Atmospheric zyxwvutsrqponmlkjihgfedcbaZYXWVUTSRQPONMLKJIHGFEDCBA Enoironment Vol. 23, No. II, pp. 2571-2595, 1989. Printed in Great Britain. o!w-6981/89 IF3.M)+o.o0 0 1989 Pergamon Pressplc NUMERICAL MODELLING OF THE LONG RANGE ATMOSPHERIC TRANSPORT OF SULPHUR DIOXIDE AND PARTICULATE SULPHATE TO THE ARCTIC TROND IVERSEN* Norwegian Institute for Air Research, P.O. Box 63, N-2001 Lillestrom, Norway (First received 1 March 1988 zyxwvutsrqponmlkjihgfedcbaZYXWVUTSRQPONMLK and injinalform 14 July 1988) Abstract-A model for the simulation zyxwvutsrqponmlkjihgfedcbaZYXWVUTSRQPONMLKJIHG of zyxwvutsrqponmlkjihgfedcbaZYXWVUTSRQPONMLKJIHGFEDC long range transport (LRT) of SO, and particulate SO:- has been constructed. It empioys 10 layers with isentropic coordinate surfaces and is Eulerian. The horizontal and vertical advection is calculated by an anti-diffusively corrected upwind scheme (Smolarkiewica, Mon. We@. Rev, 111, 479-486, 1983). The oxidation of SO, is linearized (e.g. Eliassen and Saltbones, Atmospheric Environment, 17,1457-1473,1983) with a reaction rate depending on latitude and season. Vertical turbulent diffusion and wet deposition are parameterized, and the emissions are mixed vertically to a height dependent on the local static stability in the source areas. The model includes a meteorological ‘preprocessor which estimates heating and precipitation from analyses of wind, mass field and humidity, together with knowledge ofgrouud surface properties. It has been applied to estimate the transport to the Arctic during two periods in 1983. Emission surveys were given by Semb (NILU, 1985), and separated into four main source regions in order to estimate the relative contribution from each. The model gives reasonable results as compared with measurements, however, the ground level S concentrations are somewhat exaggerated by the model. The difference in concentration level between the cold and warm seasons is reproduced, and the day to day variations are simulated reasonably well. During the simulation periods, the European and Soviet emissions are estimated to be the major contributors to Arctic air pollution. The North American emissions contribute sigticantly to upper level Arctic haze. Sources far from the Arctic contribute to upper level pollution, while close sources contribute at low levels. This pattern is much more pronounced in the cold period than during the warm, in accordance with the seasonal variation in the tropospheric static stability. The simulations indicate that the annual variation in meridional transport processes is a more important factor than the annual variation in Arctic wet deposition processes in determining the seasonal cycle in Arctic air pollution. Key word zyxwvutsrqponmlkjihgfedcbaZYXWVUTSRQPONMLKJIHGFEDCBA index: Long range transport, Arctic haze, sulphur pollution, air pollution modelling, isentropic surfaces. INTRODUCTION The scientific interest in Arctic air quality has in- creased considerably during the latest 10 years. Sev- eral me~urement campaigns have been undertaken, giving evidence of a quite polluted Arctic in the cold seasons (AGASP, 1984; Arctic Air Chemistry, 1985; Ottar et al., 1986). Ground based measurements reveal a clear seasonal cycle both in pollutant concentrations on average and in short term variability (Barrie and Hoff, 1985; Iversen and Joranger, 1985; Iversen, 1985). The maximum pollution activity occurs normally in March. Arctic air pollution also occurs in the free troposphere above the turbulent boundary layer up to 5ooO m (Joranger and Ottar, 1984; Schnell and Raatz, 1984; Raatz et al., 1985a,b; Pacyna et al., 1986). A numerical model aimed at simulating the Arctic air quality must, within reasonable accuracy, be able to reproduce these three main characteristics: the sea- sonal cycle, the episodicity and the deep vertical distribution, *Present affiliation: The Norwegian Meteorological In- stitute, P.O. Box 43, Blindern, N-0313 Oslo 3, Norway. So far there have been very few attempts at quanti- tative estimates of Arctic pollutant concentrations. Earlier models for LRT included parts of the Arctic, but the simulations were not aimed at Arctic air quality and the imputation domain did not cover sufficiently large parts of the Northern Hemisphere (e.g. Eliassen and Saltbones, 1983). Air mass trajector- ies for the lower part of the troposphere (850 hPa) computed in the same domain, was used for semi- quantitative discussions about the origins of Arctic air pollutants by Heintzenberg and Larssen (1983). A simiiar study on a larger domain was presented by Miller (1981) with 500 hPa trajectories, and trajector- ies at several atmospheric levels were investigated by Harris (1984). Another method has been to sub- jectively classify ‘transport pathways’ from medium to high latitudes by studies of weather maps (Rahn, 1979; Rahn and McCaffrey, 1980; Raatz, 1984; Raatz and Shaw, 1984), and on that basis distinguish between different source areas. This method suffers from its subjective elements and a priori assumptions about main transport layers. Air mass pathway identifica- tions have also been used together with analyses of trace element concentrations and measurements of 2571