1988 Volume 23 Issue 4 Pages 233-240
In order to select the best distributional models for size fractionated particles and sulfates concentrations, measurements were carried out with Andersen sampler for continuous 1-year period in Hiroshima City. Sampling time was about 10 days. Four distributional models were chosen for selection: Normal, Lognormal, Gamma and Inverse Gaussian. Estimation of the parameters for these models are obtained using the maximum log-likelihood method. Normal is best for fine particles (Dp ≤ 1. 1 μm), Inverse Gaussian for coarse particles (Dp≥ 2. 1 μm), fine sulfates and coarse sulfates, Gamma for suspended particulate matter (fine plus coarse particles), respectively.
The size distributions of 10, 50, 90 percentile values indicate bimodal patterns with the peak at 0.65μm and 3. 3μm or 4. 7μm for particles, monomodal patterns with the peak at 0. 65μm for sulfates, respectively. It is suggested that these results are attributed to differences of the formation mechanism between fine and coarse particles and most of sulfates are secondary constituents.
