• Amodei, M., S. Pawson, A. A. Scaife, W. Lahoz, U. Langematz, D. Li, and P. Simon, 2001: SAO and Kelvin waves in the EuroGRIPS GCMs and UK Meteorological Office analyses. Ann. Geophys, 19 , 99114.

    • Search Google Scholar
    • Export Citation
  • Boville, B. A., 1995: Middle atmosphere version of CCM2 (MACCM2): Annual cycle and interannual variability. J. Geophys. Res, 100 , 90179039.

    • Search Google Scholar
    • Export Citation
  • Butchart, N., and J. Austin, 1998: Middle atmosphere climatologies from the troposphere–stratosphere configuration of the UKMO's Unified Model. J. Atmos. Sci, 55 , 27822809.

    • Search Google Scholar
    • Export Citation
  • ——, ——, Knight, J., A. Scaife, and M. Gallani, 2000: The response of the stratospheric climate to projected changes in the concentrations of well-mixed greenhouse gases from 1992 to 2051. J. Climate, 13 , 21422159.

    • Search Google Scholar
    • Export Citation
  • Cox, P. M., R. A. Betts, C. B. Bunton, R. L. H. Essery, P. R. Rowntree, and J. Smith, 1999: The impact of new land surface physics on the GCM simulation of climate and climate sensitivity. Climate Dyn, 15 , 183203.

    • Search Google Scholar
    • Export Citation
  • Dameris, M., V. Grewe, R. Hein, and C. Schnadt, 1998: Assessment of the future development of the ozone layer. Geophys. Res. Lett, 25 , 35793582.

    • Search Google Scholar
    • Export Citation
  • Edwards, J. M., and A. Slingo, 1996: Studies with a flexible new radiation code. I: choosing a configuration for a large-scale model. Quart. J. Roy. Meteor. Soc, 122 , 689719.

    • Search Google Scholar
    • Export Citation
  • Gregory, D., R. Kershaw, and P. M. Inness, 1997: Parametrization of momentum transport by convection. II: Tests in single-column and general circulation models. Quart. J. Roy. Meteor. Soc, 123 , 11531183.

    • Search Google Scholar
    • Export Citation
  • ——, Shutts, G. J., and J. R. Mitchell, 1998: A new gravity-wave-drag scheme incorporating anisotropic orography and low-level wave breaking: Impact upon the climate of the UK Meteorological Office Unified Model. Quart. J. Roy. Meteor. Soc, 124 , 463494.

    • Search Google Scholar
    • Export Citation
  • Johns, T. C., R. E. Carnell, J. F. Crossley, J. M. Gregory, J. F. B. Mitchell, C. A. Senior, S. F. B. Tett, and R. A. Wood, 1997: The second Hadley Centre coupled ocean–atmosphere GCM: Model description, spinup and validation. Climate Dyn, 13 , 103134.

    • Search Google Scholar
    • Export Citation
  • Langematz, U., and S. Pawson, 1997: The Berlin troposphere–stratosphere–mesosphere GCM: Climatology and forcing mechanisms. Quart. J. Roy. Meteor. Soc, 123 , 10751096.

    • Search Google Scholar
    • Export Citation
  • Milton, S. E., and C. A. Wilson, 1996: The impact of parameterized subgrid-scale orographic forcing on systematic errors in a global NWP model. Mon. Wea. Rev, 124 , 20232045.

    • Search Google Scholar
    • Export Citation
  • Mitchell, J. F. B., and T. S. Hills, 1986: Sea-ice and the Antarctic winter circulation: A numerical experiment. Quart. J. Roy. Meteor. Soc, 112 , 953969.

    • Search Google Scholar
    • Export Citation
  • Morcrette, J-J., L. Smith, and Y. Fouquart, 1986: Pressure and temperature dependence of the absorption in longwave radiation parameterizations. Contrib. Atmos. Phys, 59 , 455469.

    • Search Google Scholar
    • Export Citation
  • Palmer, T. N., G. J. Shutts, and R. Swinbank, 1986: Alleviation of a systematic westerly bias in general circulation and numerical weather prediction models through an orographic gravity wave drag parametrization. Quart. J. Roy. Meteor. Soc, 112 , 10011039.

    • Search Google Scholar
    • Export Citation
  • Pawson, S., and Coauthors. 2000: The GCM-Reality Intercomparison Project for SPARC (GRIPS): Scientific issues and initial results. Bull. Amer. Meteor. Soc, 81 , 781796.

    • Search Google Scholar
    • Export Citation
  • Pope, V. D., M. L. Gallani, P. R. Rowntree, and R. A. Stratton, 1999:: The impact of new physical parametrizations in the Hadley Centre climate model—HadAM3. Climate Dyn, 16 , 123146.

    • Search Google Scholar
    • Export Citation
  • Scaife, A. A., N. Butchart, C. D. Warner, D. Stainforth, W. A. Norton, and J. Austin, 2000: Realistic Quasi-Biennial Oscillations in a simulation of the global climate. Geophys. Res. Lett, 27 , 34813484.

    • Search Google Scholar
    • Export Citation
  • Simmonds, I., and W. F. Budd, 1991: Sensitivity of the Southern Hemisphere circulation to leads in the Antarctic pack ice. Quart. J. Roy. Meteor. Soc, 117 , 10031024.

    • Search Google Scholar
    • Export Citation
  • Stratton, R. A., 1999: A high resolution AMIP integration using the Hadley Centre model HadAM2b. Climate Dyn, 15 , 928.

  • Swinbank, R., and A. O'Neill, 1994: A stratosphere–troposphere data assimilation system. Mon. Wea. Rev, 122 , 686702.

  • ——, Lahoz, W. A., A. O'Neill, C. S. Douglas, A. Heaps, and D. Podd, 1998: Middle atmosphere variability in the UK Meteorological Office Unified Model. Quart. J. Roy. Meteor. Soc, 124 , 14851525.

    • Search Google Scholar
    • Export Citation
  • Warner, C. D., and M. E. McIntyre, 1999: Towards an ultra-simple spectral gravity wave parameterization for non-orographic gravity waves. Earth Planets Space, 51 , 475484.

    • Search Google Scholar
    • Export Citation
  • View in gallery

    (a) Difference between longwave heating rates calculated by the Edwards–Slingo scheme and by the Atmospheric and Environmental Research Inc. line-by-line model (solid line), and difference between heating rates calculated by the Morcrette scheme and the line-by-line model (dotted line). In these calculations only heating from CO2, O3, and H2O is included. (b) Difference between shortwave heating rates calculated by the Edwards–Slingo scheme and the scheme used in BA. In both (a) and (b), a midlatitude summer temperature profile from the U.S. Standard Atmosphere is used

  • View in gallery

    Zonal mean fields for Jan. (a) Zonal wind for the UM control run (20-yr mean); (b) as in (a) except Met Office analyses (8-yr mean) are shown; (c) temperature for the UM control run; (d) field in (c) minus Met Office analysis temperature field. Negative values are indicated by dashed contours. Contour interval: 10 m s−1 for (a) and (b), 5 K for (c), and 2 K for (d)

  • View in gallery

    As in Fig. 2, except fields for Jul are shown and the Met Office analyses are averaged over 7 yr

  • View in gallery

    Geopotential height wavenumber-1 stationary wave amplitudes for Jul. (a) UM control run; (b) Met Office analysis. Contour interval: 50 m. The UM fields are averaged over 20 yr and the Met Office analyses over 7 yr

  • View in gallery

    (a) Daily minimum temperature at 46 hPa south of 60°S between May and Nov. Thin lines show temperatures for each model year from 1992 to 1998, and the bold lines the range of daily minimum Met Office analyses temperatures in these years. The dashed lines show the threshold temperature for the formation of type I and type II PSCs; (b) as (a), except results from a run where the orographic gravity wave scheme is switched off at 5 hPa instead of 20 hPa are shown

  • View in gallery

    Zonal mean zonal wind for Jul. Contour interval: 10 m s−1. Negative values are indicated by dashed contours. (a) Run NOCMT, (b) run OSST_SI, (c) run OSST_SI + NOCMT

  • View in gallery

    Geopotential height stationary wavenumber-1 amplitude for Jul (in meters): solid line, control run; bold line, Met Office analyses; dotted line, run OSST_SI+NOCMT; dashed line, run OSST_SI; dash–dot line, run NOCMT; long dashed line, run OGWD

All Time Past Year Past 30 Days
Abstract Views 0 0 0
Full Text Views 45 40 0
PDF Downloads 3 2 0

An Updated Climatology of the Troposphere–Stratosphere Configuration of the Met Office's Unified Model

View More View Less
  • 1 Met Office, Bracknell, United Kingdom
© Get Permissions
Full access

Abstract

In this paper results are presented from an improved version of the troposphere–stratosphere configuration of the Met Office Unified Model (UM). The new version incorporates a number of changes, including new radiation and orographic gravity wave parameterization schemes, an interannually varying sea surface temperature and sea ice climatology, and the inclusion of convective momentum transport. The UM climatology is compared with assimilated data and with results from a previous version of the UM. It is shown that the model cold biases in the January winter stratosphere and the January and July summer stratosphere are reduced, chiefly because the new radiation scheme is more accurate. The separation between subtropical and polar night jets in July is also better simulated. In addition, in the current version stratospheric planetary wave amplitudes in southern winter are less than half those in northern winter, which is in much better agreement with observations than the previous model version. Despite these improvements, the model still has a cold bias in the winter polar stratosphere, which suggests that the model representation of gravity wave drag is inadequate. Sensitivity tests were carried out and showed that the improved simulation of the separation of subtropical and polar night jets in July is due both to the different sea ice climatology and to the inclusion of convective momentum transport. The improved simulation of stationary wave amplitudes in July cannot be attributed to an individual model change, although it seems to be related to changed wave propagation and dissipation within the stratosphere rather than changes in the tropospheric forcing.

Corresponding author address: Dr. David Jackson, Room 251, Middle Atmosphere Group, Met Office, London Road, Bracknell RG12 2SZ, United Kingdom. Email: david.jackson@metoffice.com

Abstract

In this paper results are presented from an improved version of the troposphere–stratosphere configuration of the Met Office Unified Model (UM). The new version incorporates a number of changes, including new radiation and orographic gravity wave parameterization schemes, an interannually varying sea surface temperature and sea ice climatology, and the inclusion of convective momentum transport. The UM climatology is compared with assimilated data and with results from a previous version of the UM. It is shown that the model cold biases in the January winter stratosphere and the January and July summer stratosphere are reduced, chiefly because the new radiation scheme is more accurate. The separation between subtropical and polar night jets in July is also better simulated. In addition, in the current version stratospheric planetary wave amplitudes in southern winter are less than half those in northern winter, which is in much better agreement with observations than the previous model version. Despite these improvements, the model still has a cold bias in the winter polar stratosphere, which suggests that the model representation of gravity wave drag is inadequate. Sensitivity tests were carried out and showed that the improved simulation of the separation of subtropical and polar night jets in July is due both to the different sea ice climatology and to the inclusion of convective momentum transport. The improved simulation of stationary wave amplitudes in July cannot be attributed to an individual model change, although it seems to be related to changed wave propagation and dissipation within the stratosphere rather than changes in the tropospheric forcing.

Corresponding author address: Dr. David Jackson, Room 251, Middle Atmosphere Group, Met Office, London Road, Bracknell RG12 2SZ, United Kingdom. Email: david.jackson@metoffice.com

1. Introduction

The Met Office Unified Model (UM) is used in a wide range of applications, including numerical weather prediction, global climate modeling, and data assimilation. These various requirements are satisfied by using well-tested standard versions of a common basic model, and improvements incorporated in new “builds” of this common model can therefore benefit all the applications. We have incorporated recent model improvements into a troposphere–stratosphere configuration of the UM. The climatology of the previous version of the UM was described by Butchart and Austin (1998, hereafter BA), and Swinbank et al. (1998, hereafter SEA) used a similar, but shorter, model simulation to investigate middle-atmosphere variability. Here, we update BA's work by showing the climatology of the new model version. Since the UM is used not only by the Met Office, but also by the U.K. academic community, this updated climatology will be of interest to a wide range of scientists. In addition, UM simulations form an important part of many European Union, World Meteorological Organization, and other international collaborative projects.

2. Unified Model description

The troposphere–stratosphere configuration of the UM has 49 levels in the vertical extending from the surface to 0.1 hPa, with a horizontal resolution of 2.5° latitude × 3.75° longitude. The model version described here is in general similar to that described by BA, but there are a number of changes. Since the majority of these changes have already been satisfactorily tested in a climate version of the UM (which has an upper boundary at 4.6 hPa), they were all implemented in the troposphere–stratosphere configuration with little further testing. The model changes include a new land surface exchange scheme (Cox et al. 1999), a convective momentum transport scheme (Gregory et al. 1997), and slightly increased horizontal diffusion at the top two model levels [see Butchart et al. (2000) for more details]. The model was initialized with assimilated observations for 1 March 1989 and run for over 20 years. The simulation uses greenhouse gas amounts for 1995, and sea surface temperatures and sea ice amounts were taken from a 20-yr period starting in 1990 from a control integration of a coupled ocean–atmosphere climate version of the UM [HADCM2 (Johns et al. 1997)]. This differs from BA, who used a sea surface temperature and sea ice climatology calculated from observations and that had no interannual variability. The change to a model-based climatology was made so that the UM version described here could act as a reference simulation in a set of climate change experiments (Butchart et al. 2000) that use sea surface temperature–sea ice climatologies from the present day and climate change HADCM2 simulations as a lower-boundary condition.

Of possible greatest impact to the stratosphere are the changes that have been made to the treatment of orographic gravity wave drag and radiation in the new UM version. Whereas BA used the Palmer et al. (1986) gravity wave drag scheme, here a more sophisticated scheme is used that includes the effects of anisotropic orography and low-level breaking (Gregory et al. 1998) [SEA used the Milton and Wilson (1996) scheme, which is similar to the Gregory et al. scheme but has been specifically “tuned” for numerical weather prediction applications]. As in BA and SEA, the gravity wave scheme is replaced by Rayleigh friction above the 20-hPa level. BA compared their control run with a run that used increased Rayleigh friction. We use the increased Rayleigh friction, because BA showed it significantly reduced the cold bias in the July winter stratosphere, although the change in Rayleigh friction otherwise generally has little impact on the model stratosphere. In section 3 we shall compare our results with the results from BA's increased Rayleigh friction run. The radiation scheme used in BA and SEA was that described by Stratton (1999), but with alterations to the longwave part of the scheme, following Morcrette et al. (1986), to enable a more accurate representation of heating rates in the middle atmosphere. Here, this scheme has been replaced by that of Edwards and Slingo (1996), which includes the effects of CO2, H2O, O3, O2, N2O, CH4, CFC11, and CFC12. The contribution of the N2O, CH4, CFC11, and CFC12 lines to longwave heating rates in the middle atmosphere is only of the order of a few hundredths kelvin per day, and is typically two to three orders of magnitude less than the contribution from the CO2, H2O, and O3 lines. Hence, in the following discussion, longwave heating rate calculations only include contributions from the latter three constituents. Figure 1a demonstrates that the longwave heating rates (for the CO2, H2O, and O3 lines only) for the Edwards–Slingo scheme are in good agreement with those from a line-by-line model up to around the 2-hPa level and that in most of the stratosphere these heating rates are generally closer to the line-by-line model heating rates than are the Morcrette heating rates. However, in the upper stratosphere and mesosphere the cooling calculated by the Edwards–Slingo scheme is generally too small. In the lower stratosphere, Edwards–Slingo heating rates are almost identical to the line-by-line heating rates, whereas the Morcrette heating rates are up to 0.05 K day−1 greater. However, a larger impact on model temperatures will arise from differences in the shortwave heating rate (Fig. 1b). In this region Edwards–Slingo shortwave heating rates exceed those from the scheme used in BA by up to 0.25 K day−1. In addition, the difference between these heating rates increases with height, reaching a maximum of around 1 K day−1 near 5 hPa.

3. Model climatology

A full description of the UM climatology appeared in BA. It is not the intention to repeat such a description here, but rather to indicate where the updated UM climatology differs from BA, and where significant model weaknesses remain. The UM fields are compared with a climatology calculated from Met Office analyses (Swinbank and O'Neill 1994).

a. January

Figure 2 shows UM and Met Office analysis zonal mean zonal wind fields for January. The strength of the subtropical jets is reasonably simulated by the UM. However, the UM polar night jet is too strong, by over 20 m s−1 in the lower mesosphere, which suggests that the wave forcing in the winter stratosphere is too weak. In addition, the model jet maximum is located at approximately the same latitude throughout the whole stratosphere, whereas the observed maximum tilts toward the equator with increasing height. This feature is seen in numerous other middle-atmosphere GCMs (e.g., Boville 1995; Langematz and Pawson 1997; Pawson et al. 2000). The UM simulates the major observed features of the middle-atmosphere temperature field for January (Fig. 2c), such as the cold tropical tropopause and the warm summer stratopause. Compared to BA's results, the cold bias (Fig. 2d) is reduced near the tropical tropopause due to slight increases in the net heating rate. This also explains the small warm bias in much of the low- and midlatitude stratosphere. As a consequence of the too strong model polar night jet, the model winter polar stratosphere is too cold. This cold bias exceeds 14 K in the upper stratosphere, while in the region of polar stratospheric cloud (PSC) formation in the lower stratosphere it is between 2 and 6 K. This is an improvement on BA's version of the UM, which has cold biases that exceed 26 K near 5 hPa and that are between 0 and 14 K in the PSC formation region.

In the summer hemisphere the middle-atmosphere easterlies are generally weaker than observed in the mid- to high-latitude stratosphere. This is consistent with the cold bias in the UM temperature field (Fig. 2d), which increases between the subtropical and polar summer stratosphere. This cold bias is around half that reported by BA, because of the larger shortwave heating rate. Furthermore, in the low-latitude stratosphere the model easterly winds are too strong, which suggests that the westerly forcing due to model Kelvin and gravity waves here is too weak. An intercomparison suggests that Kelvin waves in the UM are weaker than those in many other middle-atmosphere GCMs (Amodei et al. 2001).

b. July

The westerly wind bias and cold bias seen in the extratropical summer hemisphere in January are also present in July (Fig. 3). The maximum temperature bias is about 6–8 K smaller than in BA, and this improvement is due to the more accurate calculation of heating rates by the Edwards–Slingo radiation scheme. However, there is a warm bias in the lower mesosphere (also seen in January), which is a result of the underestimate of longwave cooling by the Edwards–Slingo scheme. The separation between subtropical and polar night jets in July is better simulated than by the version reported by BA. Reasons for this will be further investigated in section 4. As in January, the polar night jet is too strong in the UM and its maximum is upright, and is located too close to the pole compared to the observed jet, which slopes equatorward with increasing height. The cold bias in the winter polar upper stratosphere exceeds 22 K, but in the region of PSC formation it is chiefly in the 0–4 K range. The former values are similar to those reported by BA, but in the PSC region the temperatures in BA's version of the UM are generally up to 2 K greater.

Model stationary wave amplitudes of wavenumbers 1 and 2 for January are very similar to those shown in BA. However, this is not the case for July. Figure 4 shows that the maximum southern winter stratospheric wavenumber-1 amplitudes for the UM and the analyses are similar and are less than half the Northern Hemisphere winter values. This contrasts with the results of BA, who showed that the maximum southern winter amplitudes in their version of the UM were approximately three times the corresponding analyzed values. Maximum UM wintertime stratospheric amplitudes for wavenumber 2 (not shown) are also slightly smaller in July than in January, but are approximately two-thirds the maximum analyzed amplitudes.

c. Minimum temperatures in the lower stratosphere

An examination of the seasonal cycle of stratospheric zonal wind and temperature shows that a major difference is that in the current model version the southern polar vortex persists around 1 month longer than in both the Met Office analyses and the model run of BA. This happens because the weaker wave amplitudes in the current model version may weaken eddy forcing and thus delay the breakdown of the vortex. This delayed breakdown is illustrated by a time series of the daily temperature minimum at 46 hPa and poleward of 60°S for the southern winter (Fig. 5a). One reason why the accurate simulation of these temperatures is needed is that the potential for PSC formation is temperature dependent, and thus also shown in Fig. 5a are the threshold temperatures for the formation of PSCs of nitric acid trihydrate (T = 194.9 K) and ice (T = 187.5 K), assuming the presence of 10 ppbv of nitric acid and 4.5 ppmv of water vapor. The model temperatures are within the observed range up until about the end of June. However, in July and August they are at the bottom of this range, while between September and November the temperatures for most of the model years are lower than observed. This cold bias leads to an increased capacity for PSC production in the model in late winter and early spring. Butchart and Austin (1998) also show a cold bias between August and October for most model years, but some model years have minimum temperatures within or above the observed range.

It may initially appear surprising that a more accurate simulation of planetary wave amplitudes in the southern winter and spring in the current model version leads to a poorer simulation of the evolution of the vortex than in the model version of BA. However, it should be noted that the seasonal cycle in the southern stratosphere also appears to be very sensitive to the representation of gravity wave drag. For example, the modeled southern vortex in SEA is colder and more persistent than that in BA, and this may be because SEA use a different gravity wave scheme (Milton and Wilson 1996), which tends to deposit less drag in the stratosphere than the scheme used in BA. This sensitivity is further demonstrated here via a test run in which the orographic gravity wave drag was switched off and replaced by Rayleigh friction at 5 hPa rather than 20 hPa. In the 20–5-hPa region there is more drag from the orographic scheme than from Rayleigh friction. The wavenumber-1 stationary wave amplitudes are similar to those shown in Fig. 4 up to around the 10-hPa level, but above that level are weaker (by less than 50 m at 5 hPa and by up to 150 m near 1 hPa), possibly as a result of enhanced planetary wave–gravity wave interaction. Overall, however, the eddy forcing in most or all of the 20–5-hPa region increases, and in response to this increased eddy forcing the mean meridional circulation is strengthened, and polar temperatures in the lower stratosphere increase by a few degrees compared to the control run. Figure 5b shows that the observed increase in temperatures, which starts in early September, is much better simulated in the test run than in the control run (Fig. 5a). The associated capacity for model PSC production is hence reduced. These results underline that an accurate simulation of the southern stratosphere requires not only an accurate simulation of planetary waves, but also an accurate representation of gravity wave drag. It should be noted that the changes made in the test run are rather arbitrary and that future model versions will instead use a spectral gravity wave scheme (Warner and McIntyre 1999), which includes a better representation of observed gravity waves (especially nonorographic waves).

4. Sensitivity to model changes

In section 3 we were in general able to attribute differences between the updated UM climatology and the one shown in BA to particular model changes. For example, the decreased cold bias in much of the stratosphere is largely associated with improved heating rates calculated by the Edwards–Slingo radiation scheme. However, since the new UM version has incorporated a number of changes at once, such attribution is not always possible. Two prominent examples of this are the improved separation of subtropical and polar night jets and improved stationary wave amplitudes, both in July. Since these both constitute significant improvements, it is interesting and useful to try to determine why the improvements have happened. This is done here using further model simulations, each run for 10 yr, in which each of the model changes that affect the stratosphere the most is removed. These runs are summarized in Table 1.

The separation between subtropical and polar night jets in July in runs ORAD and OGWD is very similar to the control run (Fig. 3a). In run ORAD, temperatures in the stratosphere drop by several degrees. However, the temperature changes vary little with latitude, and the meridional temperature gradient and thus thermal wind structure are not greatly changed. The results from run OGWD are consistent with Gregory et al. (1998), who showed, using a climate version of the UM with an upper boundary at 4.6 hPa, that jet separation in July is largely unchanged when their scheme replaces the Palmer et al. scheme (i.e., the scheme used in BA). It should be noted that SEA used a similar model version to BA, except that they replaced the Palmer et al. gravity wave scheme with the scheme of Milton and Wilson (1996), and that they managed to simulate the jet separation much better than BA. However, it is not clear if this is a result of using a different gravity wave scheme, or whether it is because SEA only made a 5-yr simulation and thus were unable to properly represent the overall climatology.

In run NOCMT there is a clear reduction in the separation between the subtropical and polar night jets (Fig. 6a). The effect of removing convective momentum transport is to weaken the Hadley circulation, and thus to weaken the westerly acceleration on the equatorward side of the subtropical jet. This was also noted by Pope et al. (1999). The decreased acceleration also leads to a slight poleward shift of the jet position and thus a reduction in subtropical–polar night jet separation. Subtropical and polar night jet separation is also decreased in run OSST_SI (Fig. 6b). In the Southern Hemisphere, differences between the two sea surface temperature climatologies are small (usually less than 1–2 K), but differences in sea ice near Antarctica are large. The sea ice depth in this region is up to four times smaller in the old, observations-based, climatology than in the new, model-based one, and the sea ice fraction at the northernmost extent of the sea ice (about 60°S) is also smaller. The effect of the reduced sea ice is to increase the sensible and latent heat fluxes into the atmosphere that occur when cold air flows off Antarctica and across the warmer sea ice. This leads to a zonal mean temperature increase of up to 4 K near the surface near 60°S, and an associated weakening of the westerly wind, which extends up to the stratosphere. A further impact is that in the 20°–40°S band there is a slight cooling at low levels and an associated increase in westerly wind. The size of these wind changes is less than 4 m s−1, but they are large enough to push the subtropical jet closer to the polar night jet. This pattern of UM response to changes in sea ice is generally consistent with previous studies (e.g., Mitchell and Hills 1986; Simmonds and Budd 1991), with the exception that in the UM the zonal wind changes are largely invariant from the surface up to the upper stratosphere.

A further test was run in which the old sea surface temperature–sea ice climatology was used and the convective momentum transport was switched off (run OSST_SI+NOCMT). Results (Fig. 6c) show that the July subtropical and polar night jets are smeared together even more than when the two changes are applied individually (Figs. 6a and 6b). This structure is very similar to that shown in BA. The increase in westerly wind on the poleward side of the subtropical jet between the control run and run OSST_SI+NOCMT, which accounts for the smearing of the two jets, is statistically significant at the 95% level. Thus we can conclude that convective momentum transport and the new, coupled model-based sea ice climatology appear to be the prime contributors to the jet separation seen in the new version of the UM.

By contrast, an examination of stationary wave amplitude for the test runs provides little indication of why the July amplitudes in BA's version are so much larger than in the current model version. At 100 hPa (not shown) the amplitudes for several of the test runs are between 50% and 100% larger than the control run amplitude, and the upward component of EP flux is also greater near 60°S. However, at higher levels the differences are much smaller. Figure 7 shows the stationary wavenumber-1 amplitude at 1 hPa for July for the control run, the Met Office analyses and all the other runs except run ORAD (omitted for clarity, since this amplitude is similar to the control run amplitude). Near 60°S, the largest difference between test and control run amplitudes are for those runs where convective momentum transport is switched off (runs NOCMT and OSST_SI+NOCMT). However, the maximum amplitude is around 560 m, compared to around 1300 m in BA. This suggests that the improvement in wave amplitude between BA's version and the current model version is not primarily due to changes in tropospheric forcing, but is a result of complex interactions between all of the incorporated model changes which act to change the propagation and dissipation of planetary waves within the stratosphere.

5. Conclusions

This paper demonstrates that improvements continue to be made to the climatology of the troposphere–stratosphere configuration of the UM. A particular improvement on the results shown by BA is the successful simulation of the observed seasonal cycle in winter planetary wave amplitudes, with those in July being less than half those in January. In addition, improvements to the radiation scheme lead to reductions in the cold biases near the tropical tropopause, in the summer stratosphere in January and July, and in the winter stratosphere in January. A further improvement is that the separation between subtropical and polar night jets in July is much better simulated than before. However, deficiencies remain. There is still a cold bias in the summer stratosphere, though it is less than in BA. In addition, there is a cold bias in the winter stratosphere, which suggests that, despite the improved simulation of planetary wave amplitudes, the contribution by gravity wave drag to the dynamical forcing of the winter stratosphere must be insufficient. This will be tackled in a new model version by introducing a spectral gravity wave parameterization (Warner and McIntyre 1999). This scheme should also lead to an improved simulation in the equatorial and summer stratosphere, including a successful simulation of the quasi-biennial oscillation (Scaife et al. 2000).

Sensitivity tests were carried out to determine why the jet separation and wave amplitudes in July were simulated better than before. It is clear that the subtropical–polar night jet separation is largely due to the inclusion of convective momentum transport, and the use of a different sea ice climatology. This underlines the point that a successful simulation of the stratosphere can be very much dependent on the representation of the troposphere that is used. However, it is a cause for concern that the improved jet separation resulted partially from using a model sea ice climatology that is considerably different from the old, observations-based, climatology. This highlights the benefits of doing such sensitivity tests: although we have noted an improvement in the simulation, it results partly from the use of a new, apparently less realistic, climatology, which suggests that other aspects of the model may still be inadequate. It also means that care should be taken in interpreting results from stratospheric model climate change studies (e.g., Dameris et al. 1998; Butchart et al. 2000), due to possible sensitivity to lower boundary conditions.

The sensitivity tests failed to explain why stationary planetary wave amplitudes in July are more accurately simulated in the current model version than in BA's version. However, they did suggest that the improvements were not due to changes to tropospheric forcing alone, but were more likely due to differences in wave propagation and dissipation within the stratosphere resulting from a complex interaction between many or all of the new model features. To explain this interaction fully would probably require a much larger number of tests in which varying combinations of new model schemes were replaced by the old schemes (rather than one scheme at a time being replaced, as we have generally done here). However, when processes are coupled it does not always follow that such a systematic procedure would necessarily produce unambiguous results. Hence, we are unable to conclude precisely why the model amplitudes are so much improved, other than perhaps to say it is one benefit of incorporating ever-improving representations of tropospheric and stratospheric physical processes in the model.

Acknowledgments

This work was carried out under the Core Research and Public Meteorological Services Research programs of the Met Office. We thank Adam Scaife and Richard Swinbank for useful comments and discussions. Dingmin Li provided the data for Fig. 1, and John Edwards provided further information on longwave heating rates for N2O, CH4, CFC11, and CFC12.

REFERENCES

  • Amodei, M., S. Pawson, A. A. Scaife, W. Lahoz, U. Langematz, D. Li, and P. Simon, 2001: SAO and Kelvin waves in the EuroGRIPS GCMs and UK Meteorological Office analyses. Ann. Geophys, 19 , 99114.

    • Search Google Scholar
    • Export Citation
  • Boville, B. A., 1995: Middle atmosphere version of CCM2 (MACCM2): Annual cycle and interannual variability. J. Geophys. Res, 100 , 90179039.

    • Search Google Scholar
    • Export Citation
  • Butchart, N., and J. Austin, 1998: Middle atmosphere climatologies from the troposphere–stratosphere configuration of the UKMO's Unified Model. J. Atmos. Sci, 55 , 27822809.

    • Search Google Scholar
    • Export Citation
  • ——, ——, Knight, J., A. Scaife, and M. Gallani, 2000: The response of the stratospheric climate to projected changes in the concentrations of well-mixed greenhouse gases from 1992 to 2051. J. Climate, 13 , 21422159.

    • Search Google Scholar
    • Export Citation
  • Cox, P. M., R. A. Betts, C. B. Bunton, R. L. H. Essery, P. R. Rowntree, and J. Smith, 1999: The impact of new land surface physics on the GCM simulation of climate and climate sensitivity. Climate Dyn, 15 , 183203.

    • Search Google Scholar
    • Export Citation
  • Dameris, M., V. Grewe, R. Hein, and C. Schnadt, 1998: Assessment of the future development of the ozone layer. Geophys. Res. Lett, 25 , 35793582.

    • Search Google Scholar
    • Export Citation
  • Edwards, J. M., and A. Slingo, 1996: Studies with a flexible new radiation code. I: choosing a configuration for a large-scale model. Quart. J. Roy. Meteor. Soc, 122 , 689719.

    • Search Google Scholar
    • Export Citation
  • Gregory, D., R. Kershaw, and P. M. Inness, 1997: Parametrization of momentum transport by convection. II: Tests in single-column and general circulation models. Quart. J. Roy. Meteor. Soc, 123 , 11531183.

    • Search Google Scholar
    • Export Citation
  • ——, Shutts, G. J., and J. R. Mitchell, 1998: A new gravity-wave-drag scheme incorporating anisotropic orography and low-level wave breaking: Impact upon the climate of the UK Meteorological Office Unified Model. Quart. J. Roy. Meteor. Soc, 124 , 463494.

    • Search Google Scholar
    • Export Citation
  • Johns, T. C., R. E. Carnell, J. F. Crossley, J. M. Gregory, J. F. B. Mitchell, C. A. Senior, S. F. B. Tett, and R. A. Wood, 1997: The second Hadley Centre coupled ocean–atmosphere GCM: Model description, spinup and validation. Climate Dyn, 13 , 103134.

    • Search Google Scholar
    • Export Citation
  • Langematz, U., and S. Pawson, 1997: The Berlin troposphere–stratosphere–mesosphere GCM: Climatology and forcing mechanisms. Quart. J. Roy. Meteor. Soc, 123 , 10751096.

    • Search Google Scholar
    • Export Citation
  • Milton, S. E., and C. A. Wilson, 1996: The impact of parameterized subgrid-scale orographic forcing on systematic errors in a global NWP model. Mon. Wea. Rev, 124 , 20232045.

    • Search Google Scholar
    • Export Citation
  • Mitchell, J. F. B., and T. S. Hills, 1986: Sea-ice and the Antarctic winter circulation: A numerical experiment. Quart. J. Roy. Meteor. Soc, 112 , 953969.

    • Search Google Scholar
    • Export Citation
  • Morcrette, J-J., L. Smith, and Y. Fouquart, 1986: Pressure and temperature dependence of the absorption in longwave radiation parameterizations. Contrib. Atmos. Phys, 59 , 455469.

    • Search Google Scholar
    • Export Citation
  • Palmer, T. N., G. J. Shutts, and R. Swinbank, 1986: Alleviation of a systematic westerly bias in general circulation and numerical weather prediction models through an orographic gravity wave drag parametrization. Quart. J. Roy. Meteor. Soc, 112 , 10011039.

    • Search Google Scholar
    • Export Citation
  • Pawson, S., and Coauthors. 2000: The GCM-Reality Intercomparison Project for SPARC (GRIPS): Scientific issues and initial results. Bull. Amer. Meteor. Soc, 81 , 781796.

    • Search Google Scholar
    • Export Citation
  • Pope, V. D., M. L. Gallani, P. R. Rowntree, and R. A. Stratton, 1999:: The impact of new physical parametrizations in the Hadley Centre climate model—HadAM3. Climate Dyn, 16 , 123146.

    • Search Google Scholar
    • Export Citation
  • Scaife, A. A., N. Butchart, C. D. Warner, D. Stainforth, W. A. Norton, and J. Austin, 2000: Realistic Quasi-Biennial Oscillations in a simulation of the global climate. Geophys. Res. Lett, 27 , 34813484.

    • Search Google Scholar
    • Export Citation
  • Simmonds, I., and W. F. Budd, 1991: Sensitivity of the Southern Hemisphere circulation to leads in the Antarctic pack ice. Quart. J. Roy. Meteor. Soc, 117 , 10031024.

    • Search Google Scholar
    • Export Citation
  • Stratton, R. A., 1999: A high resolution AMIP integration using the Hadley Centre model HadAM2b. Climate Dyn, 15 , 928.

  • Swinbank, R., and A. O'Neill, 1994: A stratosphere–troposphere data assimilation system. Mon. Wea. Rev, 122 , 686702.

  • ——, Lahoz, W. A., A. O'Neill, C. S. Douglas, A. Heaps, and D. Podd, 1998: Middle atmosphere variability in the UK Meteorological Office Unified Model. Quart. J. Roy. Meteor. Soc, 124 , 14851525.

    • Search Google Scholar
    • Export Citation
  • Warner, C. D., and M. E. McIntyre, 1999: Towards an ultra-simple spectral gravity wave parameterization for non-orographic gravity waves. Earth Planets Space, 51 , 475484.

    • Search Google Scholar
    • Export Citation

Fig. 1.
Fig. 1.

(a) Difference between longwave heating rates calculated by the Edwards–Slingo scheme and by the Atmospheric and Environmental Research Inc. line-by-line model (solid line), and difference between heating rates calculated by the Morcrette scheme and the line-by-line model (dotted line). In these calculations only heating from CO2, O3, and H2O is included. (b) Difference between shortwave heating rates calculated by the Edwards–Slingo scheme and the scheme used in BA. In both (a) and (b), a midlatitude summer temperature profile from the U.S. Standard Atmosphere is used

Citation: Journal of the Atmospheric Sciences 58, 14; 10.1175/1520-0469(2001)058<2000:AUCOTT>2.0.CO;2

Fig. 2.
Fig. 2.

Zonal mean fields for Jan. (a) Zonal wind for the UM control run (20-yr mean); (b) as in (a) except Met Office analyses (8-yr mean) are shown; (c) temperature for the UM control run; (d) field in (c) minus Met Office analysis temperature field. Negative values are indicated by dashed contours. Contour interval: 10 m s−1 for (a) and (b), 5 K for (c), and 2 K for (d)

Citation: Journal of the Atmospheric Sciences 58, 14; 10.1175/1520-0469(2001)058<2000:AUCOTT>2.0.CO;2

Fig. 3.
Fig. 3.

As in Fig. 2, except fields for Jul are shown and the Met Office analyses are averaged over 7 yr

Citation: Journal of the Atmospheric Sciences 58, 14; 10.1175/1520-0469(2001)058<2000:AUCOTT>2.0.CO;2

Fig. 4.
Fig. 4.

Geopotential height wavenumber-1 stationary wave amplitudes for Jul. (a) UM control run; (b) Met Office analysis. Contour interval: 50 m. The UM fields are averaged over 20 yr and the Met Office analyses over 7 yr

Citation: Journal of the Atmospheric Sciences 58, 14; 10.1175/1520-0469(2001)058<2000:AUCOTT>2.0.CO;2

Fig. 5.
Fig. 5.

(a) Daily minimum temperature at 46 hPa south of 60°S between May and Nov. Thin lines show temperatures for each model year from 1992 to 1998, and the bold lines the range of daily minimum Met Office analyses temperatures in these years. The dashed lines show the threshold temperature for the formation of type I and type II PSCs; (b) as (a), except results from a run where the orographic gravity wave scheme is switched off at 5 hPa instead of 20 hPa are shown

Citation: Journal of the Atmospheric Sciences 58, 14; 10.1175/1520-0469(2001)058<2000:AUCOTT>2.0.CO;2

Fig. 6.
Fig. 6.

Zonal mean zonal wind for Jul. Contour interval: 10 m s−1. Negative values are indicated by dashed contours. (a) Run NOCMT, (b) run OSST_SI, (c) run OSST_SI + NOCMT

Citation: Journal of the Atmospheric Sciences 58, 14; 10.1175/1520-0469(2001)058<2000:AUCOTT>2.0.CO;2

Fig. 7.
Fig. 7.

Geopotential height stationary wavenumber-1 amplitude for Jul (in meters): solid line, control run; bold line, Met Office analyses; dotted line, run OSST_SI+NOCMT; dashed line, run OSST_SI; dash–dot line, run NOCMT; long dashed line, run OGWD

Citation: Journal of the Atmospheric Sciences 58, 14; 10.1175/1520-0469(2001)058<2000:AUCOTT>2.0.CO;2

Table 1.

Description of UM sensitivity runs discussed in section 4

Table 1.
Save