The Influence of Spatial Variability of Critical Conversion Point ( CCP ) in Production of Ground Level Ozone in the Context of Tropical Climate

Critical conversion point (CCP) is a very crucial step in production of the ground level O3 chemistry. Thus, a multivariate analysis was applied on the dataset of nine selected locations in Malaysia from 1999 to 2010. It incorporated hierarchical agglomerative cluster analysis (HACA) to explore the spatial variability of CCP and principal component analysis (PCA) to determine the major sources of the air pollutants that influence ozone CCP. High variability in CCP was observed between the monitoring stations that occurred during critical conversion time (CCT) from 8:00 a.m. to 11:00 a.m. The HACA results grouped the nine monitoring stations into three different clusters, based on the characteristics of ozone concentrations during CCT period. Results of PCA for the three clusters showed that the contributions to O3 level variation during CCT by meteorological variables (UVB, temperature, relative humidity, and wind speed) are higher at 51.6%, 48.5%, and 33.3% than that of primary air pollutants (NO2, SO2, PM10) at 19.2%, 21.4%, and 15.2% for cluster 1, cluster 2, and cluster 3, respectively. Therefore, applying a targeted spatial control strategy for ground level O3 precursors during the CCT period is a crucial step.


INTRODUCTION
Ground-level ozone (O 3 ) is one of the criteria air pollutants that is always associated with degradation of air quality worldwide.It induces harmful effects on human health, crop production, material quality, and the ecosystem.As a secondary air pollutant that is produced from anthropogenic activities, the formation and accumulation of O 3 are induced by the emissions of nitrogen oxide (NO x ) and volatile organic compounds (VOCs) (Seinfeld and Pandis, 2006).O 3 formation is very responsive to changes in meteorological parameters.Thus, elevated O 3 levels are often associated with intensive solar radiation, high temperature, minimal rainfall, low wind speed, and low relative humidity (Toh et al., 2013).
The dependency of O 3 formation toward UV light causes its clear daily variations.In the presence of sunlight, nitrogen dioxide (NO 2 ) undergoes photochemical reactions to produce free oxygen atom (O), which later reacts with oxygen molecules (O 2 ) to form O 3 (Duenas et al., 2004;Azmi et al., 2010).Once O 3 is created, it is destroyed through several pathways, such as nitric titration and surface deposition (Abdul-Wahab et al., 2005).O 3 concentration variations show an interesting patterns in the morning where O 3 level reaches the lowest concentration because of the higher rate of NO titration (Jiménez-Hornero et al., 2010).Once the minimal point is reached, O 3 starts to increase with rising NO 2 concentration, thereby promoting NO 2 photolysis.When the NO 2 photolysis rate is higher than the NO titration rate, critical conversion point (CCP) occurs.Therefore, CCP is very crucial step in ground-level O 3 chemistry because the different in the chemical reaction's rate is expected to result in O 3 accumulation.
The background O 3 has increased over the last decade and is expected to continuously increase in the subsequent years (Ghosh et al., 2013).Thus, many countries, including Malaysia, monitor the current O 3 condition and have set guidelines against this air pollutant.In 2010, the Recommended Malaysian Air Quality Guideline (RMAQG) of 100 ppbv for the hourly O 3 is often exceeded in several places such as at monitoring stations in Klang Valley (Latif et al., 2012) which imperil health problems and ecological impact for millions of people lived in this region.The study of ozone variations is complex because of various possible precursors, photochemical processes, and meteorological factors (Chattopadhyay and Chattopadhyay, 2011;Toh et al., 2013).In addition, the interactions among O 3 , its precursors, and meteorological parameters occur within a wide range of temporal and spatial scales (Abdul-Wahab et al., 2005).Thus, implementing a targeted control strategy by location, such as a particular place, and by time, such as the morning rush-hour time period is crucial step to minimize O 3 precursors' emission reductions.Therefore, this study is attempted to introduce the possibilities to use CCP in explaining the production of the ground level O 3 and to explore the spatial variability of CCP in the context of tropical climate.

Study Areas Description
The dispersion and dilution of air pollution are directly influenced by local attribution such as meteorological condition as well as the location of the monitoring stations (Hosseinibalam et al., 2010).Thus, nine continuous air quality monitoring stations operated by Alam Sekitar Malaysia Sdn Bhd (ASMA) was selected in this study and covered three different types of land use: industrial (Pasir Gudang, Perai, Kemaman and Kuching), urban-residential (Kota Bharu, Kota Kinabalu, Gombak and Klang), and one reference site (Jerantut).All the stations except Jerantut are highly polluted due to industrial activity, traffic emissions and rapid development by growing populations in these areas.Fig. 1 shows the map of the study area and Table 1 shows the description of the selected monitoring stations.
Perai monitoring station (PR) is located in Seberang Perai Penang, one of the most heavily populated states in Peninsular Malaysia.The main air pollution emission contribution comes from industrial and traffic emission.In addition, power plants can also be considered as possible   , 2012).The major industries that drive the economy of Pasir Gudang are transportation and logistics, shipyard industries, petrochemical industries, as well as oil palm storage and distribution.Kemaman is situated in Terengganu (East Coast of Peninsular Malaysia), the city is relatively less developed , except a few places along the coastline where steel and petroleum plants are located (Sulong et al., 2002).Kota Bharu monitoring station was located in Kelantan.According to Shaari et al. (2012), the major land use in Kota Bharu is for agriculture, with one industrial park located at Pengkalan Chepa (Azlan et al., 2011).Kuching monitoring station is located at northeast Borneo in Sarawak state which is surrounded by industrial activities.Besides that, this station is also affected by power plants such as PPLS Power Generation Plant (coal-fired) and Sejingkat power Corporations Plant (coal-fired) (Chung et al., 2012;Dominick et al., 2012).Kota Kinabalu monitoring station is in Kota Kinabalu, the capital city of Sabah state (North Borneo), the station is surrounded by high-populated residential areas and major roads.Jerantut as a background station is located at MMS (Meteorology Monitoring Station), Batu Embun.The station is surrounded by the agricultural area and traditional Malaysian villages (Banan et al., 2013).According to Azmi et al. (2010), the source of the air pollution in Jerantut is expected to be natural forest fires, open burning, soil dust, and motor vehicles.

Weather Condition of Study Area
Climatically, Malaysia experience tropical rainforest climate distinguished by high temperature ranging from 22 to 24°C during night time and from 27 to 30°C during daytime.Seasonal variations in Malaysia are distinguished by changes in wind flow patterns and rainfall intensity (Md Yusof et al., 2010).Uniform periodic changes in wind flow patterns and rainfall intensity are described as monsoonal changes.Peninsular Malaysia has two monsoonal seasons per year, which are the northeast monsoon (NEM) (November-March) and the southwest monsoon (SWM) (June-September), and two intermoonson period occurred during April to May and October to November.The mean annual rainfall in these locations is approximately 2670 mm (Ghazali et al., 2010;Md Yusof et al., 2010), and the relative humidity ranges from 70% to 90%.Heavy seasonal rains observed during northeast monsoon (November to January) (Sulong et al., 2002), while the driest months are June and July.

Monitoring Records
Continuous hourly ground level O 3 concentrations and other air pollution levels were established across Malaysia by Department of Environment for measuring and detecting any significant changes in air quality from 1999 to 2010.Hourly O 3 concentrations are measured using the UV absorption O 3 Analyzer Model 400A, which is a microprocessor controlled device (Mohammed et al., 2013).The O 3 analyzer applies a system based on the Beer-lambert law to measure low ranges of O 3 concentration in ambient air and gaseous media (Ghazali et al., 2010).Ambient O 3 concentration is detected from internal electronic resonance of O 3 molecules using absorption of 254 nm UV light emitted from an internal mercury lamp (Teledyne, 2011).Meanwhile, hourly NO 2 and NO concentrations were collected using the NO/NO 2 / NO x analyzer model 200A (Ghazali et al., 2010).The analyzer applies the chemiluminescent detection principle to detect NO 2 concentration in ambient air.The procedures employed were adopted from the standards outlined by internationally recognized environmental organizations such as the United State Environmental Protection Agency (USEPA) (Latif et al., 2014).Meanwhile, incoming solar radiation was measured based on ultraviolet beta (UVB) rays with wavelength ranging from 280 nm to 315 nm.Hourly average temperature, relative humidity and UVB vibrations were measured with Met One 062 sensor, Met One 083D sensor, and Scintec Model UV-S-290-T, respectively.The number of each air pollutants and meteorological parameters observations in this study totals 157,680 (1461 observation per parameter × 9 stations × 12 years).The descriptive statistics of the measured 12 year data set are summarized in Table 2.

Determination of CCP Using Composite Diurnal Plot
Graphically, CCP were determined based on composite diurnal plots of O 3 , NO 2 , NO, temperature and UVB radiation.CCP was assuming to occur at point of intersection between O 3 , NO 2 , and NO diurnal plots line.If the exact intersection point cannot be obtained from the plots, estimation of the interception point was point out as the CCP.

Determination of Critical Conversion Time (CCT) Based on Ozone Production Rate
At the ground level, it is established that there is interconversion between O 3 , NO 2 , and NO concentration that is dominated by reactions as follows (Ghazali et al., 2010): With the availability of photons with wavelength shorter than 424 nm, the photo stationary state in which the concentration of NO 2 and NO were related to the O 3 concentration is given by the follows Eq. ( 4) (Seinfeld and Pandis, 2006): Assuming that [O] is not constant and rather varies with [NO 2 ] and consequently achieved instant balance between its rate of production and loss.According to the reactions before, there is a point where NO 2 is destroyed and reproduced at a very fast rate that will induced a steady-state cycle is maintained (Seinfeld and Pandis, 2006).The steadystate ozone concentration is given by the follows Eq. ( 5) (Notario et al., 2013): where j NO 2 is the rate of NO 2 photolysis; k 3 is the rate of NO titration.j NO 2 /k 3 was used to indicate the variations the rates of NO 2 photolysis and NO titration.The positive differences of j NO 2 /k 3 rates with the previous hour indicating that NO 2 photolysis rates higher than NO titration rates, while negative indicate that NO 2 photolysis were lower than NO titration.The biggest positive differences were used to point out the time for CCP.

Statistical Analysis
In this study, continuous data of selected variables are analyses according to spatial location.The mean, median, minimum, maximum and standard deviations of each variable are calculated to overview the distribution of data.For this study, no imputations methods were applied and any missing values occurred during data acquisition were omitted from analysis.

Hierarchical Agglomerative Cluster Analysis
Hierarchical agglomerative cluster analysis (HACA) is a set of multivariate techniques commonly used to group the objects into clusters so that the objects (monitoring stations) within a cluster are similar to each other while objects located in other clusters are different from each other.HACA maximizes the similarity of cases within each cluster while minimizing the dissimilarity between groups that are initially unknown (Lu et al., 2011).In this analysis, each object is considered as a separate cluster before it is connected by Ward method agglomerate techniques and squared Euclidean distance to measure the similarity between hourly ozone concentrations using Eq. ( 6).Wards method is chosen as the linkage in this study because this method used an analysis of variance approach in evaluate the distance between clusters in attempt to minimize the sum of square (SS) of any two cluster (Shrestha and Kazama, 2007).The classification of the objects can be illustrated in a dendrogram (tree diagram), which shows the measured similarity or distance between any two variables.

 
Hierarchical cluster analysis can be cut at any level.The optimum number of clusters is usually determined using the difference in distance values as the optimum point is where clear declamation between differences in distance is recorded.Then, CA is repeated using the selected number of cluster to evaluate the selected number of optimum cluster.Further, to accommodate practicality of the results as there is ample information Dmax/Dlink × 100 < 15 step is used (Shrestha and Kazama, 2007).

Principal Component Analysis (PCA)
PCA is a multivariate technique that is widely used to deal with voluminous data in monitoring studies, such as air pollution research.In the present study, this technique was applied to reduce variables and to identify the most relevant variables in O 3 variations (Dominick et al., 2012).PCA is widely known due to its capability to detect the most significant variables in dataset with minimum loss of the original information (Dominick et al., 2012;Elbayoumi et al., 2014).Principal components (PCs) were extracted, such that the first PC (PC1) accounted for the largest amount of total variation in the data set, whereas the following components accounted for the remaining variations that were not considered in PC1 (Kovac-Andric et al., 2009).PCs are generally expressed as follows: where PC i is the i th principal component, and l mi is the loading of the observed variable X m The significant variables for each component are determined based on the loading.In this study, only a factor loading that is greater than 0.4 is considered significant (Ul- Saufie et al., 2013).The sufficiency of the monitoring data for PCA was assessed using Kaiser-Meyer-Olkin (KMO) and Bartlett's tests.According to Özbay et al. (2011), these tests are applied to examine the hypothesis that the variables are uncorrelated in the population.The KMO result (0.735) showed that the value was greater than 0.5, which indicated that the data were sufficient for PCA.Meanwhile, for Bartlett's Test of sphericity (35524.2) showed that the selected variables were significantly (p > 0.001) related to one another and suitable for factor analysis.

Critical Conversion Point Based on Composite Diurnal Plots
The annual CCP at a locations were determined based on the composite diurnal plot of O 3 , NO, NO 2 concentrations, temperature and UVB in depicted in Figs. 2,  In the presence of sunlight, the NO 2 photolysis was able to complete and accumulated as the level of precursors increasing.However, at urban stations such as Gombak and Klang the CCP were later than at other places contributed to higher NO concentration during early morning (6 a.m.-8 a.m.).CCP is assumed as the interception point between O 3 , NO 2 , and NO line in the composite diurnal plots.The plots illustrated that after reached the CCP, O 3 will showed increment trends, while NO 2 and NO will showed decrement trends.It is observed that, O 3 concentrations in stations that their CCP occur around 10a.m. and later are relatively higher than in station that have their CCP earlier.The mean Due to high concentration of NO, the NO titration rate will also increase, hence increase O 3 scavenge which will profoundly reduce O 3 concentrations.The minimum O 3 concentrations in Klang, Gombak are 4.31 ppb and 1.82 ppb, respectively.However, high NO titration will ultimately produce high concentration of NO 2 , which is later converted into O 3 .In these stations, the time for CCP will be later because NO 2 photolysis rate will take longer to surpass NO titration.Alternatively, the time for CCP will be faster when NO titration rate is lower.The annual variations of CCP time were expected to occur due to the variation in the intensity of precursor's concentration.Fig. 5 illustrates the CCP during 1999-2010 based the annual average composite diurnal plots based on the differences in stations type.High variability in CCP was observed at Kemaman with the magnitudes of differences between the earliest and latest time is around 40 minutes.Result suggested that, latest time for CCP were observed at Klang during 2006.In general, at most of the studied monitoring stations, CCP was occurred at approximately between 9 a.m. and 10 a.m.Further, the majority of stations that were CCP measured at ± 9 a.m locate at east coast of Malaysia.Kemaman showed characteristically difference in CCP from other locations since at the station the concentration level of NO 2 and NO is at very low.Further, only small variation in time for CCP as constant time is observed for each year in all stations.Due to this, the effect of weather conditions is considered very minimal and hardly effected the CCP variations.In addition, the usage long period of data (12 years) influence the effect of weather condition.

Critical Conversion Point Based on Photochemical Reactions
Table 3 shows the calculated value of j NO 2 /k 3 at industrials, urban and background stations, respectively.Jenkin and Clemitshaw (2000) reported that as a result of rapid interconversion, the behaviour of NO and NO 2 is highly coupled.Thus, in the absence of any competing inter-conversion reactions at the ground level, a phototostationary state of the relations between O 3 , NO and NO 2 is obtained through Eqs.(1) to ( 5).The j NO 2 /k 3 value fluctuates throughout the day as the concentration of O 3 , NO 2 and NO also varies daily.Result in the table suggested that, the daily variations of j NO 2 /k 3 value are similar to O 3 daily variations.Han et al. (2011) also reported similar findings when calculating the j NO 2 /k 3 at Tianjian, China.Theoretically, the j NO 2 /k 3 value is supposed to be zero during the night time in absence of the photochemical reactions, however the background O 3 concentration in the atmosphere cause the j NO 2 /k 3 is at minimal around 5-7 ppb.The ratio value starting to increase during daytime and reaching maximum value at 2 PM around 20 ppb, coincide with the maximum daily O 3 concentration that were observed at most of the studied monitoring stations.After reaching the maximum point, the j NO 2 /k 3 value is decreasing as the NO 2 concentration were gradually increased reaching its evening peaks that often measured at 9 p.m.-10 p.m.
The differences in j NO 2 /k 3 value at current hour (h i ) to the previous hour (h i-1 ) denoted as ∆j NO 2 /k 3 used as indicators of the differences of the rates of NO 2 photolysis and NO titrations (Clapp and Jenkin, 2001).This value was used as indicators to different in rate of NO 2 photolysis and NO titrations and determination of CCP when the biggest positive ∆j NO 2 /k 3 value occurred.High positive ∆j NO 2 /k 3 value will lead to O 3 accumulation, while high negative ∆j NO 2 /k 3 value is expected to contribute to O 3 destruction.The result in Table 3 exhibited that, the largest ∆j NO 2 /k 3 is measured as early as 7 a.m. at Kota Kinabalu and as latest as 12 pm measured at Klang.Similar to CCP, the highest ∆j NO 2 /k 3 mostly occurred at 10 a.m.At several locations such as Kajang and Gombak the time for CCP is coincided with ∆j NO 2 /k 3 time.Meanwhile, at other locations, the CCP and ∆j NO 2 /k 3 is apart by an hour differences, which is ∆j NO 2 /k 3 were measured at an hour earlier and an hour later than the CCP time.However, at Kota Kinabalu, Jerantut, Pasir Gudang and Johor Bahru, the time for CCP and ∆j NO 2 /k 3 is minimally 2 hour apart.Technically, the calculations of ∆j NO 2 /k 3 can be used as the aided techniques in finding the CCP of O 3 formations at a certain locations.
The stations in Cluster 1 are in the main city centres of each state.These stations are surrounded by high-populated residential areas, major roads and commercial areas such as KK and KB stations.Furthermore, part of monitoring stations in cluster 1 such as KC, PG and PR stations are located in industrial areas and have the most famous ports in Malaysia namely Penang port and Port of Pasir Gudang.As example, Penang port handled a total of 6650 arrivals and departures of vessels in 2012, which is equivalent to 43 million tons of cargo and can be considered as possible sources of ozone precursors.The location of the monitoring stations is also influenced by industrial, traffic emission and power plants especially in PR and KC stations.Cluster 2 consists of stations located in the Klang Valley.In these areas, there is a very high concentration of commercial and industrial activities with heavy traffic almost entirely around the clock.Cluster 3 (Kemaman) is with a total area of 2535 km 2 and total population 166,750 in 2010.Only two major industrial sites (i.e., Kerteh Petrochemical and Gebeng Industrial Area) are located in and near Kemaman (Ismail et al., 2011).However, the higher ozone level in this location comparing to others may be due to transport of air pollutants from other locations and several meteorological parameters which always been associated with ozone variations such as wind speed and directions.Wind speed and direction are significant agent that controlling ozone transport and dilution in both daytime and night time (Ghazali et al., 2010;Kim and Guldmann, 2011;Toh et al., 2013).Awang et al. (2015) reported that high O 3 concentrations in Kemaman were observed that coincided with prevailing winds from the southerly direction.Kuantan is one of the biggest cities in east coast Peninsular Malaysia and characterized by high population and traffic density, numbers of industrials and residential establishments.

PCA Analysis during CCT
Principal component analysis was applied on nine significant parameters of the data set that influence the formation of CCP to determine the major sources of the variation in each cluster produced by HACA.They were SO 2 , PM 10 , CO, NO 2 NO, UVB, WS, temperature, and relative humidity for each cluster.These variables were selected based on their relationship with O 3 as NO 2 , NO and CO is significant O 3 precursors, PM 10 and SO 2 are primary pollutant and UVB, WS, temperature and relative humidity are important meteorological parameters that contribute to O 3 fluctuations.Most of the correlations between variables are significant at 0.01 level, thus allowed these variable to be used in PCA.
In addition, KMO and Bartlett's were used to justify the hypothesis and applicability of the selected variables for PCA.KMO value for cluster 1, cluster 2 and cluster 3 are 0.834, 0.815 and 0.661, respectively which showed that the value is higher than 0.5.According to Chattopadhyay and Chattopadhyay (2011), KMO could give an indication whether PCA is suitable for removing multicollinearity in the dataset and KMO value closer to 1 indicates that the correlation pattern is relatively compact and suitable for PCA.In the meantime, Sousa et al. (2007) reported that the null hypothesis of PCA is the variables are uncorrelated and to justify the hypothesis and applicability of the monitoring records for PCA, Bartlett's test was used.Result of the Bartlett's tests is significant as the p-value of the test is higher than 0.001 that indicate that the selected variable were related to one another and suitable for PCA.
The results of the PCA loadings after rotation are shown in Table 4.A spatial variation was observed in the loaded component of each PC and in the total variance explained by each PCA.In this study, only strong factor loadings (> 0.40) were selected for the PC interpretation.

a) Cluster 1
Cluster 1 consists of six stations that located in urban and industrial locations.The first component (PC1) explains 51.6% of the total variance which shows strong positive factor loadings for humidity (0.941), CO (0.794), NO (0.729), strong negative loadings on ambient temperature (-0.922),UVB (-0.912) and WS (-0.767).Several studies of O 3 in urban areas have shown that the variation of O 3 be influenced by a number of O 3 precursors that predominantly originate from motor vehicles such as NO and CO (Ma et al., 2012;Zhang et al., 2013).As stated previously in Eq.
(3), in the presence of both high temperatures and sunlight NO will influence O 3 formation through titration reaction (Ghazali et al., 2010).Humidity is another factor that influences formation of CCP of O 3 .High RH condition enhanced O 3 destruction as a result of the reduction in photochemical efficiency and the increase in wet deposition process (Kovac-Andric et al., 2009;Toh et al., 2013).
The second component, (PC2) explains 19.24% of total variance.It has strong positive loadings on PM 10 (0.758), NO 2 (0.712) and SO 2 (0.683).Cluster 1 is a combination of urban and industrial cities; thus the burning of fuels in automobiles and industrial facilities are suspected to be the source of air pollution (SO 2 , NO 2 ) as indicated in previous studies (Janssen et al., 2001;Dominick et al., 2012).

b) Cluster 2
Cluster 2 consists of two stations GB (Gombak) and KL (Klang).The first component (PC1) explains 48.58% of the total variance.It recorded strong positive factor loadings for humidity (0.936), NO (0.720), strong negative loadings on ambient temperature (-0.927),UVB (-0.892) and WS (-0.582).As discussed in Cluster 1 motor vehicles emissions are the major source of NO and the influence of meteorological parameters is clear on CCP formation but lower loading in cluster 2. The second component (PC2) with 21.41% of total variance shows strong positive loadings on NO 2 (0.784), PM 10 (0.778), SO 2 (0.743) and CO (0.654).The results show that the sources of pollutants in this cluster mostly related to motor vehicles and industrial emission in the area.The air monitoring stations (GB and KL) are located in dense-development urban areas in the Klang Valley and are served by several major highways, which experience heavy traffic in the morning and late afternoon rush hours.In addition, the KL station is also surrounded by industrial areas and a proximity to busy port (Port Klang) all of which have the capability to emit high amounts of pollutants into the atmosphere.According to Haris and Aris (2013), Port Klang is the busiest port in Malaysia and the 14 th busiest port in the world.In 2012, the port received and departed approximately 15,000 vessels, handling approximately 169 million tons of cargo, which is significantly higher than that of any other port in Malaysia.

c) Cluster 3
In the case of Cluster 3, there is only one station (KE station).The first component (PC1) explains 33.39% of the total variance and has strong positive loadings on ambient temperature (0.888), UVB (0.853) and WS (0.792).The second component (PC2) explains 19.4% of the total variance and shows strong positive loadings for CO (0.866) and NO (0.684), which are obviously indicative of emissions from vehicles engines such as cars and lorries.The emissions of CO arises during the incomplete combustion of fossil fuels and biomass in fumes produced by portable generators and vehicle engines (USEPA, 2013).The third first component (PC3) explains 15.20% of total variance which shows strong positive factor loadings on NO 2 (0.758), RH (0.715) and PM 10 (0.603).PM 10 and NO 2 can be formed through open burning or emitted by motor vehicles or result from transboundary pollutants around the study area (Md Yusof et al., 2010).
By comparing the results of PCAs from the three clusters, the PCs in the three clusters explained between 68% and 70.8% of the variation in ozone during CCT.Furthermore, differences between PCs in clusters 1, 2 and 3 can be observed by examining how the pollutants and meteorological variables were loaded.Meteorological parameters have a greater influence on O 3 formation in cluster 1(51.6%) and 2(48.5%)than in cluster 3(33.3%).Further, pollutant concentration parameters have a greater influence on O 3 formation differently in the three clusters in descending order 21.4%, 19.2%, 15.2% for clusters 2, 1 and 3, respectively.This difference is due to variation in the numbers of vehicles, indusial zones and number of population in each cluster.Therefore, the wise to apply a targeted spatial control strategy for ground-level ozone control during the CCT period from 8:00 to 11:00 a.m. to reduce the O 3 precursor pollutants.

CONCLUSIONS
This study analyzed the spatial variability of CCP for O 3 concentrations in nine monitoring stations located in Malaysia from 1999 to 2010.The study showed that high variability in CCP was observed between the monitoring stations which ranged from 8a.m. and 11a.m.The earliest time were measured at Kota Kinabalu due to the earlier sunrise at the location.At most of the industrial stations, the CCP were measured around 9 a.m.-10 a.m. and at most of the urban station CCP were measured around 10 a.m.-11 a.m.The results from this study show that the HACA grouped the nine air monitoring stations into three different clusters based on the selected parameters that influence CCP formation.Results of PCA for the three clusters showed that the contributions to O 3 level variation during CCT by meteorological variables (UVB, temperature, relative humidity, and wind speed) are higher at 51.6%, 48.5%, and 33.3% than that of primary air pollutants (NO 2 , SO 2 , PM 10 ) at 19.2%, 21.4%, and 15.2% for cluster 1, cluster 2, and cluster 3, respectively.The main sources of O 3 precursors that contribute in CCP in Cluster 1 and Cluster 3 are motor vehicle exhaust emissions and gases released from industrial activities.Meanwhile, the main sources of CCP of ground level ozone in Cluster 2 are predominantly from motor vehicles, emissions from industries and port related activities.In recommendation, all relevant agencies could collaborate to implement a State Implementation Plan (SIP) by location and by time which could lead to minimize in O 3 precursors' emission reductions such as motor vehicle exhaust emissions and well as in gas emissions from industry sectors.

Fig. 1 .
Fig. 1. Specific location of monitoring stations across Malaysia (map is not up to scale).

Fig. 2 .Fig. 3 .
Fig. 2. CCP based on composite diurnal plots for monitoring station located in north Malaysia.

Fig. 4 .
Fig. 4. CCP based on composite diurnal plots for monitoring station located in south Malaysia.

Fig. 5 .
Fig. 5.The annual critical conversion time of ozone formation at all station based on composite.

Table 1 .
Description of monitoring selected monitoring stations.

Table 2 .
Descriptive statistics of daily average of air pollutants and meteorological factors from 1999-2010.

Table 3 .
The calculated value of j NO 2 /k 3 for monitoring stations.

Table 4 .
Bian and Zender (2003)nalysis after varimax rotation.PM 10 is one of the most notable criteria pollutants because it can alter the photolysis rates of several trace gases.Bian and Zender (2003)claimed that high PM 10 levels in ambient air can trigger light scattering of solar radiations and reduce the solar radiation intensity that reached ground level.Reduction in solar intensity stopped photochemical reactions and diminished O 3 concentrations.