Estimating the near-surface daily fine aerosol load using hourly Radon-222 observations

We investigate the extent to which hourly radon observations can be used to estimate daily PM2.5 loading near the ground. We formulate, test and apply a model that expresses the mean daily PM2.5 load as a linear combination of observed radon concentrations and differences on a given day. The model was developed using two consecutive years of observations (2007–2008) at four sites near Sydney, Australia, instrumented with aerosol samplers and radon detectors. Model performance was subsequently evaluated against observations in 2009. After successfully reproducing mean daily radon concentrations (r2≥0.98), we used the model to estimate daily PM2.5 mass, as well as that of selected elements (Si, K, Fe, Zn, H, S and Black Carbon). When parameterizing the model for elemental mass estimates the highest r2 values were generally obtained for H, BC, K and Si. Separating results by season, the r2 values for K and BC were higher in winter for all sites, a period of time where higher concentrations of these elements are seen and a rapid estimation tool would be of particular benefit. The best overall results were obtained in winter for H and BC [r2 = 0.50, 0.68, 0.70, 0.63 (H) and 0.57, 0.57, 0.78, 0.44 (BC)], respectively for Warrawong, Lucas Heights, Richmond and Muswellbrook. Evaluation of model PM2.5 estimates was most successful for days with typical aerosol loads; loads were usually underestimated for, the less frequent, high–to–extreme pollution days. The best elemental results were obtained for BC at Richmond in winter (r2 = 0.68). However, for Warrawong and Lucas Heights r2 values increased from 0.26 to 0.60, and from 0.33 to 0.73, respectively, when several particularly high concentration events were excluded from the analysis. The model performed best at Richmond, an inland site with relatively flat terrain. However, model parameters need to be evaluated for each site.