Hanoi is the capital of Vietnam with an area of 3358 km² (after the administrative expansion in 2008) and more than 7.4 million people (2017). It is expected that Hanoi will have an area of 13,463 km² and approximately 15 million people by 2020 (General Statistics Office of Vietnam, 2018). Hanoi is situated in the Northwest of the center of the Red River Delta, characterized by a tropical monsoon climate, with an explicit difference between wet and dry seasons (Bui et al., 2018). The wet season starts from mid-April to the end of October, but heavy precipitation mainly in July and August. Besides, the air masses from North and Northeast China normally bring a cold and dry climate from October to December. From January to March or April, the air masses with high humidity move across the Pacific Ocean to Northern Vietnam and lead to fog, low clouds and drizzle. Additionally, April and October have many unusual characteristics during the time between seasons, thus they are often classified as seasons of transition (Hien et al., 2002).

In order to assess the influence of meteorological factors on PM₁₀ concentration in Hanoi, the data includes hourly meteorological factors (temperature, humidity and wind speed) and 24-h PM₁₀ concentration collected at three automatic air quality monitoring stations in 2018 (Table 1). NVC Station and MK Station are located near busy road Nguyen Van Cu and National Highway 32 respectively with very high traffic density, while TY3 Station is in a residential area surrounded by internal roads with light traffic. The three air monitoring stations are in different directions representing various predominant urban traffic characteristic associated with high, medium and low density (Figure 1). The concentration of PM₁₀ was measured in NVC station using Orthogonal Light Scattering 90˚ method while Beta-ray absorption method were used in TY3 and MK station. Weather Monitor (from ThiesClima for NVC station and Vaisala for

TY3 and MK station) was installed at roof top of each station providing essential weather conditions such as temperature, relative humidity and wind speed. The study was implemented in two seasons: dry season (from October to March) and wetseason (from April to September).

The data were categorized in different data sets in order to assess the influence of meteorological factors on PM₁₀ concentration using spreadsheets (Excel) and statistical analysis software (SPSS V.20). Means, standard deviations, and box plots were utilized to demonstrate the differences between categories. The statistical tools SPSS (Statistical Package for Social Sciences) is used to assess the correlation between PM₁₀ concentration and meteorological factors through the Spearman correlation coefficient (r). The statistically average difference of PM₁₀ concentration during dry and wet season has also constructed to show the average difference between two overall variables (Independent sample T-Test). The Sig. value of the Independent sample T-Test has the probability α = 0.05 and the reliability is 95%. When the Sig. < 0.05, there is a difference in the mean of meteorological factors and PM₁₀ concentration in the two overall variables (dry and wet seasons), conversely, in case of the Sig. ≥ 0.05, there is the similarity between the mean of the two overall variables.

In 2018, the meteorological factors have indicated that the average temperature stays at a range of 24.2˚C - 25.3˚C. The highest temperature reaches at 36.6˚C in NVC Station (July 2018), the lowest temperature is measured at 8.0˚C in TY3 Station (January 2018). The annual average humidity is about 79%, the high value of humidity during dry season lasts from April to September with an annual average value of over 80%. The dampness month is from July to August with humidity around 80% - 95%. During the winter months, there are low humidity level and rainfall, the average humidity per month ranges from 75% to 80%. Moreover, the prevailing wind direction in Hanoi is the Southeast, the average wind speed stays at 1.5 m/s in 2018 (Table 2).

The monitoring results of the annual average PM₁₀ concentration in 2018 at three automatic air quality monitoring stations have exceeded the permitted limit specified in the National Technical Regulation on Ambient Air Quality (QCVN 05:2013/BTNMT) (Figure 2). In particular, PM₁₀ concentration exceeds 1.06 time (NVC Station) to 2.00 times (MK Station) compared to the standard allowed. It can be seen that the PM₁₀ concentration in Hanoi has gone up to 156.1 - 283.9 µg/m³ in some specific days of 2018 with 3.12 - 5.68 times higher than the QCVN 05:2013/BTNMT.

The changes of monthly average PM₁₀ concentration in 2018 clearly show the differences over months of the year (Figure 3). The high PM₁₀ concentration is

mainly during dry months (October to March), the remaining months have lower PM₁₀ concentration.

Figure 4 compares the mean PM₁₀ concentration during dry and wetseason. The results showed a clear difference in PM₁₀ concentration at different monitoring sites in both seasons, with a higher concentration in dry season and lower

in wet season. This result is also consistent with the studies of Zhao et al. (2009), Padro-Martinez et al. (2012), Schleicher et al. (2013), Zhang et al. (2015), Li et al. (2017), in particular, these studies have demonstrated that air pollutant concentrations such as PM_2.5, PM₁₀, CO, SO₂, NO₂ and O₃ tend to vary significantly, and the largest concentration in winter and the lowest concentration in summer are caused by changes in the height of boundary layer and atmospheric conditions from season to season.

The evaluation of the Spearman correlation coefficient between PM₁₀ concentration and meteorological factors such as temperature, humidity and wind speed (Table 2) have indicated that the correlation coefficients are all inversely correlated at three monitoring sites. Thus, when the temperature, humidity, wind speed increases, the average PM₁₀ concentration decreases and vice versa. Wind speed has a negative correlation with the PM₁₀ concentration due to major effects associated with dust dispersion and dilution (Clements et al., 2016). This inverse correlation may indicate the predominant presence of point-sources pollution (Chaloulakou et al., 2003). During winter, the particular change in temperature may be related to air temperature inversion that reduces the height of the atmospheric boundary layer and dust diffusion, therefore increases dust concentration (Gramsch et al., 2014). Moreover, humidity has different influences on particle size distribution, for small size particles, the moisture content is usually negatively correlated as shown by the evidence of washing effect (Giri et al., 2008; Kozakova et al., 2017).

The results of correlation evaluation displayed in Table 3 with the significance level of Sig. < 0.01 and the reliability of 99% show that temperature and wind speed are negatively correlated with PM₁₀ concentration and not high at NVC Station and TY3 Station. The correlation coefficients between PM₁₀ concentration

**. Correlation coefficient at the significance level Sig. < 0.01 (99% reliability).

and temperature at NVC Station and TY3 Station are −0.587 and −0.281, respectively, and the correlation between PM₁₀ concentration and wind speed is highest at TY3 Station (−0.232). At MK station, temperature and humidity are inversely correlated with PM₁₀ concentration and fairly low at −0.199 and −0.204, respectively. According to the Giri et al. (2008), the low correlation coefficient between PM₁₀ concentration and meteorological factors has demonstrated the competition of two mechanisms, the first is the atmospheric dispersion (dust particles are removed from air pollution through the process of dry and wet deposition by rain), and the second is the aerosolization diffusion from the surface (the air emission of particulate matter from street vehicles, industrial dust and soil dust). These linear correlations have clarified the fact that most of PM₁₀ in the area has an anthropogenic origin.

The average difference test of two independent variations (Independent sample T-Test) during wet and dry season proves that the value of Sig. test < 0.05 (95% reliability) at monitoring sites has a statistical difference in the data sets of PM₁₀ concentration, temperature, humidity and wind speed during two periods (Table 4).

The meteorological conditions (wind direction and wind speed) from Ha Dong station which is the representative station for Hanoi in 2018 have shown the prevailing wind direction is the Southeast, in which the winter has the dispersed wind direction and average wind speed lower than in the summer (Figure 5). Low

wind speed and different directions have led the slow PM₁₀ dispersion thus increased air pollution. The study of Giri et al. (2008) has also indicated that wind speed is an important factor for the increase of PM₁₀ concentration in the air environment at Kathmandu valley, Nepal. The research of Padro-Martinez et al. (2012) and Zhang et al. (2015) have also shown under conditions of low wind speed and high stable atmosphere, the horizontal diffusion and vertical disturbance are less prominent, whilst air pollutant concentration will significantly increase.