Predictability of lead-210 in surface air based on multivariate analysis

Abstract

Dependence of the lead-210 activity concentration in surface air on meteorological variables and teleconnection indices is investigated using multivariate analysis, which gives the Boosted Decision Trees method as the most suitable for variable analysis. A mapped functional behaviour of the lead-210 activity concentration is further obtained, and used to test predictability of lead-210 in surface air. The results show an agreement between the predicted and measured values. The temporal evolution of the measured activities is satisfactorily matched by the prediction. The largest qualitative differences are obtained for winter months.