Keywords: ambient air, uncertainty weighting averages, time series, robust methods, pollution, metals, air quality networks, air pollution
Novel method for producing uncertainty-weighted averages from a time series of ambient air measurements
A novel method for producing uncertainty-weighted averages from time series of ambient air measurements is presented. This method uses robust regression to establish a relationship between the measured mass concentration values and their uncertainties, and then uses the residuals of individual points from this relationship to calculate a weighting parameter. This parameter can then be used to produce an uncertainty-weighted mean for the group of measurement results used to establish the relationship between the measured values and their uncertainties, or a subset of these, as required. Examples are presented of the method's application to real and synthetic data sets.