Scaling and persistence of UK pollution

Three methods are used to examine the statistical characteristics of UK pollution time series: Sigma-T (ST), Hurst rescaled range and kurtosis. Hourly observations of ozone, PM10 and PM2.5 are examined. We find evidence of high persistence or long-term memory of pollutant fluctuations up to 400 days. The ST method was found to be more sensitive than the Hurst analysis. Scaling behaviour suggesting the lack of characteristic time scales was found. Kurtosis reveals the importance of intermittent pollution episodes for particulate species, but near Gaussian behaviour for ozone.