The statistical characters of PM10 concentration in Taiwan area

摘要: Abstract The concentrations of air pollutants varied inherently with meteorological conditions and pollutant emission level. From the statistical properties (probability density) pollutants, it is easy to estimate how many times exceedance compared quality standards occurs. In this paper, three distributions (lognormal, Weibull type V Pearson distribution) were utilized simulate PM 10 concentration distribution in Taiwan areas. Air data monitoring stations, Hsin-Chu, Sha-Lu Gian-Jin, taken compare characters during a five-year period (1995–1999). Two parametric estimating methods, method moments least squares, used parameters these theoretic distributions. Therefore, frequency source reduction can be predicted from These results show that lognormal best represent daily average concentration. Between two estimation squares has more accurate than method. Hsin-Chu stations are all unimodal distributions, but Gian-Jin bimodal distribution. measured station divided into seasons, computed individually. reproduced distribution, which combined agrees well data. This result shows greatly different areas, could influenced by local seasons. addition, probabilities exceeding standard (PM >125 μg m −3 ) sources meet for successfully.