The non-parametric statistical evaluation of precipitation chemistry sampler intercomparison data

Abstract

The variability in performance of two brands of wet/dry atmospheric deposition samplers was compared for 1 year at a single site. A total of nine samplers was used. Samples were collected weekly and analyzed for pH, specific conductance, common chemical constituents and sample mass. The non-normal distribution within the data set and the non-normal distribution of residuals necessitated the application of the non-parametric Friedman test to assess the comparability of sample chemical composition and volume between and within brands of samplers. Statistically significant differences existed for most comparisons however, the test does not permit quantification of their magnitudes, except in general terms. Differences in analyzed concentrations between samplers were small.