We examined the distributions of s-hr averaged concentrations by δ
2-δ
3 diagram test and derived the following results:
(1) The distribution of s-hr averaged concentrations is the same as the hourly distribution at most of the monitoring stations.
(2) The Point (δ
2, δ
3) moves straightly to origin as increasing averaging time on δ
2-δ
3 diagram.
From these results, we propose the new arrowhead chart model which fit the observed data better than any other model when air pollutant concentrations are approximated to Pearson distributions. This model is numerically expressed as follows:
μ
1'(s)=μ
1'(1)
δ
2 (s)=δ
2 (1) G (s)/G (1)
δ
3 (s)=δ
3 (1) G (s)/G (1),
where μ
1'(1), μ
1'(s): means of hourly concentrations and s-hr averaged concentrations respectively,
δ
2(1), δ
2(s): variance coefficicnts of hourly concentrations and s-hr averaged concentrations respectively,
δ
3 (1), δ
3 (s): skewness coefficients of hourly concentrations and s-hr averaged concentrations respectively,
G (s) =2/s∫s0 (1-τ/s)γ(τ) dτ, γ(τ): auto-correlation coefficient.
Using this arrowhcad chart modcl, we investigate the method of estimating δ
2 (1) and δ
3 (1) from δ
2 (s) and δ
3 (s) when auto-correlation coefficient is approximated to exponential curve or exponential curve plus sine wave.
View full abstract