Distribution characteristics of trace elements and ionic species of aerosol collected at Canadian high arctic

Distributions of chemistry data have immense influence on the analytic techniques generally applied to source-receptor modeling. Some advanced multivariate receptor methods could be adversely affected by heavily skewed distributions. Relatively few studies addressed this issue in the use of multivariate receptor methods. In order to successfully interpret the results of a receptor modeling analysis applied to Arctic data collected by Canada for several years, distribution analyses of measured chemical composition variables, including several trace elements and ionic species, were conducted for filter samples of airborne particles at three Arctic sites, Alert, Mould Bay and Igloolik in the Northwest Territories of Canada. The data for Cl^-, NO₃^-, NH₄⁺, I, V, Al, Ca, Mg, Fe and Ti at all three sites consistently support the hypothesis that these variables are log-normally distributed. The distributions of Na⁺, K⁺, Na, Ni, Zn, and P are consistently found not to be log-normal at the three sites. For the remaining variables, the log-normal hypothesis is either accepted or rejected differently for different sites. Among these variables, sulfate did not show a log-normal distribution at Mould Bay, but did for the Alert and the Igloolik sites. Values of kurtosis indicate that sulfate has a heavier tail at Mould Bay and lighter tails at Alert and Igloolik.