Trend analysis of CTM‐derived northern hemisphere winter total ozone using self‐consistent proxies: How well can we explain dynamically induced trends?

We derive characteristic spatial patterns, and their temporal evolution, for total ozone during January, February and March in the northern hemisphere from a twenty‐year integration of a chemistry‐transport model. Our aim is to identify factors influencing ozone variability (and decadal‐scale trends) and to develop proxies for inclusion in statistical models of the ozone trend. The version of the chemistry‐transport model is set up so that the model ozone interannual variability, and any trends, are driven purely by the dynamical forcing fields. The five leading Empirical Orthogonal Functions derived using singular value decomposition are identified with the polar‐night jet, the Scandinavian pattern, the Arctic Oscillation, the North Atlantic Oscillation and the Eliassen–Palm flux respectively. We then apply trend analysis to the model results, using the corresponding principal components (PCs) as proxies in a multiple linear regression model to calculate ozone trends. As expected, the use of the five leading PCs together as explanatory variables in the regression model explains most of the northern hemisphere dynamically driven trend in the chemistry‐transport model. By comparing the performance of the trend model using the first ozone principal component and then the polar‐night jet index, we establish the importance for trend calculations of the modulation of the polar vortex which influences ozone variability in middle latitudes during late winter/spring. A suitable proxy for the polar‐night jet index should be considered in all future trend studies. Copyright © 2006 Royal Meteorological Society