Atmospheric Environment Vol. 24A, No, 4, pp. 783 788, 1990. 0(04 0981/90 $3.00 + 0.00 Printed in Great Britain. Pergamon Press plc PREDICTED AND OBSERVED CONCENTRATIONS OF SO2, SPM AND NOx OVER DELHI M. P. SINGH, P. GOYAL, T. S. PANWAR, P. AGARWALand S. NIGAM Centre for Atmospheric Sciences, Indian Institute of Technology, Hauz Khas, New Delhi 110 016, India and N. BAGCHI Central Pollution Control Board, Dr Mukherjee Nagar, Delhi, India (First received 15 February 1988 and received for publication 9 February 1989) Abstract Three air pollutants, SO2, SPM and NO~, which are identified as major pollutants in urban cities, have been considered for a comparative study between predicted and observed concentrations of these pollutants at different receptor points over Delhi. A receptor-oriented Gaussian plume model (IITLT) and Climatological Dispersion Model (CDM) have been used to estimate the long term concentration of non- reactive pollutants due to emissions from area and point sources. Modified stability parameters are used for low wind speed and calm intervals. Monthly mean concentrations of these pollutants for the period of November, December and January 1987-1988 have been computed and the results obtained from the two models are found to be close to the observed values for this period. An error analysis of 54 pairs of observed and predicted concentrations show that performances of both the models are highly satisfactory, giving an RMSE of 1.61 for IITLT Model and 1.50 for CDM. Key word index: Sourceinventory, Gaussian plume model, CDM, IITLT, wind sector, stability overpredic- tion, RMSE. INTRODUCTION In recent years 502, SPM and NOx levels over metropolitan cities of India are increasing. Any new industrial siting powered by coal or fuel oil is likely to contribute further to the ambient levels of these pollu- tants. In the context of our industrial development, the studies described below should provide a valuable tool in decision making processes and planning and management. SO 2, SPM and NOx concentrations due to local sources in Delhi have been computed by using an analytical dispersion model, IITLT and Climato- logical Dispersion Model (CDM). While the IITLT model has been developed in IIT, CDM is an operat- ional model in the U.S. Computations from both the models are compared with observations collected at Ashok Vihar, Shazadabagh, Shadara, Siri Fort, Jan- akpuri and Nizamuddin (Fig. 1). These receptors have been chosen on the basis of availability of data. The analytical dispersion model developed in this study is based on Gaussian formulation. A gridded source inventory has been developed over an area of 24 km x 24 km as shown in Fig. 1. MODEL CHARACTERISTICS A receptor oriented Gaussian plume model has been adopted to obtain the concentration of 802, SPM and NOx due to area sources along with month- ly wind roses and stability frequencies. The ground level concentration at each receptor due to area sources is the sum of all contributions of grids upwind of the receptor and is given by x/2/re -b) C=yCj - - y~ 1-~- ' - Qj(xj+, xj (1) j a(1 - b)u j where Cj is the concentration due to an area source of strength Qj located at thejth upwind grid, a and b are the stability parameters and u is the mean wind speed, Xj and Xj+ 1 are the upwind distances of the jth and (j+ 1)th grid from the receptor point (Hanna, 1974). Equation (1) requires that the grid system be aligned with the wind. By successive application of Equa- tion (1) to 16 wind directions, the model is able to give the mean monthly SO2 concentration at any receptor point after weighting it with the appropriate combina- tion of wind frequency class and stability frequency. Alignment of the source inventory grids with each of the wind direction is achieved in the manner given by Dabberdt et al. (1973). Figure 2 shows the alignment geometry. For the elevated point sources (three stacks of Indraprastha Power Plant) a simple adaption of the basic dispersion equation due to Smith (1968) is used to calculate the concentration Cp at different receptor points: Cp= _360fQ exp -2~2 ~ (2) x / 2 (~3/2 UX tTze 783