Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.
Appendix F DETECTI ON OF TREND S - IN THE VERTICAL DISTRIBUTION OF OZONE A. Barrie Pittock CSIRO Division of Atmospheric Physics Mordialloc, Australia UMKEHR METHOD Observations of the vertical distribution of ozone by the Umkehr method can be made with any Dobson spectrophoto- meter in the total ozone network. However, only about 18 stations in the network currently make regular Umkehr observations (see Table F.1). Of these, only three are in the southern hemisphere. Six stations have records extending back more than 20 years. In addition to calibration drift, the Umkehr method is subject to a number of sources of error and bias, notably the effects of tropospheric and stratospheric dust (Dave et al. 1981, De Luisi et al. 1975), and a meteorological bias due to the inability to make observations under cloudy conditions (Pittock 1970). Long-term trends in ozone concentrations at various altitudes as observed by the Umkehr technique are thus subject to major uncertainties in addition to the possibilities of random and systematic errors. The major uncertainties are due to changes in atmospheric concentra- tions of dust especially from volcanic eruptions, possible trends in cloudiness, and a serious problem of geographi- cal representativeness. Global mean concentrations of ozone at particular altitudes are, however, rather mean- ingless since the vertical distribution of ozone varies markedly with latitude and season. Umkehr-derived vertical distributions are most accurate in the middle stratosphere (around 30- to 45-km altitude) if adequate allowance can be made for stratospheric dust. Such allowance using optical depth measurements has been proposed (Dave et al. 1981) and may enable meaningful estimates of trends in the middle stratosphere at northern middle latitudes to be made. 315
316 TABLE F.1 Umkehr Stations In Operation as of 1980 Initial Year Months of Missing Station (Since 1958) Data Europ e Arosa 1961 6 Belsk 1963 46 Cairo 1978 13 Lisbon 1967 39 North America a Boulder 1978 1 Edmonton 1974 18 Japan Kagoshima 1958 88 Naha 1976 31 Sapporo 1958 96 Tateno 1958 23 India Mount Abu 1964 30 New Delhi 1965 57 Poona 1975 27 Srinagar 1976 14 Varanasi 1964 72 Australia Aspendale 1958 0 Brisbane 1959 0 Macquarie Island 1964 0 Goose Bay and Churchill have reported old data, but no data for 1979 or 1980. Statistical analysis of Umkehr data (Bloomfield et al. 1982, Penner et al. 1981) leads to estimates of the "revealed" uncertainties--that is, those detectable from the scatter or range of measurements--to which must be added estimates of the unrevealed uncertainties--that is, those due to lack of geographic coverage or to poorly estimated global trends in stratospheric dust. Combined, these analyses suggest standard deviations of global trend estimates of around 5 percent per decade in the 30- to 45-km altitude range, and somewhat less if the trend estimate is for the north temperate latitudes only (Hudson et al. 1982). An ozone depletion of about 10 percent per decade should thus be detectable at the 95 percent confidence level in the middle stratosphere of the north temperate zone. Trends of this magnitude are
317 not revealed by the existing data, which are summarized for northern middle latitudes in Figure F.1. Umkehr data for the Australian stations over the last several years are only now becoming available and have not yet been included in global trend estimates. BALLOON-BORNE OZONESONDES Balloon-borne ozonesondes are currently flown on a regular basis at a small number of stations in Western Europe, North America, Japan, and India, and at one station in the southern hemisphere (see Table F.2). These observations have high vertical resolution but are adjusted by a single factor to give absolute agree- ment with the total ozone amount measured by a nearby Dobson spectrophotometer. This single-factor adjustment is a major source of uncertainty both in individual profiles and in trend determinations because the sondes perform with lower efficiencies at low ambient pressures (high altitude). This deficiency is compensated for by using a standard pressure-dependent correction factor as well as the single-factor adjustment. However, individual sondes may differ in performance from the standard pressure-dependent correction factor, and subtle changes ; n the m~n~,facture or preparation of instruments could introduce a secular trend in this performance. Actual trends in ozone concentrations above balloon burst altitude can also cause fictitious trends in the profiles at lower altitudes to appear via the correction factor to the Dobson total amount (see Pittock 1977b). Another problem with ozonesonde measurements is that polluted tropospheric air may contaminate the intake system causing artificially low readings especially at low altitudes. Using the adjustment factor may then result in overestimates of ozone at high altitudes. For all the above reasons, the expected random error in individual ozone soundings is least between the tropopause and about 25 km altitude (standard deviation about 4 percent). The expected random error is higher (about 8 percent) in the troposphere and at 30 km and above (Hudson et al. 1982). Another major uncertainty in estimates of global trends is due to the poor spatial coverage of the ozonesonde network. This uncertainty is considerably reduced if the analysis is confined to north temperate latitudes where the spatial coverage is relatively good. However, satel
318 8 4 o 4 4 o 4 4 4 o 4 o 4 4 o 4 -8 NORTH MID-LATITUDE DOUBLE SCALE ibid ~i 3246 km Ibid 1 l FUEGO UMKEHR AGUNG TO ~ ~ In] IT s40.32DkEm ~ o No _4 _ _ Q o = ~] I ~i IT L11 I o z o - > 4 ~ HI 24-32 km UMKEHR . 16 24 km ||| III 1I 8-16 km ~ | MINI 1 1 8-16 km I- Tl. 2-8 km SON D E , ~.L- 1 1960 1970 1980 YEAR FIGURE F.1 Observed ozone variations for different layers in the troposphere and stratosphere at middle northern latitudes. The vertical bars represent approximate 95 percent confidence intervals (Angell and Korshover 1981~.
319 TABLE F.2 Balloon-borne Ozonesonde Stations in Operation as of 1980 Station Initial Year Months of Missing (Since 19 58) Data North America Churchill 1973 0 Edmonton 1973 0 Cold Lake 1977 26 Goose Bay 1969 0 Palestine 197 7 18 Toronto 1976 26 Wallops Island 1970 26 Europe Biscarrosse 1976 0 Hohenpeissenberg 1966 0 Legionowo 1980 0 Lindenberg (Tempelhof) 1967 1 1 Payerne 1968 0 Uccle 1965 20 North Polar Resolute 1966 5 Tropics Natal (Brazil) 1979 2 Australia Aspendale 196 S 6 Japan Kagoshima 196 8 19 Sapporo 1968 35 Tateno 1968 23 lite measurements and theoretical considerations reveal asymmetries between the northern and southern hemispheres; data from the north temperate latitudes cannot be extrapo- lated elsewhere. At 30 km, which is about as high as ozonesondes regularly reach, these considerations and statistical analyses of the data (Pittock 1977a, Hudson and Reed 1979, Hudson et al. 1982) lead to estimates of standard deviations of estimated global ozone trends per decade of about 5.5 percent. Thus an ozone change of about 11 percent per decade could be detected at the 95 percent confidence level. A change in ozone concentration at 30 km in the north temperate zone could be detected from the
320 ozonesonde data with confidence if it were to exceed about 8 percent per decade. No such global trend or trend in the north temperate zone has yet been detected. Trends in the ozone content of the upper troposphere are also of interest since the same theoretical models that predict ozone reduction in the middle and upper stratosphere predict ozone increases in the upper troposphere due to NOk emissions from the surface and/or aircraft exhausts. Estimates of revealed and unrevealed errors in trend determinations in the 2- to 8-km layer from ozonesonde data suggest that a change of about +18 percent at this level worldwide could be detected at the 9S percent confidence level (Hudson et al. 1982). Using data from the north temperate zone only, it should be possible to detect a trend of +14 percent per decade at the 95 percent confidence level. As most surface and aircraft emissions of NOk occur in the north temperate zone, this is a more sensible place to look for early evidence of a tropospheric trend, especially as tropospheric effects of pollutants will have a shorter lifetime than those in the stratosphere due to removal by meteorological processes. A linear regression analysis of the ozonesonde data at 2-8 km in the north temperate zone (see Figure F.1) reveals a trend during the 1970s of about +7 percent per decade (Liu et al. 1980, Angell and Korshover 1981). The revealed uncertainties as indicated by the error bars in Figure F.1 suggest that this trend might be statistically significant. However, consideration of estimates of possible unrevealed errors, as discussed above, increases the uncertainty to a standard deviation of about 7 percent per decade. Thus, with 95 percent confidence, the trend lies between -7 percent and +21 percent per decade. The most probable value of the trend thus differs from zero by about one standard deviation and has a probability of only two chances in three of being real. The data therefore are quite suggestive of an increase, but the level of confidence in the result is not high. Careful checking and stratification of the ozonesonde data and application of refined statistical techniques may result in a reduction of the uncertainty in this trend estimate.
321 SATELLITE METHODS The most extensive sets of data on ozone concentrations in the middle and upper stratosphere so far obtained by satellites have been obtained with the backscattered ultraviolet (B W) and the infrared limb emission tech- niques (Hudson et al. 1982). The usefulness of satellite data for trend analysis depends on obtaining long and essentially homogeneous time series of data. This requires continuity with the same type of, or closely comparable, instruments and inversion algorithms (the methods used to transform measured radiation intensities to the ozone distributions that give rise to them) and regular calibration by in-flight and "ground-truth" methods. Allowance must also be made for possible natural fluctuations in ozone concentration, due particularly in the upper stratosphere to possible solar cycle variations. As indicated by Panofsky (see Appendix E) in discussing total ozone measurements, the B W data from the NIMBUS-4 satellite, which commenced operation in April 1970, suffered from instrumental drift and also from a loss of spatial coverage after June 1972. Additional problems of spatial representativeness arose from interference with observations of vertical distribu- tions caused by high-energy charged particles in the vicinity of the South Atlantic magnetic anomaly. The drift problem has forced almost total reliance on Umkehr rocket, and balloon-borne ozonesondes for validation and assessment of instrument performance. The solar backscattered ultraviolet (SB W) instrument on NIMBUS-7, which commenced operation in November 1978, was designed to overcome these problems. Shorter data sets are available from the Limb Radiance Inversion Radiometer (LRIR) on NIMBUS-6 from June 1975 through January 1976, and the Limb Infrared Monitor of the Stratosphere (LIMS) on NIMBUS-7 from October 1978 through May 1979. Other data were obtained by the Stratospheric Aerosol and Gas Experiment (SAGE) on the AEM2 satellite from February 1979 to the present, an early B W type experiment on OGO-4 in 1967-1968, and a later B W instrument on AK-5, commencing in November 1975 and still operational. Data from these last two B W instruments are not yet available. According to Heath, in as-yet unpublished work (quoted in Hudson et al. (1982) and in Science, September 4, 1981, pp. 1088-1089 and submitted for publication in
322 Science), comparison of vertical profiles of ozone concentration from the NIMBUS-4 BW instrument in 1970 and 1971 with those from the SEW instrument on NIMBUS-7 in the corresponding months of 1978 and 1979 suggests that there has been some ozone depletion in the layer between 2 and 6 mbar (approximately 35- to 45-km altitude). At the altitude of maximum depletion, around 38- to 40-km altitude, the decrease averaged about 5 percent per decade (see Figure F.2). Best estimates of the revealed and unrevealed errors in global mean ozone concentrations at 40 km from the NIMBUS-7 SB W instrument alone suggest that a trend, due to whatever cause, of +1.4 percent per decade could be detected at the 95 percent confidence level with 10 years of observations (Hudson et al. 1982). The actual uncertainty in estimates of ozone reduction from the combination of NIMBUS-4 and NIMBUS-7 data is difficult to quantify but is certainly likely to be much greater than 1.4 percent per decade owing to the problems with NIMBUS-4 outlined above and the necessity to allow for a solar cycle effect above 35 km. The changing sensitivity of the NIMBUS-4 B W instrument has been taken into account by Heath using comparisons with near overpass Umkehr observations, assuming that the Umkehr network did not itself drift in calibration. The solar cycle effect was taken into account by assuming that natural concentrations of stratospheric ozone vary in phase with solar activity and that the amplitude of this effect increases monotonically with increasing altitude. Thus the smooth curves (dashed lines) in Figure F.2, which match the observations (solid lines) at the 10.0- and 0.7-mbar levels, were taken to represent the solar cycle effect, and the difference from the observations at intermediate levels was taken to represent the ozone decrease tentatively attributed by Heath to destruction by chlorofluorocarbons. Any departure of the real solar cycle effect from Heath's interpolated monotonic curves would lead to an error in the hypothesized ozone depletion profile. Heath's assumed solar cycle effect is in broad agreement with Dutsch (1979), but not with the theoretical calcula- tions of Penner and Chang (1978), nor with those of Brasseur and Simon (1981), which are based on recent solar W flux data. Neither theoretical study supports a monotonic variation of the solar cycle effect in the range between 30 and 50 km. According to Brasseur and Simon (1981), the magnitude of the solar cycle effect at
323 Global 1 0 - d E 2 0 - a' al ~0 5 0 60 7 0 80 9 0 30 ,. . . Trend analysis of \ BUV data, 1970-76 \ _ 1 ~ i t: _ I I I t iSBUVt 1979)March / BUV(1971 March ~ 1 1 1 -1 0 0 +10 Percent / Year 50 ~5 An . _ 40 I 35 FIGURE F.2 Inferred long-term ozone variations in the stratosphere from satellite observations, according to Heath (1981~. Solid lines represent observed ozone varia- tions. Dashed lines are assumed effect of solar cycle variations only. Hatched area represents decrease in ozone tentatively attributed by Heath to destruction by chloro- fluoromethanes.
324 38 to 40 km is about 8 percent, and is still about 7 percent at 30 km, where Heath assumes it to be negligible. In the light of the uncertainties of the NIMBUS-4 B W data, and more particularly of the controversial allowance for the solar cycle effect, the ozone depletion around 35- to 45-km altitude reported by Heath cannot at present be regarded as well established. Nevertheless, it is clear from the error analysis of the NIMBUS-7 system that with only another 5 to 10 years of homogeneous well- calibrated satellite data, and provided that the nature of the solar cycle effect at these altitudes can be more firmly established, it should be possible to determine whether or not significant reduction of ozone is occurring at these altitudes as current photochemical models suggest. The precise number of years of data needed to establish the existence of a statistically significant depletion within the range of theoretical possibilities will depend on the magnitude of the actual depletion and of the remaining uncertainties regarding the solar cycle effect. Error analyses for other satel- lite ozone profile measuring systems (Hudson et al. 1982) suggest similar sensitivity can be obtained from several systems using quite different physical approaches to the problem. QUESTIONS OF CAUSALITY In the absence of a detailed theoretical understanding of many of the alternative causes of ozone trends (see Appendix A), statisticians and others have attempted to assign limits to the possible magnitude of natural ozone trends and of ozone trends due to human influences other than chlorofluorocarbons (Hudson and Reed 1979, Hudson et al. 1982). These attempts have led to some disagreement, with some statisticians claiming that the variance due to long-term natural ozone variability can be estimated from total ozone data going back only a couple of decades, or in the case of one or two stations, 40 or 50 years. Other scientists who are familiar with long-term variability in other climatic variables claim that much longer records are necessary to obtain reasonable estimates of long-term natural variability. In part this disagreement rests on differing ideas about the possible nature of natural long- term climatic variability: the statisticians believe it is-essentially a manifestation of a partially cumulative short-term random variability, whereas many climatologists . .
325 maintain that climatic regimes may change with time, perhaps discontinuously, such that their statistical properties measured over intervals of a few decades may not be entirely representative of a longer time-span (e.g., see Flohn 1975). Such considerations may well apply to the stratosphere as well as to tropospheric climate. The simplistic notion that natural variability of ozone must have some upper limit, so that the detection of a real trend in ozone greater than this limit must imply human influence, is an appealing one. Drawing the line on the basis of revealed statistical variations in the data, or of an intuitive "feel" for natural varia- bility is, however, hazardous. It would be preferable if a physical approach could be adopted in which quantitative estimates were made of the various alternative causal mechanisms such as those outlined in Appendix A. Only when such an admittedly difficult and demanding course is followed will it be possible to ascribe with confidence particular causes to any observed real trend in ozone. Simultaneous measurements of other relevant variables such as temperatures, circulation parameters, the solar spectrum, and various other trace constituents and pollutants, will obviously aid the diagnostic process, and variation of effects with height and latitude provide additional means of discrimination between alternative causal mechanisms, notably solar effects, effects of the global CO2 increase, and those due to chlorofluoro- carbons or NOX emissions. The present statistically based arguments and criteria cannot be regarded as scientifically satisfactory, but must be seen as necessary interim procedures to help in the process of making decisions in the face of uncertainty (Pittock 1980). CONCLUSI ONS Current observations of the vertical distribution of ozone are severely limited as tools for the detection of ozone depletion due to (a) very poor spatial coverage by the balloon-borne ozonesonde and Umkehr observing networks, and (b) the short duration of continuous and homogeneous global data coverage by satellite. The rate of ozone depletion due to human influences is expected to vary with latitude and altitude, with maximum rates of depletion calculated to occur around 35 to 45 km
326 in altitude, and some ozone increase possible in the upper troposphere at northern middle latitudes. Since spatial coverage by balloon-borne ozonesonde and Umkehr methods is best in the northern mid-latitude zone, a focus on data analysis in this zone seems appropriate. Two possible human effects on the vertical distribu- tion of ozone have been reported to date. One is a possible increase in upper tropospheric ozone concen- tration in the north temperate zone, of about 7 percent during the 1970s (Liu et al. 1980, Angell and Korshover 1981). Given the various sources of uncertainty (Hudson et al. 1982), the probability, based on observations, that this effect is real (i.e., different from zero) is about 2 in 3. If it is real, this effect is attributed to an increase in NOX concentrations in the upper troposphere due to emissions from aircraft and surface combustion, and should not be present to an appreciable extent in the southern hemisphere. The second reported human effect is a claimed ozone depletion of the order of 5 percent per decade in the 38- to 40-km layer (Hudson et al. 1982) deduced from NIMBUS-4 B W and NIMBUS-7 SB W data over the time interval 1970 to 1979. Considering the problems experienced with the NIMBUS-4 instrument, and the uncertain but critical allowance for a possible solar cycle effect at these altitudes, this reported ozone depletion cannot at present be regarded as well established. This last uncertainty highlights the question of causality in assessing the probability of observational data reflecting ozone depletion of human origin. The question of the influence of the 11-year solar cycle on ozone concentrations above about 25-km altitude is particularly important. A definitive description of this solar cycle influence in the middle and under troposphere, which Brasseur and Simon (1981) estimate as having an amplitude of about 5 to 10 Percent in these layers. is . . . - . _ - , critical to early detection of ozone depletion at altitudes where photochemical theory indicates that the effects of chlorofluorocarbons should be greatest. Unless the solar cycle effect on ozone can be definitively described theoretically, it may be necessary to wait for accurate observations at critical altitudes over at least one whole solar cycle (11 years) in order to confidently infer that ozone depletion is due to pollution, even though the error limits in satellite vertical distribution measurements are small enough that a real trend in ozone concentration at 40-km altitude may be detected earlier.
327 A suitable set of satellite data is not currently avail able prior to 1978, and the Umkehr data are not only limited in spatial coverage but subject to uncertainty due to the need to allow for the effects of varying aerosol concentrations. Given a resolution of the solar cycle effect, satellite-based observations of ozone concentrations in the 35- to 45-km region seem to provide the best hope for early detection of ozone depletion effects. RECOMMENDATIONS In order to obtain conclusive evidence for or against the reality of significant depletion of ozone by pollutants of human origin, the following actions are recommended. 1. Satellite methods. Highest priority should be given to the maintenance of one or more continuously operating and well-calibrated homogeneous satellite systems for the determination of the vertical distribution of ozone. This should include independent ground-truth obtained from the Dobson spectrophotometer network and balloon- or rocket-borne ozonesondes. Data obtained by more than one independent satellite system operating simultaneously, using different physical principles (e.g., backscattered W and limb-scanning systems), would add greatly to confidence in any conclusions reached. 2. Focus on zones. Attention should be focused on those altitudes and latitudes where theoretical effects of pollution are greatest and nonsatellite data coverage is best. This implies a focus on altitudes in the range of 35 to 45 km, and the upper troposphere in the north temperate zone. Observations in the south temperate zone would provide a useful check, especially as upper tropo- spheric effects are expected to be negligible in the southern hemisphere. 3. Solar cycle effect. A definitive description of the effect on stratospheric ozone of the 11-year solar cycle, especially at 35 to 45 km, is urgently needed. Efforts should be directed to (a) theoretical analysis of the solar cycle effects, (b) monitoring of solar ultraviolet radiation, solar protons, and any other solar outputs likely to affect ozone concentrations, and
328 (c) analysis of zonally representative ozone profile data at 35 to 45 km over at least one whole 11-year solar cycle, including further attempts to refine existing Umkehr data from the north temperate zone with proper allowance for variable aerosol effects. 4. Umkehr method. The spatial coverage by the Umkehr method should be increased, especially in the north and south temperate zones. Since many Dobson spectrophoto- meters are already in place that could but do not at present make Umkehr measurements, this should not be unduly difficult to achieve. Effort must also be made to make proper allowance for the effects of varying concen- trations of tropospheric and stratospheric particulate matter using actual particle concentrations as suggested by Dave et al. (1981). 5. Ozonesondes. The balloon-borne ozonesonde network in the north temperate zone should be maintained and if possible improved, and that in the south temperate zone (currently one station only) increased. More refined statistical techniques and critical data analysis should be applied to the existing north temperate zone ozone- sonde data. 6. Tropospheric ozone. More theoretical work is needed on the distribution of ozone in the troposphere, including especially the effect of NOk from aircraft and surface emissions and the chronological evolution of these effects using emission data. 7. Monitoring other variables. In view not only of the solar cycle effect, but also of the effect of changing temperature (due to increasing carbon dioxide concentrations) and variations in atmospheric circula- tion, other relevant meteorological variables and chemical constituents must be monitored in order both to test photochemical theory and reaction rates more critically and to enable a useful reduction in background variance due to causes other than pollution (e.g., see Bloomfield et al. 1981, Pittock 1973). REFERENCES Angell, J.K. and J. Korshover (1981) Update of ozone variations through 1979. Pages 393-396, Proceedings of the Quadrennial International Ozone Symposium, August 4-9, 1980. Boulder, Colo.: National Center for Atmospheric Research.
329 Bloomfield, P., M.L. Thompson, G.S. Watson, and S. Zeger (1981) The association of ozone with meteorological variables. Pages 306-313, Proceedings of the Quadrennial International Ozone Symposium, August 4-9, 1980. Boulder, Colo.: National Center for Atmospheric Research. Bloomfield, P., M.L. Thompson, G.S. Watson, and S. Zeger (1982) Frequency Domain Estimation of Trends in Stratospheric Ozone e Technical Report No. 182, Department of Statistics, Princeton University. (Submitted for publication to Journal of Geophysical Research.) Brasseur, G. and P.C. Simon (1981) Stratospheric chemical and thermal response to long-term variability in solar UV irradiance. Journal of Geophysical Research 86(C8):7343-7362. Dave, J.V., C.L. Mateer, and J.J. De Luisi (1981) An examination of the effect of haze on the short Umkehr method for deducing the vertical distribution of ozone. Pages 222-229, Proceedings of the Quadrennial International Ozone Symposium, August 4-9, 1980. Boulder, Colo.: National Center for Atmospheric Research. De Luisi, J.J., B.M. Herman, R.S. Browning, and R.K. Sato (1975) Theoretically determined multiple-scattering effects of dust on Umkehr observations. Quarterly Journal of the Royal Meteorological Society 101:325-331. Dutsch, H.U. (1979) The search for solar cycle-ozone relationships. Journal of Atmospheric and Terrestria 1 Physics 41:771-785. Flohn, H. (1975) History and intransitivity of climate. Pages 106-118, The Physical Basis of Climate and Climate Modelling. GARP Publication No. 16. Geneva: World Meteorological Organization. Heath, D. (1981) Secular changes in stratospheric ozone from satellite observations (1970-1979). Unpublished manuscript. (Submitted to Science.) Hudson, R.D. and E.I. Reed (1979) The Stratosphere: Present and Future. NASA 1049. Washington, D.C.: National Aeronautics and Space Administration. Hudson, R.D., et al., eds. (1982) The Stratosphere 1981: Theory and Measurements. WMO Global Research and Monitoring Project Report No. 11. Geneva: World Meteorological Organization. (Available from National Aeronautics and Space Administration, Code 963, Greenbelt, Md. 20771.) ~A ~
330 Liu, S.C., D. Kley, M. McFarland, J.D. Mahlman, and H. Levy II (1980) On the origin of tropospheric ozone. Journal of Geophysical Research 85:7546-7552. Penner, J.E. and J.E. Chang (1978) Possible variations in atmospheric ozone related to the eleven-year solar cycle. Geophysical Research Letters 5:817-820. Penner, J.E., L.P. Golen, and R.W. Mensing (1981) A time series analysis of Umkehr data from Arosa. UCRL Reprint 85420. Livermore, Cal.: Lawrence Livermore Laboratory. Pittock, A.B. (1970) On the representativeness of mean ozone distributions. Quarterly Journal of the Royal Meteorological Society 96:32-39. Pittock, A.B. (1973) Global meridional interactions in stratosphere and troposphere. Quarterly Journal of the Royal Meteorological Society 99:424-437. Pittock, A.B. (1977a) Climatology of the vertical distribution of ozone over Aspendale (38°S, 145°E) . Quarterly Journal of the Royal Meteorological Society 103:575-584. Pittock, A.B. (1977b) Ozone sounding correction procedures and their implications. Quarterly Journal of the Royal Meteorological Society 103:809-810. Pittock, A.B. (1980) Monitoring, causality and uncertainty in a stratospheric context. Pure and Applied Geophysics 118:643-661.