Seasonal Variation of Criteria Pollutant in an Urban Coastal Environment: Kuala Terengganu

The aim of this study are (i) to determine the yearly and monsoonal variations of the criteria pollutants in Kuala Terengganu, (ii) identify the influences of the meteorological factors and other minor criteria pollutants that impose on the concentration of the most significant criteria pollutant using Principal Component Analysis (PCA). The hourly data for the criteria pollutants (PM10, CO, O3, NO2, and SO2) and meteorological factors (temperature, relative humidity, windspeed) for the duration of 10 years procured from the Department of Environment, Malaysia were analyzed. PM10 is the only pollutant that frequently recorded concentrations exceeding the MAAQG in the length of the year 2001 to 2010 and the criteria pollutants reported differ significantly between the monsoon seasons. PCA showed that sources contributing to the most significant criteria pollutant (PM10) are meteorological factor influences and industrial emissions.


Introduction
More than two-thirds of the world's largest cities are located on the coastline and around half of the human population inhabits within 150 km of a coast.The rapid development of the coastal cities also causes the gradual air quality degradation which in turn increases the health threat to both human and environment.Various studies have been done over the years to identify the attribution of air pollutions towards people's health and the adverse effects brought forward to the environment [1][2][3][4].
In order to categorized the health impact of air pollution, United States Environmental Protection Agency (US EPA) has established the Air Quality Index (AQI) to signify the air quality in a particular region in which the index is mainly focused on the measurement of five major air pollutants, which are carbon monoxide (CO), sulphur dioxide (SO 2 ), nitrogen dioxide (NO 2 ), particulate matter (PM) and ozone (O 3 ) [5].In Malaysia, there is a similar system to indicate the overall quality of air called Air Pollution Index (API), [6].Based on the Environmental Quality Report of the year 2013, the concentration of the criteria pollutants often exceeds the Malaysia Ambient Air Quality Guidelines and the most prominent pollutant that frequently causes the moderate air quality days in Malaysia is particulate matter.Motor vehicles, abrasion of tires and brakes, marine aerosols, agricultural activities and even cooking contributes to a significant amount of PM, biomass burnings produces CO, whereas industrial activities and power plants are the major contributors of aerosols such as SO 2 , NO 2 and O 3 forming pollutants [7]- [10].The ambient air quality management policies are the product of series of complex and conservative processes, which include the air quality monitoring, preparation of emission inventory and control strategies delineation, and long-term compliance monitoring [11].In Malaysia, the air quality status monitoring was done under the responsibility of the Department of Environment (DOE).
Air quality is often influenced by three major factors, i.e. emission sources, prevailing meteorological factors and terrain morphology [12].The study was conducted in Kuala Terengganu, one of the capital cities on the East Coast of Peninsular Malaysia, which is less than 2 km from the shoreline.As the study area is closely associated with the sea, the sources of air pollution originate from offshore as well as onshore with a flat terrain.Kuala Terengganu experienced two major monsoons annually and two inter-monsoons in between [13].This coastal urban area has been developing rapidly in the past years and has achieved the urbanization level of 59.1% in the year 2010 [14] and thus further exposing the residents to the bearings of air pollution.To assess the efficacy of the control and preventive regulations of the air pollution, the trend, and status of the air quality must be studied thoroughly.The aim of this study is to interpret the variations of the criteria pollutant in Kuala Terengganu and determine the seasonal changes of the criteria pollutants.The influences of the meteorological factors and other minor criteria pollutants that impose on the concentration of the most significant criteria pollutant, which in our case is PM 10 , will also be identified.The source apportionment analysis may provide insight for the identification of the pollution emission sources and the amount of impact inflicted to the air pollution.

Study Area and Data Collection
Kuala Terengganu is the capital city of Terengganu state.It is located on the East Coast of the Peninsular Malaysia with less than 2 km from the coastline.There are a total number of three air quality monitoring stations in the Terengganu state, each located in a different district, managed by the Department of Environment.The air quality monitoring station chosen for this study is situated in the capital of Terengganu state which is located at Sekolah Men.Keb.Chabang Tiga (N05°18.487",E103°07.226")(Figure 1).The topography of the area is characterized by lowlands due to the proximity to the coastline.
The data used for the analysis were acquired from the Department of Environment (DOE).The hourly data for the criteria pollutants (PM 10 , CO, O 3 , NO 2, and SO 2 ) and meteorological factors (temperature, relative humidity, windspeed) for the duration of 10 years starting from January 2001 to December 2010 were procured.

Statistical Analysis
The pollutant concentration and meteorological factor trends along the 10 years of data were sorted and calculated by using Microsoft Excel®.The data were the depicted in monthly average trends.Statistical analysis was executed on the data set to extract descriptive information.Analysis of Variance, Multiple Range Test, Kruskal-Wallis test and Standard Kurtosis Test were done to test the significance amongst the means of the data, to determine which mean is significantly different from each other, identify the outliers and test for normality.The missing values during the study period were omitted in the statistical computations.

Principal Component Analysis
Principal Component Analysis (PCA) is a source identification method that enables us to certify that the variables were ideally correlated with one sole component only without correlating with other component and produce a new set of variables called principal component (PCs) [15], [16].The number of principal components will be less than or equals to the original variables.The most significant Pc allow us to determine the relationship between the variables and the dataset, thus deducing the possible sources.
The highest concentration for PM 10 was recorded in January 2002.Although Malaysia is experiencing moderate Northeast monsoon, the Northern ASEAN regions were dominated by dry weather conditions where hotspots were detected in Cambodia, Laos PDR, Myanmar, and Thailand due to the El Nino occurrence.Prevailing North-easterly winds brings over smoke and haze from the northern region to the study area [17].
As we can see in Table 2, the annual variations for the criteria pollutants does not fluctuates much in the past 10 years except for PM 10 .The concentration for PM 10 exceeded the MAAQG of 50 μg/m 3 in eight out of the ten years whereas other criteria pollutants recorded concentrations way below the MAAQG (Table 2).As the study area is an urban coastal environment, the main contributors of PM 10 are not only local sources such as traffic aerosols, biomass burning, soil dust, vehicle exhaust and secondary aerosols but a high amount of the concentration were originating from the sea sprays [18], [19].
The trend of the monthly average concentration of the criteria pollutants was illustrated in Figure 2. PM 10 and O 3 have the highest concentration in December and January while NO 2 and SO 2 recorded higher concentrations in May, and July.As for CO, two peaks that fall in month February and August were noticed.

Monsoonal Variation
There are two major monsoons in Malaysia and two intermonsoons in between.The wet monsoon, also known as the Northeast monsoon (NEM) blows from November to March, followed by the April inter-monsoon (AIM), where the northeasterly winds change to southwesterly winds.The dry monsoon, otherwise the Southwest monsoon (SWM) starts from May and ends in September and replace by the October inter-monsoon (OIM) [13], [20].
The wind rose of hourly wind direction, and windspeed for the four monsoons was illustrated in Figure 3.The wind is strongest during the NEM, and the wind originates mostly from Northeast whereas, in AIM, the dominant wind direction gradually changes to Southwest and last until the OIM.The southwesterly wind during the SWM is generally milder but happens more frequently [21].
The criteria pollutants recorded at the study areas differ significantly between the distinctive monsoons as the pvalue for each pollutant are less than 0.05 at the 95.0% confidence level (Table 3).This is mostly due to the distinguishable meteorological conditions portrays by the different monsoons.The NEM is characterized by heavy rainfall and strong winds while the SWM are accompanied by cloudless skies and drier weather.
To identify the differences between the monsoons, the Multiple Range Test was performed, and the results were tabulated in Table 4.The lower and upper limits of the average of the concentration of the criteria pollutants for the different monsoons were calculated, and the range of the limits was referred as intervals.These intervals are calculated based on the Fisher's least significant difference (LSD) procedure, and the overlapping intervals signify similar means, which denotes that there is no significant difference between the two monsoons.
Referring to Table 4, the average concentration of PM 10 is 52.1 μg/m 3 for NEM, 53.1 μg/m 3 for SWM and 49.2 μg/m 3 for the inter-monsoons.As for the NO 2 , the reported average concentration were 0.0060 ppm, 0.0054 ppm and 0.0049 ppm for SWM, inter-monsoons, and NEM respectively whereas for SO 2 , the highest average concentration were 0.0010 ppm for SWM, followed by 0.0008 ppm and 0.0007 ppm for NEM and inter-monsoons accordingly.The concentration of PM 10 , NO 2 and SO 2 are higher during the SWM.This could be due to the stable drier meteorological conditions that cause hindrance to dispersion and removal of the pollutants from the atmosphere [22,23].
The highest average concentration of CO was recorded in the SWM, followed by the inter-monsoons and NEM with the values of 0.467 ppm, 0.454 ppm and 0.446 ppm respectively.The intervals for the three different monsoon categories do not overlap thus, indicating that all three monsoons have statistically significant different concentrations of CO.The primary sources of CO are incomplete combustion of fossil fuels and biomass burning [24,25].Hydroxyl groups function as sinks to remove CO from the atmosphere and due to the substantial deposition phenomena during the NEM, hydroxyl groups are available in abundance thus, pairing up with the CO in the air and produces CO 2 [25,26].
Of all the criteria pollutants, only O 3 show an overlapping interval in the NEM and SWM.The average concentration of O 3 are highest in SWM, with the value of 0.01480 ppm, coming next is 0.01478 ppm in NEM and 0.01201 ppm in the inter-monsoons.

Pincipal Component Analysis of Particulate Matter Concentration
The most significant pollutant observed in the study area is PM 10 , and thus, the PCA of PM 10 with other criteria pollutants and meteorological factors were performed for the two main monsoons, the NEM and SWM in order to further understand the dynamics and influences on the PM 10 concentration.
To execute the PCA, the individual variable and the set of variables must obtain the Kaiser-Meyer-Olkin (KMO) of Sampling Adequacy value of greater than 0.50 and the probability associated with Bartlett's Test of Sphericity be less than the level of significance [27].The KMO value for the NEM and SWM dataset was 0.822 and 0.753 respectively while the probability associated with the Bartlett test is <0.001 for both datasets, which satisfies the basic requirement for the PCA.
The list of eigenvalues associated with each linear component before extraction, after extraction and after rotation for the NEM and SWM is presented in Table 5.Before extraction, 8 linear components are identified within both data sets.The eigenvalues associated with each factor represent the variance explained by the particular linear component and also displays their eigenvalue in terms of the percentage of variance explained.All factors with the eigenvalues greater than 1 were extracted but for our case, component 3 for both monsoons has eigenvalue closed to 1, hence, the third component were taken into consideration.The percentage variability is about 80.7% for NEM and 73.5% for SWM by using the three factor loadings.The rotation has the effect of optimizing the factor structure and one consequence for these data is that the relative importance of the three factors is equalized.The factor scores obtained from the PCA for the NEM and SWM were displayed in Table 6.There are three principal components (PC) that take accounts for the concentration of PM 10 in both NEM and SWM.For NEM, PC1 (52.6% of the total variance) was loaded with windspeed, relative and humidity and O 3 which grouped all the meteorological factors in one.O 3 has shown a strong correlation with the meteorological factors [28][29][30][31][32][33][34][35] and thus, falls into the same group.The relative humidity shows a negative score which indicates increasing relative humidity brings negative contributions towards PM 10 concentration as high water vapor content in the air provides washing effects which reduce the amount of particles in the atmosphere [36][37].PC2 (16.3% of the total variance) consist of CO and NO 2 , which points towards vehicular emissions [7], [38].PC3 (11.7% of total variance) was weighted with only SO 2, which indicates industrial emissions and open burning [39].A slight difference among the factors distribution was noticed in the principal components for SWM compared to the NEM.PC1 (40.3% of total variance) was predominantly weighted by windspeed, temperature, and O 3 , which expressed that meteorological factors were the main contributors due to the fact that increased windspeed and temperature can cause resuspension of road dust and soil deposition [40].PC2 of SWM (20.3% of total variance) demonstrate the similar factors loading as the PC2 in NEM, which is CO and NO 2 that implies vehicular emission as the possible sources.PC3 (11.7% of total variance) are made out of relative humidity and SO 2 , which indicates meteorological factors and open burning sources.

CONCLUSION
PM 10 is the only pollutant that frequently recorded concentrations exceeding the MAAQG in the duration of the year 2001 to 2010.The criteria pollutant of these 10 years for the Northeast monsoon (NEM), Southwest monsoon (SWM) and inter-monsoons (IM) were compared, and the results show that the concentration of pollutants reported differs significantly between the monsoon seasons.PM 10 , NO 2 , and SO 2 are notably higher during the SWM due to the stable drier meteorological conditions that cause hindrance to dispersion and removal of the pollutants from the atmosphere.The concentration of CO was mainly lower during the NEM.This may be due to high precipitation frequency during the NEM, which increases the amount of hydroxyl radicals in the atmosphere that acts as the sink for CO and in turn reducing the levels of CO in the air.Principal component analysis showed that sources contributing to the PM 10 concentration which is meteorological factor influences, and industrial emissions.

Figure 1 .
Figure 1.Location of study area

Table 1 .
Summary of criteria pollutants data in Kuala

Table 2 .
Annual averaging values of criteria pollutant(Year 2001(Year   -2010)   ) *Standard Deviation a Malaysian Ambient Air Quality Guidelines b World Health Organization Air Quality Guidelines c European Commission Air Quality Standards Figure 2. Monthly average of criteria pollutant (year 2001-2010)

Table 3 .
Analysis of variance if criteria pollutants

Table 4 .
Multiple Range Test: Average and Least Significance Difference (LSD) Intervals for criteria pollutants in different monsoons

Table 5 .
Total Variance and Eigenvalues for (a) NEM and (b) SWM Total Variance and Eigenvalues of NEM

Table 6
Factor loadings after PCA varimax-rotation for both monsoons