Baseline air mass selection at Cape Point, South Africa:: application of 222Rn and other filter criteria to CO2

Concentrations of Rn-222 (March 1999-August 2002) from a recently installed analyser at the station Cape Point (34degreesS, 18degreesE), South Africa, were analysed statistically. The combination of Rn-222 with information on wind direction and carbon monoxide (CO) permitted a classification of air masses into continental, marine, and mixtures of both. The ability to select trace gas data representing purely maritime conditions is shown through application to carbon dioxide (CO2) data. Rn-222 levels at Cape Point ranged from near zero to above 5000 mBq m(-3). Monthly percentiles show practically no seasonal dependence for values up to the 25th percentile (P25), corresponding to Rn-222 < 100 mBq m(-3), which is considered typical for marine air. In contrast, Rn-222 percentiles exceeding P50 reveal an austral winter maximum, related to a higher incidence of continental air at that time of the year. The wind sector pattern for Rn-222 concentrations largely coincides with that observed for CO, but covers a wider sector to the east, reflecting continental source areas. Air masses with Rn-222 levels between 100 and 250 mBq m(-3) were found to be still affected by terrestrial sources. Our routinely used percentile-based statistical filter applied to CO2 accepted <75% of the data compared to the radon criterion for maritime air (Rn-222 < 100 mBq m(-3)), which only yielded 18%, but is more stringent in excluding terrestrial influences. CO2 data filtered by using a combination of various selection parameters agreed well with data obtained for Rn-222 < 100 mBq m(-3), confirming that this Rn-222 threshold is a suitable criterion for purely maritime data. Nonetheless, for CO2 the statistical filter, which does not depend on other species and has better data coverage, still retains its applicability for routine trace gas filtering with respect to baseline concentrations. (C) 2004 Elsevier Ltd. All rights reserved.