We use the same version of the merged satellite data set as in F&S 2005. The set is prepared by NASA and combines version 8 of the TOMS and SBUV data for the time period from November 1978 to December 2003 (Frith et al., 2004). The zonal mean total ozone values cover up to 80 N from April to September in the NH and up to 80 S from October toMarch in the SH. The data for August and September 1995 as well as May and June 1996 are missing. We use estimates of total ozone from ground based measurements to fill the gaps (Fioletov et al., 2002). The 24-year time series of each month consists of monthly and zonal means which are area weighted and averaged over certain latitude bands.
The CMAM is a three-dimensional chemistry-climate model with comprehensive physical parameterisations (Beagley et al., 1997; de Grandpr´e et al., 2000). The version used here has a domain from the surface of the earth to approximately 97 km and T32 spectral truncation. The model includes a fully interactive stratospheric chemistry with all the relevant catalytic ozone loss cycles and heterogeneous reactions for sulphate aerosols, liquid ternary solutions (the socalled Type 1b Polar Stratospheric Clouds, PSCs) and water ice (Type 2 PSCs). There is no parameterisation of nitric acid trihydrate PSCs (Type 1a PSCs) or any associated denitrification. To calculate the persistence and photochemical decay of total ozone anomalies in CMAM we use ozone data from a 45-year transient run of the model. The run (Eyring et al., 2006) was performed for the period 1960-2004 and was forced by observed changes in well-mixed greenhouse gases, halogens, and sea-surface temperatures.
The total ozone autocorrelations were calculated for the entire extratropics for both hemispheres (35o–80o latitude) as a function of time lag for each month of the year. The autocorrelation is thus
where n is the number of years in the record, t is a particular month, t + is a subsequent month lagged by months, and fi is the deviation from the mean in the year i. The correlation coefficients between ozone values for the same month in different years are low. Thus each year can be considered as independent and correlation coefficients greater than 0.4 are statistically significant at the 95% confidence level for the observations, and greater than 0.3 for CMAM.
As in F&S 2005 we scale the equivalent effective stratospheric chlorine (EESC) (WMO, 2003) trend to be one unit per year during the 1980s and use the scaled EESC loading as a proxy for the long term trend of each time series. Thus we can express the trend coefficients in DU/year during the 1980s. We remove the EESC fit prior to the anomaly analysis.