Multiple symptoms of total ozone recovery inside the Antarctic vortex during austral spring

• Published in: Atmospheric Chemistry and Physics
• Main Authors: A. Pazmiño, S. Godin-Beekmann, A. Hauchecorne, C. Claud, S. Khaykin, F. Goutail, E. Wolfram, J. Salvador, E. Quel
• Format: Article in Journal/Newspaper
• Language: English
• Published: Copernicus Publications 2018
• Subjects: Physics; QC1-999; Chemistry; QD1-999; Antarctic; The Antarctic; Austral; Antarc*
• Online Access: https://doi.org/10.5194/acp-18-7557-2018; https://doaj.org/article/6ade51e3fb8842f4bdb035b22b0c8069

The long-term evolution of total ozone column inside the Antarctic polar vortex is investigated over the 1980–2017 period. Trend analyses are performed using a multilinear regression (MLR) model based on various proxies for the evaluation of ozone interannual variability (heat flux, quasi-biennial oscillation, solar flux, Antarctic oscillation and aerosols). Annual total ozone column measurements corresponding to the mean monthly values inside the vortex in September and during the period of maximum ozone depletion from 15 September to 15 October are used. Total ozone columns from the Multi-Sensor Reanalysis version 2 (MSR-2) dataset and from a combined record based on TOMS and OMI satellite datasets with gaps filled by MSR-2 (1993–1995) are considered in the study. Ozone trends are computed by a piece-wise trend (PWT) proxy that includes two linear functions before and after the turnaround year in 2001 and a parabolic function to account for the saturation of the polar ozone destruction. In order to evaluate average total ozone within the vortex, two classification methods are used, based on the potential vorticity gradient as a function of equivalent latitude. The first standard one considers this gradient at a single isentropic level (475 or 550 K), while the second one uses a range of isentropic levels between 400 and 600 K. The regression model includes a new proxy (GRAD) linked to the gradient of potential vorticity as a function of equivalent latitude and representing the stability of the vortex during the studied month. The determination coefficient ( R 2 ) between observations and modelled values increases by ∼ 0.05 when this proxy is included in the MLR model. Highest R 2 (0.92–0.95) and minimum residuals are obtained for the second classification method for both datasets and months. Trends in September over the 2001–2017 period are statistically significant at 2 σ level with values ranging between 1.84 ± 1.03 and 2.83 ± 1.48 DU yr −1 depending on the methods and considered proxies. This result ...