Special Issue on Air Pollution and its Impact in South and Southeast Asia (IX)

Nafisa Islam  1, Tarik Reza Toha  2, Mohammad Maksimul Islam3, Tanvir Ahmed This email address is being protected from spambots. You need JavaScript enabled to view it.4 

1 Department of Civil Engineering, Bangladesh University of Engineering and Technology, Dhaka, Bangladesh
2 Department of Computer Science and Engineering, Bangladesh University of Engineering and Technology, Dhaka, Bangladesh
3 Department of Civil, Construction, and Environmental Engineering, North Carolina State University, Raleigh, NC, USA
4 Department of Civil Engineering, Bangladesh University of Engineering and Technology, Dhaka, Bangladesh


Received: February 16, 2022
Revised: November 16, 2022
Accepted: December 5, 2022

 Copyright The Author(s). This is an open access article distributed under the terms of the Creative Commons Attribution License (CC BY 4.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are cited.


Download Citation: ||https://doi.org/10.4209/aaqr.220082  


Cite this article:

Islam, N., Toha, T.R., Islam, M.Ma., Ahmed, T. (2023). Spatio-temporal Variation of Meteorological Influence on PM2.5 and PM10 over Major Urban Cities of Bangladesh. Aerosol Air Qual. Res. 23, 220082. https://doi.org/10.4209/aaqr.220082


HIGHLIGHTS

  • Meteorological influence on PM variation depends on temporal and regional factors.
  • Cross correlation indicates varied lagged response of PM to meteorological variation.
  • Including meteorological lagged terms in MNLR increases the PM prediction accuracy.
 

ABSTRACT


A detrended seasonal analysis on a 6-year (2013–2018) dataset of daily Particulate Matter (PM) concentration and meteorological parameters is performed to understand the spatio-temporal variation of PM and the seasonal influence of meteorological factors on PM pollution over 6 major urban cities of Bangladesh. Cross-correlation and multiple non-linear regression (MNLR) of air quality and meteorological data were used to explore the meteorology-PM interactions and their spatio-temporal variability. Meteorological influence on PM was found to be stronger in the southern part of the country relative to the northwestern part. MNLR analysis implied that meteorological parameters could explain up to 39% of daily PM variability during high pollution days. The deposition effect of relative humidity was prominent during the premonsoon season, while rainfall impact becomes dominant in the monsoon season, specifically in the northeastern region. Wind speed was observed to have a dilatory effect on PM variation, although wind seemed to carry sea aerosol in southern regions. In addition, the northwestern wind appeared to contribute to PM rise in most urban areas by carrying PM loading. In winter, the low temperature was found to favor PM accumulation, while in monsoon, high temperature causes PM rise, possibly by assisting atmospheric secondary aerosol formation. Solar radiation positively influenced atmospheric PM formation, and the influence was stronger in the northwestern region. It is shown that meteorological parameters influence the seasonal variability of PM, but the extent of this influence varies depending on temporal and regional factors.


Keywords: Particulate matter, Meteorology, Seasonal variation, Cross-correlation, Multiple non-linear regression


1 INTRODUCTION


Air pollution in major cities of Bangladesh has become quite acute in recent years due to rapid industrialization and urbanization. Dhaka, the capital of Bangladesh, was ranked as the 2nd most polluted city globally in 2019 (AirVisual, 2019). Besides, Gazipur and Barisal—two other major cities of Bangladesh—were ranked as the 38th and 41st most polluted cities globally in 2018 (CBS News, 2019). Road and building construction activities to support urbanization, mushrooming of brick kilns to facilitate construction, increased traffic and resulting vehicular emissions, and ongoing infrastructure megaprojects are responsible for exacerbating the air quality in urban areas of Bangladesh. Among all air pollutants, particulate matter (PM2.5 and PM10) is considered one of the most harmful elements due to its link with human health. Because of its size, PM can enter the bloodstreams of respiratory tracts, causing respiratory diseases and even death (WHO, 2016). It also affects visibility and is responsible for global and regional climate change (Won et al., 2021).

Since PM is a crucial factor in air quality degradation, it has become an important research topic for the scientific community (Mahapatra et al., 2018; Li et al., 2017a; Munir et al., 2013). Many studies have been performed to understand the physical, chemical, and optical properties of PM and to explore the driving factors influencing the spatial variability of PM (Islam et al., 2022b; Hajiloo et al., 2019; Mancilla et al., 2015; Ganguly et al., 2006). Meteorology has been found to be one of the primary determinants of PM variation in the atmosphere (Liu and Cui, 2014; Trivedi et al., 2014). Studies have shown that meteorological parameters such as wind speed, temperature, relative humidity, solar radiation, rainfall, atmospheric boundary layer, pressure system play a significant role in explaining PM variation in the USA, China, and India (Islam et al., 2022a; Singh et al., 2021; Ilias et al., 2020; Wang et al., 2013). Hassan et al. (2020) investigated the long-term PM relationship with meteorological parameters in Malaysia and identified that PM is strongly correlated with temperature, precipitation, and relative humidity. Mahapatra et al. (2018) examined a three-year dataset of PM and meteorology and observed that local meteorology and long-range transport significantly contribute to aerosol particle variation in eastern India. Chen et al. (2018) proposed a non-linear lag-response model for predicting PM2.5 using meteorological information in China. Li et al. (2017b) reviewed major advances in aerosol-boundary layer interaction processes and stated that planetary boundary layer plays a vital role for aerosol accumulation in the atmosphere. In addition, studies have found that wind shear, temperature inversion, and long-range transport are physically relevant to the variation of PM concentration (Bai et al., 2022; Braun et al., 2020; Guo et al., 2020; Zhang et al., 2020; Xu et al., 2019; Guo et al., 2017; Wu et al., 2014). Available literature corroborates the need for understanding meteorology to comprehend the spatio-temporal variability of PM.

Since the pattern of atmospheric circulation and corresponding pressure systems vary regionally, the atmospheric influence on PM is likely to vary from region to region (Chen et al., 2020). Li et al. (2019) observed that the meteorological influence on PM was different in different regions of China. Azmi et al. (2010) stated that the variation of PM10 might be attributed to regional tropical factors in Malaysia. Therefore, region-based study is necessary to understand a comprehensive meteorology-PM relationship for a certain area. As for Bangladesh, studies investigating the meteorology-PM interaction mostly focused Dhaka city. For example, Islam et al. (2015, 2022d) explored the meteorological influence on PM in Dhaka. Afrin et al. (2021) developed a meteorology-based PM prediction model and it was for Dhaka city too. Although some studies explored the air pollution scenario of other major cities of Bangladesh, meteorological impacts on air quality were not assessed. For example, Rahman et al. (2019) investigated the intra-site and inter-site air pollutant correlations between Dhaka, Gazipur and Narayanganj city. Shahriar et al. (2020) determined the efficacy of several machine learning algorithms to predict air pollutants in the city of Dhaka, Chittagong, Sylhet, and Rajshahi. However, as mentioned earlier, the impact of meteorological parameters (wind speed, temperature, relative humidity etc.) on air quality was not investigated at those cities. A recent study attempted to determine the correlation between meteorology and PM at six major cities of Bangladesh using back trajectory and conditional probability function analysis (Rana and Khan, 2020). However, the time dependent non-linear relationship of meteorology and PM was not explored here. Thus, the spatial and temporal variability of PM in Bangladesh and its interaction with meteorology is still poorly constrained.

We performed annual and seasonal cross-correlation analysis of meteorological parameters and PM concentrations—measured in eight monitoring stations of Bangladesh—to accomplish three objectives: (a) present spatio-temporal variability of PM pollution in Bangladesh, (b) assess the seasonality in meteorological influences on PM, and (c) compare the meteorological influences on PM among 5 major urban cities of Bangladesh. In the cross-correlation analysis, we also explored the relationship between wind azimuths and PM and observed their relationship for different regions. In addition, we performed seasonal multiple non-linear regression analysis to assess the temporal meteorological impact on PM variation in different regions. In this analysis, we incorporated the lagged terms of meteorological parameters which, from cross-correlation analysis, was noted to have strong influence on PM variation. We also considered the interaction terms of meteorological parameters to assess their internal association in affecting the PM variation. Finally, we validate our regression analysis by applying them for a recent dataset in one of the monitoring stations.

 
2 METHODS


 
2.1 Study Area

This study encompasses five out of eight divisions (main administrative units) and six major cities (within the divisions) of Bangladesh. Air quality and meteorological data of these cities were collected from the Continuous Air Monitoring Stations (CAMSs) in those cities. There are eleven CAMSs in Bangladesh where real-time monitoring of PM concentration and meteorological parameters is performed. PM concentration is monitored using the beta attenuation method (Model BAM-1020) by Met One Instrument Inc., USA. Due to poor quality data, three CAMS were discarded from our analysis. The remaining eight stations—located in six cities, as mentioned above—were the data source of this study. The location of these stations is presented in Fig. 1, and the details are provided in Table S1.

Fig. 1. Location of CAMSs in major urban cities of Bangladesh.Fig. 1. Location of CAMSs in major urban cities of Bangladesh.
 

 
2.2 Climatology of Bangladesh

Bangladesh has four seasons with marked seasonal variation: Winter (December–February), Premonsoon (March–May), Monsoon (June–September), and Post-monsoon (October–November). During winter, cold-dry calm wind (0–2 m s1) from the northwestern direction prevails over the country. However, the direction of wind flow is southwesterly in the hilly eastern region. Low humidity, rainfall, temperature, and solar radiation persist in this season. After winter, premonsoon comes. In this season, wind flows at a maximum speed ranging from 2 to 4 m s1 and has variable directions. This season is characterized by the hottest weather and thunderstorms, which are called Norwesters. These short-span thunderstorm events occur from April to May and shower this territory with gusty wind and torrential rain. Besides, temperature, radiation, humidity, and rainfall start to increase during this season. After the premonsoon, monsoon comes. During monsoon, moist warm high-speed wind (3–4 m s1) flows from the southwestern direction. This season is characterized by high humidity and persistent rainfall. After monsoon, postmonsoon arrives. In this season, precipitation, humidity, temperature, radiation, and wind speed start to decrease. After postmonsoon, winter comes again. A summary statistic of the meteorological parameters—categorized by seasons—in each CAMS is reported in Table S2, and the wind rose diagrams are shown in Fig. S2.

 
2.3 Data Pre-processing and Filtering

The raw dataset contained a significant number of outliers and missing values. Dataset was pre-processed through outlier removal (using the iterative method) and missing value imputation (using the 'k-NN' algorithm). The dataset is filtered by the Kalman Filter algorithm, which is implemented in PyCharm Community Edition, version 2020.2.3. The processes are described in detail in Supplementary Materials.

 
2.4 Data Structure and Cross-correlation Analysis

The data being analyzed contained eight linear variables—PM2.5, PM10, wind speed, temperature, relative humidity, solar radiation, rainfall amount and duration, and one angular variable—wind direction. Wind direction—measured clockwise from the geographical north and expressed in degrees—was converted to trigonometric cosine and sine of wind azimuths, representing the north-south and east-west components, respectively. Finally, cross-correlation analysis was performed on the filtered time-series data (discussed in Section 2.3 and Section S2 of the Supplementary Material) of PM concentration (PM2.5 and PM10) and meteorological parameters (wind speed, cosine and sine component of wind speed, temperature, relative humidity, solar radiation, rainfall amount and duration). The basic cross-correlation formula is presented in Eq. (1) and Eq. (2).

 

Here, Cxy(k) is the cross-correlation function, x(t) was concerned meteorological parameter at time t, y(t + k) was PM concentration at time (t + k), k was lag between meteorological parameter and PM concentration, and T was the length of data-series. The cross-correlation coefficient was determined using Eq. (3) from the cross-correlation function:

 

Here, $S_x=\sqrt{{C_{xx}}(0)}$ and $S_y=\sqrt{{C_{yy}}(0)}$. If the cross-correlation coefficient rxy is positive, PM and meteorological parameters vary proportionally, and vice versa. If lag k is positive, meteorological parameters lead PM and vice versa.

 
2.5 Multiple Non-linear Regression (MNLR)

We used multiple non-linear regression (MNLR) to determine the influence of combined meteorological parameters on PM variation. The model used Log-transformed PM time series (as the dependent variable) because PM concentration had been mathematically shown to have log-normal distribution (Ott, 1990). Many studies—investigating indoor (Islam et al., 2022c; Young et al., 2019) and outdoor PM concentrations (Beckerman et al., 2013) and PM emissions (Islam et al., 2021)—found the same and used logarithmic PM distribution in models/analyses. In our dataset, the logarithm of daily PM concentration follows normal distribution determined by Kolmogorov-Smirnov test (p = 0.0049 and 0.0710 for PM2.5 and log(PM2.5), respectively and p = 0.0035 and 0.0821 for PM10 and log(PM10), respectively). Therefore, our approach to selecting the logarithmic regression model aligns with the previous findings. Studies have also shown that transforming PM to logarithmic distribution helps reduce skewness and, in this way, helps in improving accuracy (Song et al., 2014; Tian and Chen, 2010; Liu et al., 2007). In MNLR models, the significance of independent variables was assessed at a 99% confidence level. We explored three approaches to develop regression models for each season. As the first approach, each meteorological parameter was added as an independent variable along with the lead or lagged terms—obtained from the seasonal cross-correlation analysis. Next, non-linear regression (up to two degrees) was incorporated (2nd approach). Finally, interaction terms of the independent variables (i.e., meteorological parameters) were added (3rd approach). The model/approach with the lowest PRESS and highest adjusted r2 value (discussed below) was selected for each case.

The general form of the MNLR model used is presented in Eq. (4).

 

Here, n was the number of meteorological parameters (n = 1–8), y was PM concentration, x1 and x2 both were meteorological parameters, x1x2 was the interaction term of meteorological parameters, β was regression coefficient, and ε was error term, where ε = (yi  ŷ), yi = observed y values and ŷ = y values given by Eq. (4). The coefficient of determination (r2) measured the extent of meteorological influence on PM variation, determined by Eq. (5).

 

Here, St and Sr represented the total sum of squares ( $\displaystyle\sum_{i=1}^{n}{(y_i-\overline{y})}^2$) and error sum of squares $\displaystyle\sum_{i=1}^{n}{(y_i-\widehat{y})}^2$, respectively. However, r2 might reduce the regression efficiency by increasing its values with increasing terms to the model. Hence, r2 was adjusted based on the added terms, determined using Eq. (6).

 

Here, m was the element number in a series, and n was the total number of considered meteorological parameters. We also used Prediction Error Sum of Squares (PRESS) to assess the efficacy of regression by detecting the presence of any excess independent variable. A smaller PRESS value indicates better efficacy of the regression analysis, which was determined by Eq. (7).

 

Here, hii was ith 'Leverage Value,' which only relied on the meteorological parameters. For n = 2 and sample size of T,

 


3 RESULTS AND DISCUSSION


 
3.1 Annual Trend

The annual average concentration of PM2.5 and PM10 over eight stations for the period of 2013–2018 is shown in Fig. 2. Here, it can be observed that the PM concentration of all stations exceeds the annual Bangladesh National Ambient Air Quality Standard (Hossen and Hoque, 2018). Among all stations, the highest PM concentration is observed at Darus Salam, Dhaka. In general, PM levels started decreasing in 2014, then started rising again in 2017. A similar trend is observed for other stations located in Dhaka, Gazipur, and Chittagong. This may be explained by the enactment of the Brick Kiln Control Act 2013 at the end of 2014 (DoE, 2017), after which the emission control activities (punitive actions by the Department of Environment) increased. However, PM concentration shows a sharp rise in 2018, when the 11th National Parliament Election was held (The Daily Star, 2018). Before the election, construction activities of some mega projects were accelerated, namely, Dhaka Metro Rail Project, Dhaka-Gazipur Four Lane Expressway Project and Improvement of Agrabad Access Road Project in Chittagong (Islam, 2018). As part of these construction works, brick kiln operation increased, which, in combination with construction works, possibly led to the rise in PM concentrations in 2018. In addition to interannual trend analysis, the spatial variability of seasonal average PM concentration is also investigated. From the seasonal spatial distribution maps of PM2.5 and PM10 (Fig. S3 and Fig. S4), it can be observed that PM2.5 concentration remains high to moderate in the central to southern region and low in the northern region of Bangladesh. However, PM10 concentration remains high in the northwestern and central region of Bangladesh. Northwestern region is the entrance point of long-range transported pollutants in Bangladesh (Begum et al., 2016), and high PM10 is observed here.

Fig. 2. The annual average concentration of (a) PM2.5 and (b) PM10. Here, error bars indicate 95% confidence interval. The black dashed lines represent the Annual Bangladesh National Ambient Air Quality Standard of PM2.5 (15 µg m–3) and PM10 (50 µg m–3).Fig. 2. The annual average concentration of (a) PM2.5 and (b) PM10. Here, error bars indicate 95% confidence interval. The black dashed lines represent the Annual Bangladesh National Ambient Air Quality Standard of PM2.5 (15 µg m–3) and PM10 (50 µg m–3).

 
3.2 Annual Cross-correlation Between PM and Meteorology

Annual cross-correlation analysis is performed between the daily dataset of PM and meteorological parameters from 2013–2018, and the summary is presented in Table 1. Here, wind speed, relative humidity, and rainfall show negative, and solar radiation shows a positive correlation with PM. Temperature shows both positive and negative correlations with PM depending on stations (Table 1). A negative correlation between most of the meteorological parameters and PM implies that an increase in those meteorological parameters will cause a decrease in PM and vice versa. In addition, Table 1 shows that the meteorological parameters show correlation with PM within 0–3 days lag (enclosed in parenthesis). This lag value represents the tentative time taken by the meteorological parameters to exert noticeable influence on PM. Zero (0) lag means, the concerned meteorological parameter of a specific day is related to the PM concentration of the same day. Positive lag (such as, +1) means the meteorological parameter of a specific day is related to the PM concentration of the following (one day later) day. Negative lag (such as, –1) means meteorological parameter of a specific day is related to the PM concentration of the previous (one day earlier) day. Therefore, positive lag means meteorological parameter leads PM, negative lag means PM leads meteorological parameter and ‘0’ lag means simultaneous interaction.

Table 1. Annual cross-correlation coefficients between meteorological parameters and PM over eight stations showing maximum correlation coefficients and corresponding lag period in days (in parenthesis).

 
3.2.1 Wind speed and direction

Among all the meteorological parameters, wind speed shows the highest negative correlation with PM fractions for all stations (Table 1). A negative correlation between wind speed and PM implies that high wind speed favors PM dissipation and low wind speed favors PM accumulation. In all stations except Sangsad Bhaban and Barisal, PM2.5 shows a higher negative correlation with wind speed than PM10. High wind speed, in addition to PM dissipation, can sometimes cause an increase in PM10 by resuspending road dust, leading to a weaker negative correlation with PM10 than PM2.5 (Elminir, 2005; Jung et al., 2002). However, in Sangsad Bhaban and Barisal, the correlation of wind speed is higher with PM10 than PM2.5. In Sangsad Bhaban premises, traffic emission is low, and the station is installed over 100 m away from nearby traffic road. Moreover, high grass cover exists in Sangsad Bhaban premises, making the road dust resuspension minimal due to increased friction (Li et al., 2017a; Hamed et al., 2011). In Barisal, a major PM2.5 pollution source is the water vessel emissions in the nearby Kirtonkhola river port (Rana and Khan, 2020; Cheng and Li, 2010). Irrespective of the magnitude of wind speed, the contribution of river vessels to PM2.5 emission remains consistent throughout the year, resulting in a lower negative correlation between wind speed and PM2.5. For the same reason, PM2.5 is less dependent on other meteorological parameters compared to PM10 in Barisal. The negative correlation between PM10 and wind speed is lowest in Rajshahi, which is located in the northwestern part of Bangladesh. Here, the wind comes from the continental regions of the Arabian Peninsula, which is characterized by haze (Begum et al., 2016). This transboundary airflow contains natural dust, which primarily consists of coarser-sized particles (PM10) (Fuzzi et al., 2015). This transboundary pollution might result in a lower negative correlation between wind speed and PM10 in Rajshahi.

The annual cross-correlation coefficients between wind direction and PM are presented in Table S3 of supplementary materials. Table S3 shows that the cosine and sine components of wind speed show positive and negative correlations with PM, respectively, for majority of stations. Positive cosine values are associated with the wind coming from the 270°–0°–90° directions, and negative sine values are associated with the wind from the 180°–270°–360° directions. Therefore, the correlation pattern may imply that the airmass contains higher PM loading when it comes from the northwestern direction. However, the cosine and sine components of wind azimuths are negatively correlated with PM in Sylhet (Table S3). Such correlation may infer that PM loading remains high in the southwestern wind here. For the majority of stations, the components of wind speed show higher correlations with PM10 relative to PM2.5, suggesting that in addition to local sources, a significant contribution of PM10 comes from long-range transportation.

 
3.2.2 Temperature

Temperature shows a negative correlation with PM for most stations (Table 1). This may refer to the PM dilution event caused by the increase in boundary layer height preceded by ambient temperature rise. However, in high vegetative areas, such as Sylhet and Chittagong, a positive correlation is observed between PM10 and temperature. In such areas, non-volatile organic compound (NVOC) is one of the major sources of atmospheric particles (Holmes, 2007). Since, NVOC emission is largely modulated by temperature (Kim et al., 2020), temperature increase might eventually lead to coarser sized particle increase in the atmosphere.

 
3.2.3 Relative humidity and rainfall

Relative humidity, rainfall amount, and rainfall duration show a negative correlation with PM for all stations, which may represent wet deposition, flocculation, and subsequent gravitational settling of PM. For most stations, the correlation of relative humidity and PM is stronger than that of rainfall properties (amount and duration) and PM. Since, in these areas, precipitation events are irregular, the effect of rainfall is not that much prominent on PM. However, in Sylhet, rainfall exerts the strongest negative correlation with PM. Since the rainfall events are heavy and consistent in Sylhet, the scrubbing effect of precipitation is predominant here. Rainfall duration is found to have a higher negative correlation with PM than rainfall amount for all stations, indicating that consistency of rainfall has a more reducing impact on PM than that of rainfall intensity. Therefore, we used only the rainfall duration parameter to represent rainfall property in our subsequent analysis.

 
3.2.4 Solar radiation

Solar radiation correlates positively with PM fractions, and the correlation is stronger for coarser particles (PM10). This may refer to the fact that, solar radiation induces photochemical reactions which propitiate the daytime O3, sulfate particles and secondary organic aerosol formation through oxidation process. The enhanced secondary aerosol production and condensation of atmospheric sulfuric acid can combinedly contribute to new particle formation (Wang et al., 2016). Literature showed that solar radiation is the driving factor to produce new particles in the atmosphere by catalyzing photochemical reactions of various pre-existing ions with OH radical. In the presence of high-temperature conditions, these photochemical reactions get boosted up (Wonaschütz et al., 2015; Hamed et al., 2011; Vehkamaki et al., 2004). Under a moist atmosphere, these new particles instantly undergo hygroscopic growth and become coarser in size (Seinfeld and Pandis, 2016). Therefore, under the combination of high solar radiation, temperature, and relative humidity, the secondary formation and growth of coarser particles increases. This phenomenon might lead to a high positive correlation between PM10 and solar radiation. Nonetheless, the highest positive correlation between solar radiation and PM is observed in Rajshahi. Here, the extent of the sunshine hour and incident solar radiation is maximum (Datta et al., 2014), resulting in a strong positive correlation between PM and solar radiation.

 
3.3 Aerosol-meteorology Interaction

In order to understand how PM and meteorological variables vary over time, a summary statistic of PM and meteorological variables for different seasons is presented in Table S2 of supplementary materials. From Table S2, it is evident that PM level is the highest during winter, the lowest during monsoon and moderate during premonsoon and postmonsoon seasons. We investigated how this PM variation can be attributed to meteorological influences through seasonal cross-correlation analysis and presented the seasonal cross-correlation coefficients between PM and meteorological parameters in Table 2 and Table 3, where the lag periods are shown in days within parenthesis. These tables implicate the seasonal influence and lag effect of meteorological parameters on PM concentration.

Table 2. Seasonal cross-correlation coefficients between meteorological parameters and PM2.5 over eight stations showing maximum correlation coefficients and corresponding lag period in days (in parenthesis).

Table 3. Seasonal cross-correlation coefficients between meteorological parameters and PM10 over eight stations showing maximum correlation coefficients and corresponding lag period in days (in parenthesis).

 
3.3.1 Wind speed

A strong negative correlation between wind speed and PM is observed during winter and monsoon season (Table 2 and Table 3). The calm wind of winter and the high-speed wind of monsoon might influence such a correlation (Jacobson and Kaufman, 2006). However, in Chittagong, the lowest negative correlation between PM10 and wind speed is observed in monsoon. Since Chittagong is a coastal city, here inflow of sea-aerosol by high-speed monsoon wind is very prominent. During premonsoon, the occurrence of Norwester (a thunderstorm event) is prevalent, which kicks up the soil particles causing resuspension of PM10. Hence, a lower negative correlation between premonsoon PM10 and wind speed is observed. Additionally, the negative correlation between premonsoon PM10 and wind speed is noticeably low in Barisal. Barisal is in the southern part of the country, and it encounters the strongest surge of Norwester every year, leading to more frequent dust resuspension. This phenomenon might result in the poor negative correlation between premonsoon PM10 and wind speed in Barisal. The highest negative correlation between wind speed and PM (PM2.5 and PM10) is observed in Darus Salam, Dhaka and Gazipur, which are the zones of highest PM pollution. It suggests that PM variation is strongly modulated by wind speed in regions with higher pollution levels.

In order to understand the influence of wind direction on high PM loading, wind rose diagrams for eight stations of Bangladesh are plotted for the winter season in Fig. S2. It shows that wind flow from the north-westerly direction affects PM variation in the majority of stations in winter. In addition to wind rose diagrams, seasonal cross-correlation coefficients between wind direction and PM are also determined and presented in Table S4 of supplementary materials. Table S4 shows that, in addition to the northwestern wind, high PM loading of Gazipur and Barisal are also associated with the northeastern wind direction in winter (Table S4). The long-range transport of pollution in Gazipur may play a role in this regard. Gazipur is observed to have PM contribution from the eastern state of Tripura and the northeastern state of Assam/Meghalaya of India (Rana et al., 2016). Therefore, when the winter wind blows from these directions, it brings airmass high in PM loading to Gazipur. In contrast, in Barisal, primary pollution sources are local and are in the northern direction of Barisal station. As a result, the wind might influence PM variation from this direction. In Sylhet, where the southwestern wind is prevalent in winter, a moderate correlation between the southwestern wind component and PM2.5 is observed (Table S4). The predominant southwesterly wind conveys air from Dhaka-Gazipur and thus contributes to PM loading in the urban air of Sylhet.

 
3.3.2 Temperature

Temperature and PM are negatively correlated during winter and positively correlated during other seasons, as shown in Table 2 and Table 3. Here, it can be observed that the negative correlation is stronger for PM2.5 than PM10. This trend can be explained by the phenomena of brick kilns activities nearby the monitoring stations. The total number of legally registered brick kilns in Dhaka, Gazipur city, Chittagong, Sylhet, Rajshahi, and Barisal are 2295, 199, 1663, 257, 1176, and 366, respectively (FSFM, 2019). These brick kilns, located in the suburbs of major cities, operate fully during winter and primarily emit fine particles (PM2.5). Therefore, in winter, when the temperature is lower, the emission of PM2.5 to the atmosphere is more intense than that of PM10, leading to a stronger negative correlation between PM2.5 and temperature. Another noticeable fact is that, these correlations are obtained at a positive lag of 0–2 days, which indicates that it takes around 0–2 days to get a notable effect of temperature decrease on PM variation.

During the monsoon season, the highest positive correlation between temperature and PM is observed with no lag or no lead, indicating an immediate response of PM to the change of temperature. High-temperature conditions and a warm rainy season usually favor the formation of bioaerosols (Ahmed et al., 2013), and probably due to this reason, a positive correlation between temperature and PM during these seasons is observed. Since daytime temperature is more conducive to biogenic volatile organic compound (BVOC) formulation amidst forest environment (Liu et al., 2015; Perlwitz and Miller, 2010; Holmes, 2007), this might be responsible for a stronger (relative to other stations) positive correlation between temperature and PM in Sylhet.

Postmonsoon shows a weak positive correlation between temperature and PM for most stations, except Chittagong and Barisal, where a moderately strong positive correlation is observed (especially for PM10). Such a strong correlation might attribute to the coastal proximity of these stations. Sea-surface cools more slowly than land-surface, requiring more time to lower the ambient temperature (Jayamurugan et al., 2013). As a result, the climate of postmonsoon remains almost the same as that of monsoon in these two locations. Owing to this reason, the correlation coefficient of postmonsoon temperature and PM is possibly similar to that of monsoon season.

 
3.3.3 Relative humidity and rainfall duration

A strong negative correlation between relative humidity and PM occurs during premonsoon and postmonsoon. Although high humidity persists in monsoon, this season exerts a weaker correlation than premonsoon. In addition, the frequency and durability of precipitation events increase rapidly during the monsoon season, making the influence of relative humidity insignificant on PM. Besides, PM starts to reduce during premonsoon, and at the end of the season, the ambient PM level becomes moderate or insignificant. Therefore, the effect of relative humidity on PM removal in monsoon cannot become as prominent as in premonsoon since most of the PM removal has already been taken place in the premonsoon season.

On the contrary, the highest negative correlation between rainfall duration and PM occurs during monsoon (Table 2 and Table 3). Since monsoon season is characterized by heavy and persistent rainfall, it results in higher PM removal in this season by wet deposition, leading to a negative association between the pair.

 
3.3.4 Solar radiation

All stations show a positive correlation between solar radiation and PM throughout the year, as shown in Table 2 and Table 3. Here, it can be observed that the highest positive correlation mainly occurs during the monsoon season. Monsoon is characterized by high incident solar radiation, which favors frequent new particle formation (NPF) events. Additionally, condensation sink remains low in monsoon as PM concentration (pre-existing aerosol) becomes lower, leading to a favorable environment for NPF (Wonaschütz et al., 2015; Vehkamaki et al., 2004). This phenomenon may result in a stronger positive correlation between PM and solar radiation during monsoon. The second highest correlation is observed in the postmonsoon season. The environment conducive to NPF in monsoon still prevails in the postmonsoon. Additionally, after monsoon rainfall, vegetation and grass cover growth also favors particle formation events (Salma et al., 2021; Ganguly et al., 2006).

 
3.4 Multiple Non-linear Regression Analysis (MNLR)

The results of MNLR analysis for daily PM and meteorological variables are shown in Table 4. Besides, the details of selected seasonal MNLR models for PM2.5 and PM10 with adjusted r2 values, PRESS values, F-statistics, p-values and Durbin Watson values are presented in Table S6 and Table S7. Here, the model with the lowest PRESS and highest adjusted r2 value is selected for each case. For ease of explanation, wind speed, temperature, relative humidity, solar radiation, and rainfall duration are denoted as WS, T, RH, SR, and RD, respectively. Table S6 and Table S7 show that inclusion of lag and interaction terms notably increase the efficacy of the model. For example, prediction efficacy increases from 23% to 36%, when lead term of temperature and interaction terms are included into the model for winter PM2.5 prediction in Darus Salam station of Dhaka (CAMS-3). Similar results are observed for other cases. This finding is aligned with an earlier study that meteorological information of previous days can be used as important factors for the prediction of PM variation (Chen et al., 2018).

Table 4. Results of MNLR analysis for daily PM and meteorological variables

Table 4 shows that the adjusted r2 is higher during post-monsoon (70%–31%) and pre-monsoon (64%–29%), whereas it is moderate during winter (39%–11%) and monsoon (49%–13%). This indicates that PM variability is substantially influenced by meteorology throughout the year. However, the rest of the PM variation is modulated by other unknown factors that are not considered in our model. Examining the generated models for all cases (total 64 models), it is observed that the interaction term of wind speed and relative humidity (WS × RH) prevails throughout the year. On the other hand, T × RH is present in winter and postmonsoon, and SR × RH and SR × RD are prominent in premonsoon and monsoon. Therefore, solar radiation individually may not affect PM variation strongly, but it exerts a powerful influence in association with other parameters. The presence of interaction terms SR × RH and SR × RD during monsoon indirectly supports our hypothesis on NPF events that solar radiation accelerates particle formation in the presence of moisture during monsoon.

The most substantial meteorological influence is observed in Chittagong, followed by Barisal from the station-wise analysis. Strong meteorological influence on PM variation suggests that regulatory measures should focus on climatological influence, transboundary pollution and anthropogenic emission to sufficiently reduce PM pollution in these areas. On the other hand, lower meteorological influence is observed in Rajshahi and Dhaka (Sangsad Bhaban and Farmgate). This suggests that regulatory measures should primarily focus on anthropogenic emission reduction in these locations.

Using dominance analysis (Afrin et al., 2021; Tonidandel and LeBreton, 2011; Azen and Budescu, 2003), we determined the relative contribution of each meteorological parameter to the total variation of PM in each season to demonstrate how the relative contribution of meteorological parameters vary over seasons for the PM fractions. We performed this analysis for the MNLR models without the interaction terms, considering Chittagong station (CAMS-7) as a sample location. The result of this analysis is presented in Fig. 3. In general, temperature, relative humidity and wind seemed to have major contribution to total variation of both PM2.5 and PM10. Temperature showed higher contribution to PM2.5 variation than PM10 in all seasons, while the opposite is observed for relative humidity. A similar finding is observed in a previous study- based on Dhaka city, Bangladesh (Afrin et al., 2021). Fig. 3 also shows the change in contribution from meteorological parameter over seasons. For both PM fractions, temperature had the highest contribution in winter, whereas for relative humidity, it was in the postmonsoon season. During the other two seasons (i.e., premonsoon and monsoon), wind speed and direction dominated the overall PM variation. Solar radiation and rainfall duration seemed to be minimal contribution, although the contribution rose for PM10 during the monsoon season.

Fig. 3. The relative contribution of each meteorological factors in explaining the total PM2.5 and PM10 variability in CAMS-7 of Chittagong.Fig. 3. The relative contribution of each meteorological factors in explaining the total PM2.5 and PM10 variability in CAMS-7 of Chittagong.

Although presenting PM prediction models was not an objective of this study, we performed model validation of our MNLR models considering Chittagong station (CAMS-7) as a sample location for a recent dataset from 2019 to 2021. Fig. 4 and Fig. 5 represent the results of the validation process. The time series plots of observed and modeled PM2.5 and PM10 concentrations are shown in Figs. 4(a) and 4(b). These figures imply that the model prediction was better for PM10 compared to PM2.5. The season wise scatter plots (Fig. 5) of modeled versus observed PM concentrations indicate that model predictions vary over seasons, as observed in a previous study (Afrin et al., 2021). Overall, the time series plots of the actual and predicted PM (Fig. 4) and the moderate correlation coefficients (r = 0.33–0.76) between them (Fig. 5) indicate that these MNLR models can capture the seasonal variability of PM to some extent.

Fig. 4. Time series plot of actual and fitted (a) PM2.5 and (b) PM10 concentration in CAMS-7 of Chittagong.Fig. 4. Time series plot of actual and fitted (a) PM2.5 and (b) PM10 concentration in CAMS-7 of Chittagong.

 Fig. 5. Scatter plot of Fitted vs. actual (a–d) PM2.5 and (e–h) PM10 concentration for four seasons in CAMS-7 of Chittagong.Fig. 5. Scatter plot of Fitted vs. actual (a–d) PM2.5 and (e–h) PM10 concentration for four seasons in CAMS-7 of Chittagong.
 


4 CONCLUSION


 
This study analyzed temporal and spatial variation of meteorological influence on PM concentration over major urban areas of Bangladesh from 2013 to 2018. Annual and seasonal cross-correlation analyses were performed to identify major influential parameters on PM variation at different times of the year. In addition, seasonal multiple non-linear regression analysis (MNLR) using interaction and lagged terms was performed to understand the internal association of meteorological parameters with each other and their combined effect on particulate matter.

Annual cross-correlation analysis shows that wind speed, humidity, and rainfall are negatively correlated and solar radiation is positively correlated with PM concentration. Temperature shows a negative correlation with PM for most stations. However, the pair is positively correlated for stations located in the eastern part of the country. Wind flow coming from the northwestern direction is presumed to carry higher PM loading. Brick kilns and transboundary pollution are identified as potential aerosol sources coming from this direction.

Seasonal cross-correlation analysis shows that wind speed, humidity, and rainfall are negatively correlated, while solar radiation is positively correlated with PM throughout the year. However, for temperature and PM pair, a negative correlation is observed during winter and a positive correlation is observed during other seasons. The coastal proximity generation of monsoon sea-aerosol in Chittagong, premonsoon Norwester in Barisal, high plantation cover in Sylhet are presumed to influence PM concentration. Relative humidity and PM strongly correlate inversely during premonsoon and postmonsoon whereas, rainfall is prominent on PM during monsoon. The effect of rainfall is most substantial in Sylhet, the zone of maximum rainfall in the country. Finally, solar radiation and PM show maximum positive correlation during monsoon, indicating a combined influence of solar radiation and atmospheric moisture on new particle formation. MNLR analysis indicates that meteorological influence is the strongest in Chittagong and Barisal and the weakest in Dhaka and Rajshahi. The study indicates that while meteorological factors can influence the seasonal variability of PM, the extent of the influence can be modulated by local and regional factors. Therefore, a single meteorology-based PM prediction model may not be useful for different spatial scales, where topographic and other differences exist. This study is the first of its kind to identify spatial heterogeneity of meteorological influences on seasonal PM variation in different cities of Bangladesh.

 
ACKNOWLEDGMENTS


The authors would like to show gratitude to the Department of Environment (DoE) of the Ministry of Environment, Forest, and Climate Change (MoEFCC) for providing the ambient air quality and meteorological data. Graduate student funding from the CASR, BUET is also acknowledged.


REFERENCES


  1. Afrin, S., Islam, M.M., Ahmed, T. (2021). A meteorology based particulate matter prediction model for megacity Dhaka. Aerosol Air Qual. Res. 20, 200371. https://doi.org/10.4209/aaqr.​2020.07.0371

  2. Ahmed, J., Hossain, K.S., Bashar, M.A. (2013). Aeromycoflora of the Dhaka University campus. Bangladesh Journal of Botany 42, 273–278. https://doi.org/10.3329/bjb.v42i2.18029

  3. AirVisual (2019). World air quality report: Region & city PM2.5 ranking. (accessed 15 January 2022).

  4. Azen, R., Budescu, D.V. (2003). The dominance analysis approach for comparing predictors in multiple regression. Psychol. Methods 8, 129. https://doi.org/10.1037/1082-989X.8.2.129

  5. Azmi, S.Z., Latif, M.T., Ismail, A.S., Juneng, L., Jemain, A.A. (2010). Trend and status of air quality at three different monitoring stations in the Klang Valley, Malaysia. Air Qual. Atmos. Health 3, 53–64. https://doi.org/10.1007/s11869-009-0051-1

  6. Bai, K., Li, K., Guo, J., Cheng, W., Xu, X. (2022). Do more frequent temperature inversions aggravate haze pollution in China? Geophys. Res. Lett. 49, e2021GL096458. https://doi.org/​10.1029/2021GL096458

  7. Beckerman, B.S., Jerrett, M., Martin, R.V., van Donkelaar, A., Ross, Z., Burnett, R.T. (2013). Application of the deletion/substitution/addition algorithm to selecting land use regression models for interpolating air pollution measurements in California. Atmos. Environ. 77, 172–177. https://doi.org/10.1016/j.atmosenv.2013.04.024

  8. Begum, B.A., Hopke, P.A., Markwitz, A. (2016). An approach for quantitative estimation of long range transport of fine particulate matter entering Bangladesh. Int. J. Integr. Sci. Technol. 2, 34–38.

  9. Braun, R.A., Aghdam, M.A., Bañaga, P.A., Betito, G., Cambaliza, M.O., Cruz, M.T., Lorenzo, G.R., MacDonald, A.B., Simpas, J.B., Stahl, C., Sorooshian, A. (2020). Long-range aerosol transport and impacts on size-resolved aerosol composition in Metro Manila, Philippines. Atmos. Chem. Phys. 20, 2387–2405. https://doi.org/10.5194/acp-20-2387-2020

  10. CBS News (2019). The most polluted cities in the world, ranked. (accessed 14 January 2022).

  11. Chen, Z., Chen, D., Zhao, C., Kwan, M.P., Cai, J., Zhuang, Y., Zhao, B., Wang, X., Chen, B., Yang, J., Li, R. (2020). Influence of meteorological conditions on PM2.5 concentrations across China: A review of methodology and mechanism. Environ. Int. 139, 105558. https://doi.org/10.1016/j.​envint.2020.105558

  12. Chen, Z.Y., Zhang, T.H., Zhang, R., Zhu, Z.M., Ou, C.Q., Guo, Y. (2018). Estimating PM2.5 concentrations based on non-linear exposure-lag-response associations with aerosol optical depth and meteorological measures. Atmos. Environ. 173, 30–37. https://doi.org/10.1016/j.​atmosenv.2017.10.055

  13. Cheng, Y.H., Li, Y.S. (2010). Influences of traffic emissions and meteorological conditions on ambient PM10 and PM2.5 levels at a highway toll station. Aerosol Air Qual. Res. 10, 456–462. https://doi.org/10.4209/aaqr.2010.04.0025

  14. Datta, D., Chowdhury, S., Saha, A.K., Islam, M.M., Rahman, M.M. (2014). Tilted and horizontal solar radiation for 6 zones in Bangladesh. Int. J. Sci. Technol. Res. 3, 92–96.

  15. Department of Environment (DoE) (2017). National strategy for sustainable brick production in Bangladesh. Department of Environment, Ministry of Environemnt and Forests, Government of the People’s Republic of Bangladesh.

  16. Elminir, H.K. (2005). Dependence of urban air pollutants on meteorology. Sci. Total Environ. 350, 225–237. https://doi.org/10.1016/j.scitotenv.2005.01.043

  17. Frankfurt School of Finance and Management Bangladesh (FSFM) (2019). Bangladesh Brick Sector Roadmap 2019-2030: Climate and Clean Air Coalition to Reduce Short-Lived Climate Pollutants. FS-UNEP Collaborating Center for Climate and Sustainable Energy Finance, gGmbH, Adickesallee 32–34, 60322 Frankfurt am Main, Germany.

  18. Fuzzi, S., Baltensperger, U., Carslaw, K., Decesari, S., Facchini, M.C., Fowler, D., Koren, I., Langford, B., Lohmann, U., Nemitz, E., Pandis, S., Riipinen, I., Rudich, Y., CSchaap, M., Slowik, J.G., Spracklen, D.V., Vignati, E., Wild, M., Williams, M., Gilardoni, S. (2015). Particulate matter, air quality and climate: lessons learned and future needs. Atmos. Chem. Phys. 15, 8217–8299. https://doi.org/10.5194/acp-15-8217-2015M

  19. Ganguly, D., Jayaraman, A., Gadhavi, H. (2006). Physical and optical properties of aerosols over an urban location in western India: Seasonal variabilities. J. Geophys. Res. 111, D24206. https://doi.org/10.1029/2006JD007392

  20. Guo, J., Lou, M., Miao, Y., Wang, Y., Zeng, Z., Liu, H., He, J., Xu, H., Wang, F., Min, M., Zhai, P. (2017). Trans-Pacific transport of dust aerosols from East Asia: Insights gained from multiple observations and modeling. Environ. Pollut. 230, 1030–1039. https://doi.org/10.1016/j.envpol.​2017.07.062

  21. Guo, J., Chen, X, Su, T., Liu, L., Zheng, Y., Chen, D., Li, J., Xu, H., Lv, Y., He, B., Li, Y., Hu, X.M., Ding, A., Zhai, P. (2020). The climatology of lower tropospheric temperature inversions in China from radiosonde measurements: roles of black carbon, local meteorology, and large-scale subsidence. J. Clim. 33, 9327–9350. https://doi.org/10.1175/JCLI-D-19-0278.1

  22. Hajiloo, F., Hamzeh, S., Gheysari, M. (2019). Impact assessment of meteorological and environmental parameters on PM2.5 concentrations using remote sensing data and GWR analysis (case study of Tehran). Environ. Sci. Pollut. Res. 26, 24331–24345. https://doi.org/10.1007/s11356-018-1277-y

  23. Hamed, A., Korhonen, H., Sihto, S.L., Joutsensaari, J., Järvinen, H., Petäjä, T., Arnold, F., Nieminen, T., Kulmala, M., Smith, J.N., Lehtinen, K.E.J., Laaksonen, A. (2011). The role of relative humidity in continental new particle formation. J. Geophys. Res. 116, D03202. https://doi.org/10.1029/​2010JD014186

  24. Hassan, H., Latif, M.T., Juneng, L., Amil, N., Khan, M.F., Yik, D.J., Abdullah, N.A. (2020). Interaction of PM10 concentrations with local and synoptic meteorological conditions at different temporal scales. Atmos. Res. 241, 104975. https://doi.org/10.1016/j.atmosres.2020.104975

  25. Holmes, N. (2007). A review of particle formation events and growth in the atmosphere in the various environments and discussion of mechanistic implications. Atmos. Environ. 41, 2183–2201. https://doi.org/10.1016/j.atmosenv.2006.10.058

  26. Hossen, M.A., Hoque, A. (2018). Variation of ambient air quality scenario in Chittagong City: A case study of air pollution. J. Civ. Constr. Environ. Eng. 3, 10–16. https://doi.org/10.11648/j.​jccee.20180301.13

  27. Ilias, A., Russell, E., Loikith, P.C. (2020). How Meteorology can Determine Ambient Air Pollution Impact on Population of Concerns. AGU Fall Meeting Abstracts 2020, A150–0012. Bibcode: 2020AGUFMA150.0012I

  28. Islam, M., Afrin, S., Ahmed, T., Ali, M.A. (2015). Meteorological and seasonal influences in ambient air quality parameters of Dhaka City. J. Civ. Eng. 43, 67–77.

  29. Islam, M.M., Wathore, R., Zerriffi, H., Marshall, J.D., Bailis, R., Grieshop, A.P. (2021). In-use emissions from biomass and LPG stoves measured during a large, multi-year cookstove intervention study in rural India. Sci. Total Environ. 758, 143698. https://doi.org/10.1016/j.​scitotenv.2020.143698

  30. Islam, M.M., Meskhidze, N., Satheesh, A.R., Petters, M.D. (2022a). Turbulent flux measurements of the near‐surface and residual‐layer small particle events. J. Geophys. Res. 127, e2021JD036289. https://doi.org/10.1029/2021JD036289

  31. Islam, M.M., Neyestani, S.E., Saleh, R., Grieshop, A.P. (2022b). Quantifying brown carbon light absorption in real-world biofuel combustion emissions. Aerosol Sci. Technol. 56, 502–516. https://doi.org/10.1080/02786826.2022.2051425

  32. Islam, M.M., Wathore, R., Zerriffi, H., Marshall, J.D., Bailis, R., Grieshop, A.P. (2022). Assessing the effects of stove use patterns and kitchen chimneys on indoor air quality during a multiyear cookstove randomized control trial in rural India. Environ. Sci. Technol. 56, 8326–8337. https://doi.org/10.1021/acs.est.1c07571

  33. Islam, N., Toha, T.R., Islam, M.M., Ahmed, T. (2022d). The association between particulate matter concentration and meteorological parameters in Dhaka, Bangladesh. Meteorol. Atmos. Phys. 134, 64. https://doi.org/10.1007/s00703-022-00898-2

  34. Islam, S. (2018). The government's 10 mega projects are gaining importance ahead of the elections. shorturl.at/fwHN7 (accessed 16 January 2022).

  35. Jacobson, M.Z., Kaufman, Y.J. (2006). Wind reduction by aerosol particles. Geophys. Res. Lett. 33, L24814. https://doi.org/10.1029/2006GL027838

  36. Jayamurugan, R., Kumaravel, B., Palanivelraja, S., Chockalingam, M.P. (2013). Influence of temperature, relative humidity and seasonal variability on ambient air quality in a coastal urban area. International J. Atmos. Sci. 2013, 264046. https://doi.org/10.1155/2013/264046

  37. Jung, I., Kumar, S., John, K. (2002). Impact of meteorology on the fine particulate matter distribution in Central and Southeastern Ohio. Proceedings of the American Meteorological Society: 12th Joint Conference on Applications of Air Pollution Meteorology with the Air and Waste Management Association, Norfolk. 

  38. Kim, G., Park, S., Kwak, D. (2020). Is it possible to predict the concentration of natural volatile organic compounds in forest atmosphere? Int. J. Environ. Res. Public Health 17, 7875. https://doi.org/10.3390/ijerph17217875

  39. Li, X., Ma, Y., Wang, Y., Liu, N., Hong, Y. (2017a). Temporal and spatial analyses of particulate matter (PM10 and PM2.5) and its relationship with meteorological parameters over an urban city in northeast China. Atmos. Res. 198, 185–193. https://doi.org/10.1016/j.atmosres.2017.​08.023

  40. Li, Z., Guo, J., Ding, A., Liao, H., Liu, J., Sun, Y., Wang, T., Xue, H., Zhang, H., Zhu, B. (2017b). Aerosol and boundary-layer interactions and impact on air quality. Natl. Sci. Rev. 4, 810–833. https://doi.org/10.1093/nsr/nwx117

  41. Li, X., Song, H., Zhai, S., Lu, S., Kong, Y., Xia, H., Zhao, H. (2019). Particulate matter pollution in Chinese cities: Areal-temporal variations and their relationships with meteorological conditions (2015–2017). Environ. Pollut. 246, 11–18. https://doi.org/10.1016/j.envpol.2018.11.103

  42. Liu, J., Cui, S. (2014). Meteorological influences on seasonal variation of fine particulate matter in cities over Southern Ontario, Canada. Adv. Meteorol. 2014, 169476. https://doi.org/10.1155/​2014/169476

  43. Liu, X., Yu, X., Zhang, Z. (2015). PM2.5 concentration differences between various forest types and its correlation with forest structure. Atmosphere 6, 1801–1815. https://doi.org/10.3390/​atmos6111801

  44. Liu, Y., Franklin, M., Kahn, R., Koutrakis, P. (2007). Using aerosol optical thickness to predict ground-level PM2.5 concentrations in the St. Louis area: A comparison between MISR and MODIS. Remote Sens. Environ. 107, 33–44. https://doi.org/10.1016/j.rse.2006.05.022

  45. Mahapatra, P.S., Sinha, P.R., Boopathy, R., Das, T., Mohanty, S., Sahu, S.C., Gurjar, B.R. (2018). Seasonal progression of atmospheric particulate matter over an urban coastal region in peninsular India: Role of local meteorology and long-range transport. Atmos. Res. 199, 145–158. https://doi.org/10.1016/j.atmosres.2017.09.001

  46. Mancilla, Y., Herckes, P., Fraser, M.P., Mendoza, A. (2015). Secondary organic aerosol contributions to PM2.5 in Monterrey, Mexico: Temporal and seasonal variation. Atmos. Res. 153, 348–359. https://doi.org/10.1016/j.atmosres.2014.09.009

  47. Munir, S., Habeebullah, T.M., Seroji, A.R., Morsy, E.A., Mohammed, A.M.F., Saud, W.A., Abdou, A.E.A., Awad, A.H. (2013). Modeling particulate matter concentrations in Makkah, applying a statistical modeling approach. Aerosol Air Qual. Res. 13, 901–910. https://doi.org/10.4209/​aaqr.2012.11.0314

  48. Ott, W.R. (1990). A physical explanation of the lognormality of pollutant concentrations. J. Air Waste Manage. Assoc. 40, 1378–1383. https://doi.org/10.1080/10473289.1990.10466789

  49. Perlwitz, J., Miller, R.L. (2010). Cloud cover increase with increasing aerosol absorptivity: A counterexample to the conventional semidirect aerosol effect. J. Geophys. Res. 115, D08203. https://doi.org/10.1029/2009JD012637

  50. Rahman, M.M., Mahamud, S., Thurston, G.D. (2019). Recent spatial gradients and time trends in Dhaka, Bangladesh, air pollution and their human health implications. J. Air Waste Manage. Assoc. 69, 478–501. https://doi.org/10.1080/10962247.2018.1548388

  51. Rana, M.M., Khan, M.H. (2020). Trend characteristics of atmospheric particulate matters in major urban areas of Bangladesh. Asian J. Atmos. Environ. 14, 47–61. https://doi.org/10.5572/ajae.​2020.14.1.047

  52. Rana, M.M., Sulaiman, N., Sivertsen, B., Khan, M.F., Nasreen, S. (2016). Trends in atmospheric particulate matter in Dhaka, Bangladesh, and the vicinity. Environ. Sci. Pollut. Res. 23, 17393–17403. https://doi.org/10.1007/s11356-016-6950-4

  53. Salma, I., Thén, W., Aalto, P., Kerminen, V.M., Kern, A., Barcza, Z., Kulmala, M., Petala, T., Kulmala, M. (2021). Influence of vegetation on occurrence and time distributions of regional new aerosol particle formation and growth. Atmos. Chem. Phys. 21, 2861–2880. https://doi.org/10.5194/​acp-21-2861-2021

  54. Seinfeld, J.H., Pandis, S.N. (2016). Atmospheric Chemistry and Physics: From Air Pollution to Climate Change, 1st ed. John Wiley & Sons Inc., ProQuest Ebook Central.

  55. Shahriar, S.A., Kayes, I., Hasan, K., Salam, M.A., Chowdhury, S. (2020). Applicability of machine learning in modeling of atmospheric particle pollution in Bangladesh. Air Qual. Atmos. Health 13, 1247–1256. https://doi.org/10.1007/s11869-020-00878-8

  56. Singh, B.P., Singh, D., Kumar, K., Jain, V.K. (2021). Study of seasonal variation of PM2.5 concentration associated with meteorological parameters at residential sites in Delhi, India. J. Atmos. Chem. 78, 161–176. https://doi.org/10.1007/s10874-021-09419-8

  57. Song, W., Jia, H., Huang, J., Zhang, Y. (2014). Song, W., Jia, H., Huang, J., Zhang, Y., 2014. A satellite-based geographically weighted regression model for regional PM2.5 estimation over the Pearl River Delta region in China. Remote Sens. Environ. 154, 1–7. https://doi.org/10.1016/​j.rse.2014.08.008

  58. The Daily Star (2018). Road to the 11th Parliamentary Election. (accessed 16 January 2022).

  59. Tian, J., Chen, D. (2010). Tian, J., Chen, D. (2010). A semi-empirical model for predicting hourly ground-level fine particulate matter (PM2.5) concentration in southern Ontario from satellite remote sensing and ground-based meteorological measurements. Remote Sens. Environ. 114, 221–229. https://doi.org/10.1016/j.rse.2009.09.011

  60. Tonidandel, S., LeBreton, J.M. (2011). Relative importance analysis: A useful supplement to regression analysis. J. Bus. Psychol. 26, 1–9. https://doi.org/10.1007/s10869-010-9204-3

  61. Trivedi, D.K., Ali, K., Beig, G. (2014). Impact of meteorological parameters on the development of fine and coarse particles over Delhi. Sci. Total Environ. 478, 175–183. https://doi.org/10.1016/​j.scitotenv.2014.01.101

  62. Vehkamaki, H., Dal Maso, M., Hussein, T., Flanagan, R., Hyvarinen, A., Lauros, J., Merikanto, J., Monkkonen, P., Pihlatie, M., Salminen, K., Sogacheva, L., Thum, T., Ruuskanen, T.M., Keronen, P., Aalto, P.P., Hari, P., Lehtinen, K.E.J., Rannik, U., Kulmala, M. (2004). Atmospheric particle formation events at Varrio measurement station in Finnish Lapland 1998–2002. Atmos. Chem. Phys. 4, 2015–2023. https://doi.org/10.5194/acp-4-2015-2004

  63. Wang, D., Zhou, B., Fu, Q., Zhao, Q., Zhang, Q., Chen, J., Li, J. (2016). Intense secondary aerosol formation due to strong atmospheric photochemical reactions in summer: Observations at a rural site in eastern Yangtze River Delta of China. Sci. Total Environ. 571, 1454–1466. https://doi.org/10.1016/j.scitotenv.2016.06.212

  64. Wang, J., Wang, Y., Liu, H., Yang, Y., Zhang, X., Li, Y., Zhang, Y., Deng, G. (2013). Diagnostic identification of the impact of meteorological conditions on PM2.5 concentrations in Beijing. Atmos. Environ. 81, 158–165. https://doi.org/10.1016/j.atmosenv.2013.08.033

  65. World Health Organization (WHO) (2016). Ambient air pollution: A global assessment of exposure and burden of disease. World Health Organization, Geneva, Switzerland.

  66. Won, W.S., Oh, R., Lee, W., Ku, S., Su, P.C., Yoon, Y.J. (2021). Hygroscopic properties of particulate matter and effects of their interactions with weather on visibility. Sci. Rep. 11, 1–12. https://doi.org/10.1038/s41598-021-95834-6

  67. Wonaschütz, A., Demattio, A., Wagner, R., Burkart, J., Zíková, N., Vodička, P., Hitzenberger, R. (2015). Seasonality of new particle formation in Vienna, Austria–Influence of air mass origin and aerosol chemical composition. Atmos. Environ. 118, 118–126. https://doi.org/10.1016/j.​atmosenv.2015.07.035

  68. Wu, W., Zha, Y., Zhang, J., Gao, J., He, J. (2014). A temperature inversion-induced air pollution process as analyzed from Mie LiDAR data. Sci. Total Environ. 479–480, 102–108. https://doi.org/10.1016/j.scitotenv.2014.01.112

  69. Xu, T., Song, Y., Liu, M., Cai, X., Zhang, H., Guo, J., Zhu, T. (2019). Temperature inversions in severe polluted days derived from radiosonde data in North China from 2011 to 2016. Sci. Total Environ. 647, 1011–1020. https://doi.org/10.1016/j.scitotenv.2018.08.088

  70. Young, B.N., Clark, M.L., Rajkumar, S., Benka‐Coker, M. L., B., A., B., Peel, J.L. (2019). Exposure to household air pollution from biomass cookstoves and blood pressure among women in rural Honduras: A cross‐sectional study. Indoor Air 29, 130–142. https://doi.org/10.1111/ina.12507

  71. Zhang, Y., Guo, J., Yang, Y., Wang, Y., Yim, S.H.L. (2020). Vertical wind shear modulates particulate matter pollutions: A perspective from radar wind profiler observations in Beijing, China. Remote Sens. 12, 546. https://doi.org/https://doi.org/10.3390/rs12030546


Share this article with your colleagues 

 

Subscribe to our Newsletter 

Aerosol and Air Quality Research has published over 2,000 peer-reviewed articles. Enter your email address to receive latest updates and research articles to your inbox every second week.

7.3
2022CiteScore
 
 
77st percentile
Powered by
Scopus
 
   SCImago Journal & Country Rank

2022 Impact Factor: 4.0
5-Year Impact Factor: 3.4

Aerosol and Air Quality Research partners with Publons

CLOCKSS system has permission to ingest, preserve, and serve this Archival Unit
CLOCKSS system has permission to ingest, preserve, and serve this Archival Unit

Aerosol and Air Quality Research (AAQR) is an independently-run non-profit journal that promotes submissions of high-quality research and strives to be one of the leading aerosol and air quality open-access journals in the world. We use cookies on this website to personalize content to improve your user experience and analyze our traffic. By using this site you agree to its use of cookies.