Long-term variations in total ozone derived from Dobson and satellite data

Abstract

Total ozone growth rates are calculated using flexible ‘tendency curves’ that can follow ozone variations on all timescales greater than that of the quasi-biennial oscillation. This method improves on traditional trend analysis using straight line fits because it follows ozone variations more closely, providing visual information about the timing and global distribution of ozone variations. Results are compared from long-running Dobson sites and from two homogenized satellite data sets, one constructed at NASA/Goddard Space Flight Center and the other developed at New Zealand's National Institute of Water and Atmospheric Research. Although the most negative ozone trends in the Southern Hemisphere appear to be linked to polar vortex chemistry, those in the Northern Hemisphere have occurred between 35°N and 40°N and may be related to dynamical trends and/or chemistry on episodically occurring volcanic aerosols. A quasi-decadal cycle in total ozone was present since the mid-1920s and hence is independent of halogen chemistry. Its cause remains unknown. Including the deseasonalized and detrended local temperature in the ozone trend model decreases the standard error of the ozone trend over most of the globe.

Introduction

Variations in total ozone have received much scientific scrutiny since the 1970s when it was first proposed that man-made chemicals such as chlorofluorocarbons (CFC) could impact the Earth's protective ozone shield (Molina and Rowland, 1974). Decreases in the stratospheric ozone column allow more ultraviolet radiation in the 280–320 nm range to reach the surface where it is harmful to living organisms. Trend studies gained prominence when it was recognized that an ozone ‘hole’ was forming every spring over Antarctica (Farman et al., 1985). The word trend has implied a ‘generally consistent, continuing change over a given period’ (WMO (World Meteorological Organization), 1995). Bojkov et al. (1996) referred to “a process of ozone decline that has a stable and continuous cause”. Some past trend studies have looked for a linear response beginning in 1970 to increases in CFC emissions such as one might expect from gas phase halogen chemistry. However, calculated ozone trends were much larger than the 1% change expected from homogeneous chemistry above 30 km. Furthermore, the majority of ozone loss was in the lower stratosphere below 25 km (Stolarski et al., 1992). Past trend studies also found that the interval chosen for trend determination often had a significant impact on trend values, so that total ozone variations with time were in fact not linear. Solomon et al. (1996) identified large changes in stratospheric aerosol abundance in the presence of halogens as a source for some of those non-linear ozone losses. In the latter study the magnitude of modeled changes was 50% smaller than those observed in total ozone, but the timing and location of the losses was correct. Thus, the accelerated losses in total ozone in the Northern Hemisphere midlatitudes seen after the Pinatubo volcanic eruption may have been caused by heterogeneous chlorine and bromine chemistry occurring on liquid sulfate particles in the lower stratosphere.

Another source for ozone depletion that does not follow a linear model is dilution from the Polar Regions. This process occurs when filaments of ozone depleted air are transported into midlatitudes when the vortex erodes in the spring. It is especially evident in the Southern Hemisphere, where it may have a cumulative effect from one year to the next which is estimated to add about 20% to the depletion (Prather et al., 1990).

The Montreal Protocol and its amendments have led to a reduction in halogen emissions so that concentrations in the stratosphere are no longer increasing in a linear fashion. Aside from halogen chemistry, other effects that one would not expect to be linear are those driven by changes in atmospheric dynamics. Recent studies provide evidence for the role of dynamics in long-term midlatitude total ozone variations (Reid et al., 2000; Thompson et al., 2000; Appenzeller et al., 2000). Increases in greenhouse gases and water vapor may contribute to depletion by causing radiative cooling of the stratosphere (Shindell et al., 1998; Dvortsov and Solomon, 2001).

The unfolding knowledge of different sources for midlatitude total ozone variations bring into question the assumption that long-term total ozone variations should be modeled with a straight line. Fioletov et al. (2002) has dealt with the issue of nonlinear ozone variations by calculating sliding 11-year trends. Their method successfully shows how trends change with time, although it still assumes linear behavior over each 11-year period. In this study a flexible curve is fitted to ozone residuals after known natural variations are removed. This curve, referred to here as the total ozone tendency curve, can follow decreases, increases, and accelerations so that it gives more visual information than a straight line. The derivative of this flexible curve is the growth rate curve, whose average value is comparable to the trend values of past studies. Although we no longer assume that the ozone trends solely reflect the effect of homogeneous halogen chemistry on ozone, calculating these trends is still an important task and allows comparison with other studies. The flexible tendency and growth rate curves give specific information in terms of the timing and location of total ozone variations over the globe. Applied to satellite data, this method illuminates geographical differences in ozone variation, thereby giving clues to possible processes affecting total ozone. In the future these visual tools may be used to monitor for unexpected changes, eventual recovery, and possible effects of climate change.