We now turn to the future of the stratospheric ozone layer. As we saw in Chapter 11 and have alluded to in many of the other Chapters, including this present one, manmade chlorofluorcarbon compounds (CFCs) are contributing to serious ozone loss in the stratosphere. The "ozone hole" is a seasonal phenomenon over Antarctica, but it has worldwide ramifications. Though just a "trace gas," the presence of ozone is critical to the existence of life on the land surface of Earth, since ozone shields the surface from biologically destructive ultraviolet radiation from the sun. The modeling assessments and trend predictions that we have discussed here are an essential tool for forecasting ozone levels over the long term. They allow us to conduct experiments that we would not otherwise want to conduct on the real atmosphere. On this basis, they allow us to make sound public policy decisions. What type of decisions are ultimately made then becomes a political question. Science enters into the realm of the policymaking and politics.

4.1 CFC Production Phase-Out: the Political Process

With the developing awareness of the potential impact on the stratospheric ozone layer due to unchecked CFC emissions, the international political community began to take action. The first meeting to discuss the issue was held in 1985 at the Vienna Convection. While nations took no formal action at this point, they did agree that the matter warranted further scientific investigation and possibly required future political remedies.

Two years later in 1987, 27 nations signed the Montreal Protocol, which require the signers to agree to a 50% reduction in CFC production by 1999. As the evidence of the damage to the ozone layer accumulated, nations realized that this agreement would be insufficient. They met again in London in 1990. This time, over 80 nations signed on to the complete elimination of CFC production by the year 2000, a radical advancement over the Montreal agreement.

Despite the 1990 London agreement, new scientific evidence indicated that even this action would not be soon enough to stop some destruction of the ozone layer. So once again the international community met to revise CFC policy. The result was the Copenhagen Amendment of 1992, in which the international community agreed to the complete phase-out of CFCs by 1996 and a reduction or phase-out of HCFCs (a less destructive replacement chemical) by the year 2030.

For each of these political actions, the impact on chlorine concentrations in the atmosphere could be calculated using long-term model forecasts of the sort described in this Chapter. Figure 12.16 shows how each of these agreements affected the future chlorine loading of the stratosphere. Recall that CFCs provide the chlorine to the stratosphere that then destroys ozone. It is clear from the figure that unregulated CFC production and even production consistent with the early agreements would have had serious consequences for the ozone layer.

In the case of stratospheric ozone depletion, the political and scientific processes seem to have worked together in helping resolve the ozone/CFC problem. Note how particularly useful in this process were the set of model-based predictions from scientists of the consequences of a particular set of actions. For example, when member nations signed the Montreal Protocol, politicians may have asked, "What are the consequences of unchecked CFC release on stratospheric ozone? Another question might have been "How much of an impact would a 50% reduction in CFC production have on stratospheric ozone?" A third question might have been "How long would it take to restore the damaged ozone layer if CFC production were phased out altogether?" To answer such questions, we turn our attention to the models.

4.2 Long-Term Ozone Trend Model Predictions

To produce scientifically reliable long-term predictions, which may then serve as the basis for sound policy decisions, models need to be built with sound chemistry and physics. This is true whether the model is forecasting ozone trends, carbon dioxide and global temperature trends, or any other earth system trend. But even a well constructed model can produce meaningless predictions if the inputs to the model are poor.

In the case of a model trying to predict the future of the stratospheric ozone amounts, inputs to the model include the amounts of the various other chemical species that impact ozone. We have already seen in Figure 12.16 how different amounts of chlorine loading may change in this future. Other species, such as inorganic bromine compounds (Bry), nitrous oxide (N2O), and methane (CH4) also play key roles in determining stratospheric ozone distribution. We need to have scenarios of their likely future concentrations based on different predictions of future amount.

4.2.1 World Meteorological Organization (WMO) assessments -- The World Meteorological Organization (WMO) has attempted to make accurate forecasts of ozone trends based on different scenarios involving the chemical species listed above. The plots in Figure 12.17 are based on the latest WMO assessment document. They show the past observed and model-predicted future values of species key to controlling stratospheric ozone concentrations. These include inorganic chlorine (Cly), inorganic bromine (Bry), nitrous oxide (N2O), and methane (CH4). The time period is 1970-2050, representing observed past and predicted future values. Two different estimates are shown; one is associated with the 1994 WMO assessment document, represented by the solid line, and another associated with the current 1998 assessment, represented by the dashed line.

The change in the estimates between the two assessments shown in Figure 12.17 is not insignificant. For example, the historical amount of bromine in 1980 has been reduced by over 30% between the 1994 and 1998 assessments. We also note a 25% reduction in the predicted future level of methane in the year 2050. While we believe that our estimates and predictions are better in 1998 than they were in 1994, the magnitude of the changes over this 4-year time period are suggestive of the uncertainties that may exist in our model inputs. We must be mindful of these uncertainties as we interpret the results.

4.2.2 Trend predictions based on ensembles of models -- In addition to the uncertainty in the model inputs discussed above, there are also uncertainties related to particular models. Each model performs its physical and chemical calculations slightly differently. There are so-called systematic biases inherent to each model. As a result, the predictions of various models do not always agree. The best interpretation of model results, therefore, may be gained by looking at an ensemble or group of models and searching out the predictions upon which most or all agree. Although weather forecasters also use ensemble weather models for forecasting the weather for the next few days or week, forecasting ozone for several decades is much more akin to "forecasting" the climate. Our models give us forecasts based on different scenarios, sometimes extrapolating the past into the future, sometimes making wholly different assumptions.

a. Predicting the past (hindcasting) -- A good place to start is by examining a case in which the models attempt to "predict the past" or hindcast (see Section 2.1.2). The data input includes actual measurements of various trace constituent amounts. Figure 12.18 show the hindcast predictions of an ensemble of ten 2-D models for ozone from 1979 - 2000 based on the real-world data put into the model. We have also have real ozone data from 1979 to 1997 from the TOMS satellite instruments. We can use this TOMS data to validate the predictions.

As we can see in Figure 12.18, the models produce quite a range of predictions. This is particularly evident around the period during which the Mount Pinatubo aerosols were prevalent in the stratosphere (1991-1994). (Recall that the models are given real world constituent gas and aerosol data.) During the Pinatubo period, the models predict anywhere from -2.5% to -8.7% change relative to 1979 ozone, while TOMS indicates a change of about -6.2%. For the most part, the model predictions tend to form an envelope around the real data, although in the mid-1980s, the models all seem to predict changes of smaller magnitude than those actually observed by TOMS. The models, incidentally, did not include the solar cycle, which has an affect on ozone concentrations (see Chapter 8).

b. Predicting future trends by extrapolating hindcast trends -- The type of analysis we've discussed provides evidence for the credibility of the models. The fact that models predictions generally cluster around historical data enhances our confidence in their power to predict the future. Such studies also provide a sense of the uncertainty inherent in the model predictions. Figure 12.19 replots the data shown in Figure 12.18, but with the analysis extended forward in time another 50 years to the year 2050. The data is again compared to the 1979-1997 TOMS measured data for validation.

Although the models show a variety of predictions for ozone in 2050 relative to 1979, they all agree that the difference will still be negative: ozone will not recover globally to the 1979 levels until after the year 2050, according to all the models included in the WMO assessment.

c. Predicting future trends based on changing model inputs -- Changing the model inputs, however, can have a rather dramatic impact on the results. Figure 12.20 shows that a variety of results are possible if the input amounts of Cly, Bry, N2O, and CH4 are changed. By 2050, ozone levels can vary anywhere between +1.5% to -0.5% relative to 1979 values.

d. Limitations and uncertainties in predicting future trends -- In addition to the release of chemical species into the atmosphere which we can predict (such as methane and CFCs), natural processes are also responsible for releasing chemicals and aerosols that can have serious impacts on stratospheric ozone. The model predictions shown above make certain assumptions that there will be no major perturbations to atmospheric trace gas or aerosol content. Major perturbations typically refer to volcanic eruptions of the Mount Pinatubo sort. The models cannot possibly account for the occasional, unforecastable major volcanic eruption. They merely assume that we will not have another volcanic eruption of the magnitude and type of Mount Pinatubo anytime in the next 50 years. The Pinatubo eruption clearly had a major impact on global ozone levels through its large contribution of aerosols to the stratosphere. Future eruptions could delay ozone recovery. It is worth noting, though, that most models indicate that as chlorine concentrations decrease (due to the reduction in chlorine-containing emissions), future eruptions will not have the magnitude of impact on ozone that Mt. Pinatubo had.

In addition, recent studies have been examining the impact of climate change on ozone. Global warming scenarios usually predict, along with the surface warming that is reported in the media, a cooling in the stratosphere. Cooler temperatures and a related change in stratospheric dynamics could lead to more persistent polar stratospheric clouds and more polar ozone loss during the winter/spring seasons. Research indicates that the impact may be particularly noticeable in the Arctic. Such cooling could delay recovery of Arctic ozone past 2045.

Scientists are not yet confident in their ability to predict volcanic eruptions or the precise impacts of global change. Therefore, we must temper our confidence in interpretation of model predictions for future ozone amounts with an understanding of our lack of knowledge in other parts of the global climate system which affect ozone.

e. Predicting Antarctic and Arctic ozone loss and recovery -- Before we end our discussion of the model predictions of the future of ozone, let's look at predictions for ozone levels over the South Pole during September. As we learned in Chapter 11, this is the month when the ozone hole rapidly develops due to a combination of photochemistry, cold temperatures, and heterogeneous chemistry on the polar stratospheric clouds. Figure 12.21 shows model predictions for ozone over the South Pole from 1980-2050. Actual South Pole ozonesonde readings are plotted for comparison and validation of the model. The model predictions are hindcasts, like that shown in Figures 12.18 and 12.19, back to 1979 and forecasts out to 2050.

The results of the models shown in Figure 12.21 appear to overestimate systematically the losses occurring in the 1990s as compared to the ozonesonde data. This means that there is some problem in the model. Recall the ClO overproduction problem from Figure 12.14, and the additional reaction discovered recently that was not included in previous models.

The model also shows that recovery to 1980 levels are not predicted to occur until after the year 2050. All the models seem to indicate that in the Antarctic, things are currently about as bad as they will ever be and that recovery should be detectable around the year 2020. In the Arctic, most models predict that things will get worse before they get better, with levels lower than 1995 levels occurring through 2015.

4.2.3 Jet travel and rocket launch impact on stratospheric ozone -- Models will continue to be applied to address issues related to anthropogenic effects on stratospheric ozone. This includes not just the CFC issue, but also other possible manmade effects. Of current interest is the impact of a new fleet of high speed, supersonic transports that would reduce travel times across the Pacific and Atlantic by a factor of two, but would directly release chemicals into the lower stratosphere, including nitrogen species and water vapor. Model predictions will affect the political decisions related to the production of such a fleet of aircraft.

Models have already been used to assess the impact of shuttle and rocket launches on stratospheric ozone. It was suggested by one study that losses of greater than 8% in local total ozone occur in the neighborhood of space shuttle launches. However, the cumulative impact of a series of shuttle and rocket launches on the stratosphere is rather small, as shown in other studies, which showed a maximum total ozone decrease of 0.12%. Such studies will certainly be important in future political and economic decisions related to ozone depleting chemicals.