Temporal and Spatial Variations in Ambient Air Quality during 1996 – 2009 in Bangkok , Thailand

The severe air pollution in Bangkok (BKK) is an important issue in Thailand. The Bangkok air quality and meteorological data used in this study were collected by the Pollution Control Department of the Ministry of Natural Resources and Environment, Thailand, during 1996–2009. Measurements of hourly air quality and meteorological data were derived from 10 residential and seven roadside sites. Pearson’s chi-square cross tabulation statistics show that the 24-hour mean PM10 concentrations at both roadside and residential sites were to be significantly higher than the Thai National ambient air quality standards (NAAQS) and World Health Organization (WHO) guidelines. The daily 1-hour maximum O3 (O3-1hr) concentration was higher than the Thai NAAQS at both sites. However, the concentrations of 8-hour time-weighting average of CO (CO8hr) were lower than the Thai NAAQS at both sites. The 24-hour average SO2 concentration and the daily 1-hour mean concentration of NO2 were higher than the WHO guidelines, but complied with the Thai NAAQS at both sites. A stepwise multiple linear regression model was used to analyze the significant factors affecting PM10, CO8hr, O3-1hr, NO2, and SO2 levels at both sites, and the results showed a decreased association with meteorological parameters and an increased association with the area studied and seasons. In contrast, O3-1hr levels exhibited a decreased association with the area studied. This study found that traffic emissions are the major factor causing the spatial variation in air pollutants in BKK, Thailand, while the meteorological parameters might be the main factors that affect the temporal variations. Our findings show the heterogeneously spatiotemporal characteristics of air pollution in BKK, and can be used to help mitigate this problem.


INTRODUCTION
The air quality in the Bangkok Metropolitan Region (BMR) is considered to be the worst area in Thailand (PCD, 2006(PCD, , 2007(PCD, , 2010)).Air pollution in Bangkok (BKK) is mainly due to automobile and industry sources (Jinsart et al., 2002;Chuersuwan et al., 2008).The total number of vehicles registered to the Department of Land Transport (DLT) in 2007 was 5. 57 million, and in 200957 million, and in , it was 6.22 million (DLT, 200757 million, and in , 2009)).The road network in BKK has not been adequately expanded, which causes traffic jams and forces vehicles to move at slower speed.Road traffic was a significant source of air pollution in BKK, and the number of vehicles on the road was increasing every year.The DLT reported that the number of vehicles in the capital city in 2009 was nearly doubled (6,103,719 vs. 3,549,082) by comparing with that in 1996 (DLT, 2009).This had led to increase the particulate matter (PM) emissions into the ambient, especially those of smaller size particles (PM with aerodynamic diameter less than 10 microns or PM 10 ) emitted from the diesel engine-powered buses and trucks.The air pollution emitted from the roads was identified as a public health risk in BKK because the recorded level was consistently higher than the accepted standard (Ostro et al., 1999;BMA, 2003BMA, , 2008)).Traffic emissions were considered to be a major problem because the emissions contain many toxic substances, e.g., PM 10 , CO, ozone (O 3 ), NO x , SO x , and aromatic hydrocarbons (Jinsart et al., 2002;Kim Oanh and Zhang, 2004;Ruchirawat et al., 2005;Chuersuwan et al., 2008).
Systematic ambient air quality measurements in BKK, Thailand began in 1996 (Ostro et al., 1999).Since September of 1996, the continuous air quality data were available for PM 10 , CO, O 3 , NO 2 , and SO 2 , as well as the supplementary meteorological parameters such as wind speed (WS), wind direction (WD), temperature, and relative humidity (RH).The monitoring data was automatically restored and recorded as database.The database was expected to meet the Thai National ambient air quality standards (NAAQS) of 2007 (Gazette, 2007).However, only few studies had discussed the long-term trends or spatiotemporal variations for air quality in BBK (Jinsart et al., 2002;Kim Oanh and Zhang, 2004;Chuersuwan et al., 2008).
Based on the evidence, the above-mentioned analyses were needed to be the bases for control strategies or other policies making by related authorities.The aims of this study were three fold: (1) to investigate the trends of air pollutants during 1996-2009, (2) to determine the temporal and spatial variations of the air quality data, and (3) to establish a stepwise multiple linear regression model to depict these data adopted from both residential and roadside sites in BKK.

Descriptions of Locations and Seasons
Bangkok (BKK) is the capital city of Thailand and ranks 68 th in size out of the country's 76 provinces with an area of 1,568 km 2 .The city is split into 50 districts and 154 subdistricts.Currently, there are approximately 8,160,522 registered residents in BKK (DOPA, 2006).However, the actual number of people who stay in the city can reach 15 million.The BKK has a population density of 3,637 persons per km 2 .Land use in BKK consists of 700 km 2 urbanised area, 379.0 km 2 of agricultural land (23.6% of the land), and 453.5 km 2 area for commercial, industrial and government uses (29.6% of the land).BKK is situated on the low flat plain of the Chao Phraya River, which extends to the Gulf of Thailand.BKK is located 13°45' north in latitude, and 100°31' east in longitude (Kim Oanh and Zhang, 2004;Chuersuwan et al., 2008).The elevation is approximately 2.3 meters above sea level.The area surrounding BKK and the nearby provinces comprise a series of plains and river deltas that lead into the Bay of BKK approximately 30 km south of the city centre.
BKK has a tropical monsoon climate with two main seasons: wet or rainy and dry.The wet season extends from 16 May to 15 October.The dry season can be classified into two periods.The first period (16 October-15 February) is characterised by the mild weather of the winter monsoon and is known as the local winter.The second period (16 February-15 May) is known as the local summer and is extremely hot (Ostro et al., 1999).Meteorological statistics from 1990-1999 show the annual average temperature of BKK to be 25-33°C, with an average WS of 1.3 m/s and a high RH year-round (Kim Oanh, 2004).

Air Quality and Meteorological Data
The BKK air quality and meteorological data used in this study were collected by the Pollution Control Department of the Ministry of Natural Resources and Environment, Bangkok, Thailand, from 1996to 2009(PCD, 2010).Hourly-based air pollution and meteorological data were collected from 17 air-monitoring stations, which included 10 ambientresidential and 7 ambient-roadside sites (Fig. 1).Detail information for selected sites was also shown in Fig. 1.Residential sites are the representative sites for residential areas, where are densely populated areas with a population of more than 50,000 persons.The residential monitoring sites are located approximately 100 m to 2 km far from a major road.However, the roadside sites are defined as closing to a major road approximately 10-50 m.
PM 10 , CO, O 3 , NO 2 , SO 2 , temperature, RH, WS, and WD are monitored using Beta Ray absorption, non-dispersive infrared detection, ultraviolet (UV) absorption photometry, chemiluminescence, UV-fluorescence, a thermistor, a thinfilm capacitor, a cup propeller, and potentiometer wind vanes, respectively.These instruments are connected to a data logger that transmits the hourly average values from a modem to a central computer situated in the building of the station.The instruments are calibrated every month.For NO 2 (USEPA Federal Reference Method RFNA-0691-082), measured by using chemiluminescence monitors equipped with molybdenum/photolytic convertors, within 10% uncertainty of measurements had been shown in the literature (Dunlea et al., 2007;Steibacher et al., 2007).In this study, we especially focused on eight-hour time-weighting average of CO (CO 8hr ) and daily 1-hour maximum O 3 (O 3-1hr ) because these two pollutants represent the severe conditions in rush hours and the potential damages to people in BKK.

Statistical Analysis
The SAS version 9.2 statistical software for windows (SAS Institute Inc., NC, USA) was used to conduct a descriptive analysis, and two independent sample t-tests were used to compare air pollution concentrations and meteorological parameters between roadside and residential sites.An one-way analysis of variance (ANOVA) test was used to compare spatial and temporal variations of pooled data of PM 10 , CO 8hr , O 3-1hr, NO 2 , and SO 2 between sites and seasons at roadside and residential sites in the BKK during 1996-2009.Pearson's chi-square (χ 2 ) cross tabulation statistic was used to test correlations between the reported values of PM 10 , CO 8hr , O 3-1hr , NO 2 , and SO 2 that exceeded the thresholds of the Thai NAAQS of 2007 (Gazette, 2007) and the WHO guidelines (WHO, 2005), with weekday and weekend days, seasons and areas studied.Additionally, this statistic was used to test the relationship between air pollutants in different areas and the correlations between air pollutants and meteorological factors.We also developed a stepwise multiple linear regression model to determine the associations among air pollution concentrations, meteorological factors, areas studied, and seasonal parameters at roadside and residential sites in BKK in 2009.The general regression model can be expressed as: where b 0 is the regression constant and b i and b j are the regression coefficients of continuous (X i ) and category (Z j )  (2006,2007,2010)).
variables, respectively.The values of the constants and the coefficients can be determined by using the least-squares method.Finally, e shown in the regression equation is the error term.The significance level of the constant and coefficients are statistically tested using the t distribution.
A generally used measure of the goodness of fit of a linear model is r 2 (Akpinar et al., 2008).PM 10 , CO 8hr , O 3-1hr , NO 2 , and SO 2 concentrations were considered to be the dependent variables, while the meteorological parameters, such as temperature, RH, and WS, were considered to be the independent and continuous variables.Our multiple linear regression model also take into account the spatial variations, such as roadside and residential (reference) sites, and temporal variations, such as the winter season, summer season, and rainy season (reference), as independent and category variables.

Air Quality and Standards
The daily mean concentrations of PM 10 at roadside and residential sites were found to be significantly higher than the Thai NAAQS since 2007 and the WHO guidelines since 2005 (Table 1).The daily maximum concentration of O 3-1hr was higher than the Thai NAAQS in both sites.Additionally, the daily mean concentrations of CO 8hr was lower than the Thai NAAQS in both sites and the daily mean concentration of NO 2 and SO 2 were higher than the WHO guidelines, but complied with Thai NAAQS in both sites.According to the WHO guidelines, the daily (24-hour) mean concentration threshold of PM 10 was 50 µg/m 3 .The overall recorded levels exceeded this threshold in 3918 of 4900 cases in roadside sites (80.0%) and in 2216 of 4889 cases in residential sites (45.3%) during 1996-2009.For NO 2 , results also showed the whole records exceeded the threshold (WHO guideline: 106 ppb) in 2072 (42.3%) and 695 (14.2%) cases in roadside and residential sites, respectively.However, only 443 of 4900 (9.0%) cases and 143 of 4889 (2.9%) cases were confirmed to exceed the Thai NAAQS (120 µg/m 3 ) for roadside and residential sites, respectively.The daily one-hour mean concentration threshold of SO 2 was 8 ppb, which was exceeded in 1786 of 4892 cases in roadside sites (36.5%) and in 387 of 5114 cases in residential sites (7.6%).Jinsart et al. (2002) compared the daily average PM 10 in BKK with the Thai NAAQS and showed that the PM 10 exceedance rate during April 1997 to March 2000 from 27.3% declines to 10.5% for roadside sites and from 8.6% decreases to 2.7% for general sites, respectively.Jinsart et al. (2002) also indicated that the mass concentration ratio of PM 2.5 (PM with aerodynamic diameter less than 2.5 μm) to PM 10 in high polluted area (0.8 ± 0.08) is higher than that in low-polluted area (0.65 ± 0.04).By comparing with previous study (Jinsart et al., 2002), our data is longer (14 yrs. vs. 3 yrs.)and wider (17 sites vs. 8 sites) than previous study and shows more representative in BKK.We also provided the new WHO guideline and showed that more exceedance rates were considered.

Spatial Trends
The spatial trends of the overall average concentrations   (c)-2(j)), indicated that they had an inverse trend (higher concentrations and lower variations in winter) at both sites.In BKK, the overall and seasonal averages of air pollutants at roadside sites higher than that at residential sites were summarized, yet the seasonal effects might be significant than spatial effect on inter-roadside and inter-residential sites air pollutants variations.More details for site-specific variations with higher spatial resolutions in BKK similar to the previous study (Kim Oanh and Zhang, 2004) did not discussed here.However, the different temporal variations in BKK were discussed later in this study.

Long-Term Trends
Table 2 summarises the fourteen-year average and seasonspecific concentrations of air pollutants and meteorological parameters during 1996-2009 between roadside and residential sites.The overall average concentrations of PM 10 , CO 8hr , NO 2 , and SO 2 in roadside sites were higher than those in residential sites (74.5 ± 31.6 µg/m 3 , 1.8 ± 0.7 ppm, 34.3 ± 11.9 ppb, and 7.7 ± 3.9 ppb vs. 55.0 ± 25.4 µg/m 3 , 0.8 ± 0.4 ppm, 22.0 ± 9.2 ppb, and 5.1 ± 2.0 ppb).However, the overall average concentration of O 3-1hr in roadside sites was lower than those in residential sites (37.6 ± 19.9 ppb vs. 61.0± 29.1 ppb).The overall average concentration of O 3-1hr was lower than Thai NAAQS in both sites.The average concentrations of PM 10 , CO 8hr , NO 2 , and SO 2 in roadside and residential sites were comparable in magnitude to those previously reported by the PCD (PCD, 2007) for the same air monitoring sites.Since 1993, the Office of National Energy Policy has formulated a 10-year policy for fuel improvement (Thavisin, 2001); therefore, the level of air pollution is getting better.
Since 1996, the annual average concentrations of PM 10 , CO 8hr , and SO 2 were decreased; however, the NO 2 and O 3-1hr levels were slightly increased (Figs. 3(a) and (b)).Decreasing the NO 2 and O 3-1hr levels is currently impractical because the Thai government promotes the use of gasohol fuel in cars and motorbikes.PM 10 and CO 8hr levels at roadsides and residential sites were slightly increased compared to previous years (Fig. 3(a)) due to construction activities to improve road conditions in 2004, which resulted in traffic congestion in BKK, leading to the accumulation of air pollutants emitted from vehicles (PCD, 2006).Ozone is a secondary pollutant formed by this reaction between volatile organic compounds (VOCs) and nitrogen oxides in the presence of sunlight.There are many factors that make controlling the formation of ozone difficult in BKK.The various measures in place to reduce ozone levels do not reduce the amount of ozone down to the standard criteria.The main sources of ozone in the atmosphere were confirmed from VOCs produced by gas stations, motorcycles and 2-stroke motor tricycles, car spray paint garages, metal plating factories, electronics factories that use solvents for cleaning, and barbecue grills (CSD, 2008).
Additionally, the PCD measured the air pollution for gasohol cars and motorcycles during [2003][2004].The result of air quality tests from the five air measurement stations in BKK found that the THC levels decreased 5-25%, the CO levels decreased 15-30%, and the NO x levels slightly increased.These results were consistent with those of a Brazilian study demonstrating that THC levels were reduced by 12%, CO levels were reduced by 32%, and NO x levels slightly increased during the same period (PCD, 2007).The average of the air pollution concentrations and meteorological parameters were found to be significantly different between the studied sites (p < 0.0001), indicating that high traffic sites had higher air pollution levels than residential sites.Our results were similar with previous reports (Grivas et al., 2004; Kim Oanh et al., 2006; Chuersuwan et  al., 2008; Azmi et al., 2010).
An one-way ANOVA was applied to the data to elucidate temporal and spatial variations.Using a p-value of < 0.0001, the average concentrations of air pollutants and meteorology parameters during the three seasons at roadside and residential sites were found to differ significantly from each other.At the urban traffic areas, there was an intensive seasonal variability, with higher concentrations of PM 10 , CO 8hr , O 3-1hr , NO 2 , and SO 2 during the winter months.This finding is similar to that of previous studies (Tsai et al., 2008;Juneng et al., 2009;Azmi et al., 2010;Rose et al., 2011;Vlachogianni et al., 2011).
Figs. 3(c) and 3(d) shows the average monthly concentrations of PM 10 , CO 8hr , O 3-1hr, NO 2 , and SO 2 during the 14-year monitored period.The concentrations of PM 10 , CO 8hr , NO 2 , and SO 2 exhibited the same seasonal pattern at roadside and residential sites, reaching maximum during the winter period and minimum during the rainy period.Note that during the dry season, a weak influence from the highpressure ridge could be felt in BKK, which might reduce pollutant dispersion and cause a lack of rain scavenging (Kim  Oanh et al., 2006; Chuersuwan et al., 2008).This pollutant variation was associated with the seasonal variation of the relevant meteorological dispersion, similar to the more common occurrence of stable and extremely stable low wind speed situations during the winter period (Vlachogianni et al., 2011).The O 3-1hr concentrations had a different seasonal pattern from the PM 10 , CO 8hr , NO 2 , and SO 2 concentrations.The maximum O 3-1hr concentrations occurred during the winter, and a second peak occurred during the summer.The lowest O 3-1hr concentrations occurred during the rainy  season at both roadside and residential sites due to the influence of sunlight on photochemical reactions.
3(e) and 3(f) shows that the daily concentration variations of PM 10 , CO 8hr , NO 2 , and SO 2 in roadside sites are higher than those in residential sites, which is the opposite of the ozone variations.The diurnal patterns of PM 10 , CO 8hr , and NO 2 display double peaks occurring during the morning rush hour traffic peak around 07:00 to 09:00 and during the evening rush hour after 19:00 in both sites.In contrast, the peak ozone concentration occurred 14:00 in the afternoon, in accordance with the diurnal cycle of solar radiation.This is because sunlight is the most important factor affecting the photochemical reactions that produce ozone.These results did not reflect the results of the other gases.The temperature inversion emerging near the road surface at night also helped to augment and prolong the evening peaks, which persisted until the traffic flow subsided late at night, causing the pollutant concentrations to decrease accordingly.
Moreover, atmospheric stability and WS factors influencing PM 10 , CO 8hr , NO 2 , and SO 2 concentrations were likely to exhibit spatial and temporal variations.For instance, local contributions around roadside sites accounted for approximately one-third of the contributions in residential sites, most likely from high traffic and surrounding activities.The recorded daily means of air pollution concentrations were found to differ significantly between roadside and residential sites (p < 0.0001), indicating that, in addition to the negative influence of prevailing regional meteorological conditions, traffic emission patterns affected the entire area (Grivas et al., 2004; Tsai et al., 2008; Kim Oanh et al.,  2009; Azmi et al., 2010; Genc et al., 2010; Vlachogianni et  al., 2011).Moderate correlations were found between PM 10 in roadside sites and PM 10 and NO 2 in residential sites (r = 0.403, p < 0.0001 and r = 0.308, p < 0.0001, respectively), indicating the importance of direct road traffic emissions.Generally, there were lower correlations from frequently reported results for monitoring networks spanning over a large area as the distance between stations increases (Grivas et al., 2004).

Correlations between Air Pollutants and Meteorological Factors
The correlations between air pollutants and meteorological parameters are shown in Table 3.All air pollutants were negatively correlated with WS because suspended air pollutants were accumulated during lower WS and diluted/ removed during higher WS (Grivas et al., 2004;Kim Oanh et al., 2009).Data showed that PM 10 was positively correlated with NO 2 and CO 8hr (r = 0.73, p < 0.0001 and r = 0.64, p < 0.0001, respectively) and moderately correlated with SO 2 (r = 0.54, p < 0.0001).However, PM 10 was negatively correlated with WS (r = -0.44,p < 0.0001).In addition, CO 8hr and NO 2 were strongly correlated (r = 0.66, p < 0.0001) indicating that a major source of their emissions came from vehicles (Grivas et al., 2004;Azmi et al., 2010).

Regression Models
The stepwise multiple linear regression models were used to find the factors that affected air pollution in BKK.The regression models for air pollutants with meteorological, area, and seasonal variables in 2009 were showed in Table 4.The regression analysis was conducted on significant factors affecting PM 10 , NO 2 , and SO 2 levels, including the areas studied, season, RH, temperature, and WS.Temperature was not a significant factor affecting O 3-1hr , while RH and season were not significant factors affecting CO 8hr levels.The coefficient of determination (r 2 ) for PM 10 , CO 8hr , O 3-1hr , NO 2 and SO 2 was 0.54, 0.68, 0.51, 0.64, and 0.22, respectively.The regression models for PM 10 , CO 8hr , O 3-1hr , and NO 2 decreased moderately with decreasing WS, temperature, and RH.On the other hand, there was a weak association between SO 2 concentrations and meteorological parameters, similar to a previous finding (Akpinar et al., 2008).
Air pollution concentrations decreased with increasing WS and circulation of airflow because pollutants dilute through dispersion when WS is higher.The volume and dilution of air pollution were controlled by WS and WD.Moreover, RH is inversely related to air pollution because it controls the absorption rate of air pollutants.The coefficients of the season dummy variable indicated higher concentrations of PM 10 , CO 8hr , O 3-1hr , NO 2 and SO 2 during the winter and summer seasons, using the rainy season as a reference.Further, the coefficients of the areas studied dummy variable indicated higher concentrations of PM 10 , CO 8hr , NO 2, and SO 2 in roadside sites, using residential sites as a reference.The inverse was found for O 3-1hr .It is demonstrated by both spatial and temporal variations in the PM 10 , CO 8hr , O 3-1hr , NO 2 and SO 2 models.There were marked significant effects of seasonal, temporal, and meteorological factors on air pollution concentrations, and the findings were similar to previous studies (Kim Oanh et al., 2009; Genc et al., 2010;  Rose et al., 2011).Fig. 4 shows the daily-basis predicted and monitoring data of PM 10 and O 3-1hr concentrations at both roadside and residential sites in BKK in 2009.Results showed that the prediction performances in PM 10 and ozone at residential sites were better than that at roadside sites in 2009 (PM 10 : r 2 = 0.58 vs. r 2 = 0.37; O 3-1hr : r 2 = 0.45 vs. r 2 = 0.43).However, the predicted model might not performed very well and still needed to improve with adding more detail affecting factors such as land-use, vehicle density, distance to monitoring site, seasonality, and so on.Further, if the resources were allowable, a GIS-based land-use regression model (Liu et al., 2009;Chen et al., 2012) might be applied for predicting the air quality in BKK.

Weekday-Weekend Variations and Potential Ozone Estimates
Table 5 shows the weekday (WKD) and weekend (WKE) variations at both roadside and residential sites in different seasons in BKK during 1996-2009.All the five pollutants including PM 10 , CO 8hr , O 3-1hr , NO 2 , and SO 2 did not show the significantly different between WKD and WKE durations at both roadside and residential sites for overall and single season.Table 5 only indicated that the air pollutants influenced by season change and location variables.For ozone, the phenomenon was differed from previous studies in India (Debaje and Kakade, 2006) and Japan (Sadanaga et al., 2012).Lontati et al. (2006) had compared the fine PM data during WKD and WKE periods in Milan, Italy, and showed that the traffic emissions played a significant role to influence the PM sizes and concentrations at residential, traffic, and suburban sites during different seasons.Additionally, one study in Taipei, Taiwan (Chang and Lee, 2007), polled five monitoring sites and showed that the WKD-WKE variations of air pollutants were not significant different.However, our findings were not consistent with the Milan study but that in Taipei study.
Potential ozone or oxidant (O x = O 3 + NO 2 ) (Kley et al., 1994;Chou et al., 2006) estimates distributions between roadside and residential sites in BKK during 1996-2009 are described in Table 6.For annual trends of O x , there existed peaks for roadside sites in [2003][2004] and that for residential sites in 2008-2009.Overall, the annual O x estimates at roadside sites were larger than that at residential sites but not significant.By monthly comparisons, they showed a peak occurred from November to January (next year) at both sites.The average differences between these two kinds of sites ranged from 5.9 ppb (in August) to 11.3 ppb (in February).By comparing with one study in Tokyo,   Japan (Sadanaga et al., 2012), the O x production (Here defined as O x = O 3 + NO 2 -0.1 NO x ) in annual variation between study sites (urban and suburban) was little.The seasonal change was the major dominant factor for O x production.In addition, the diurnal trends at both sites were similar and showed that peak duration occurred from 12:00 to 17:00 everyday.In the peak duration, however, the O x estimates at roadside sites had higher than that at residential sites (54.1-58.0ppb vs. 42.2-44.2ppb) but not significant which due to the higher variations shown at both sites.

CONCLUSIONS
With the exception of O 3-1hr , the annual averages of PM 10 , CO, NO 2 , and SO 2 concentrations in roadside sites were higher than those in residential sites in BKK.In winter, all the air pollutants concentrations were higher than those during summer and rainy seasons.Interestingly, there were no differences between weekday and weekend durations for the same site category and season in our observations.Furthermore, we proposed stepwise multiple linear regression models with meteorological, spatial, and temporal factors to five air pollutants at both roadside and residential sites in 2009.Results also showed good performances with r 2 ranged between 0.51 and 0.68 except SO 2 .Compared with the standard in Thailand and guideline by WHO, the severe impacts caused by air pollutants should be noticed and concerned.Since 1993, however, the levels of air pollutions in BKK have decreased because the improvement of air quality have been included, by the national government, in the policy on the protection of public health, which was formulated in accordance with the National Economic and Social Development Plan (Kunchornrat et al., 2008;Vichit-Vadakan and Vajanapoom, 2011).Our findings showed the heterogeneously spatiotemporal characteristics in BKK and supported the future improving direction of air pollution control.

Fig. 4 .
Fig. 4. Scatter plot showing the predicted and monitoring data of (a, b) PM 10 and (c, d) O 3-1hr concentrations at both roadside and residential sites in BKK in 2009.

Table 1 .
Number of exceedances WHO guideline and Thai daily standard of the five air pollutions monitored in workday (WKD) and weekend (WKE) classified by season and areas.

Table 2 .
Descriptive statistics of different air pollution and meteorology at roadside and residential sites inBKK during 1996BKK during  -2009.   .

Table 4 .
Regression model of air pollution concentrations, with a regression coefficient for meteorological conditions, the areas studied, and seasonal parameters in BKK in 2009.

Table 5 .
Weekday and weekend variations at both roadside and residential sites in different seasons inBKK during 1996BKK during  - 2009.   .

Table 6 .
Potential ozone (O x ) estimates distributions between roadside and residential sites inBKK during 1996BKK during  -2009.   .