Experimental study of the air pollution transport by synchronised monitoring of atmospheric aerosols

Particulate pollution transport is estimated by means of cross-correlation and regression analyses. The aerosol particle size number spectrum in the f orm of 12 fraction concentrations is measured in units cm(-3) (particles pe r cm(3)) every 10 minutes during three approximately 20-day measurement cam paigns simultaneously in two measurement points in Estonia. The distance be tween these points is approximately 100 km for two campaigns, and 7 km for the third campaign. Two electrical aerosol spectrometers designed at Tartu University, having a wide particle diameter range(10 nm-10 mu m), are used. The spectrometer's record is the mean particle spectrum for the 10 min mea surement time. The air pollution transport is investigated during the time intervals when the air mass moves from one measurement point towards the ot her. The time series of aerosol size fraction concentrations for both locat ions are prewhitened to eliminate autocorrelation and to achieve stationary series of the ARIMA residuals. Then the cross-correlation function of thes e two series of residuals is calculated The time lag corresponding to the m ode of this function is treated as the mean time of pollution transport fro m the windward measurement point to the leeward one. For the submicron aero sol fraction (d = 60 nm-1 mu m) 3-5 h time lags are found. Mostly these tim e lags coincide with the mean wind velocities on some of the pressure level s (ground, 850 and 700 hPa) available in the study. In cross-wind cases the fraction concentrations measured in two points separated by 100 km were un correlated, but for the two points separated by 7 km there was quite a high correlation with zero time lag. The part of local and distant sources in t he formation of the particle concentration in the leeward location is estim ated by regression analysis. (C) 1999 Elsevier Science Ltd. All rights rese rved.