The long-term change in total ozone from observations and from the 45-year transient CMAM run associated with the EESC loading is shown in Fig. 8. The trend coefficients can be expressed in terms of the linear trend during the 1980s (see Sect. 2) and are estimated separately for each month of the year. The magnitude of the trends are weaker in CMAM than in the observations, especially in the northern hemisphere. This is not unusual for CCMs (Eyring et al., 2006). Overall the trend functions of ozone show similar characteristics in the observations and CMAM. The seasonal cycle of the ozone trends have a maximum in spring which is especially strong in the polar regions, reflecting the polar springtime ozone depletion. The polar springtime ozone trends are clearly stronger in the SH. In both hemispheres the polar trends amplify the seasonal cycle of the ozone trends for the entire extratropical region compared to the midlatitudes. However the summertime trends are nearly identical over middle and polar latitudes.
F&S 2003 demonstrated that in the NH midlatitudes the observed long-term trends are determined by the trends in the winter/spring buildup. There is no need to invoke summertime ozone chemistry or springtime polar ozone depletion to explain the summertime ozone trends in the NH. In contrast the seasonality of observed trends in the southern midlatitudes cannot be explained by the springtime trends there. As noted in F&S 2005 the mechanism works better in the SH if the entire extratropical region (35o–80oS) is considered, implying that springtime polar ozone depletion and transport contribute to the summer ozone trend over SH midlatitudes. We now examine if the trend function of total ozone from CMAM from summer through to early autumn can be explained by the springtime trends. Therefore we estimate the trend for each month by multiplying the trend value for a specific selected springtime month with the regression coefficient between the two months. Figures 9c and d show the results for the SH midlatitude and extratropical regions based on the trend values for October, November or December. The actual trend function is also displayed. In the southern midlatitudes the actual trend in summer is significantly stronger than the estimated trends. If, however, the entire extratropical region is considered then the estimates of summertime trends from springtime trends work much better and show the same seasonality as the actual trend function. These results are consistent with the observations.
In the same way, the long-term trends are in line with interannual variability over the entire northern extratropics. This is illustrated by Fig. 9b, which shows the actual trend function from late spring through early autumn together with the estimates of the trend based on the actual trend values for March, April, or May and the regression coefficients for the interannual anomalies. However there are differences between CMAM and the observations for northern midlatitudes. In the observations, F&S 2003 showed that interannual anomalies in NH midlatitudes persist through summer, despite the mixing in of polar air following the vortex breakdown. This is because in terms of total ozone mass, the midlatitude anomalies dominate over the polar anomalies; the anomalies seen in Fig. 5 are multiplied by the area of the region. However in CMAM the midlatitude anomalies decay rapidly after the vortex breakdown (not shown here), because of the much larger impact of the CMAM polar anomalies. As in the SH the estimated trends underpredict the summertime trends in NH midlatitudes in CMAM (Fig. 9a). This is in contrast to the observations. The implication is that in CMAM the relative importance of Arctic ozone loss is greater than in the observations, just as the interannual Arctic anomalies in CMAM have an unrealistically large relative impact on the overall extratropical anomalies (Fig. 5).