 100-m-high meteorology towers were divided into 10 layers, i.e., 10, 13, 17, 22, 28, 36, 46, 60, 77, and 100 m. Observations of 10-min averaged wind speed and wind direction were made from June 2013 to May 2014 by using a ZQZ-TF type wind direction and wind speed sensor (Jiangsu Radio Science Research Institute Co. Ltd, China). Data were acquired suing a ZQZ-CII-SE data acquisition system (Jiangsu Radio Science Research Institute Co. Ltd, China). A series of quality control procedures were used on the wind speed data, such as logical extreme checks, static checks, time consistency checks, and similarity checks. Likewise, the logical extreme and static check methods were used on the wind direction data (Jin et al., 2016). Statistics showed that the proportions of eligible wind direction and wind speed data were 96.2% and 95.6%, respectively. The unqualified data were excluded.

The frequencies of wind direction (16 wind directions and static wind) were calculated at the annual, seasonal, day and night, and diurnal timescales at each tower. Note that the sample number could be different for different timescales. For example, on the annual timescale (Fig. 4), the number of wind direction samples was 431530 at MD, but on the diurnal timescale in spring (Fig. 7) the number was only 61449. Therefore, different results could occur on different timescales.

Figure 2 Comparison of averaged diurnal variations of (a) 2-m temperature (℃), (b) 10-m wind speed (m s-1), and (c) 2-m specific humidity (g kg-1) between model simulations (black solid dots) and observations (white hollow dots) from 0000 LST (Local Standard Time) 31 October to 2 November 2013, for all stations.
Figure 3 Annual wind directions and their distributions with height at the five 100-m towers in Urumqi City.
Figure 4 Seasonal wind directions and their distributions with height at the five 100-m-high towers in Urumqi City. (a) Spring, (b) Summer, (c) Autumn, and (d) Winter.
Figure 5 Wind directions (a) during the day and (b) at night, and their distributions with height at the five 100-m-high towers in Urumqi City.
Figure 6 Diurnal variations of prevailing winds and their distributions with height at the five 100-m-high towers in Urumqi City (circles represent static wind; upward-pointing arrows represent northerly wind; downward-pointing arrows represent southerly wind). (a, b, c, d) YNLJ, (e, f, g, h) STS, (i, j, k, l) LYS, (m, n, o, p) HGS, and (q, r, s, t) MD; (a, e, i, m, q) spring, (b, f, j, n, r) summer, (c, g, k, o, s) autumn, and (d, h, l, p, t) winter.
Figure 7 Monthly average wind speed and its distributions with height at the five 100-m-high towers in Urumqi City.

The seasonal changes in the surface flow field and mountain-valley wind were analyzed for Urumqi City using the hourly means of wind speed and wind direction data obtained from 39 regional automatic stations from June 2013 to March 2014. The same quality control procedures as for the five meteorology towers were applied. The erroneous data were excluded.

In the analysis, vector wind (u, v) was calculated according to Eqs. (1) and (2):

(1)
(2)

where u (m s-1) and v (m s-1) are the vector wind, U (m s-1) is the wind speed observation, and W (0°-360°) is the wind direction observation. A spatial distribution characteristic of the vector wind was performed, interpolating data for the wind field by using Kriging interpolation method. An evenly distributed wind field was obtained for the entire innermost WRF domain and Urumqi urban area with the apply of Surfer 11.0 and ArcGIS 10.0 software. Seasonal and clear day at 0000 and 1200 LST (Local Standard Time) spatial distribution features of the wind field were established by using the vector wind.

2.4Model

Owing to the lack of observations, which prevented us from properly capturing the impacts of terrain on near-surface wind distributions based on observations alone, we also employed high-resolution WRF model numerical simulations to reveal such influences.

Specifically, version 3.6.1 of the WRF model was used, and NCEP-NCAR data (horizontal resolution: 1°× 1°) were used to generate the initial and boundary field values. In Urumqi, the region defined by 42.5°-45.0°N, 86.5°-89.1°E was the main research area. The model simulation area focused on 43.78°N, 87.65°E. The discretization of the model area used three layers of nesting (Fig. 1). The resolutions of the static terrain data were 10 m, 2 m, and 30 s, respectively. The corresponding nested grid horizontal resolutions were 30, 10, and 3.3 km, respectively. The grid numbers were 120 × 120 for d01, 106 × 106 for d02, and 94 × 94 for d03. All were configured with 28 vertical layers. The key physical parameterizations are shown in Table 1. Time integration employed a three-order Runge-Kutta scheme. The model operation time was 48 h and the integration period was 1800 UTC 30 October to 1 November 2013. The time step was 180 s, the background field was updated every 6 h, and the time output was calculated every 3 h.

Table 1 Model physical parameterizations for all domains

Figure 2 shows the WRF-simulated 2-m temperature, 10-m wind speed, and 2-m specific humidity, compared with observations, respectively. The observed range of 2-m temperature diurnal variations was underestimated by the model, although both shared the same patterns. Despite the fact the model could, by and large, capture the diurnal variations of surface wind and specific humidity, it systematically overestimated the surface wind speed and underestimated surface specific humidity.

In order to analyze the influence of terrain on the wind field in the boundary layer, we carried out a terrain sensitivity experiment, in which we compared the results with actual terrain and no terrain. The original terrain height within the innermost domain (d03) in Urumqi City was changed to sea level.

3Results 3.1Observed vertical structure of near-surface wind 3.1.1Wind direction 3.1.1.1Annual wind direction

The annual prevailing winds and their characteristics in the surface layer of the Urumqi urban area are shown in Fig. 3. In general, northerly and southerly winds prevailed throughout the whole year at YNLJ, STS, LYS, and HGS, whereas easterly winds prevailed below 36 m and southeasterly and northwesterly winds prevailed above 46 m at MD. In other words, the largest wind frequency was southerly or northerly in Urumqi’s southern suburbs, urban area, and northern suburbs. Therefore, the bell-shaped valley terrain distribution plays a significant role in the large differences in observed wind directions among the Urumqi urban area and its southern and northern suburbs.

The annual prevailing winds were southerly and northerly in the surface layer at YNLJ. In particular, the prevailing winds were southerly at 17-36 m in the surface layer at a frequency of 17.9%-19%; whereas, at other heights in the surface layer, the prevailing winds were northerly with a frequency of 15.2%-16.9%.

The annual prevailing winds at STS and LYS were complicated by the surrounding terrain and buildings. At STS, the prevailing winds were northerly at a height of 10-22 m in the surface layer with a frequency of 15.2%-17.8%, southerly at a height of 28 m in the surface layer with a frequency of 16.6%, and northerly and northwesterly at a height of 36-100 m in the surface layer with frequencies of 15.2%-16.1% and 15.2%-16.4%, respectively. By contrast, at LYS, the frequency of northerly winds was 11.4%-13%, the frequency of north-northwesterly (NNW) winds was 11.3%-11.9%, and the frequency of west-northwesterly (WNW) winds was 26.8%.

The annual prevailing winds were complex at HGS. The prevailing winds were south-southeasterly (SSE) below a height of 28 m in the surface layer with a frequency of 15.1%-19%, whereas they were NNW above a height of 28 m in the surface layer with a frequency of 13.1%-14.1%. Easterly and westerly winds were rarely found to occur in this region. Such a result is largely related to the influences of local terrain, which can extend to the north of HGS (with a relative height of about 27 m). In contrast, to the south, east, and west of HGS, the terrain is high overall, and forms a blocking effect.

The winds at MD differed from those at the other sites. The prevailing winds were easterly below a height of 46 m in the surface layer with a frequency of 11.8%-22.1%, whereas they were southerly at a height above 46 m in the surface layer with a frequency of 10.3%-10.8%. Such winds are related mainly to the geographical position of MD, i.e., its situation to the west side of large tracts of farmland and low-lying areas on the east side of the undulating Gobi desert, and on the south side of the village.

3.1.1.2Seasonal wind direction

We conducted a seasonal wind direction analysis for the five towers (Fig. 4). The maximum wind frequency in winter is shown in Table 2.

Table 2 The most frequent wind directions in winter at different height levels of the five 100-m-high towers in Urumqi City

Horizontally, the prevailing winds were northerly and NNW in the surface layer of the Urumqi urban area and in the southern suburbs during summer, and easterly and westerly in the northern suburbs. During autumn, the southerly wind increased in the urban area and southern suburbs, where southerly, SSE, and south-southwesterly (SSW) winds were most frequent. During winter, static wind appeared in the urban area and northern suburbs, while the wind direction was northerly in the urban area and southerly in the southern suburbs. Thus, atmospheric pollutants readily accumulated. During spring, there was an increase in the frequency of northerly wind in the urban area.

Comparison of the longitudinal cross-section results for the region revealed the following features: (1) The prevailing winds were largely easterly below 28 m in the surface layer during all seasons at MD, particularly in autumn. By contrast, the prevailing winds above that layer exhibited seasonal variations, being mainly westerly during summer, SSW in autumn, and static in winter. (2) At YNLJ, the prevailing winds were northerly in the near-surface layer during summer, but turned to a dominant southerly at low levels in the surface layer. During autumn, the northerly wind became less prevalent at low levels in the surface layer, whereas the dominance of southerly wind enhanced, and continued to do so until winter. During winter, the prevailing winds were southerly in the surface layer, or there was a dominant northerly wind. During spring, the southerly wind became less prevalent at low levels in the surface layer, and vice versa for the northerly wind. (3) The prevailing wind was SSE below 17 m in the surface layer during summer, autumn, and spring at HGS, with the highest frequency in autumn. The prevailing winds were NNW and southerly above 28 m in the surface layer during summer, autumn, and spring. In winter, they were NNW in the surface layer, where a static wind was most frequent. (4) The prevailing winds were complex at STS and LSY, owing to the effects of buildings with various heights. The prevailing wind was NNW in the surface layer during summer at LYS, with southerly and SSW winds in autumn and northeasterly and northerly winds in winter. The prevailing winds were southerly below 22 m in the surface layer during spring and summer, but northerly and NNW above 22 m in the surface layer. The prevailing winds were northerly and NNW in the surface layer during spring and summer at STS, southerly and SSW during autumn, and NNE during winter.

Our analysis shows that a northerly wind is the most frequent in the surface layer during summer in Urumqi City. An easterly wind occurs during all seasons in northern rural areas. A static wind is the most frequent during winter in the northern suburbs and rural areas, with a southerly in spring, along with complex patterns. Similarly, it demonstrates that the wind characteristics of the Urumqi valley are mainly composed of a frequent mountain breeze during autumn and a valley breeze in spring and summer (being most frequent in the latter of these two seasons).

3.1.1.3Day-and night-time variations in wind direction

To illustrate the distributional characteristics of the vertical structure of near-surface winds, we analyzed the variations in wind direction during the day and at night at the five towers (Fig. 5). In general, the prevailing wind direction in the urban and northern suburbs at night possessed a similar frequency to that in the day, but the wind directions were opposite. During the day, the wind direction was mainly northerly, whereas it was mainly southerly at night.

Figure 5 shows that the prevailing wind at YNLJ during the day was southerly below 77 m and SSE above 77 m, with frequencies of 17.8%-30.2% and 11.3%-12.8%, respectively. At night, the prevailing wind was northerly in the 100-m surface layer, with a frequency of 24.1%-26%. The prevailing wind was northerly in the day and NNW in the 100-m surface layer at STS, with frequencies of 23%-25.6% and 22.9%-25.5%, respectively. At night, the northerly wind frequency decreased, whereas the southerly and SSW, wind frequencies increased (21.9%-26.8% and 15.1%-23.4%, respectively). The prevailing wind at LYS in the day was mainly northerly in the 100-m surface layer, with a frequency of 17.4%-20%. At night, the northerly wind frequency decreased and the southerly wind frequency increased; the prevailing wind was mainly SSW, with a frequency of 15.4%-18.9%. The prevailing wind was NNW in the day in the 100-m surface layer at HGS, with a frequency of 19.3%-25.2%. At night, the northerly wind frequency decreased and the prevailing wind was SSE below 22 m but southerly above 22 m, with frequencies of 25.1%-31.8% and 16.9-22.2%, respectively. The prevailing wind in the day at MD was northerly below 28 m and westerly above 28 m, with frequencies of 11.7%-15% and 11.6%-12.1%, respectively. At night, the northerly and easterly wind frequencies decreased and the winds were mainly easterly in the lower 100-m surface layer, with a frequency of 29.3%-37%, and SSE in the upper 100-m surface layer. As a result, the frequencies in the urban and northern suburb areas were similar during the day and at night, but the wind directions were opposite. The prevailing wind was generally northerly in the day in the 100-m surface layer at Urumqi but southerly at night, with a frequency of 24%-55%.

3.1.1.4Diurnal variation of prevailing winds

The mountain-valley wind of Urumqi is illustrated in Figs. 4 and 5. The mountain breeze was found to occur in autumn, and the valley wind more often in summer and spring. The mountain breeze was southerly, and the valley wind northerly.

Figure 6 shows that, at YNLJ, the valley wind was more likely to occur during 1000-1800, 0800-1900, and 1000-1800 LST in spring, summer, and autumn, respectively. At STS, the valley wind was more likely to occur during 0900-1900 and 0900-2000 LST in spring and summer, respectively. At LYS, the valley wind was more likely to occur during 1000-2000, 0900-1900, and 1000-1800 LST in spring, summer, and autumn, respectively. At HGS, the valley wind was more likely to occur during 0900-1900, 0900-2000, and 0900-1800 LST in spring, summer, and autumn, respectively. Thus, the valley wind occurs more often in spring and summer, for 10-11 h, in the urban and northern suburbs of Urumqi. In contrast to the valley wind, the mountain breeze occurs at night during spring, summer, and autumn at all stations, but especially in the southern suburbs.

Statistical analysis showed that the time when the mountain wind changes to a valley wind was 1-2 h earlier in summer compared with that in spring and autumn; whereas, the valley wind changed to a mountain wind 1-2 h later in summer compared with that in spring and autumn. The conversion time was found to be from 0800 to 1000 LST, attributable to solar radiation continuing to strengthen after sunrise such that a valley wind would soon form, develop, and continue until 1900 to 2100 LST, before the next conversion time. After sunset, when solar radiation gradually weakens, the valley wind will decline, before the mountain breeze develops again from 0800 to 1000 LST.

Figure 6 also shows that, in winter, static wind occurred more frequently at STS, LYS, HGS, and MD, but especially at STS. Static wind was more frequent from the evening until the morning, and at the height of more than 22 m in winter. At MD in summer, a switch time of 0900 LST for wind direction was found, with an easterly wind at night and a westerly wind in the day at heights of below 28 m. Meanwhile, the westerly wind in the day switched to an SSE wind at night at the heights of more than 36 m. Static wind occurred more frequently at this site during spring, autumn, and winter.

3.1.2Wind speed

Figure 7 shows the changes in wind speed at the five 100-m-high towers in Urumqi City from June 2013 to April 2014.

The monthly average wind speeds were in the ranges of 1.9-5.5 (YNLJ), 1.1-3.6 (STS), 1.2-4.3 (LYS), 1.2-4.3 (HGS), and 1.1-3.5 (MD) m s-1, respectively. At YNLJ, the seasonal wind speed variation was relatively weak, associated with the high topography of this region. However, significant seasonal variations were observed at STS, LYS HGS, and MD. Monthly wind speed was higher in April, June, July, August, and September. For these five months (April, June, July, August, and September), the average wind speed values were 1.9-3.4 (STS), 2.2-3.9 (LYS), 2.8-3.9 (HGS), and 2.3-3.4 (MD) m s-1, respectively.

The wind speed at YNLJ was the lowest during winter, at less than 2 m s-1, reflecting the effect of terrain on the wind speed in Urumqi City, which is located at the top of a bell-mouthed valley basin that narrows from Dabancheng to Chaiwopu in the valley. The prevailing winds were southerly in Dabancheng and the southern suburbs during winter (Fig. 4 also shows that the prevailing winds were southerly at YNLJ). By contrast, the wind speed was low during winter in urban areas due to the stability of the inversion layer and the topography east of the barrier.

Therefore, the seasonal changes in the wind speed characteristics of Urumqi City are affected mainly by the large-scale background wind field, thereby reflecting the effects of the dynamic and thermodynamic features of the local terrain. During spring, the ground temperature increases rapidly because the air is not stable, and thus wind forms readily due to cold air intrusion. During summer, these cold air forces weaken and there are fewer strong winds compared with spring. During winter, an inversion layer is present in Urumqi, with cold air intrusion from the basin inversion layer above, so the surface wind speed is low (Zhang and Zhang, 2006).

Table 3 shows the seasonal changes in wind speed characteristics in Urumqi City and the differences compared with those in other cities where the wind speed is also strong in summer and weak in winter. Thus, we can see how the specific terrain affects the wind speed.

Table 3 Comparison of wind speeds in different cities
3.2Distributional characteristics of the nearsurface flow field 3.2.1Observed wind field

The low-level wind field is an important factor related to the control of atmospheric pollution in a specific region (Lindsey et al., 1999; Jin et al., 2000). Thus, determining the wind field is essential for the establishment of an atmospheric dispersion model (Wang and Jiang, 2006). Many studies have shown that the terrain is an important factor that affects the low-level wind field, e.g., in Beijing, Qingdao, and Lan-zhou citys (Wang and Jiang, 2006; Li et al., 2008; Wang et al., 2012). Therefore, the seasonal changes in the surface flow field and valley wind were analyzed for Urumqi City using the wind speed and wind direction data obtained from 39 regional automatic stations between June 2013 and March 2014.

Figure 8 shows the spatial distributions of the near-surface wind in the mountain-valley area during spring, summer, and autumn from the 39 stations data; the Kriging interpolation method, by using Surfer and ArcGIS software, was employed. A number of statistical interpolation methods are available for surface wind interpolation, but the study by Luo et al. (2008) suggested that the Kriging method is more sophisticated because it explicitly accounts for spatial variance. This method was therefore selected for the present analysis. The statistics showed that the probability of a mountain wind was the highest in autumn, whereas the probability of a valley wind was the highest during spring and summer. The flow field was found to be straight in the main urban area, but scattered outside this area. Moreover, the wind direction was disordered at night in the main urban area.

Figure 8 Average surface-layer flow (u, v; m s-1) during (a, e) spring, (b, f) summer, (c, g) autumn, and (d, h) winter, at (a-d) 0000 LST and (e-h) 1200 LST, in Urumqi City.

During summer, the horizontal flow field was divergent and scattered in the main urban area, and the highest in Dabancheng. Furthermore, the streamline was curved near the Tianshan Mountains due to the effect of topography. Thus, in the main urban area, the flow field was scattered at 0000 LST and northerly at 1200 LST, with the wind speed being higher at this latter time than that at night. By contrast, during winter, the surface wind speed was less than 1 m s-1, and was usually static. The flow fields in spring and autumn were similar to that in summer, and there was weak convergence.

Thus, the flow characteristics comprise more airflow around the mountain, where the convergence and divergence are dominated by the terrain in eastern and southwestern Urumqi. The diurnal variation in the wind field is significant during spring, summer, and autumn in the main urban area, where the wind speed is strong during the day (the mountain wind is a southerly wind) and weak at night (the valley wind is a northerly wind). The flow is scattered due to the effect of the city.

According to the results shown in Fig. 8, we can summarize the low-layer flow field based on the mountain’s forcing of the flow. The influence of a narrow pipe effect, located in Dabancheng valley (northwest to southeast) between two sections of the Tianshan Mountains, is strong. The wind deflects and the wind speed increases in the valley. The northwest wind is higher in the narrow pipe than in the urban areas.

This feature is most obvious during summer, followed by spring and autumn, and not at all obvious in winter. In addition, the deflection of the airflow has a significant effect on the urban area. The mountain-valley wind in Urumqi is affected significantly by the underlying surface and topography, where there is a valley wind during the day from Junggar Basin and a mountain wind at night from the Tianshan Mountains (Lin and Li, 1985). In the present study, Fig. 9 shows that the valley wind was northerly during the day and the mountain wind was southerly at night. Thus, the directions of the mountain and valley winds were opposite, by almost 180°, and the two wind directions alternated twice during the day. Therefore, the valley wind exhibited a diurnal pattern. The differences in the thermodynamics of the underlying surface and the dynamic nature of the mountains and the city make the circulation strong during the day but weak at night.

Figure 9 Characteristics of the mountain-valley wind during a clear day on (a, e) 13 April 2013, (b, f) 28 August, (c, g) 1 November 2013, and (d, h) 11 January 2014, at (a-d) 0000 and (e-h) 1200 LST, in Urumqi City (valley wind during the day: northerly; mountain wind at night: southerly).
3.2.1WRF-simulated wind field

Figure 10 compares the WRF-simulated 10-m wind field results at 0000 LST 1 and 1200 LST 2 November 2013. The results clearly demonstrate the influence of the terrain on the wind field in Urumqi, as follows: (1) In Urumqi City, there is a mountain-valley wind, where the wind blows from the valley to hillsides during the day but from the mountain to the valley at night. (2) The narrow pipe effect of the valley is obvious in the Dabancheng valley. There is a gap in the Tianshan Mountains from Urumqi to Dabancheng, where the airflow narrows in the channel, and thus much of the air is retained in the vicinity of Dabancheng, thereby increasing the difference in pressure between the two regions, producing high winds. (3) The air around the mountain exhibits obvious deflection and a flow phenomenon, where the leeward wind speed is low. (4) The wind speed decreases as the air passes over the mountain, due to the effect of friction.

Figure 10 The WRF-simulated 10-m wind field at (a, c) 1200 and (b, d) 0000 LST, based on (a, b) the actual terrain and (c, d) no terrain (m s-1).

Figure 11 shows the WRF-simulated wind field at 950, 850, 700, and 500 hPa at 1200 LST 1 and 0000 LST 2 November 2013, based on the actual terrain. The results indicate that at 950 hPa, the mountain-valley wind is able to affect MD in Urumqi up to Fukang in the Changji area. At 850 hPa, the nature of the mountain-valley wind circulation of the horizontal wind field changes significantly. At 0000 LST, the air around the Tianshan Mountains converges to the bottom of the valley of Urumqi, so the wind through the gap is enhanced. The wind speed increases and the wind direction is mainly southeasterly and southerly. At 1200 LST, the wind speed is also strong in the valley of Urumqi and the wind direction is mainly northwesterly and northerly. During the day, the southerly wind weakens and the northerly wind strengthens at 700 hPa. The strong northerly wind makes it possible to affect Turpan Basin. The characteristics of the flow near the mountain slope are also obvious. At 500 hPa, the system wind (northwesterly and westerly wind) dominates during the day, and the wind speed is low ( < 4 m s-1) at night.

Figure 11 WRF-simulated wind field at (a, c, e, g) 1200 and (b, d, f, h) 0000 LST, at (a, b) 950, (c, d) 850, (e, f) 700, and (g, h) 500 hPa, based on the actual terrain (m s-1). Weather station at CJ: Changji; FK: Fukang; MD: Midong; and DBC: Dabancheng.

Figure 12 shows the WRF-simulated vertical velocity at 1200 LST 1 and 0000 LST 2 November 2013 over the Urumqi area (87°390E). The results show that the terrain is high in the south and low in the north at 87°390E, where 42.6°-43.2°N represents the top of the mountain and the highest elevation. Meanwhile, there is a small valley that declines continuously over 43°-44°N in the hillside area, and 44°-45.1°N covers the alluvial fan with little difference in elevation.

Figure 12 Updraft strength in a vertical section over the Urumqi region at (a) 1200 LST 1 November and (b) 0000 LST 2 November 2013 (m s-1; blank denotes terrain region).

The vertical velocity is generally consistent during the day and at night, while the vertical wind speed is generally larger during the day than at night. The main difference is the denser airflow at the top of the southern area, which increases from 1200 LST during the day while the airflow sinks. There is a negative center near 42.9°N at 400 hPa (central value of -0.2 m s-1), which indicates the absence of obvious upward flow due to the increased airflow. At night (0000 LST), there is descending motion. The vertical velocity in the valley is almost 0 m s-1 during the day and at night, and the thermal effect is weak, where the vertical movement is caused mainly by the effect of topography.

4Conceptual model

A conceptual model for the characteristics of wind over Urumqi was established on the basis of the analysis of the vertical wind structure above 100 m in the surface layer and the variation of wind flows (Fig. 13). In Urumqi, the wind direction (except in northern rural areas) is mainly northerly and southerly in the sur-face layer above 100 m, which is consistent with the strike of the Urumqi gorge. At YNLJ, the wind speed is 1.9-5.5 m s-1 higher than in the urban, northern suburban, and northern rural areas. The flow characteristics comprise more airflow around the mountain. The narrow pipe wind is obvious between the two sections of the Tianshan Mountains. Mountain winds (southerly wind) and valley winds (northerly wind) are often observed in spring, summer, and autumn, while static wind is often observed in winter.

Figure 13 Conceptual model of the wind system in the surface layer over Urumqi. MD: Midong; YNLJ: Yannanlijiao.
5Conclusions

This study employed a year of continuous observational data obtained from five 100-m-high meteorological towers and 39 regional automatic stations, as well numerical simulation results, to analyze the characteristics of the wind field over Urumqi. The study’s findings can be summarized as follows:

(1) Urumqi is surrounded on three sides by mountainous terrain. The vertical structure of wind direction and wind speed above 100 m in the surface layer is significantly affected by the topography. The annual and seasonal wind directions (except in winter) are mainly northerly and southerly in the surface layer above 100 m in urban, southern suburban, and northern suburban areas, which is consistent with the strike of the Urumqi gorge. Static wind frequently occurs in urban, northern suburban, and northern rural areas, but infrequently in southern suburbs. No obvious seasonal variation in wind speed can be found in the southern suburbs, where wind speed is relatively high throughout the year.

(2) The wind field in Urumqi is also significantly affected by the topography. The flow characteristics comprise more airflow around the mountain, where the convergence and divergence are dominated by the terrain in eastern and southwestern Urumqi. The diurnal variation in the wind field is significant during spring, summer, and autumn in the main urban area, where the wind speed is strong during the day (northerly valley wind) and weak at night (southerly mountain wind).

(3) The wind during the day over Urumqi is a northerly valley wind, while at night it is a southerly mountain wind. The valley wind occurs more often in spring and summer for 10-11 h in urban and northern suburban areas of Urumqi. The conversion from the mountain wind to the valley wind occurs during 0800-1000 LST, whereas the conversion from the valley wind to the mountain wind occurs during 1900-2100 LST. The mountain to valley wind conversion time is 1-2 h earlier in summer than in spring and autumn, whereas the valley to mountain wind conversion time is 1-2 h later in summer than in spring and autumn. The mountain-valley wind can affect the MD area of Urumqi as far as Fukang in the Changji area. The characteristics of the mountain-valley wind are particularly obvious at 850 hPa, according to the results from the WRF model simulation.

Acknowledgments: We appreciate the suggestions and comments from the three anonymous reviewers, which were helpful in improving the overall quality of the paper. Particular thanks are given to Ali Mamtimin, who has provided the five 100-m meteorology towers data in Urumqi City.
References
Acharya S., Neupane S., Shrestha R., et al ,2014: Early monsoon time local flow characteristics over the Het-auda valley and its implications. Journal of Institute of Science and Technology , 19 , 109–117.
DOI:10.1175/1520-0450(1997)036<0119:AIOFRA>2.0.CO;2 Bossert J. E. ,1997: An investigation of flow regimes affecting the Mexico City region. J. Appl. Meteor. , 36 , 119–140. DOI:10.1175/1520-0450(1997)036<0119:AIOFRA>2.0.CO;2
DOI:10.1007/s00704-008-0054-0 Chen Y., Jiang W. M., Zhang N., et al ,2009: Numerical simulation of the anthropogenic heat effect on urban boundary layer structure. Theor. Appl. Climatol. , 97 , 123–134. DOI:10.1007/s00704-008-0054-0
DOI:10.1016/j.atmosres.2005.09.005 Cogliati M. G., Mazzeo N. A. ,2006: Air flow analysis in the upper Rıo Negro Valley (Argentina). Atmos. Res. , 80 , 263–279. DOI:10.1016/j.atmosres.2005.09.005
DOI:10.1175/JAMC-D-10-05034.1 De Wekker S. F. J., Godwin K. S., Emmitt G. D., et al ,2012: Airborne Doppler Lidar measurements of valley flows in complex coastal terrain. Journal of Applied Meteorology and Climatology , 51 , 1558–1574. DOI:10.1175/JAMC-D-10-05034.1
Dou Jingjing, Wang Yingchun, Miao Shiguang ,2014: Fine spatial and temporal characteristics of humid-ity and wind in Beijing urban area. J. Appl. Meteor. Sci. , 25 , 559–569.
DOI:10.1175/1520-0493(2000)128<1106:DWITHK>2.0.CO;2 Egger J., Bajrachaya S., Egger U., et al ,2000: Diurnal winds in the Himalayan Kali Gandaki Valley.Part I:Observations. Mon. Wea. Rev. , 128 , 1106–1122. DOI:10.1175/1520-0493(2000)128<1106:DWITHK>2.0.CO;2
Hu Xiaoming, Liu Shuhua, Liang Fuming, et al ,2005: Observational study of wind fields, temperature fields over Beijing area in summer and winter. Acta Scientiarum Naturalium Universitatis Pekinensis , 41 , 399–407.
Jin Jianjun, Zhang Lei, Chen Changhe, et al ,2000: Characteristics of surface wind over the eastern part of Lanzhou basin in winter. Journal of Lanzhou University (Natural Sciences) , 36 , 113–120.
DOI:10.7519/j.issn.1000-0526.2016.06.009.(inChinese) Jin Lili, He Qing, Li Zhenjie, et al ,2016: Quality con-trol method and treatment for Urumqi meteorology tower gradient observation data. Meteor. Mon. , 42 , 732–742. DOI:10.7519/j.issn.1000-0526.2016.06.009.(inChinese)
Jü Lixia, Wang Qingeng, Zhang Meigen, et al ,2003: A simulation of heat island circulation and mountain-valley breeze in Jinan. Climatic Environ. Res. , 8 , 467–474.
DOI:10.1016/j.atmosres.2013.07.006 Laiti L., Zardi D., de Franceschi M., et al ,2013: Atmospheric boundary layer structures associ-ated with the Ora del Garda wind in the Alps as revealed from airborne and surface measure-ments. Atmos. Res. , 132-133 , 473–489. DOI:10.1016/j.atmosres.2013.07.006
Li Qingbao, Miao Shiguang, Sun Guiping, et al,2008: Terrain impact on the wind field in the racing ar-eas for Beijing 2008 Olympic Sailing Competition. Weather and Climate Section under the Influence of Complex Terrain of Chinese Meteorological Society Meeting. Beijing: Chinese Meteorological Society: 188-198.
Li Qingbao, Miao Shiguang, Liu Xuegang, et al ,2010: On the impacts of boundary layer flows and the terrain on the afternoon sea-breeze in the Qingdao Olympic sailing venue. Acta Meteor. Sinica , 68 , 985–997.
DOI:10.1080/10473289.2011.617628 Li X., Xia X. A., Xin Y., et al ,2012: An examina-tion of boundary layer structure under the influ-ence of the gap winds in Urumqi, China, during air pollution episode in winter. Journal of the Air & Waste Management Association , 62 , 26–37. DOI:10.1080/10473289.2011.617628
DOI:10.1002/2014JD022778 Li X., Xia X. A., Wang L., et al ,2015: The role of foehn in the formation of heavy air pollution events in Urumqi, China. J. Geophys. Res. , 120 , 5371–5384. DOI:10.1002/2014JD022778
Li Xia, 2013:The research on formation mechanism of the severe air pollution in gap town Urumqi in winter. Ph. D. dissertation, University of Chinese Academy of Sciences, Beijing, China, 135 pp.(in Chinese)
Li Xiaoli, Bi Baogui, Li Zechun ,2005: Simula-tion study of formation mechanism of winter urban boundary layer structure over Beijing area. Acta Meteor. Sinica , 63 , 889–902.
Lin Zhiguang, Li Yingjiang ,1985: The climatological research of the mountain-valley breeze in Tianshan Mountains. Geographical Research , 4 , 63–70.
DOI:10.1175/1520-0450(1999)038<1031:MPATTO>2.0.CO;2 Lindsey C. G., Chen J., Dye T. S., et al ,1999: Mete-orological processes affecting the transport of emis-sions from the Navajo Generating station to Grand Canyon national park. J. Appl. Meteor. , 38 , 1031–1048. DOI:10.1175/1520-0450(1999)038<1031:MPATTO>2.0.CO;2
Liu Xuefeng, Ren Guoyu, Liang Xiuhui, et al ,2009: The characteristics of wind speed variation at different altitudes of boundary layer in Hebei Province. Me-teor. Mon. , 35 , 46–53.
Liu Zhenxin, 2014:Assessing the urbanization effects by applying an urban canopy model and coupling it with weather research and forecasting. Ph. D. dissertation, Peking University, Beijing, China, 120-132.(in Chinese)
DOI:10.1002/joc.1583 Luo W., Taylor M. C., Parker S. R. ,2008: A com-parison of spatial interpolation methods to estimate continuous wind speed surfaces using irregularly distributed data from England and Wales. Int. J. Climatol. , 28 , 947–959. DOI:10.1002/joc.1583
DOI:10.3369/tethys.2008.5.02 Martínez D., Cuxart J., Cunillera J. ,2008: Con-ditioned climatology for stably stratified nights in the Lleida area. Tethys , 5 , 13–24. DOI:10.3369/tethys.2008.5.02
DOI:10.1016/j.atmosres.2008.02.013 Miao S. G., Chen F. ,2008: Formation of horizontal convective rolls in urban areas. Atmos. Res. , 89 , 298–304. DOI:10.1016/j.atmosres.2008.02.013
DOI:10.1175/2008JAMC1909.1 Miao S. G., Chen F., LeMone M. A., et al ,2009: An ob-servational and modeling study of characteristics of urban heat island and boundary layer structures in Beijing. Journal of Applied Meteorology and Clima-tology , 48 , 484–501. DOI:10.1175/2008JAMC1909.1
DOI:10.1007/s00376-013-2234-9 Miao Yucong, Liu Shuhua, Chen Bicheng, et al ,2013: Simulating urban flow and dispersion in Beijing by coupling a CFD model with the WRF model. Adv. Atmos. Sci. , 30 , 1663–1678. DOI:10.1007/s00376-013-2234-9
DOI:10.1023/A:1026090103764 Oliveira A. P., Bornstein R. D., Soares J. ,2003: Annual and diurnal wind patterns in the city of Sao Paulo. Water, Air and Soil Pollution:Focus , 3 , 3–15. DOI:10.1023/A:1026090103764
Qin Jun, Yuan Yechang, Li Yan, et al ,2001: A study of wind system in a sophisticated relief in mountain area. Climatic Environ. Res. , 6 , 493–497.
DOI:10.1175/JAS-D-11-0129.1 Schmidli J., Rotunno R. ,2012: Influence of the valley surroundings on valley wind dynamics. J. Atmos. Sci. , 69 , 561–577. DOI:10.1175/JAS-D-11-0129.1
DOI:10.1175/2010MWR3523.1 Schmidli J., Billings B., Chow F. K., et al ,2011: Inter-comparison of mesoscale model simulations of the daytime valley wind system. Mon. Wea. Rev. , 139 , 1389–1409. DOI:10.1175/2010MWR3523.1
DOI:10.1155/2011/569475 Seto D., Clements C. B. ,2011: Fire whirl evolu-tion observed during a valley wind-sea breeze re-versal. Journal of Combustion , 569475. DOI:10.1155/2011/569475
Shou Shaowen,1993: Mesoscale Meteorology. Beijing: Meteoro-logical Press.
DOI:10.1177/030913339902300407 Sturman A. P., McGowan H. A., Spronken-Smith R. A. ,1999: Mesoscale and local climates in New Zealand. Progress in Physical Geography , 23 , 611–635. DOI:10.1177/030913339902300407
Wang Chunhua, Lü Aihua, Yu Xiaoli, et al ,2010: Anal-ysis on current status of atmospheric pollution and its affecting factors in Urumqi. Journal of Xinjiang Agricultural University , 33 , 349–353.
Wang Guiling, Jiang Weimei ,2006: Characteristics of wind field in low layer over complex terrain. Jour-nal of PLA University of Science and Technology , 7 , 491–495.
Wang Jin, Zhang Lei, Wang Tengjiao, et al ,2012: Com-parative analysis of mountain-valley wind circu-lation characteristics over semi-arid areas nearby Lanzhou. Journal of Arid Meteorology , 30 , 169–177.
DOI:10.1007/s10546-006-9142-9 Weigel A. P., Chow F. K., Rotach M. W. ,2007: On the nature of turbulent kinetic energy in a steep and narrow Alpine valley. Bound.-Layer Meteor. , 123 , 177–199. DOI:10.1007/s10546-006-9142-9
Whiteman C. D.,2000: Mountain Meteorology:Funda-mentals and Applications. Oxford: Oxford University Press: 355.
Wu Qingmei, Zhang Shengjun ,2010: The pollution influencing cause analysis of a fog-haze process. Me-teor. Environ. Sci. , 33 , 12–16.
Yu Hongmin, Ren Guoyu, Liu Yulian ,2013: The characteristics of wind speed variation at different altitudes of boundary layer in Heilongjiang Province. Journal of Natural Resource , 28 , 1718–1730.
DOI:10.1007/s00703-003-0056-5 Zöngl G. ,2004: A reexamination of the valley wind system in the alpine inn valley with numerical sim-ulations. Meteor. Atmos. Phys. , 87 , 241–256. DOI:10.1007/s00703-003-0056-5
Zhang Jiabao, Su Qiyuan, Sun Shenqing,1986: The Guidance Handbook of Short-term Weather Fore-casting in Xinjiang. Urumqi: Xinjiang People Press: 171-205.
Zhang Qiang, Lü Shihua, Zhang Guangshu ,2003: The structure of atmospheric boundary layer over valley city and its transfer ability. Plateau Meteor. , 22 , 346–353.
Zhang Xuewen, Zhang Jiabao,2006: Xinjiang Me-teorological Handbook. Beijing: China Meteorological Press: 127-131.
Zhang Renwen, Fan Shaojia, Li Yingmin ,2012: Observational study of mountain-valley breeze over Conghua in 2008 fall. J. Trop. Meteor. , 28 , 134–139.