Establishing the Association between Quarterly/Seasonal Air Pollution Exposure and Asthma Using Geospatial Approach

Although it is known that air pollution may lead to increased asthma prevalence, no clear scientific evidence of direct association between air pollution and asthma rate has been reported. In the present study, a Geographical Information System (GIS) approach was developed to determine the association between asthma hospital discharge rate (ADR) and seasonal exposure to specific ambient air pollutants in eastern Texas, USA, during the period 2009 to 2011. Quarterly asthma data were obtained from Texas State Department of Health, National Asthma Survey surveillance of Texas State, USA. Quarterly mean concentrations of fine particular matter (PM2.5), nitrogen dioxide (NO2), and ozone (O3) were determined from the corresponding measured daily data collected by various air quality monitoring stations distributed in different counties in the study area. Using Pearson correlation analysis, quarterly average of air pollutant concentrations was compared to quarterly Asthma discharge rate (ADR). The results revealed that the association between quarterly exposure of air pollution and ADR was not statistically significant in the study area. During the study period, a negative correlation coefficient was observed between the quarterly mean concentration of ozone and NO2 with the quarterly ADR. However, in most of the cases a positive correlation coefficient was observed between the quarterly mean concentration of PM2.5 and the quarterly ADR, indicating a probable association between ambient air pollution exposure and asthma prevalence.


INTRODUCTION
Air pollution is one of the most serious environmental threats to urban populations (Cohen et al., 2005).It has been reported to cause adverse health impacts on people of all ages.Exposure to common urban air pollutants has been linked to a wide range of adverse health outcomes including respiratory and cardiovascular diseases, asthma exacerbation, reduced lung function and premature death (U.S. EPA, 2006EPA, , 2009)).Over the past decade, many epidemiologic studies have demonstrated positive associations between air pollution and mortality (Levy et al., 2000;Goodman et al., 2004;Pope et al., 2004;Schwartz, 2004;Analitis et al., 2006).But, many questions regarding the effects of air pollution remain unanswered and overall the effects of air pollution have not been fully quantified.Most of the studies of air pollution have used time-series analysis to relate daily asthma rates to daily air pollution levels for short-term associations between air pollution and health.A substantial number of epidemiological studies reported associations between mortality/morbidity with air pollution levels (Rosas et al., 1998;Dales et al., 2000;Levy et al., 2000;Chen et al., 2004;Goodman et al., 2004;Pope et al., 2004;Schwartz, 2004;Chang et al., 2005;Analitis et al., 2006;Dominici et al., 2006;Hinwood et al., 2006;Yang et al., 2007;Yang, 2008;Halonen et al., 2008;Chiu et al., 2009;Halonen et al., 2009;Belleudi, 2010;Hsieh et al., 2010;Gorai et al., 2014).
The evidence on adverse effects of air pollution on public health has led to more stringent standards for levels of outdoor air pollutants in many countries including USA.Airborne pollutants may influence the symptoms of asthma patients (Delfino et al., 1996(Delfino et al., , 1997)).Asthma is a burden on communities, with significant public health and financial consequences.The number of complaints increases day by day due to increasing trend of air pollution.Asthma is a respiratory medical problem and considered one of the major health issues for all age groups worldwide.Asthma is defined by the World Health Organization (WHO) as one of the chronic respiratory diseases (International Classification of Disease 9th revision, code 493; ICD9-493).The disease is characterized by bronchial inflammation, with an exaggerated response of the lower airways and limited air flow in these airways.The prevalence of asthma in different countries varies widely, but the disparity is narrowing due to rising prevalence in low and middle income countries and flat trend in high income countries (WAO, 2011).An estimated 300 million people worldwide suffer from asthma, with 250,000 annual deaths attributed to the disease (WHO, 2007).Asthma is not only related to genetic and environmental factors, but also it is believed to be affected by air pollutants such as fine particulate matter (PM 2.5 ), ozone (O 3 ), and nitrogen dioxide (NO 2 ).Many studies have been conducted on exposure of air pollution and its association with health, but inconsistent results have been reported regarding its association with asthma rate.Though most of the studies have pointed out that air pollution triggers the asthma rate; few have reported that there is no correlation between air pollution and asthma rate (Rosas et al., 1998;Dales et al., 2000;Chen et al., 2004).Asthma is a chronic disease linked with considerable morbidity, mortality, and health care use.The available literature on asthma studies shows a large geographic variation from local/community level all the way to country level.The studies on asthma and other epidemics have raised some important questions as to what factors contribute to the emergence of asthma outbreak.So there is a dire need to identify these factors and develop a model to establish a relationship that will help to predict asthma outbreak in advance.Such a model will also help public health officials to take preventive measures to minimize asthma prevalence or eradicate the disease completely.The present study attempts to use Geographical Information System (GIS) for analyzing the seasonal association of air pollution (PM 2.5 , NO 2 , and O 3 ) exposure and ADR in eastern part of Texas during three consecutive years from 2009 to 2011.The application of GIS facilitates quantifying the results toward better understanding of the interplay between air quality and asthma and deducing the solutions to mitigate asthma prevalence (Croner et al., 1996;Jerrett et al., 2005).

STUDY AREA
The spatial distribution of air pollutants is affected by emission sources and the atmospheric conditions.Eastern part of Texas (shown in Fig. 1) was considered for the spatial distribution analysis of air pollutants (PM 2.5 , NO 2 , and O 3 ) and asthma discharge rate and their associations.Due to insufficient number of air pollution monitoring stations in the western part of the state, only eastern part was considered for the study.Texas State located in the west-south-central region of the United States.The longitude and latitude of the state are 71°47'25"W to 79°45'54"W and 40°29'40"N to 45°0'42"N respectively.It is the second most populous (25,145,561), and the twenty ninth most densely populated (96.3 inhabitants per square mile of land area) state of the 50 United States (U.S. Census Bureau: Resident Population Data, 2010).Texas State covers 261,797.12square miles of land area and ranks as the 2nd state by size (U.S. Census Bureau, State Area Measurement).Texas is bordered on the north by Oklahoma and Arkansas (with part of the line formed by the Red River); on the east by Arkansas and Louisiana (with part of the Louisiana line defined by the Sabine River); on the south-east by the Gulf of Mexico; on the south-west by the Mexican states of Tamaulipas, Nuevo León, Coahuila, and Chihuahua (with the line formed by the Rio Grande); and on the west by New Mexico.In general, the climate in Texas State varies widely, from arid in the west to humid in the east.There is significant variation in the geography from one region of the state to another.There are coastal regions, mountains, deserts and wide open planes.In coastal regions, the weather is neither particularly hot in the summer nor particularly cold during the winter.East Texas has the humid subtropical climate typical of the Southeast, occasionally interrupted by intrusions of cold air from the north.

MATERIALS AND METHODS
Geographic Information Systems (GIS) is one of the scientific tools for health data processing, analysis of geographical distribution and variation of diseases, mapping, monitoring and management of health epidemics (Johnson and Johnson, 2001).It is also very helpful for air pollution distribution analysis and mapping.In the past, GIS has been applied to estimate the spatial concentrations of air pollutants (Liao et al., 2006) and many epidemiologic studies have adopted GIS to explore the health impact of air pollutants on asthma (English et al., 1999;Oyana, 2004Oyana, , 2005;;Sahsuvaroglu et al., 2009).
Although different counties in Texas State might have varying air pollutant concentrations, the air pollution data in each county was not available.In order to make an exposure assessment for the study area (eastern part of Texas, USA), using geospatial technique we linked the quarterly exposure of air pollution level and corresponding asthma hospital discharge rate (ADR) to estimate the impact of air pollution on ADR.In the present study, GIS is used to estimate the association between air pollutants [particulate matter less and equal to 2.5 μm size (PM 2.5 ), nitrogen dioxide (NO 2 ) and ozone (O 3 )] and ADR.All the three pollutants (PM 2.5 , NO 2 , and O 3 ) are among the highest important parameter of environmental hazards and are at significant levels in air to adversely affect human health (WHO, 2003).The methodology is applied in two stages to deduce the association between air pollution exposure and ADR.First, we estimated the pollutants concentration profile and ADR profile within the defined study area by constructing a spatial model using quarterly average pollutant concentration data and ADR data.Second, we linked air pollutants concentration to asthma rate within the defined study area.

Asthma Hospital Discharge Data
For the period 2009 through 2011, county-wise quarterly asthma hospital discharge numbers and estimated population Texas State, with very high population density, had approximately 19.3 million residents during 2009-2011(Texas State Asthma Survey Report, 2009).ADR indicates the number of asthma-related hospital discharges per 10,000 populations for a specified period of time.Quarterly ADR for each quarter of 2009, 2010, and 2011 were calculated by dividing the number of quarterly asthma hospital discharge numbers by the estimated population for that time period in a particular zone and then multiplying by 10000.The estimated rates represent crude rate on the basis of estimated population of the county.
ADRs vary depending on the region and county of residence.The annual numbers of asthma hospital discharges in Texas State were 28,044, 25,596, and 24,405 in 2009, 2010, and 2011, respectively whereas the respective numbers were 24,982, 22,526, and 21,646 in the selected study area (eastern part of Texas).
The quarterly numbers of asthma hospital discharges were 7064, 5605, 4476, and 6674, respectively in four quarters of 2009.Similarly, the quarterly asthma discharges numbers in 2010 were 6765, 4978, 3950, and 5805, respectively for four quarters and the respective numbers for 2011 were 6831, 4419, 3747, and 5361.The counties reported less than reported and thus the annual number of asthma discharges may not match with the sum of the quarterly numbers in the corresponding years.The asthma hospital discharges numbers indicate that every year the minimum number of cases was observed in third quarter and the maximum number of cases observed in first quarter.
The county wise quarterly ADR for 2009, 2010, and 2011 are represented in Figs.2(a)-2(c) respectively.The names of the counties corresponding to the numbers shown in Figs.2(a)-(c) are listed in supplementary file.The maximum ADR in four quarters of 2009 were observed respectively in the counties of Trinity, San Jacinto, Red River, and Red River.Similarly, the maximum ADR in four quarters of 2010 were observed respectively in the counties of Red River, Red River, Panola and Red River.In 2011, the same data were observed in counties of Red River, Bee, Lamar and Bell.

Air Pollution Data
Air pollution data collected by U.S. EPA's Air Quality System (AQS) at the various monitoring stations located in different counties of eastern part of Texas for the 3 years from 2009 to 2011 were used for the study.The air pollution data used in this study were taken from the United States Environmental Protection Agency (U.S. EPA) air quality system data mart (Source: http://www.epa.gov/airdata/ad_rep_mon.html).The concentrations of three criteria air pollutant parameters (NO 2 , PM 2.5 , and O 3 ) at various monitoring stations located in different counties were retrieved for a three-year period from 2009 to 2011.NO 2 , PM 2.5 and ozone concentrations were obtained respectively for thirty five, forty, and fifty nine monitoring stations.The characteristics of the raw data collected from the website are daily average (24 hrs) concentrations of PM 2.5 , daily maximum 8 hours average concentrations of ozone and daily maximum 1 hour average concentrations of NO 2 .The daily data for each monitoring station were used for determination of monthly average concentrations.Monthly average data were used for determination of quarterly averages calculation.The quarterly average PM 2.5 concentrations are graphically represented in Figs.3(a The descriptive spatial statistics of the three pollutants are represented in Table 1.The minimum quarterly average concentrations of PM 2.5 in four quarters of 2009 were 5.9 μg/m 3 , 7.8 μg/m 3 , 5.5 μg/m 3 , and 6.8 μg/m 3 , respectively and these values were observed in Wichita County.The maximum average concentrations of PM 2.5 in four quarters of 2009 were 13.3 μg/m 3 , 13.7 μg/m 3 , 14.7 μg/m 3 , and 16.1 μg/m 3 , respectively and these values were observed respectively in the Harris County except in second quarter.In second quarter, the maximum concentration was observed in Cameron County.The mean spatial concentrations in four quarters of 2009 were 8.9 μg/m 3 , 11 μg/m 3 , 11.7 μg/m 3 , and 7.8 μg/m 3 , respectively.
Similarly, the minimum quarterly average concentrations of PM 2.5 in four quarters of 2010 were 5.5 μg/m 3 , 6.8 μg/m 3 , 8.7 μg/m 3 , and 6.1 μg/m 3 , respectively and these values were observed in the Wichita County except in third quarter.In the third quarter, the minimum concentration was observed in Fayette County.The maximum average concentrations of PM 2.5 in four quarters of 2010 were 11.7 μg/m 3 , 13.2 μg/m 3 , 14.6 μg/m 3 , and 11.3 μg/m 3 , respectively and these values were observed respectively in the Harris County except in third quarter.In the third quarter, the maximum concentration was observed in Bowie County.The mean spatial concentrations in four quarters of 2010 were found to be 8.3 μg/m 3 , 10.5 μg/m 3 , 11 μg/m 3 , and 9.1 μg/m 3 , respectively.
In 2011, the minimum quarterly average concentrations of PM 2.5 in four quarters were 4.8 μg/m 3 , 9.9 μg/m 3 , 6.9 μg/m 3 , and 4.7 μg/m 3 , respectively and these values were observed respectively in the counties of Nueces, Wichita, Ellis, and McLennan.The maximum average concentrations of PM 2.5 in four quarters were 17.3 μg/m 3 , 17.5 μg/m 3 , 13.2 μg/m 3 , and 10.8 μg/m 3 , respectively and these values were observed in the Harris County except in first quarter (observed in Cameron County).The mean spatial concentrations in four quarters of 2011 were found to be 8.9 μg/m 3 , 11 μg/m 3 , 11.7 μg/m 3 , and 7.8 μg/m 3 , respectively.
The minimum quarterly average concentrations of NO 2 in four quarters of 2009 were 7.3 ppb, 3.9 ppb, 4.9 ppb, and 7.1 ppb, respectively and these values were observed respectively in the counties of Smith, Mclennan, Mclennan, and Harrison.The maximum average concentrations of NO 2 in four quarters of 2009 were 34.5 ppb, 23.4 ppb, 24 ppb, and 30.5 ppb respectively and these values were observed in Harris County except in third quarter (observed in Tarrant County).The mean spatial concentrations in four quarters of 2009 were 19.8 ppb, 13.1 ppb, 21.1 ppb, and 17.9 ppb, respectively.
Similarly, the minimum quarterly average concentrations of NO 2 in four quarters of 2010 were 8.7 ppb, 4.8 ppb, 5.5 ppb, and 9.2 ppb, respectively and these values were observed respectively in the counties of Smith, Galveston, Galveston, and Harrison.The maximum average concentrations of NO 2 in four quarters of 2010 were 36.2 ppb, 25.5 ppb, 23.7 ppb, and 36.9 ppb, respectively and these values were observed respectively in the Harris County.The mean spatial concentrations in four quarters of 2010 were 21.1 ppb, 13.4 ppb, 13.2 ppb, and 20.9 ppb, respectively.
In 2011, the minimum quarterly average concentrations of NO 2 in four quarters were 7.5 ppb, 3.2 ppb, 4.7 ppb, and 10 ppb, respectively and these values were observed respectively in the counties Smith, Galveston, Galveston, and Smith.The maximum average concentrations of NO 2 in four quarters were 33.7 ppb, 24.3 ppb, 27.8 ppb, and 36.2 ppb, respectively and these values were observed respectively in the Harris County except in third quarter (observed in Dallas County).The mean spatial concentrations in four quarters of 2011 were 19.1 ppb, 12 ppb, 14.9 ppb, and 19.8 ppb, respectively.
The minimum quarterly average concentrations of O 3 in four quarters of 2009 were 29.3 ppb, 32.8 ppb, 23.1 ppb, and 23.3 ppb, respectively and these values were observed respectively in the counties of Tarrant, Dallas, Cameron Similarly, the minimum quarterly average concentrations of O 3 in four quarters of 2010 were 30.2 ppb, 33.3 ppb, 25.8 ppb, and 31.9 ppb, respectively and these values were observed respectively in the counties of Harris, Hidalgo, Cameron and Harris.The maximum average concentrations of O 3 in four quarters of 2010 were 42.5 ppb, 49.7 ppb, 49.8 ppb, and 42.8 ppb, respectively and these values were observed in the counties of Orange, Denton, Denton, and Smith.The mean spatial concentrations in four quarters of 2010 were 37 ppb, 42.2 ppb, 38.4 ppb, and 38.3 ppb, respectively.
In 2011, the minimum quarterly average concentrations of O 3 in four quarters were 27.9 ppb, 35.3 ppb, 27.3 ppb, and 30.4 ppb, respectively and these values were observed respectively in the counties Harris, Hidalgo, Cameron, and Harris.The maximum average concentrations of O 3 in four quarters were 41.4 ppb, 52.9 ppb, 65.8 ppb, and 41 ppb, respectively and these values were observed respectively in the counties of Parker, Tarrant, Tarrant, and Jefferson.The

SPATIAL ANALYSIS USING GIS
Geospatial analysis of data not only provides spatial relationship but also helps to deduce association between the multivariate data and to resolve complex issues among the data.During last decade, GIS-based pollution mapping using interpolation techniques such as inverse distance weighted, Kriging, and land use regression modeling (Jerrett et al., 2005) was explored by many researchers for epidemiological studies.The outbreak of asthma has drawn much attention in the past two decades since data all around the world have shown a high rate of asthma morbidity and mortality despite the availability of effective symptomatic treatment.A quick survey of the literature shows that geospatial techniques for statistical analysis and modeling were rarely used in air quality and asthma analysis.In the present study, we demonstrate the use of GIS for analyzing the spatial pattern of ADR and air pollutants to deduce seasonal association between air pollution exposure and ADR, and to visualize major threat areas in the form of maps.
The source of air pollution data for GIS is obtained from measurements of air pollutants that were routinely collected at 134 U.S. EPA administered monitoring stations (thirty five for NO 2 , forty for PM 2.5 , and fifty nine for Ozone) distributed in different counties as shown in Fig. 6.All point data (PM 2.5 , NO 2 , and O 3 ) were entered into a Geographic Information System using ArcGIS software from Environmental Systems Research Inc. (ESRI, 2001).The first stage involved determining the location (latitude and longitude) of air pollution monitoring stations given by U.S. EPA monitoring website for the corresponding stations.The spatial locations of each of the selected monitoring stations along with the pollutant concentrations were fed into the GIS system.For the distribution analysis of asthma discharge rate, we have designed a point shape file by considering a location at the centroid of each county in eastern part of Texas.The attributes entered to particular centroid point were the ADR calculated for the corresponding county.Though, there are various types of interpolation techniques (inverse distance weighted method, Kriging, Cokriging, Radial Basis Functions etc.) for spatial mapping, the present study used inverse distance weighted (IDW) method for spatial mapping in each cases.The software used for the analysis is Geostatistical Analyst Extension module of ArcGIS version 10.2.In IDW interpolation method, a smooth surface is estimated from irregularly spaced data points based on the assumptions that the spatial variation in the feature (O 3 , PM 2.5 , NO 2 , and ADR) is homogeneous over the domain but depends only on the distance between sites.
The method interpolates the point data obtained for various monitoring stations in the study area to predict the concentration in each grid cell over a spatial domain.Root mean square error (RMSE) and correlation coefficient (R) values were used to select accuracy of the model fit which estimate the distribution of air pollutants.The RMSE and R values of each individual model are listed in Table 2.The cross-validation results (RMSE and R) of the three air pollutants (PM 2.5 , NO 2 , and O 3 ) and ADR prediction model for each quarter of three years (2009 to 2011) were determined using ArcGIS Geostatistical extension software.The RMSE should be low and coefficient of correlation (R) should be close to 1 for reliable prediction for the areas where the concentrations of air pollutants and asthma rate is not known.The results shown in Table 2 indicate that R values for each case are very close to 1 and thus can ensure the reliable prediction.RMSE values also relatively low in each of the cases.Since the present work emphasized on the association between the relative concentration level and its association with asthma rate, the results are not influences much on the final outcomes.

RESULTS AND DISCUSSION
In this study, ambient O 3 , PM 2.5 , and NO 2 profiles in the eastern part of Texas were estimated and these values were linked to ADR derived from recorded and interpolated data.The spatial distributions of quarterly average of daily maximum eight hours ozone concentrations were derived using GIS and inverse distance weighted (IDW) techniques.The results of this study clearly illustrate the complex nature of spatial variation in ozone concentrations, and confirm the marked variation in dispersions and precursor's emissions characteristics.The maximum quarterly average of maximum 8 hours ozone concentrations were observed during third quarter (July to September) of each year.In first and fourth quarter of three years, the maximum ozone concentrations were observed in the southern part of the study area.
Similarly, the spatial distribution maps of quarterly average NO 2 concentrations obtained from IDW method are represented in Figs.8(a)-8(l) respectively for twelve quarters of three years (2009 to 2011).Spatial distributions of quarterly average concentration indicate that the maximum concentrations were observed in the counties of Tarrant, Dallas and Harris in most of the occasions.The significant level of NO 2 was observed in the counties situated close to the three major emission zones [Beaumont/Port Arthur (BPA), Dallas/Fort Worth (DFW), and Houston/ Galveston/Brazoria (HGB)].This spatial pattern reflects most likely the aggregated density of emission source.In third quarter of 2011, the higher concentration was observed in all the counties situated in north-west part of the study area.In many occasions the higher concentration was observed in Harris County and extended towards the south-west directions.This may be due to atmospheric conditions.
The spatial distribution maps for PM 2.5 obtained from IDW method are represented in Figs.9(a)-9(l) respectively for twelve quarters of three years (2009 to 2011).Spatial distributions of quarterly average of PM 2.5 concentrations [shown in Figs.9(a)-9(l)] indicate more or less similar distribution patterns in each quarter of three years.The higher concentrations were observed in the counties situated in and around the Harris county, north-west corner (Bowie County) and also in the south-west border of the study area.The lower concentrations were observed in the counties situated in north-west corner of the defined study area.
The spatial distributions of ADR obtained from IDW analysis are represented in Figs.10(a)-10(l) for the twelve quarters of three years (2009, 2010, and 2011) respectively.Spatial distributions of asthma discharge rate [shown in Figs.10(a)-10(l)] indicate that the asthma discharge rates had more or less similar spatial patterns in each quarters or seasons during three years.That is, during the years the higher asthma cases observed in the same regions and there was no significant shift in regions of higher asthma cases.The maximum ADR was observed in and around the counties of Dallas, Harris, Houston, Orange, Red river, Bowie, Orange, Jefferson, and Nueces.

Cross Correlation Analyses
The study sought to investigate the spatio-temporal association between quarterly air pollution exposure level and ADR.To understand the inter-relationships among predictor variables, point data correlation analyses were carried out.
The extracted point data of ADR were compared with the extracted values of pollutant concentrations (NO 2 , PM 2.5 , and O 3 ) for understanding the association between seasonal 21.The cross correlation analyses results are presented in Table 3.
The results (shown in Table 3) indicate that there is no significant association or relation between asthma discharge rate and exposure of air pollutants (PM 2.5 , O 3 , and NO 2 ).Although the results showed uniform sign of correlations between individual air pollutants and ADR in most of the cases, the correlations were not statistically significant.Ozone and NO 2 levels showed negative correlations with ADR in most cases while the PM 2.5 concentrations consistently showed positive correlations with ADR.The correlation coefficients between ADR and ozone concentration are -0.082,-0.086, -0.056, and -0.047, respectively in four quarters of 2009.In 2010, the correlation coefficients between ADR and ozone concentration are 0.096, 0.047, 0.004, and 0.083, respectively in four quarters.
Similarly, the correlation coefficients between ADR and ozone concentration are -0.034,-0.066, -0.007, and -0.020, respectively in four quarters of 2011.Thus, it can be inferred that ozone is negatively correlated with asthma discharge rate in every quarters of 2009 and 2011.In 2010, the results   Year-wise variation in correlation coefficients between asthma discharge and quarterly average pollutant concentration (PM 2.5 , NO 2 , and Ozone) indicates a consistent trend except in the case of ozone in 2010.

Auto Correlation Analyses
Auto correlation analyses sought to investigate the spatial association between quarterly averages of individual air pollution level and asthma rate.Auto correlation analyses were conducted in similar ways as cross-correlation analyses.The correlation results for four variables (PM 2.5 , NO 2 , O 3 , and ADR) are presented in Table 4.The correlation analyses results for asthma discharge rate (ADR) clearly indicate that the correlation coefficients among different quarters are statistically significant at 1 percent significance level in each of the cases.Thus, the occurrences of asthma cases are spatially constant with seasons and years.Similarly, the correlation analyses results for NO 2 indicate that the correlation coefficients among different quarters are statistically significant at 1 percent significance level in each of the cases.Thus, the spatial distributions of NO 2 in each quarter of three years had nearly uniform patterns indicating the distribution is mainly influenced by the emission characteristics.The correlation coefficients between different quarters of PM 2.5 average concentrations were found to be statistically significant at 1 percent significance level in each case except four (first quarter of 2010 and second quarter of 2011, first quarter of 2011 and second quarter of 2011, third quarter of 2010 & second quarter of 2011, and third quarter of 2011 and fourth quarter of 2011).This indicates that the distributions patterns in each quarter of the three study years are more or less uniform.
The correlations coefficients between quarterly average concentrations of ozone were also statistically significant at 1 percent significance level in most of the cases but there is significant number of occasions when the correlations

Fig. 6 .
Fig. 6.Air Pollution Monitoring Station in eastern part of Texas.
Figs. 7(a)-7(l) depicts the spatial patterns of quarterly average O 3 concentrations for twelve quarters during 2009 to 2011 respectively.Figs.7(a)-7(l) clearly indicates that the spatial patterns of O 3 concentrations were not uniform during the corresponding quarters in three years except in second and third quarter.In second and third quarter of three years, the spatial patterns were more or less uniform and the maximum concentrations found in the counties situated within the north-east part of the defined study area.Furthermore, the spatial patterns are also varied in different quarters of same year.This is due to seasonal changes in the local climatic conditions (temperature, wind speed and wind directions) in the regions.Due to change in climatic conditions, the formations and dispersions of ozone also varied in the regions and hence showed no uniform spatial patterns of ozone concentrations.The variability of Texas' climate is a consequence of interactions between the state's unique geographic location on the North American continent and several factors that result because of the state's location.The movements of seasonal air masses such as arctic fronts from Canada and subtropical west winds from the Pacific Ocean and northern Mexico influence the dispersion of ozone concentrations.The monthly windrose diagrams for three locations (Austin, Beaumont/Port Arthur, and Houston) were downloaded from the website of Texas Commission on Environmental Quality (Source: http://www.tceq.state.tx.us/ airquality/monops/windroses.html).The windrose diagrams indicate the hourly wind speed and wind direction from 1984 to 1992.These plots revealed that wind speed and direction significantly varied in three locations.The most interesting facts observed from the analyses that high wind speed observed during the months of June to August.During the months of December to March, the wind speed was relatively low in the region.Thus, the dispersion of air pollutants was significantly influenced by local climatic conditions.

Fig. 7 .
Fig. 7. Spatial distribution of quarterly average of maximum daily eight hours ozone concentrations during 2009 to 2011 (a) first quarter of 2009 (b) second quarter of 2009 (c) third quarter of 2009 (d) fourth quarter of 2009 (e) first quarter of 2010 (f) second quarter of 2010 (g) third quarter of 2010 (h) fourth quarter of 2010 (i) first quarter of 2011 (j) second quarter of 2011 (k) third quarter of 2011 (l) fourth quarter of 2011.

Fig. 8 .
Fig. 8. Spatial distribution of quarterly average of maximum daily one hour NO 2 concentrations during 2009 to 2011 (a) first quarter of 2009 (b) second quarter of 2009 (c) third quarter of 2009 (d) fourth quarter of 2009 (e) first quarter of 2010 (f) second quarter of 2010 (g) third quarter of 2010 (h) fourth quarter of 2010 (i) first quarter of 2011 (j) second quarter of 2011 (k) third quarter of 2011 (l) fourth quarter of 2011.

Fig. 9 .
Fig. 9. Spatial distribution of quarterly average of 24 hours PM 2.5 concentrations during 2009 to 2011 (a) first quarter of 2009 (b) second quarter of 2009 (c) third quarter of 2009 (d) fourth quarter of 2009 (e) first quarter of 2010 (f) second quarter of 2010 (g) third quarter of 2010 (h) fourth quarter of 2010 (i) first quarter of 2011 (j) second quarter of 2011 (k) third quarter of 2011 (l) fourth quarter of 2011.

Fig. 10 .
Fig. 10.Spatial distribution of asthma rate during 2009 to 2011 (a) first quarter of 2009 (b) second quarter of 2009 (c) third quarter of 2009 (d) fourth quarter of 2009 (e) first quarter of 2010 (f) second quarter of 2010 (g) third quarter of 2010 (h) fourth quarter of 2010 (i) first quarter of 2011 (j) second quarter of 2011 (k) third quarter of 2011 (l) fourth quarter of 2011.

Table 1 .
Descriptive statistics of annual average pollution concentrations and asthma rate in eastern part of Texas.

Table 2 .
Cross validation results of prediction models.