Recent, unprecedented, Arctic ozone losses: climate change or large inter-annual variability?

The record ozone loss over the Arctic in the spring of 2011 [e.g., Manney et al., 2011] highlights the importance of a detailed understanding of the connection between cold polar temperatures, polar stratospheric clouds (PSCs) and column ozone. Several studies have analyzed the empirical relationship between PSC volume (VPSC) and ozone loss in the Arctic [e.g., Rex et al., 2006], and put forward the hypothesis that the coldest Arctic winters are getting colder and therefore anomalous ozone losses in the Arctic are likely to increase in the coming decades. In the present study we analyze trends and variability in polar temperatures, VPSC and column ozone, using three different reanalysis products (ERA-INTERIM, NCEP-NCAR, NASA-MERRA) and numerical model output (from selected models participating in the Chemistry-Climate Model Validation Activity Phase 2). We compute VPSC for latitudes north of 60°N based on T < TNAT at 50 and 30 hPa levels, following the approach of Rex et al. [2006]. We employ a variety of statistical measures for extremes in order to identify possible changes in the frequency distribution of polar temperatures and VPSC, and attempt to determine whether the recent occurrences of record ozone loss are indicative of statistically significant trends or simply a reflection of large natural variability. The results show a good agreement in VPSC among the individual reanalysis data sets during 1980-2011 and a clustering of record low and high VPSC values in recent years in all three reanalysis products. While inter-annual variability is widely within ± 1 standard deviation prior to 1995, it increases in the second half of the records, suggesting a recent widening of the statistical distribution function in both the lower and the upper tail. However, if we test the statistical distribution functions of VPSC for 1980-1995 and 1996-2011 for statistically significant differences, the applied tests (Kolmogorov-Smirnov and k-sample Anderson-Darling) do not yield significant results, thus showing that VPSC in both time periods can be considered to stem from the same underlying probability distribution. Also we do not find statistically significant trends in VPSC in any of the reanalyses, besides for 5-yr maximum values in NCEP-NCAR. We conclude that VPSC in the winter of 2010/11, although large, lies well within the range of natural variability and can from the current observational record not be interpreted as a signal of Arctic climate change.

References:

Manney, G. L., et al. (2011), Unprecedented Arctic ozone loss in 2011, Nature, 478(7370).

Rex, M., et al. (2006), Arctic winter 2005: Implications for stratospheric ozone loss and climate change, Geophysical Research Letters, 33(23).