Possible effects of climate change of wind on aerosol variation during winter in Shanghai, China

Particuology 20 (2015) 80–88

Contents lists available at ScienceDirect

Particuology journal homepage: www.elsevier.com/locate/partic

Possible effects of climate change of wind on aerosol variation during winter in Shanghai, China Weidong Zhou a , Xuexi Tie b,c,∗ , Guangqiang Zhou a , Ping Liang d a

Shanghai Center for Urban Environmental Meteorology, Shanghai 200135, China Key Laboratory of Aerosol Science and Technology, SKLLQG, Institute of Earth Environment, Chinese Academy of Sciences, Xian 710075, China c National Center for Atmospheric Research, Boulder, CO 80307, USA d Shanghai Climate Center, Shanghai 200030, China b

a r t i c l e

i n f o

Article history: Received 13 February 2014 Received in revised form 12 August 2014 Accepted 13 August 2014 Keywords: Aerosol Wind Climate change Winter monsoon Shanghai

a b s t r a c t Several data sets were introduced to investigate the possible effects of climate-change-related variation of wind on aerosol concentration during winter in Shanghai, China. These data sets included the daily wind speed, wind direction, visibility, and precipitation from 1956 to 2010, hourly PM10 concentration from 2008 to 2010, and the NCEP/NCAR reanalysis data of global atmospheric circulation from 1956 to 2010. The trend of aerosol concentration and its correlations with wind speed and wind direction in winter were analyzed. Results indicated that there was an increase in the number of haze days in winter of 2.1 days/decade. Aerosol concentration, represented by PM10 in this study, was highly correlated to both wind speed and direction in winter. The PM10 concentration increased as wind speed decreased, reaching maximum values under static wind conditions. The PM10 concentration was relatively lower under easterly winds and higher under westerly winds. The analysis showed that weaker East Asia winter monsoons have resulted in a reduction of wind speed, increase in static wind frequency, and decline in the frequency of northerly winds since the 1980s. Moreover, the rapid expansion of urban construction in Shanghai has changed the underlying surface considerably, which has led to a reduction in wind speed. Finally, a wind factor was defined to estimate the combined effects of wind speed and wind direction on aerosol concentrations in Shanghai. The analysis of this factor indicated that changes in atmosphere circulation and urbanization have had important effects on the number of winter haze days in Shanghai. © 2014 Chinese Society of Particuology and Institute of Process Engineering, Chinese Academy of Sciences. Published by Elsevier B.V. All rights reserved.

Introduction Atmospheric aerosols include both natural and anthropogenic aerosols. The study by Zheng, Luo, Zhao, Chen, and Kang (2012) shows that the major aerosols in Eastern China are anthropogenic, and satellite measurements show that there is an increasing trend of aerosol optical depth (AOD) in the Shanghai region, suggesting that anthropogenic emissions are increasing significantly (Luo, Lu, Li, & Zhou, 2000). In addition to increased emissions, many studies have indicated that aerosol pollution events are closely related to meteorological conditions (Wang, Lin, Cai, & Chen, 2008; Yang, Wang, & Huang, 1994; Zhou, Qi, Gan, & Gao, 2010), and that the effect of

∗ Corresponding author at: National Center for Atmospheric Research, Boulder, CO 80307, USA. Tel.: +1 303 497 1470; fax: +1 303 497 1400. E-mail address: [email protected] (X. Tie).

meteorological conditions on aerosol concentrations has a short time scale (days–week) (Tie, Geng, Peng, Gao, & Zhao, 2009). Ke and Tang (2007) studied the aerosol scattering coefficient in Beijing in both autumn and winter, and their results showed that wind direction had considerable impact on the aerosol scatting coefficient. For example, the southwesterly wind led to an increase in the aerosol scattering coefficient, whereas the northeasterly wind produced a reduction. The study by Qiu, Sheng, Fang, and Gao (2004) suggests that southerly wind conditions result in an increase in the AOD in Qingdao, enhancing the scattering of solar radiation. Work by Xu, Geng, Zhen, and Gao (2010) shows that different surface winds lead to distinctively different effects on the diffusion and transportation of aerosols. The aerosol scattering coefficient decreases with wind speed when the prevalent wind direction is easterly; however, the converse is true under the effect of westerly winds. Additionally, Xu et al. (2005) proposed that the pollution diffusion process in urban regions might form a local downwind “plume” effect, enhancing the regional aerosol influence due to climatic characteristics of local

http://dx.doi.org/10.1016/j.partic.2014.08.008 1674-2001/© 2014 Chinese Society of Particuology and Institute of Process Engineering, Chinese Academy of Sciences. Published by Elsevier B.V. All rights reserved.

W. Zhou et al. / Particuology 20 (2015) 80–88

81

and 2010 were analyzed. The data were collected at Pudong Station (31◦ 14 N, 121◦ 32 E). To focus on the impact of wind on aerosol pollution, days with precipitation were excluded from the analysis. Historical daily measurements of wind speed and direction from 1956 to 2010 were used to investigate the long-term characteristics of winds in Shanghai. The number of haze days was estimated using measurements of relative humidity, visibility, and precipitation data. The emission data of dust and concentrations of SO2 and NO2 were obtained from the Shanghai Environment Monitoring Center. Additionally, reanalysis data from NCEP/NCAR (US National Centers for Environmental Prediction/US National Center for Atmospheric Research) with spatial resolution of 2.5◦ × 2.5◦ were used to study the general circulation patterns. The atmospheric circulation characteristic index (used for the parameter of meridional circulation) from the National Climate Center of China Meteorological Administration was also applied in the study. The remote sensing data from Chinese Academy of Sciences Data Center for resources and Environmental Sciences. PM10 concentrations were measured using a GRIMM-180 Stationary Aerosol Monitor (GRIMM Technologies, Inc., Germany) with precision of 1 ␮g/m3 and a flow rate of 72 L/h, which was calibrated periodically and properly maintained. Contact anemorumbometer, precipitation gauge, and psychrometer were applied to measure meteorological elements until 1999, and the MILOS 500 auto weather station (Vaisala Inc., Finland) has been implemented since 2000. Parallel observational experiments were performed for two years to evaluate the quality and ensure the consistency of the data; this was executed to the standards of the China Meteorological Administration (CMA). Visibility and weather phenomena were recorded by weather observers using the CMA specifications for ground observation (China Meteorological Administration, 2003). Fig. 1. Map of Shanghai and nearby cities.

Statistical methods dynamic effects. Furthermore, the study by Song and Lu (2006) suggests that the AOD of aerosols in Shanghai reaches a maximum in summer and a minimum in spring. Seasonal differences in ground visibility show good consistency with the above features of atmospheric aerosols, which has been proven by other related research (Lei, Zhang, He, & Streets, 2011; Luo, Lu, Zhou, Li, & He, 2001; Streets et al., 2008). However, few studies have focused on the changes of urban climate on the aerosol concentrations in Shanghai, especially in winter, which is the most polluted season. During winter, the frequency of precipitation is low, and wind speed and direction play important roles in controlling the local air pollution (Zhang, Zhen, Tan, & Yin, 2010). In this study, we analyze the effects of longterm changes in wind speed and direction on aerosol pollution in Shanghai during winter. Data and methodology

The statistics analysis system (SAS) was used for the normal distribution test. The Spearman correlation analysis of the hourly PM10 concentrations and wind speed, long-term trend of PM10 , wind direction, and wind speed in Shanghai was analyzed using linear regression. The Mann–Kendall method (Fu & Wang, 1992) (with confidence of 0.05) was applied to detect the wind mutation. Following the method of Li and Fang (2005), the Siberian highpressure index (SHPI) and the East Asian winter monsoon intensity index (EAWMII), using the NCEP/NCAR reanalysis data, were calculated by analyzing sea surface pressure. The correlation of these indices with wind speed and direction were calculated to obtain insight into the relationship between large general circulations and local winds. The detailed methodology for calculating the SHPI and EAWMII are as follows. The SHPI can be obtained by calculating the average sea level pressure (ASLP) within the region 40–60◦ N (j = 53–61) and 75–115◦ E (i = 31–47) in Eq. (1):

Meteorological and aerosol data Shanghai is a coastal megacity with an average altitude of 4 m. It is one of the biggest cities within the region of the East Asia monsoon, although there are many large- and medium-sized cities to its west. It is bordered by Jiangsu and Zhejiang provinces to the west and by the East China Sea to the east (Fig. 1). The geographical environment results in the local meteorological conditions having significant influence on aerosol concentrations over Shanghai. The hourly concentrations of PM10 in winter (Dec–Feb) and the corresponding data on wind speed and direction between 2008

 ASLP =

i=31,47,j=53,61

SLPi,j

(47 − 31 + 1)(61 − 53 + 1)

.

(1)

The EAWMII (in Eq. (2)) is calculated as follows: (a) calculate the sea level pressure (SLP) difference between 120◦ E and 150◦ E (i.e., subscripts 1 and 2), (b) normalize the values (i.e., superscript * in Eq. (2)) and calculate the sum within 30–60◦ N (i.e., j = 49–61), and (c) normalize again (i.e., superscript **). In Eq. (2), t = 1–55 represents the years from 1956 to 2010. The normalizing operator in Eq. (2)

82

W. Zhou et al. / Particuology 20 (2015) 80–88

was calculated using Eq. (3), where x denotes the difference in SLP and  denotes the standard error calculated in Eq. (4). EAWMIIt =

NMLt =

⎧ 61 ⎨ ⎩

[(SLP1,j,t − SLP2,j,t )∗]

⎭

j=49

xt − x¯ , 

 =

⎫ ⎬

55 (x t=1 t

54

∗ ∗.

(2)

(3) − x¯ )2

.

(4)

Four wind direction categories are used in this paper, i.e., east, south, west, and north. The east wind includes East-Northeast (ENE), East (E), East-Southeast (ESE), and Southeast (SE). The south wind includes South-Southeast (SSE), South (S), SouthSouthwest (SSW), and Southwest (SW). The west wind includes West-Southwest (WSW), West (W), West-Northwest (WNW), and Northwest (NW). The north wind includes North-Northwest (NNW), North (N), North-Northeast (NNE), and Northeast (NE). The wind speed and frequency of direction for each wind category are given by the sums of the values from all the sub-wind directions. Incidentally, aerosol concentrations are influenced considerably by factories closing during the Spring Festival vacation in February, which causes significantly different results from those in December and January. Therefore, PM10 concentrations in November, December, and January were selected to represent the winter aerosol concentrations in this study.

Fig. 2. Long-term change in number of winter haze days in Shanghai (1983–2010), emissions of particles and SO2 (1991–2010), and annual mean NO2 concentration (2000–2010).

Results and discussion Long-term aerosol variation The variation in haze days was employed to denote the aerosol trend because of the missing long-term measurements of aerosols. Haze is one type of weather phenomenon that results in low visibility, caused by suspended particulates in conditions of relative high humidity. Therefore, the severity of haze can be taken, to some extent, to represent the level of aerosol pollution. A haze day is defined as a day during which the daily average visibility is <10 km and daily average relative humidity is <90%, but phenomena that might reduce the visibility, i.e., precipitation, blowing snow, snow storms, blowing sand, dust storms, floating dust, and smoke, do not occur (Deng et al., 2008; Wu et al., 2006). Fig. 2 shows that between 1983 and 2010, the average number of winter haze days in Shanghai was 10.3 with a maximum of 22 days in 2004 and a minimum of 3 days in 1986. There is an obvious increasing trend in the number of haze days during this period, i.e., about 2.1 days/decade (the linear trend passed the 0.05 confidence test). The long-term change in the number of haze days can be separated into two obvious phases: 1983–2001, and 2002–2010. The number of haze days in 2002 (15 days) is obviously higher compared with 2001 (4 days). The average number of haze days during the second phase is 13.4, which is 4.7 more than during the first phase. The variation in the occurrence of haze days shows distinct interannual variability and periodic cycles, which might reflect the characteristics of long-term changes of aerosol concentrations. Fig. 2 also shows the variation of annual emissions of particles and SO2 (Shanghai Statistical Yearbook, 2009, 2011). The results indicate that emissions of particles and SO2 have both decreased during the two decades since the early 1990s, suggesting that the increase in the number of haze days cannot be ascribed to emissions of particles and SO2 . Furthermore, the annual mean NO2 concentration (Shanghai Statistical Yearbook, 2005, 2007, 2009, 2011) in urban Shanghai has been decreasing since 2000, suggesting that

Fig. 3. Average wind speed in winter in Shanghai from 1956 to 2010.

it too, is not connected with the increase in the number of haze days. Therefore, the change in the occurrence of haze related to emissions or pollutant concentrations might not be the major reason behind the increase in the number of haze days and thus, some other factors such as changes in atmospheric circulation and/or the local wind field might need to be considered. Characteristics of long-term average winter wind speed in Shanghai The average wind speed in Shanghai in winter from 1956 to 2010 was 3.2 m/s with a maximum speed of 4.3 m/s in 1982 and a minimum speed of 2.0 m/s in 1998. Fig. 3 shows an obvious decreasing trend in winter wind speed of about 0.33 m/s every 10 years (the linear trend passed the 0.001 confidence test). The wind speed shows a decreasing trend before the 1980s, during which the lowest speed appears in 1980. During the early 1980s, the wind speed initially increases, reaching a peak in 1982–1984, but then it decreases rapidly. A slight increase can be observed in the early 2000s, following which it then decreases steadily. We also used the Mann–Kendall method to study the mutation testing of average wind speed in winter in Shanghai. Two feature

W. Zhou et al. / Particuology 20 (2015) 80–88

83

Fig. 5. Static winter wind frequency in Shanghai (1956–2010).

Fig. 4. Winter wind direction frequency from 1956 to 2010 in Shanghai.

lines (within the 0.05 confidence range) intersect in 1988, which means the mutation of the average wind speed appears in 1988. We compared the average wind speed before and after 1988, and the result showed that it decreased by 1.2 m/s. Characteristics of long-term winter wind direction frequency in Shanghai Shanghai is located within the subtropical monsoon zone and the prevailing winter wind direction is northwesterly. The frequency of wind direction (Fig. 4) shows that the dominant wind direction in winter is NNW, which has a frequency of 14.4%, followed by NW and WNW, with frequencies of 9.7% and 9.4%, respectively. Wind directions of SW and SSW have the lowest frequencies of 2.1% and 2.3%, respectively. To show the long-term variation of the wind direction more intuitively, we estimated the average wind direction frequency of the east, south, west, and north winds from 1956 to 2010 using statistical method, which produced frequencies of 18%, 9%, 29%, and 38%, respectively. This result indicates that the frequency of each wind direction has declined, especially the frequency of the west wind, which has decreased by about 1% every 10 years. The other wind direction frequencies have decreased by less than 0.2%; however, none of the decreasing trends passed the significance test. The frequency of each wind direction has decreased differently, but the frequency of static wind conditions has increased obviously since the early 1980s (Fig. 5), which is closely related to the reduction in wind speed in the late 1980s. The long-term trend of the static wind frequency first shows a decrease and then an increase. The frequency decrease begins in the late 1950s and reaches a minimum value in the mid-1970s and mid-1980s. It then increases and reaches a maximum value (18%) in 1998, reduces during 2007–2009, and then increases rapidly in 2010. Characteristics of short-term aerosol concentration and its correlation with wind To illustrate the relationship between aerosol concentration and wind speed and direction, we analyzed the hourly winter PM10 concentrations and wind observations on November 2008 (550

samples) using the SAS statistical software. The probability distribution function figure and Kolmogorov–Smirnov test results show that the PM10 concentrations and wind speed sample follow a nonnormal distribution. We further analyzed their correlation using the Spearman correlation analysis method, which has no requirement for the sample to have a normal distribution. The results show an obvious negative correlation between PM10 concentration and wind speed; the correlation coefficient is −0.3784 and it passed the significance test (0.001). Similar results were found for the other two winter months (December 2008 and January 2009). This indicates that wind speed has significant influence on PM10 concentration, and that PM10 concentration would decrease if wind speed were to increase. To analyze the influence of wind speed on PM10 concentration, 5083 samples of hourly wind speed from 2008 to 2010 were divided into six levels based on the criteria of surface meteorological observations. Detailed information regarding the levels of wind speed and corresponding average PM10 concentrations is shown in Table 1. Table 1 shows that the PM10 concentrations reach a maximum (106.1 ␮g/m3 ) under static wind conditions, and decrease obviously as the wind speed increases. For level 6 winds, the PM10 concentration decreases by 17.3 ␮g/m3 compared with level 5, which is considerably larger than the concentration reduction when the wind speed is low; this indicates that the effect of wind speed on PM10 concentration is greater at higher speeds. The PM10 concentration for level 6 winds is 45.7 ␮g/m3 , which is 57% smaller than the concentration under level 1 winds. These results indicate that wind speed has significant short-term impact on the PM10 concentration.

Table 1 Wind speed levels and corresponding average PM10 concentrations and sample numbers from 2008 to 2010 in Shanghai. Wind level

Wind speed (m/s)

Average concentration (␮g/m3 )

Sample number

Level 6 Level 5 Level 4 Level 3 Level 2 Level 1

>4 3–4 2–3 1–2 0.3–1 <0.3

45.7 63.0 71.5 82.0 91.8 106.1

13 161 923 1785 1368 833

84

W. Zhou et al. / Particuology 20 (2015) 80–88 Table 3 Winter wind direction influence factor from 2008 to 2010 in Shanghai.

Fig. 6. Winter wind directions and their corresponding PM10 concentrations in Shanghai from 2008 to 2010. Table 2 Winter wind speed influence factor from 2008 to 2010 in Shanghai. Wind speed (m/s)

Wind speed influence factor

>4 3–4 2–3 1–2 0.3–1 <0.3

0.43 0.59 0.60 0.77 0.87 1.00

Shanghai is a coastal city with many large- and medium-sized cities to its west and ocean to its east. Therefore, wind direction plays an important role in PM10 concentration. The average winter PM10 concentrations for different wind directions from 2008 to 2010 in Shanghai (Fig. 6) indicate that the lowest PM10 concentration appears (63.8 ␮g/m3 ) when the wind direction is ESE, followed by ENE and E with PM10 concentrations of 65.4 and 68.2 ␮g/m3 , respectively. The PM10 concentration is very high when the wind direction is W, WSW, SW, and SSW; the maximum concentration is 103.0 ␮g/m3 under SSW wind conditions. These results suggest that when the wind direction is from the east, PM10 concentration is reduced because of the effect of clean air from the ocean. When the wind direction is from the west, pollutants from upwind cities are transported to Shanghai, increasing its PM10 concentration.

Wind direction

Wind direction influence factor

East wind South wind West wind North wind Static wind

0.61 0.79 0.90 0.76 1.00

for the Fwind-speed calculation. The PM10 concentration under static wind conditions is selected as the standard, and the ratio of PM10 concentration under the different wind direction categories to the standard is set as the Fwind-speed value (Table 3). Under these definitions, both large values of Fwind-speed and Fwind-direction represent conditions of bad diffusion or good transportation of pollutants in Shanghai. Furthermore, the combination of Fwind-speed and Fwind-direction represents the aggregate effect of wind on aerosol concentrations. The relationship of Fwind with Fwind-speed and Fwind-direction is shown in Eq. (5). Fwind = Fwind-speed Fwind-direction

(5)

The analysis was performed to evaluate the applicability of using hourly samples from 2008 to 2010. The combined classification of wind speed and wind direction was used with 21 categories, including static wind (SS) and five wind speed levels for each wind direction. For example, E1–E5 represent wind speed levels 2–6 in Table 1, respectively, for an east wind. The combined factors and their corresponding PM10 concentrations exhibit a close relationship (Fig. 7) with a correlation coefficient of 0.8015 at the 0.001 significance level. This suggests that the Fwind factor has good applicability for the estimation of the integrated influence of wind speed and wind direction on aerosol dispersion. Therefore, the Fwind factor can be applied to analyze the long-term influence of wind on aerosol concentrations. Incidentally, the PM10 concentrations in lower wind speed levels for the west and south direction are larger than in static wind conditions, indicating an increase of aerosols due to transportation from upwind regions. This suggests a switch between the dominant effect of transportation and dispersion under different wind speeds, which needs to be studied more closely.

Quantitative estimation of the impact of wind on aerosol concentration The previous analysis has indicated that the influence of wind on aerosol concentration is a combined effect of wind direction and wind speed. To estimate the influence of wind speed, a wind factor (Fwind ) is defined, which can represent the aggregate impact of both wind speed (Fwind-speed ) and direction (Fwind-direction ). The Fwind-speed factor represents the influence of the different wind speed levels on PM10 concentration. The value of Fwind-speed is defined as the ratio of the average PM10 concentration at each wind speed level to the standard PM10 concentration (static wind PM10 concentration). The details of the Fwind-speed factor are listed in Table 2. The Fwind-direction factor is the other part of the wind factor used for the quantitative estimation of the influence of wind direction on PM10 concentration. The methodology adopted is similar to that

Fig. 7. Comparison of the Fwind factor and PM10 concentration for different categories. E, S, W, and N are the east, south, west, and north wind directions, respectively; 1–5 represent levels 2–6 of the wind speed presented in Table 1; SS is static wind.

W. Zhou et al. / Particuology 20 (2015) 80–88

Fig. 8. Winter Fwind factor in Shanghai (1956–2010).

85

Fig. 9. Siberian high-pressure index (SHPI) (1956–2010).

Long-term characteristics of comprehensive wind factor in Shanghai Eq. (5) was expanded to calculate the long-term comprehensive factor of the winter monsoon (Fwind ) in Shanghai. The results (Fig. 8) indicate that the trend of Fwind is generally increasing and significant at the 0.001 significance level. The factor begins with a minimum value of 0.336 in 1956 and displays an increasing trend until 1970. Two large values over 0.4 can be seen in 1968 and 1970 during the first period of increase. The Fwind factor then shows a slight decreasing trend, reaching the second minimum of 0.341 in 1987. Subsequently, an obvious increasing trend can be seen, which reaches a maximum value of 0.531 in 2009. The rapid increase of the Fwind factor is related to the change of wind speed since the late 1980s over Shanghai and to the weakening of the winter wind speed and the increasing frequency of static wind conditions. A short-term decrease is also found around 2000, which is connected with a weak amplification of the wind speed during the corresponding years. Table 4 shows the correlation of wind speed, static wind frequency, and the Fwind factor, which shows there is positive correlation between the comprehensive wind factor and static wind frequency (0.3781); however, the correlation between the comprehensive impact factor and wind speed is negative (−0.7265). The

Fig. 10. East Asian winter monsoon intensity index (EAWMII) (1956–2010).

two tests are significant at the 0.01 and 0.001 levels, respectively. The major reason for the increment in the static wind frequency is the reduction of wind speed, and the correlation between the two factors is −0.5753. The long-term characteristics of the Fwind factor

Fig. 11. (a) Number of buildings over 30 floors and (b) corresponding construction area from 1998 to 2010 in Shanghai.

86

W. Zhou et al. / Particuology 20 (2015) 80–88

Fig. 12. Land use map of Shanghai: (a) 1980 and (b) 2008.

W. Zhou et al. / Particuology 20 (2015) 80–88

87

Table 4 Correlation of the Fwind factor, static wind frequency, and average wind speed. Fwind Fwind Static wind frequency Average wind speed * +

1 0.3781* –0.7265+

Static wind frequency *

0.3781 1 –0.5753+

Average wind speed –0.7265+ –0.5753+ 1

Stands for results that passed the 0.01 significance test. Stands for results that passed the 0.001 significance test.

reflect its impact on aerosol concentration in Shanghai; the correlation coefficient between the winter Fwind factor and the number of haze days from 1983 to 2010 is 0.4941 (0.01 significance level). Change of atmospheric circulation and its impact on wind The Siberian high-pressure area is one of the most important surface circulation systems of the Asian continent, and it has considerable impact on wind speed in winter in China. The results show slight downward trend of SHPI (Fig. 9) with a correlation coefficient between SHPI and winter wind speed in Shanghai of 0.2045 (this did not pass the 0.05 confidence test). Because there is a mutation in wind speed in 1988, we separated the study period into two and analyzed their respective correlations between SHPI and winter wind speed. The correlation coefficient during the first phase is 0.4513 (it passed the 0.01 significance test), but the result of the second phase did not pass the significance test. This indicates that changes in wind speed during the first phase are attributable mainly to the Siberian high-pressure atmospheric circulation, because the Siberian high-pressure system affects the activity of cold air, changing the northerly wind speed over highlatitude regions of mainland China. However, the result for the second phase indicates that influence from anthropogenic activities is also very important. Fig. 10 shows a decreasing trend of the EAWMII since the 1950s. It is positively correlated with winter wind speed and influences the wind direction frequency in Shanghai. The correlation coefficient between the EAWMII and north wind frequency is 0.2909 (it passed the 0.05 significance test), which indicates that the weaker East Asia monsoon has led to a decrease in the frequency of the north wind. Furthermore, results of the meridional circulation index show a downward trend, indicating that the intensity of the southward movement of cold air has declined, resulting in a reduction of the north wind speed. Urbanization and its impact on wind speed The urbanization of Shanghai has accelerated since the 1980s. Data from the Shanghai Statistical Yearbook (2000, 2002, 2003, 2005, 2007, 2009, 2011) shows that the permanently resident population increased from 11.47 million in 1980 to 14.12 million in 2010; an average annual increase of 88 thousand. However, because of the construction needs to accommodate this growth, the influx of migrant workers has resulted in a population explosion with the total population recorded as 23.03 million at the end of 2010. Energy consumption in Shanghai has increased considerably because of the rapid population growth and city construction. It has increased by about 3.5 times since the early 1990s, from 31.91 million tons of standard coal in 1990 to 112.01 million tons of standard coal in 2010. The number of high-rise buildings (over 30 floors) has also increased rapidly during this urbanization (Fig. 11). There were no high-rise buildings in 1980 and only 15 in 1990; however, the number of high-rise buildings has soared since 1990 with an annual increment of 72 buildings. The area given over to the construction of high-rise buildings has increased from 9.32 million m2

in 1999 to 32.08 million m2 in 2010 with an annual increment of 2.07 million m2 . The rapid development of high-rise buildings has increased the surface roughness of the city landscape, decreasing the surface wind speed considerably. Fig. 12 indicates that land use in Shanghai has changed significantly since the 1980s. Compared with 1980, farmland had decreased by 28.6% (1497.8 km2 ) and construction land has increased by 221.5% (1420 km2 ). Other land use types such as forest, grassland, and water bodies have increased by 37% (159 km2 ) in total. The areas of construction land were 641 and 2061 km2 in 1980 and 2008, respectively, corresponding to proportions of the total area of Shanghai of 10.0% and 32.3%. This increase in construction land has led to obvious effects on meteorological elements (Zhou, Zhu, & Liang, 2010), such as wind (Jiang, Luo, & Zhao, 2008). Conclusions In this study, we analyzed the relationships of PM10 concentrations with wind direction, wind speed, and the correlation between the long-term wind change and aerosols. The major conclusions are as follows: (1) There is a trend of increase in the number of winter haze days in Shanghai. The annual average number of haze days between 1983 and 2010 was 10.3 days with an increasing trend of 2.1 days/decade. The change in the number of haze days was separated into two phases: 1983–2001, and 2002–2010. The annual average number of haze days was 13.4 in the second phase, which is 4.7 days more than in the first phase. (2) PM10 concentration is closely related to wind direction and wind speed. The maximum PM10 concentration occurred (106.1 ␮g/m3 ) under static wind conditions, and it decreased obviously when the wind speed increased. The wind direction also has considerable impact on PM10 concentration; the lowest concentration appeared for ESE winds, while the highest concentration occurred for SSW winds. (3) Average winter wind speed was 3.2 m/s in Shanghai with an obvious reducing trend of 0.33 m/s every 10 years. Mutation of winter wind speed happened in 1988, and the wind speed decreased 1.2 m/s afterward. The prevailing winter wind direction was northwest with a frequency of 14.4%; the frequencies of the east, south, west, and north winds were 18%, 9%, 29%, and 38%, respectively. The frequency of wind in all directions declined, whereas that of static wind increased. (4) The wind factor Fwind was defined to denote the aggregate impact of wind speed and wind direction on aerosol concentration. The Fwind factor initially decreased after 1956 to its lowest value in 1984, and then it increased rapidly until 2010. The major reason for the Fwind increment during the late 1980s is the reduction of wind speed and increase of static wind frequency. (5) Before the late 1980s, winter wind speed was highly related to the Siberian high-pressure area, but urbanization has played an important role in terms of the rapid increases of the population, energy consumption, and high-rise construction. This result is identical to that of Jiang, Luo, Zhao, and Tao (2010). The East Asia

88

W. Zhou et al. / Particuology 20 (2015) 80–88

winter monsoon is another cause of wind change in Shanghai. A weaker East Asia winter monsoon reduces the frequency of the winter north wind. The comprehensive impact leads to an increase of the wind factor, suggesting an increase of aerosol concentration. In fact, there is much uncertainty regarding the effects of climate change and urbanization on aerosol variation, and it is difficult to estimate quantitatively the impact of each factor. Moreover, many other factors can influence long-term aerosol variation. Acknowledgments The authors thank Dr. Shi Jun from the Shanghai Climate Center for his help regarding the processing of remote sensing data, and the anonymous reviewers for their constructive comments. This work was supported by the National Natural Science Foundation of China (NSFC) under Grant nos. 41275186, 41430424, and the CAS Pilot Special Project (Grant XDA05090204). The National Center for Atmospheric Research is sponsored by the National Science Foundation. References China Meteorological Administration. (2003). Criterion of surface meteorological observation. Beijing: China Meteorological Press (in Chinese). Deng, X., Tie, X., Wu, D., Zhou, X., Bi, X., Tan, H., et al. (2008). Long-term trend of visibility and its characterizations in the Pearl River Delta (PRD) region, China. Atmospheric Environment, 42, 1424–1435. Fu, Z., & Wang, Q. (1992). The definition and detection of the abrupt climatic change. Scientia Atmospherica Sinica, 16(4), 482–493. Jiang, Y., Luo, Y., & Zhao, Z. (2008). Characteristics of wind direction change in China during recent 50 years. Journal of Applied Meteorological Science, 19(6), 666–672 (in Chinese). Jiang, Y., Luo, Y., Zhao, Z., & Tao, S. (2010). Changes in wind speed over China during 1956–2004. Theoretical and Applied Climatology, 99(3–4), 421–430. Ke, Z., & Tang, J. (2007). An observation study of the scattering properties of aerosols over Shangdianzi, Beijing. Chinese Journal of Atmospheric Sciences, 31(3), 553–559 (in Chinese). Lei, Y., Zhang, Q., He, K., & Streets, D. G. (2011). Primary anthropogenic aerosol emission trends for China, 1990–2005. Atmospheric Chemistry and Physics, 11(3), 931–954.

Li, C., & Fang, Z. (2005). Linkage of arctic oscillation and winter temperature in northeast China. Plateau Meteorology, 24(6), 927–934 (in Chinese). Luo, Y., Lu, D., Li, W., & Zhou, X. (2000). Characteristic of atmosphere aerosol optical depth change in China in the past 30 years. Chinese Science Bulletin, 45(5), 549–554 (in Chinese). Luo, Y., Lu, D., Zhou, X., Li, W., & He, Q. (2001). Characteristics of the spatial distribution and yearly variation of aerosol optical depth over China in last 30 years. Journal of Geophysical Research: Atmospheres, 106(D13), 14501–14513. Qiu, M., Sheng, L., Fang, Y., & Gao, Y. (2004). The effect of weather situation on the aerosol optical properties in Qingdao. Periodical of Ocean University of China, 34(6), 925–930 (in Chinese). Shanghai Municipal Statistics Bureau. (2000, 2002, 2003, 2005, 2007, 2009, 2011). Shanghai statistical yearbook. China Statistics Press (in Chinese). Song, L., & Lu, D. (2006). Investigating of atmospheric optical characteristics over Shanghai region. Climatic and Environmental Research, 11(2), 203–206 (in Chinese). Streets, D., Yu, C., Wu, Y., Chin, M., Zhao, Z., Hayasaka, T., et al. (2008). Aerosol trends over China, 1980–2000. Atmospheric Research, 88, 174–182. Tie, X., Geng, F., Peng, L., Gao, W., & Zhao, C. (2009). Measurement and modeling of O3 variability in Shanghai, China: Application of the WRF-Chem model. Atmospheric Environment, 43(28), 4289–4302. Wang, H., Lin, C., Cai, Y., & Chen, B. (2008). The relationship between mutation characteristics of PM10 and meteorological conditions in Fuzhou City. Journal of Tropical Meteorology, 24(5), 564–568 (in Chinese). Wu, D., Bi, X., Deng, X., Li, F., Tan, H., Liao, G., et al. (2006). Effect of atmospheric haze on the deterioration of visibility over the Pearl River Delta. Acta Meteorologica Sinica, 64(4), 510–517. Xu, J., Geng, F., Zhen, C., & Gao, W. (2010). Aerosol scattering coefficients and the factors affecting them in Shanghai Pudong. Acta Scientiae Circumstantiae, 34(1), 211–216 (in Chinese). Xu, X., Shi, X., Zhang, S., Ding, G., Miao, Q., & Zhou, L. (2005). Influence of aerosol in Beijing and the surrounding city group and the related climate effect. Chinese Science Bulletin, 50(22), 2522–2530 (in Chinese). Yang, D., Wang, S., & Huang, J. (1994). The relations between the meteorological conditions and the air pollution over the city of Lanzhou. Journal of Lanzhou University, 30(1), 132–136 (in Chinese). Zhang, G., Zhen, X., Tan, J., & Yin, J. (2010). The analysis of the relationship between the air quality in Shanghai and surface pressure patterns and meteorological factors. Journal of Tropical Meteorology, 26(1), 124–128 (in Chinese). Zheng, X., Luo, Y., Zhao, T., Chen, J., & Kang, W. (2012). Geographical and climatological characterization of aerosol distribution in China. Scientia Geographica Sinica, 32(3), 265–272. Zhou, D., Qi, Y., Gan, L., & Gao, J. (2010). Analysis on PM10 pollution and meteorological conditions in Changchun City. Journal of Anhui Agricultural Sciences, 38(15), 8034–8036 (in Chinese). Zhou, W., Zhu, J., & Liang, P. (2010). Characteristics of winter temperature variation in Shanghai during 1873–2006 and their probable causes. Journal of Tropical Meteorology, 26(2), 211–217 (in Chinese).