Detecting variability changes in Arctic total ozone column
Résumé
To better assess inter-annual variance changes in ozone time series, we study and propose a statistical procedure based on recent advances in wavelet multi-resolution analysis. This approach, novel to the field of ozone analysis, has the advantages of detecting significant changes in variance and of characterizing the distribution of these changes, including not only their timings but also their strengths.
As a test case for our method, we study total ozone column time series from the TOMS (Total Ozone Mapping Spectrometer) Version 7 in the Arctic region over the period 1981–1992. This statistical procedure allows us to analyse the well-known winter–spring transition and the inter-annual variability behaviour. For this specific example, two distinct types of winter–spring transition can be clearly identified for the ozone series. As expected from dynamical studies of the vortex evolution, the first type is constituted of the years for which there exists a very abrupt and temporally localized ozone variability change. More precisely, we detect a significant peak in ozone variance at two multi-resolution wavelet levels. The second type is characterized by smooth ozone variance changes over time and space. Moreover, the Arctic vortex breakdown timing seems to occur, on average, 1 month earlier than our detected changes in ozone variability. These results suggest a potential link between these two phenomena.