Research Journal of Recent Sciences ______ ______________________________ ______ ____ ___ ISSN 2277 - 2502 Vol. 2 ( ISC - 2012 ), 285 - 289 (201 3 ) Res.J. Recent .Sci. International Science Congress Association 285 Regression modeling of G aseous A ir P ollutants and M eteorological P arameters in a S teel C ity, Rourkela , India Kavuri N.C., Paul K.K. and Roy N. De partment of C ivil E ngineering, N ational I nstitue of T echnology, R oukela 769008, O disha, INDIA Available onli ne at: www.isca.in Received 5 th Novem ber 2012 , revised 2 6 th December 2012 , accepted 23 th January 201 3 Abstract Traditional algorithms such as diffusion model employed for estimating the distribution of pollutants in amb ient air are complicated involving the solution of complex differential equations. Employing multivariate statistical models which attempt to find the underlying relationships between a set of inputs and outputs may give an easy way to predict these gaseou s pollutants. A multiple linear regression model has been developed for predicting sulphur dioxide, oxides of nitrogen, ammonia and carbon monoxide in a steel city using the meteorological parameters like temperature, relative humidity, wind speed and wind direction. Results have shown a good correlation between predictors and predicted values (R 2 ≈0.7). A unifor ffct of the meteorological parameters in distributing these gaseous pollutants has been observed. Keywords: R egression modelling, steel city, c orrelation analysis. Introduction Air pollution is one of the basic problems being faced in urban areas. Exposure to ambient air pollution has been associated with a number of different health outcomes such as, skin irritation, respiratory and cardiov ascular diseases. Besides deleterious effects on human health, it also causes evident negative effects on ecosystems, materials, and the visibility 1,2 . These effects are giving urban air quality an increasing attention these days. The main constituents of a ir pollution path of the urban atmosphere are emission and transmission of air pollutants resulting in the ambient air pollution 3 . Although, urban air pollution was considered as a problem mainly associated with domestic heating and industrial emissions in itially, contribution of traffic emissions are also considered at present 4 . All emitted pollutants are dispersed and diluted in the atmosphere and these processes are strongly affected by meteorological conditions, such as rainfall, air temperature, wind s peed (WS), and wind direction (WD) 5 . Therefore, air quality in the urban areas has been generally interpreted with the combination of various meteorological factors 6 . For estimation of flow of energy and performance of systems analytical computer codes ar e often used. The algorithms employed are usually complicated involving the solution of complex differential equations. These programs usually require a large computer power and need a considerable amount of time for accurate predictions. One approach to p redict atmospheric air quality is to use a detailed atmospheric diffusion model. Such models aim to resolve the underlying physical and chemical equations controlling pollutant concentrations and therefore require detailed emissions data and meteorological fields. Collet and Oduyemi provided a detailed review of this particular type of model 7 . The second approach is to devise statistical models which attempt to determine the underlying relationship between a set of input data and targets. Regression modelli ng is such a statistical approach that has been applied to air quality modelling and prediction in a number of studies 8, 9 . Rourkela, well known steel city, where one of the major industrial complexes of SAIL (Steel Authority of India Limited) for steel p roduction is located has been considered for the present study. Four gaseous pollutants namely sulphur dioxide (SO 2 ), oxides of nitrogen (NO x ), ammonia (NH 3 ) and carbon monoxide (CO) were considered for the regression modelling using meteorological paramet ers such as wind direction (WD), wind speed (WS), temperature (T) and relative humidity (RH). Multiple linear regression (MLR) was used for studying the linear correlation between individual gaseous pollutants and meteorological parameters. Study area: A steel city, Rourkela (2212' N, 8454' E) situated at 219 m above msl is selected as a study area in the present research work. It is one of the most important industrial cities in the Sundargarh district of the state of Odisha, India. It is located at th e heart of a rich mineral belt and surrounded by a range of hills and encircled by rivers. As per 2011 census report of India, population of Rourkela is 6,89,298. The city is spread over an area of 121.7km 2 in close proximity of iron ore, dolomite, lime st one and coal belts. The perennial Koel River flows through this valley and meets another perennial river Sankh at Vedavyas on the outskirts of Rourkela. It has a tropical climate having average annual rainfall between 160 – 200 cm. Three monitoring sites h ave been selected for the present ambient air quality study. Indira Gandhi Park, Udit Nagar and Jalda which are in the proximity of the steel plant emissions as shown in f igure 1. Research Journal of Recent Sciences ______ _ _ __ _____________________________ ______________ _ ________ ISSN 2277 - 2502 Vol. 2 ( ISC - 2012 ), 285 - 289 (201 3 ) Res.J.Recent.Sci International Science Congress Association 286 Figure - 1 Study area. Three sampling stations were chosen, 1) IG Park 2) U dit Nagar 3) Jalda Material and Methods Sampling Protocol: The concentrations of gaseous pollutants (NO x , SO 2 , and NH 3 ) have been measured by using absorbing solution in their respective impingers attached to a respirable dust sampler. Sampled air is al lowed to pass through each impinger at a flow rate of 0.6lpm for a duration of 8h. The preparation of absorbing solutions and their analysis for NO x and SO 2 have been performed according to the ASTM standards, D1607 - 91 and D2914 – 01 respectively. Analysis o f NH 3 was done following the Nesselerization method prescribed by American Public Health Association (APHA, 1985) 10 . The concentration of CO is measured by using the help o f an instrument called HT - 1000 c arbon monoxide metre (HT - 1000). Sampling has been performed twice in a week, preferably one weekday and another weekend. All the samples were collected during January to December 2011. A total of 77 samples have been collected during the study period. Multiple Linear Regression : Multiple linear regression attempts have been made to model the relationship between two or more explanatory variables and a response variable by fitting a linear equation to observed data. Every value of the independent variable x is associated with a value of the dependent variab le y . The population regression line for p explanatory variables x 1 , x 2 , ... ,x p This line describes how the mean response y changes with the explanatory variables. The observed values for y vary about their means y and are assumed to have the same standard deviation σ . The fitted values b 0 , b 1 , ..., b p estimate the parameters β 0 , β 1 ,... β p of the population regression line. Since the observed values for y vary about their means y , the multiple regression model includes a term for this variation. In words, the model i s expressed as Research Journal of Recent Sciences ______ _ _ __ _____________________________ ______________ _ ________ ISSN 2277 - 2502 Vol. 2 ( ISC - 2012 ), 285 - 289 (201 3 ) Res.J.Recent.Sci International Science Congress Association 287 Data = Fit + Residual where the "FIT" term represents the expression . The " Residual " term represents the deviations of the observed values y from their means y , which are normally distributed with mean 0 and varianc e σ 2 . The notation for the model deviations is ε . Formally, the model for multiple linear regression, given n observations, is for All the regression analysis was done by using IBM - SPSS software. Results and Dis cussion During the study period the average concentration of SO 2 , NO x , NH 3 and CO ar found to  26.9μg/ 3 , 38.21μg/ m 3 , 366.77μg/ m 3 and 572.11μg/ m 3 respectively. The descriptive statistics of both gaseous pollutants and meteorological parameters are pre sented in table 1. Monthly averages of gaseous pollutant concentrations were presented in figures 2 and 3. Very low concentrations of gaseous pollutants can be observed during the monsoon season of the year and higher concentrations during spring. This ca n be attributed to the atmospheric stability and the inversion effects that take place during spring season and increase in the wind turbulences during the monsoon season. Table - 1 Descriptive statistics of gaseous pollutants and meteorological pa rameter during the study period WS (m/s) Temperature ( o C) RH WD (degrees) SO 2 (g/m 3 ) NO x (g/m 3 ) NH 3 (g/m 3 ) CO (g/m 3 ) N 92 92 92 92 92 92 92 92 Mean .70 26.0132 65.5868 216.12 26.9043 38.2163 366.7761 572.1108 Std. Deviation .849 3.49877 12.16226 61.928 6. 98646 10.63080 165.60986 268.44990 Range 4 15.80 50.40 225 26.77 45.54 765.58 1033.43 Minimum 0 17.00 35.20 45 14.86 20.53 117.92 104.45 Maximum 4 32.80 85.60 270 41.63 66.07 883.50 1137.89 Figure - 2 Monthly average concentrations of SO 2 and NO x during the study period Figure - 3 Monthly average concentrations of NH 3 and CO during the study period Research Journal of Recent Sciences ______ _ _ __ _____________________________ ______________ _ ________ ISSN 2277 - 2502 Vol. 2 ( ISC - 2012 ), 285 - 289 (201 3 ) Res.J.Recent.Sci International Science Congress Association 288 The regression analysis of gaseous pollutants with the meteorological parameters has shown a good correlation between them ( t able 2 and 3). Almost al l of the gaseous pollutants have resulted in similar regression coefficient values indicating that the effect of meteorological parameters on these four gaseous pollutants is almost uniform. SO 2 has shown a relatively low linear correlation with the meteoro logical parameters. This can be due to the presence of various excluded atmospheric reactions involving SO 2 . For, e.g., well - known photo oxidation reactions resulting with ozone formation influence SO 2 concentrations considerably according to the following equations 11 : Such disregarded atmospheric reactions of SO 2 may cause some decrease in the efficiency of regression model run with the meteorological factors. Conclusion The monthly average concentrations of four gaseous pollutants taken twice a week (one weekday and one weekend) for one year have shown higher concentrations during spring and lower concentrations during monsoon which may be due to the inversion effect in spring and the wash out during monsoon. The regression analysis of gaseous pollutants with meteorological parameters has shown a good correlation between them. One of the interesting observations that can be made is similar values of the regression coefficients which may be indicating the unifo rm effect of the meteorological parameters in distributing these gaseous pollutants.SO 2 has shown a relatively low linear correlation with the meteorological parameters which can be attributed to the photooxidation reactionsresulting with ozone. Acknowledg ments Authors acknowledge Department of Science and Technology (DST), Government of India, New Delhi for sponsoring and funding this study as Fast Track Scheme for Young Scientist. Table - 2 Regression Model summery for different gaseous pollutants with the meteorological parameters. Model R R Square Adjusted R Square Std. Error of the Estimate Residuals NO x .849 a .720 .704 5.78058 .982 SO 2 .834 a .696 .679 3.96064 1.322 NH 3 .857 a .734 .720 87.70595 .910 CO .853 a .727 .712 144.10201 .962 Table - 3 R egression model coefficients of meteorological parameters for different gaseous pollutants. Model Unstandardized Coefficients Standardized Coefficients t Sig. B Std. Error Beta NO x (Constant) 94.159 12.352 7.623 .000 WS - .474 .798 - .038 - .594 .554 RH - .468 .474 - .535 - .988 .327 WD - .003 .012 - .016 - .228 .820 Temperature - .936 1.661 - .308 - .563 .575 SO 2 (Constant) 64.469 8.463 7.618 .000 WS - .549 .547 - .067 - 1.005 .318 RH - .246 .325 - .428 - .757 .452 WD - .002 .008 - .015 - .217 .829 Temp erature - .795 1.138 - .398 - .699 .487 NH 3 (Constant) 1188.449 187.406 6.342 .000 WS - 2.714 12.106 - .014 - .224 .823 RH - 9.642 7.188 - .708 - 1.341 .184 WD - .041 .178 - .015 - .231 .818 Temperature - 6.863 25.197 - .145 - .272 .786 CO (Constant) 1930.312 307.910 6.269 .000 WS - 14.436 19.890 - .046 - .726 .470 RH - 14.241 11.809 - .645 - 1.206 .232 WD - .038 .292 - .009 - .129 .898 Temperature - 15.606 41.399 - .203 - .377 .707 Research Journal of Recent Sciences ______ _ _ __ _____________________________ ______________ _ ________ ISSN 2277 - 2502 Vol. 2 ( ISC - 2012 ), 285 - 289 (201 3 ) Res.J.Recent.Sci International Science Congress Association 289 References 1. Gurjar B.R., Butler T.M., Lawrence M.G. and Lelieveld J., Evaluatio n of Emissions and Air Quality in Megacities, Atmos. Environ. , (42) , 1593 ( 2008 ) 2. Ozden O., Dogeroglu T. and Kara S., Assessment of Ambient Air Quality in Eskis ehir, Turkey, Environ. Int. , ( 34 ) , 678 ( 2008 ) 3. Mayer H., Air Pollution in Cities, Atmos. Environ ., 33, 4029 (1999) 4. Venegas L.E. and Mazzeo N.A., Modelling of Urban Background Pollution in Buenos Aires City Argentina, Environ. Modell. Softw., (21) , 577 (2006) 5. Cuhadaroglu B., Demirci E., Influence of Some Meteorological Factors on Air Pollution in Trab zon City, Energy Build, (25), 79 (1996) 6. Banerjee T., Singh S.B. and Srivastava R.K., Development and Performance Evaluation of Statistical Models Correlating Air Pollutants and Meteorological Variables at Pantnagar, India, Atmos. Res., 99, 505 (2011) 7. Coll et R., Oduyemi K., Air Quality Modelling , A Technical Review of Mathematical Approaches , Meteorological Applications, 4, (1997) 8. Shi J., Harrison R., Regression Modelling of Hourly NO x and NO 2 Concentrations in Urban Air in London, Atmospheric Environment, (24), 31 (1997) 9. Ziomass I., Melas D., Zerefos C. and Bais A., Forecasting Peak Pollutant Levels from Meteorological Var iable, Atmospheric Environment, (24), 29 (1995) 10. APHA, Nitrogen, Standard Methods for the Examination of Water and Wastewater, 18 th Edition, American Public Health Association , Method 417 , B, 379 - 381 (1985) 11. Sidebottom H.W., Badcock C.C., Jackson G.E., Calvert J.G., Reinhardt G.W. and Damon E.K., Photooxidation of Sulfur Dioxide, Environ. Sci. Technol., 6 (1), 72 (1972)