We determined the performance of different Circulation Type Classifications (CTCs) to stratify air pollutants concentrations in Polish cities in winter. Our analysis is based on 15 CTCs calculated by COST 733 as well as on 5 manual universally used manual weather type classifications. For this purpose we compared and tested the explained variation (EV) and within-type standard deviation (WSD) methods. Finally, EV method has been chosen for evaluating classifications for daily values of SO2, NO2, PM I O and CO as well as vertical dispersion conditions obtained from SODAR data. We also presented the methodology of choosing smog episode days based on 90-percentile values. For the winter smog episodes data from Krakow different classifications have been compared using Gini coefficient method. The best results for separate air pollution data series as well as for smog episode days were obtained for Hess-Brezowski Gro/3wetterlagen classification (HBGWL). Moreover, good results were obtained for the based on principal component analysis PCACA classification, Polish Niedzwiedz TCN2I, modified Polish Litynski LITTc, modified Lamb LWT2, and three modified HBGWL (GWTC26, OGWL, OGWLSLP) classifications. The same classifications except for HBGWL are good for SODAR data. For the best CTCs, the differences between various classes are visible, however a big scattering is still observed. Main urban air pollution problems arise in situations when flow with Southerly component is observed. Correlations between air pollution data and SODAR data (calculated for marginal means obtained for different classes) confirm a negative role of both low height of the ground-based inversion and long duration of the low-level elevated inversion in urban areas.

The paper presents a method of identifying distant emission sources of fine particulate matter PM2.5 affecting significantly PM2.5 concentrations at a given location. The method involves spatial analysis of aggregate information about PM2.5 concentrations measured at the location and air masses backward trajectories calculated by HYSPLIT model. The method was examined for three locations of PM2.5 measurement stations (Diabla Góra, Gdańsk, and Katowice) which represented different environmental conditions. The backward trajectories were calculated starting from different heights (30, 50, 100 and 150 m a. g. l.). All points of a single backward trajectory were assigned to the PM2.5 concentration corresponding to the date and the site of the beginning of trajectory calculation. Daily average concentrations of PM2.5 were used, and in the case of Gdańsk also hourly ones. It enabled to assess the effectiveness of the presented method using daily averages if hourly ones were not available. Locations of distant sources of fine particulate matter emission were determined by assigning to each grid node a mean value of PM2.5 concentrations associated with the trajectories points located within the so-called search ellipse. Nearby sources of fine particulate matter emission were eliminated by filtering the trajectories points located close to each other (so-called duplicates). The analyses covered the period of January-March 2010. The results indicated the different origin of air masses in the northern and southern Poland. In Diabla Góra and Gdańsk the distant sources of fine particulate matter emission are identified in Belarus and Russia. In Katowice the impact of the Belarusian PM2.5 emission sources was also noted but as the most important fine particulate matter emission sources were considered those located in the area of Romania, Hungary, Slovakia and Ukraine.