Relative contributions of population genetic processes to molecular evolution have been an important and controversial question in evolutionary genetics. We quantify the impacts of population genetic processes underlying evolution of human influenza virus H3N2 on reducing the effective population size of HA (hemagglutinin) segment. For this analysis, we use computer simulation of viral population reproducing in discrete generations where each virus sequence represents the state of viruses infecting one host. We used mutation rate and selection coefficients for beneficial and deleterious alleles estimated by observed changes of variant allele frequencies and nonsynonymous-to-synonymous substitution ratios. The effective size of population in which selective sweeps occur is inferred by the rate of soft selective sweeps. This size relative to the census size of viral population is used to estimate how much reduction in the effective size occurs by other processes: background selection and meta-population dynamics. The variation-reducing power of background selection and meta-population dynamics is greater than that of recurrent positive selection and must be crucial in explaining the observed level of sequence diversity in H3N2. We also found that per-site synonymous diversity of genomic segments varies with segment length and the rate of adaptive evolution, which can be explained by interplay between negative and positive selection. The joint analysis of these processes on the genomic pattern of variation provides insight on the rate of reassortments in H3N2 viruses, which might otherwise be difficult to obtain.