Statistical modelling to answer key questions in atmospheric chemistry: Three case studies

On the basis of three examples we show the high potential of the application of statistical models in atmospheric chemistry. The topics include: (i) homogenisation and long-term trend analysis of the world's longest total ozone series from Arosa (Switzerland) by multiple regression modelling; (ii) determination of emission factors of road vehicle classes by multiple regression models from measurements in a road tunnel; (iii) contributions of different (anthropogenic) emission sources to ambient air concentrations by Principal Component Analysis and visual inspection, using ambient air measurements (17 hydrocarbons, nitrogen oxides and sulfur dioxide) of Zurich and its surroundings.