1. Role of the Stratosphere in Climate Modelling: The Connection Between the Hadley and the Brewer-Dobson Circulation M. A. Giorgetta (1), E. Manzini (2), M. Esch (1) and E. Roeckner (1) (1) Max Planck Institute for Meteorology, Hamburg, Germany (2) Istituto di Geofisica e Vulcanologia and Centro Euro- Mediterraneo per i Cambiamenti Climatici, Bologna, Italy

2. Motivation The tropospheric mean circulation in a GCM depends on the representation of the stratosphere [Boville 1984] Tropospheric weather is sensitive to the state of the stratosphere [Baldwin et al. 2003] Are climate change projections sensitive to the stratospheric representation? (Most AOGCMs used for AR4 do not fully resolve the stratosphere)  Investigate and demonstrate effects of different models of the stratosphere on the tropospheric climate in GCM experiments  Contribute to SPARC DynVar “Top”

3. Aims of this work  Explore effects of “Low Top” vs. “High Top” GCMs  Low top atmosphere: troposphere + lower stratosphere lower stratosphere = upper boundary region of AGCM  High top atmosphere: trop. + strat. + lower mesosphere Use coupled atmosphere ocean GCM to explore effects of different stratospheric representations on the tropospheric climate Use atmospheric GCMs with prescribed lower boundary conditions

4. Experimental design MPI-M AGCMs and AOGCMs: ECHAM5 Low Top atmosphere, p top = 10 hPa (Roeckner et al. 2006) MAECHAM5High Top atmosphere, p top = 0.01 hPa (Manzini et al. 2006) ECHAM5/MPIOMLow top atmosphere / ocean (Jungclaus et al. 2006) MAECHAM5/MPIOMHigh top atmosphere / ocean AM-LOW ECHAM5(T63L31) AMIP2 SST+ice (1978-1999) AM-HIGH MAECHAM5(T63L47) AMIP2 SST+ice (1978-1999) CM-LOW ECHAM5(T63L31)/MPIOM(GR1.5L40) 100 years (CTRL exp. for IPCC AR4) CM-HIGH MAECHAM5(T63L47)/MPIOM(GR1.5L40) 100 years

5. Experimental design Common features of all 4 experiments Horizontal atmospheric resolution T63 / ~1.9°x1.9° Troposph. vertical grid: 26levels in [surface, 110hPa] Dynamics and processes in troposphere Ocean model: ~1.5° resolution, 40 levels Differences Vertical resolutions from ~110 hPa to 0 hPa Low top:31 levels, 5 levels in ]110,10] hPa High top:47 levels, 9 levels in ]110, 10] hPa +12 levels in ]10, 0.01] hPa

6. Experimental design Differences (cont.) Horizontal diffusion: dx/dt = -(-1) q ∙K x ∙ ∇ 2q x, 2q=8 Low top:  To avoid spurious wave reflection at the upper boundary, the order of hyper-diffusion is reduced in the stratosphere: 2q=(6,4,2,2,2) at (90, 70, 50, 30, 10 hPa)  Acts on waves, incl. large scale waves, and zonal mean High top:  Equal order of hyper-diffusion 2q=8 at all levels Gravity wave drag parameterization Low top:  Orographic GWD (Lott and Miller, 1999) High top:  Orographic GWD (Lott and Miller, 1999)  GWD from a spectrum of gravity wave with atmospheric sources. (Hines, 1997)

7. Coupled experiments Low top: CM31 is a ~500 year control experiment for CMIP3 High top: CM47 is started from an ocean state of the CM31 simulation, the atmosphere is initialized at the new vertical resolution Initial drift of CM47 over ~60 years Compare years 61 to 160 of CM47 with a 100 year period of CM31

8. Questions Is the tropospheric climate different between the Low Top and High Top CM simulations? What differences occur if the lower boundary conditions (SST+ice) are prescribed – and how much do these changes correspond to changes in the coupled system? Which mechanisms induce these changes?

9. Coupled model Annual mean temperature T (K)

10. Coupled model Annual mean U (m/s)

11. Coupled model Annual mean residual vertical velocity w* (mm/s)

12. Coupled model vs. uncoupled model Annual mean temperature T (K)

13.  Common in coupled and AMIP experiments  T is significantly changed in the stratosphere and upper tropical troposphere  Hadley circulation stronger in high top model  Brewer-Dobson circulation stronger in high top model  Differences  AMIP: dT in troposphere is ~0 below 300 hPa  Coupled: dT = ~0.5 K in troposphere below 200 hPa  Differences between coupled experiments must be explained by different stratospheric forcing terms and resolution effects

14. Dynamical forcing terms in the stratosphere dU/dt|dyn = dU/dt by Div. of EP-flux

15. Conclusions A low and high top AGCM has been used for uncoupled and coupled experiments to explore effects of different models of the stratosphere on the troposphere Using identical resolution in the troposphere and the same tropospheric parameterizations, the tropospheric climate changes under the influence of the stratospheric dynamics. The analysis of dynamical forcing terms shows: Horizontal diffusion acting on large scale waves becomes visible as a strong difference in EP flux divergence at 50 hPa Horizontal diffusion is also non-negligible in the zonal mean Between 50 hPa and 10 hPa, the change in Div.F drives the changes in the Brewer-Dobson circulation and thereafter the Hadley circulation changes  Stratospheric representation matters for tropospheric climate N.B.:Resolution effects would be much larger for better resolution. Use of MAECHAM5 with ~90 layers would generate QBO  Amplification of interannual variability See also poster of Shaw and Shepherd

16. END