Testing the ability of trajectory statistics to reproduce emission inventories of air pollutants in cases of negligible measurement and transport errors

Methods of trajectory statistics have often been used in combination with a ir quality measurements to identify source areas of air pollutants. Due to many limitations, it is not clear which share of the deviations between res ults of the statistics and (independent) emission inventories could really be attributed to the methods applied, and which share is due to problems wi th the input data, for example wrong measurements and erroneous air traject ories. In this study, we tested two well-known statistical methods under co nditions of negligible measurement and transport errors by applying them to concentrations of total reactive nitrogen computed by a Lagrangian box mod el using exactly the same trajectories as input. The resulting concentratio n fields of total reactive nitrogen were compared to the emission inventory of nitric oxides used by the model. Two test data sets were constructed, n amely one "ideal" set neglecting temporal variations of emissions and depos ition, and one "real" set considering these processes. As a result, it was shown that both methods were in principle successful in reproducing the mos t important features of the emission inventory. It can be concluded that su ch statistical methods, with some limitations, are helpful in investigating source-receptor relationships of air pollutants, provided that the measure ments are accurate and representative and that the trajectories reasonably well describe pollutant transport. (C) 1999 Elsevier Science Ltd. All right s reserved.