J. C. Farman, B. G. Gardiner & J. D. Shanklin
British Antarctic Survey, Natural Environment Research Council, High Cross, Madingley Road, Cambridge CB3 OET, UK
Total 03 has been measured at the British Antarctic Survey stations, Argentine Islands 65deg. S 64deg. W and Halley Bay 76deg. S 27deg. W, since 1957. Figure 1a shows data from Halley Bay. The mean and extreme daily values from October 1957 to March 1973 and the supporting calibrations have been discussed elsewhere [4,5]. The mean daily value for the four latest complete observing seasons (October 1980-March 1984) and the individual daily values for the current observing season are detailed in Fig.1. The more recent data are provisional values. Very generous bounds for possible corrections would be +/-30 matm cm. There was a changeover of spectrophotometers at the station in January 1982; the replacement instrument had been calibrated against the UK Meteorological Office standard in June 1981. Thus, two spectrophotometers have shown October values of total 03 to be much lower than March values, a feature entirely lacking in the 1957-73 data set. To interpret this difference as a seasonal instrumental effect would be inconsistent with the results of routine checks using standard lamps. Instrument temperatures (recorded for each observation) show that the March and October operating conditions were practically identical. Whatever the absolute error of the recent values may be, within the bounds quoted, the annual variation of total 03 at Halley Bay has undergone a dramatic change.
Figure 1b shows data from Argentine Islands in a similar form, except that for clarity the extreme values for 1957-73 have been omitted. The values for 1980 to the present are provisional, the extreme error bounds again being +30 matm cm. The changes are similar to those seen at Halley Bay, but are much smaller in magnitude.
Upper-air temperatures and winds are available for these stations from 1956. There are no indications of recent departures from established mean values sufficient to attribute the changes in total O3 to changes in the circulation. The present-day atmosphere differs most prominently from that of previous decades in the higher concentrations of halocarbons. Figure 2a shows the monthly mean total O3 in October at Halley Bay, for 1957-84, and Fig. 2b that in February, 1958-84. Tropospheric concentrations of the halocarbons F-11 (CFCl3) and F-12 (CF2Cl2) in the Southern Hemisphere  are also shown, plotted to give greatest emphasis to a possible relationship. Their growth, from which increase of stratospheric ClX is inferred, is not evidently dependent on season. The contrast between spring and autumn O3 losses and the striking enhancement of spring loss at Halley Bay need to be explained. In Antarctica, the lower stratosphere is ~40 K colder in October than in February. The stratosphere over Halley Bay experiences a polar night and a polar day (many weeks of darkness, and of continuous photolysis, respectively); that over Argentine Islands does not. Figure 3 shows calculated amounts of NOx in the polar night and the partitioning between the species . Of these, only NO3 and NO2 are dissociated rapidly by visible light. The major reservoir, N2O5, which only absorbs strongly below 280 nm, should be relatively long-lived. Daytime levels of NO and NO2 should be much less in early spring, following the polar night, than in autumn, following the polar day. Recent measurements  support these inferences. The effect of these seasonal variations on the strongly interdependent ClOx and NOx cycles is examined below.
The 03 loss rate resulting from NOx and Cl0x may be written 
L accounts for over 85% of O3 destruction in the altitude range 20-40 km. At 40 km, N and C are roughly equal. Lower down, C decreases rapidly to 10% of L at 30 km, 3% at 20 km (refs 6, 8). Equation (1) is based on two steady-state approximations, (see Table 1a for the reactions involved)
valid in daytime, with [O] in steady state with . Reaction (4) has a negative temperature coefficient, whereas reaction (1) has large positive activation energy , with the result that u is strongly dependent on [Cl0] at low temperature, as shown in Fig. 4. [Cl0] is not simply proportional to total ClX, because ClONO2 formation (reaction (10)) intervenes. Throughout the stratosphere, X<<1, so that [ClO]~[Cl+ClO]. From a steady-state analysis of the reactions given in Table 1b,
Values of u, X and [Cl+Cl0] obtained from equations (2), (3) and (4) are in good accord with full one-dimensional model results for late summer in Antarctica . Neglecting seasonal effects other than those resulting from temperature and from variation of [NO+ NO2], it is possible to solve simultaneously for [NO2] and [Cl0], and to derive L. Results are shown in Table 2 as relaxation times , /L, for various conditions. The spring values (lines 2, 3 and 4) are highly dependent on ClX amount (compare columns a and b), the autumn values (line 1) much less so. At Argentine Islands, the sensitivity to ClX growth should resemble that seen in line 2, attributable solely to low temperature. Lines 3 and 4 show the enhanced sensitivity possible at stations within the Antarctic Circle, such as Halley Bay, arising from slow release of [NO+NO2] following the polar night. It remains to be shown how stable 03 budgets were achieved with the relaxation times for the lower chlorine level (Table 2, a).
Much 03 destruction is driven by visible light, but production requires radiation below 242 nm. On the dates shown (Table 2), destruction persists for some 11 h, while, because of the long UV paths, production is weak (except around noon) at 29 km, and is virtually absent below that altitude. Line 1 of Table 2 then demands 03 transport in autumn from the upper to the lower stratosphere, which is consistent with inferred thermally-driven lagrangian-mean circulations . A mean vertical velocity of 45 m per day is in good accord with calculations of net diabatic cooling  and gives a realistic total 03 decay rate in an otherwise conventional one-dimensional model . The short relaxation times in the lower stratosphere in autumn are tolerable, with adequate transport compensating for lack of 03 production.
In early spring, on the other hand, wave activity scarcely penetrates the cold dense core of the Antarctic polar vortex and with very low temperatures the net diabatic cooling is very weak . Lagrangian transport in the vortex should then be almost negligible. (The virtual exclusion of Agung dust from the vortex supports this view .) The final warming signals the end of this period of inactivity and is accompanied by large dynamically induced changes in 03 distribution. However, before the warming, with low chlorine, total 03 was in a state of near-neutral equilibrium, sustained primarily by the long relaxation times. With higher chlorine, relaxation times of the order seen in line 4, Table 2, entail more rapid 03 losses. With negligible production below 29 km and only weak transport, large total 03 perturbation is possible. The extreme effects could be highly localized, restricted to the period with diurnal photolysis between polar night and the earlier of either the onset of polar day or the final spring warming. At the pole [NO+NO2] rises continuously after the polar night, with the Sun. The final warming always begins over east Antarctica and spreads westwards across the pole. At Halley Bay the warming is typically some 14 days later than at the pole. Maximum O3 depletion could be confined to the Atlantic half of the zone bordered roughly by latitudes 70 and 80deg. S.
Comparable effects should not be expected in the Northern Hemisphere, where the winter polar stratospheric vortex is less cold and less stable than its southern counterpart. The vortex is broken down, usually well before the end of the polar night, by major warmings. These are accompanied by large-scale subsidence and strong mixing, in the course of which peak O3 values for the year are attained. Hence, sensitivity to ClX growth should be minimal if, as suggested above, this primarily results from 03 destruction at low temperatures in regions where 03 transport is weak.
We have shown how additional chlorine might enhance 03 destruction in the cold spring Antarctic stratosphere. At this time of the year, the long slant paths for sunlight make reservoir species absorbing strongly only below 280 nm, such as N2O5, ClONO2 and HO2NO2, relatively long-lived. The role of these reservoir species should be more readily demonstrated in Antarctica, particularly the way in which they hold the balance between the NOx and ClOx cycles. An intriguing feature could be the homogeneous reaction (Table 1c) between HCl and ClONO2. If this process has a rate constant as large as 10(-16) cm3 s(-1) (ref. 2) and a negligible temperature coefficient, the reaction would go almost to completion in the polar night, leaving inorganic chlorine partitioned between HCl and Cl2, almost equally at 22 km for example. Photolysis of Cl2 at near-visible wavelengths would provide a rapid source of [Cl + ClO] at sunrise, not treated in equation (4). The polar-night boundary is, therefore, the natural testing ground for the theory of linear response to chlorine [1, 2]. It might be asked whether a nonlinear response is already evident (Fig.2a). An intensive programme of trace-species measurements on the polar-night boundary could add greatly to our understanding of stratospheric chemistry, and thereby improve considerably the prediction of effects on the ozone layer of future halocarbon releases.
1. Cicerone, R. J., Walters, S. & Liu, S. C. J geophys. Res. l8, 3647-3661 (1983).
2. Prather, M. J., McElroy, M. B. & Wofsy, S. C. Nature 312, 227-231 (1984).
3. The Stratosphere 1981, Theory and Measurements (WMO Global Ozone Research Monitoring Project Rep. No. 11, 1981).
4. Farman, J. C. & Hamilton, R. A. Br. Antarct. Surv. Sci. Rep. No 90 (1975).
5. Farman, J. C. Phil. Trans. R. Soc. B279, 261-271 (1977).
6. Farman, J. C., Murgatroyd, R. J., Silnickas, A. M. & Thrush, B. A. Q. Jl R. met. Soc.(submitted).
7. McKenzie, R. L. & Johnston, P. V. Geophys. Res. Lett, 11, 73-75 (1984).
8. Johnston, H. S. & Podolske, J. Rev. Geophys. Space Phys. 16, 491-519 (1978).
9. Chemical Kinetics and Photochemical Data for Use in Stratospheric Modelling, Evaluation No. 6 (JPL Publ. 83-62, 1983).
10. Dunkerton, T. J. atmos. Sci. 35, 2325-2333 (1978).
11. Dopplick, T. G. J. atmos. Sci 29, 1278-1294 (1972).