• Introduction
• Description of model and experiment performed
• Total ozone amounts and trends
• Temperature trends
• Water vapour trends
• Discussion and conclusion
• References

FIGURES

• Figure 1
• Figure 2
• Figure 3
• Figure 4
• Figure 5
• Figure 6
• Figure 7
• Figure 8

1. Introduction

Stratospheric cooling and stratospheric ozone loss have both been observed since at least the start of the 1980s (e.g. WMO, 1999, Chapters 4 and 5 and references therein). In turn, the stratosphere has cooled further due to both the observed ozone changes as well as increases in greenhouse gas concentrations. Alongside the changes in ozone and temperature, changes in water vapour (Oltmans and Hofmann, 1995) have also occurred. While this qualitative picture is now broadly accepted, a more quantitative understanding needs to be established if future predictions are to be considered reliable. A general summary of some recent results is described in WMO (1999), Chapter 12. One of the main difficulties is that the temperature in the Arctic winter lower stratosphere is close to that required for heterogeneous chemical processes to operate and therefore small temperature reductions could trigger significant chemical change. It is quite likely, then that two-dimensional models and other simplified models are unsuitable for future ozone predictions. Instead it is suggested that three-dimensional coupled chemistry climate models are required to simulate accurately the past stratosphere and to determine the likely future ozone amounts and stratospheric temperature variation.

Here, the first steps towards detailed model verification against past observations are commenced, by running a general circulation model (GCM) with coupled chemistry continuously from March 1979 to January 2000, thus providing a 20-year database of model results. The results are here analysed to provide model trends in total ozone, temperature, water vapour. The main purpose of this paper is to examine the results critically to determine whether they are good enough to warrant long simulations of the future ozone layer using such a computing-intensive coupled model.

2. Description of model and experiment performed

The Met. Office Unified Model (UM) (Cullen, 1993) is here used in extended form with 64 levels (upper boundary 0.01 hPa, Figure 1) and with a latitude-longitude resolution of 2.5 by 3.75. The model contains a complete range of stratospheric chemical reactions allowing representation of all the main ozone formation and destruction processes. The chemistry is also coupled to the model radiation scheme. By resolving the whole of the stratosphere and most of the mesosphere the model is able to simulate possible dynamical feedbacks, such as changes in global mean meridional circulation, important for ozone transport. Running on our local massively parallel Cray T3E, the model takes about 40 hours per model year using 36 processors. The model has undergone many improvements (see Austin, 2000) since the earlier version of (Austin et al., 2000a,b).

Table 1. Long-lived tracer amounts used in the model simulations.

Year (1 Jan)	1980	1985	1990	1995	2000	Notes
Cl (ppb)	1.63	2.21	2.72	3.26	3.53	1
Br (ppt)	9.4	10.1	11.5	13.6	15.5	1
Effective halogen (ppb)	2.10	2.72	3.30	3.94	4.31	1,2
NO (ppb)	18.4	18.6	18.8	19.1	19.4	3
CFC11 (ppt)	173	222	263	291	289	4
CFC12 (ppt)	295	382	477	532	545	4
NO (ppb)	302	306	310	314	319	4
CH (ppb)	1603	1651	1700	1749	1810	4
CO (ppm)	337	346	355	365	375	4

Notes: 1. Values are for the equator at the top model level. 2. The effective halogen is determined by Cl + 50.0 Br, and represents crudely the net impact of halogens on ozone depletion if all the Bromine were in the form of Chlorine. 3. The peak value is given. 4. The values for the greenhouse gases are for the troposphere. The model was initialised on 1 March 1979 as in Austin et al. (2000a) and a simulation of 20 years 10 months was completed with specified sea surface temperatures and sea ice from the Atmospheric Model Intercomparison Project (AMIP) (Gates et al., 1999). Concentrations of the key molecules and greenhouse gases (Houghton et al., 1996, Table 2.5 and Figure 2.3) are indicated in Table 1 for the period 1980-2000.

3. Total ozone amounts and trends

Figure 2 compares the mean value for the period with the mean from the Total Ozone Mapping Spectrometer (TOMS) data (Stolarski, 1993). All the main observed features are very well reproduced, including low tropical ozone in January and February, and a broad peak in Arctic ozone in April. Arctic ozone also reaches a minimum in autumn as observed. In the southern hemisphere, the Antarctic ozone hole and the mid-latitude maximum are present in October, as observed. In spite of this agreement, the model is generally about 25 DU too high except in the Arctic spring (up to 15 DU too low), the Antarctic spring (about 50 DU too low), and a more substantial 75DU too high in southern mid-latitudes during spring.

The globally averaged total ozone for the model as a function of time is compared with TOMS data in Figure 3. A clear solar signal is present in the observations whereas these processes have not been incorporated into the physics of the model. The model results are biassed high relative to the observations. The first 10 months of the model results should be ignored, as the model was still spinning up. For the remaining period of the run, the model trend was -3.3 0.9 DU/decade in the annual average in agreement with the observed trend of approximately -2.0 1.4 (2 error bars) after removing the solar cycle (4.5 1.4 DU per 100 units of F flux). However, the observed trend is sensitive to the period chosen -- e.g. over the period 1980--1998 the observed trend was -3.1 1.4 DU/decade. Also, the observed trends have been computed using data from different satellites which could have had a significant impact on the results.

The total ozone trend (DU/decade) as a function of month and latitude is illustrated in Fig. 4 in comparison with TOMS data. The model results are qualitatively in agreement with observations in showing large springtime losses over the polar regions but the model Antarctic ozone hole is present well into the southern hemisphere summer. This is consistent with the known problems of climate models in simulating a polar vortex that is too strong and which is present for too long (Pawson et al., 2000, and references therein). This is further accentuated by radiative feedback in a coupled model. In middle and high latitudes of the northern hemisphere, the ozone depletion is less than observed, due in part to the absence of aerosol chemistry in this simulation (e.g. Solomon et al., 1998).

Figure 5 shows the results for the local minima in the Arctic and Antarctic spring periods, together with the maximum size of the ozone hole, as indicated by the area within the 220DU contour. Comparisons with our previous (49-level) model simulations (Austin et al., 2000a,b) are also included. In the Arctic, the model generally agrees with observations although it is biassed slightly too high whereas the 49-level model was more nearly in agreement with observations. Also, whereas the model has a negligible trend in the minimum of -1 16 DU/decade the observed trend is just statistically significant at -23 18 DU/decade. In the Antarctic, the 64-level model results are biassed slightly low, compared with a slight opposite bias in the 49-level results. The trend in the minimum for the 64-level model, of -51 16 DU/decade agrees with the observed trend of -64 12 DU/decade where 2 error bars are used throughout. The size of the model ozone hole, by contrast, has a different trend to that observed, being too large initially and does not increase in size as rapidly as in the observations. The cause of this error is most likely related to the dynamics of the vortex, a consequence of the use of Rayleigh friction to mimic gravity wave drag in the stratosphere and mesosphere.

4. Temperature trends

The annual mean model temperature trends are shown in Fig. 6 in comparison with trends determined from Stratospheric Sounding Unit (SSU)/Microwave Sounding Unit (MSU) data (Scaife et al., 2000a) for the slightly shorter period Jan 1980 - Dec 1997. The solar cycle has been removed from the observations (see e.g. Scaife et al., 2000a), but other processes such as the quasi-biennial oscillation and the impacts of aerosol have not been accounted for. In the upper stratosphere the model is broadly in agreement with observations but diverges from them both in the mesosphere and in the middle and lower stratosphere. The temperature trends are to a first approximation independent of latitude. Significant deviations from this occur in the lower stratosphere in southern middle latitudes where the 0.0 and -0.25 contours buckle upwards. The feature is absent in the observations but its cause in the model may be related directly to the ozone trend which also shows an increase in that region. Over Antarctica, where the SSU trends are less reliable, the observations indicate the cooling occurs in the lower stratosphere due to the ozone hole and this is followed by an upper region of warming. Qualitatively, this also occurs in the model, but the influence of the ozone hole does not extend as far north as indicated in the SSU/MSU data.

A quantitative comparison of the globally and annually averaged temperature trends at selected levels (Fig. 7) confirms many of these points. In the figure, comparisons are also made with our previous results from a model without coupled chemistry (Butchart et al., 2000, small circles). Although this simulation covered a different period (1992 to 2051), the changes in GHG amounts and sea conditions were approximately linear throughout the period and changed at virtually the same rate as for the period studied here (1980-2000). Thus the Butchart et al. (2000) simulation provides an important benchmark for model comparisons. The inclusion of (model computed) ozone trends increases model cooling rates significantly throughout the stratosphere, in agreement with Langematz (2000) and Rosier and Shine (2000). This leads to better agreement with observations in the lower stratosphere (indicated by the results for 46 and 100 hPa). In the middle stratosphere, the modelled trends are significantly larger than observed and inclusion of ozone results in poorer agreement with observations. In the upper stratosphere, the previous underprediction of observed trends is replaced by a slight overprediction (within the error bars). In the Figure the quoted error bars (2) are the errors in determining the trend from the sequence of annual temperatures, and in the case of the observations do not include any error due to removal of the solar cycle or instrument drift.

5. Water vapour trends

Figure 8 shows the model water vapour concentrations and trends. During the first four years of the integration, the water vapour in the model had not reached steady state and hence this period is ignored in the calculation of the trends. The model water vapour is in good agreement with observations (e.g. Randel et al., 1998) with mesospheric values exceeding 6ppmv and the tropical minimum of about 3.5 ppmv in the annual average. The model also simulates realistically low values over Antarctica (minimum 3.0 ppmv in the annual average) due to the impact of dehydration in the winter and spring seasons. The trends (Fig. 8) are generally very small in the stratosphere with increases of order 1% per decade over most of the domain. Descent over the Arctic combined with methane oxidation increases the water vapour by almost 3% per decade at 10 hPa, while a slight reduction occurs over Antarctica due to the dominance of the cooling. The water vapour trends in this model simulation are controlled by the slight cooling at the tropopause (Fig. 11) rather than any effects due to the increase in strength of the Brewer-Dobson circulation.

6. Discussion and conclusion

This work is new in completing a long (20-year) simulation of the past chemistry and dynamics of the stratosphere in a detailed chemistry-climate model with observed greenhouse gases, halogen loadings, sea surface temperatures and sea ice amounts. Critical comparisons have been made with observations over the period 1980-2000. It can be concluded that the annual ozone variability is reproducible and that the values are reasonably consistent with observations. The globally averaged total ozone trend is also in agreement with observations when the solar cycle is removed from the latter. However, there are some important discrepancies, with Arctic ozone depletion smaller than observed and the impact of the Antarctic ozone hole lasting too long into the summer. Also, although the depth of the Antarctic ozone hole is reasonably well simulated it is too small and subject to too much interannual variability in the model.

The strengths and weaknesses of the modelled ozone are closely related to the temperature trends which in general are consistent with trends observed from SSU/MSU data. Also, the model temperature trends and interannual variability for the coupled chemistry simulation presented here are generally in better agreement with observations (Randel and Wu, 1999a, Scaife et al., 2000a) than a simulation of a similar version of the model without chemistry (Butchart et al., 2000). However, there is a systematic underprediction of the cooling rates in the lower stratosphere which could be related to the underprediction of model water vapour trends (Forster and Shine, 1999). These trends are small in the model (up to 3% increase per decade locally) but in the Antarctic lower stratosphere are negative due to reductions in temperature. Further, the absence of a water vapour trend in the model clearly contradicts observations (e.g. Oltmans and Hofmann, 1995).

To return to the original question of whether the model is good enough to predict accurate stratospheric ozone and temperature trends for the next 20 years, the results suggest first the need for further model development. Such comments apply equally well to more simplified models, which conceivably could provide a misleading picture of future stratospheric trends in view of the complexity of the many conflicting processes.

Acknowledgments. This work was supported by the U.K. Public Met. Service Research and Development Programme, the U.K. Department of Environment Transport and the Regions (contract EPG/1/1/83), and the CEC project `European project on Stratospheric Processes and their Impact on Climate and the Environment' (EuroSPICE) which commenced in March 2000. I would like to thank Jeff Cole (U. Reading, UK) for supplying updated AMIP sea surface temperatures and sea ice and Jeff Knight (The Met. Office) for implementing some of the model improvements. Dave Jackson (The Met. Office) kindly supplied the methane oxidation scheme, John Nash (The Met. Office) provided SSU/MSU temperature anomalies.

References

Austin, J., A three-dimensional coupled chemistry-climate model simulation of past stratospheric trends, J. Atmos. Sci., Submitted, 2000.

Austin, J., J.R. Knight, and N. Butchart, 2000a: Three-dimensional chemical model simulations of the ozone layer: 1979-2015. Q. J. R. Meteorol. Soc., 126, 1533-1556.

Austin, J., J.R. Knight, and N. Butchart, 2000b: Three-dimensional chemical model simulations of the ozone layer: 2015-2055. Q. J. R. Meteorol. Soc., In press.

Butchart, N., J. Austin, J.R. Knight, A.A. Scaife, and M.L. Gallani, 2000: The response of the stratospheric climate to projected changes in the concentrations of the well-mixed greenhouse gases from 1992 to 2051, J. Climate, 13, 2142-2159.

Cullen, M.J.P., 1993: The unified forecast/climate model, Meteorol. Mag., 122, 81-94.

Forster, P.M de F. and K.P. Shine, 1999: Stratospheric water vapour changes as a possible contributor to observed stratospheric cooling. Geophys. Res. Lett., 26, 3309-3312.

Gates, W.L. et al., 1999: An overview of the results of the Atmospheric Model Intercomparison Project (AMIP). Bull. Amer. Meteor. Soc., 80, 29-55.

Houghton,J.T., L.G. Meira Filho, B.A. Callander, N. Harris, A. Kattenberg, and K. Maskell (eds.)., 1996: Climate Change 1995, The Science of Climate Change, Cambridge University Press, Cambridge.

Langematz, U., 2000: An estimate of the impact of observed ozone losses on stratospheric temperature. Geophys. Res. Lett., 27, 2077-2080.

Oltmans, S.J. and D.J. Hofmann, 1995: Increase in lower-stratospheric water vapour at a mid-latitude northern hemisphere site from 1981-1994. Nature, 374, 146-149.

Pawson, S. et al., 2000: The GCM-reality intercomparison project for SPARC (GRIPS): Scientific issues and initial results. Bull. Am. Met. Soc., 81, 781-796.

Randel, W.J. and F. Wu. 1999: Cooling of the Arctic and Antarctic polar stratospheres due to ozone depletion. J. Clim., 12, 1467-1479.

Randel, W.J., F. Wu, J.M. Russell III, and A. Roche, 1998: Seasonal cycles and QBO variations in stratospheric CH and HO observed in UARS HALOE data. J. Atmos. Sci., 55, 163-185.

Rosier, S. and K. Shine, 2000: The effect of two decades of ozone change on stratospheric temperature as indicated by a general circulation model. Geophys. Res. Lett., 27, In press.

Scaife, A.A., J. Austin, N. Butchart, S. Pawson, M. Keil, J. Nash, and I.N. James, 2000: Seasonal and interannual variability of the stratosphere diagnosed from UKMO TOVS analyses. Q.J.R. Meteorol. Soc., 126, 2585-2604.

Solomon, S., R.W. Portmann, R.R. Garcia, W. Randel, F. Wu, R. Nagatani, J. Gleason, L. Thomason, L.R. Poole, and M.P. McCormick, 1998: Ozone depletion at mid-latitudes: Coupling of volcanic aerosols and temperature variability to anthropogenic chlorine. Geophys. Res. Lett., 25, 1871-1874.

Stolarski, R.S., 1993: Monitoring stratospheric ozone from space. In: The role of the stratosphere in climate change. NATO ASI I8, Springer Verlag, 319-346.

World Meteorological Organization (WMO)/United Nations Environment Programme (UNEP), 1999: Scientific Assessment of Ozone Depletion: 1998, World Meteorological Organisation, Global Ozone Research and Monitoring Project -- Report No. 44, obtainable from WMO GLobal Ozone Observing System, PO Box 2300, 1211-Geneva-2, Switzerland.