Methodology for the simulation of the long-term average concentration of air pollutants based on a mesoscale meteorological model coupled with an Eulerian dispersion model is discussed. The procedure of the method is, (1) days in the examined term are classified by some indices characterizing the weather pattern in the region of interest through the course of the day, and occurrence frequencies of the classified groups are counted, (2) numerical simulations of pollutant distribution are carried out for days representative of major weather patterns, and (3) the simulation results are averaged with weighting by the occurrence frequencies. The results obtained by this aggregation are regarded as the average distribution in the examined term.
In the present study, tentative indices for the classification of weather patterns have been defined, as well as the procedure for the selection of representative days to be simulated. The method has been applied to the meteorological data during the three year period from April 1994 to March 1996 in the Kanto district. As a result, half of the days in the examined years were contained in fourteen major weather patterns, each of which was divided into two subgroups, and their representatives, 28 days in total, were determined.
Simulations of local weather and NOx dispersion for these days, and the aggregation of long-term average concentrations, were attempted using the meso-meteorological model ANEMOS coupled with an Eulerian transport model. The correlation coefficients between the annual average or three-year average concentrations estimated by the aggregation and those measured consistently exceeded 0.8, revealing good feasibility of the method.
Furthermore, the application of this method to different types of pollutants as ozone will be discussed in Part II.
View full abstract