Can we trust the simulated gravity-wave response to

climate change?

Ted Shepherd

Department of Physics

University of Toronto

NCAR TIIMES Gravity-Wave Retreat, 2006

• GW parameterizations are highly tuned to reproduce current climate– So why should we trust their response to

climate change?

• There are two issues:– Changes in source characteristics– Changes in propagation and dissipation

• This talk will only address the latter

• Will focus primarily on polar vortex

Impact of parameterized mesospheric GW drag on downwelling and temperature over the winter

pole in a zonal mean modelDashed line is without GW drag, solid line is with GW drag

From Garcia & Boville(1994 JAS)

From Beagley et al. (1997 Atmos.-Ocean)

Cumulative contribution of resolved and parameterized wave drag at various altitudes on polar downwelling at 10 hPa in CMAM with

only orographic GW drag

Parameterized Resolved

Total

From Austin et al. (2003 ACP)

Wave driving vs polar temperature in the Antarctic

Heat flux at 100 hPa estimates the (resolved) wave activity entering the stratosphere

More wave forcing implies more polar downwelling and a warmer pole

Differences reflect GW drag

• Holton (1983 JAS) explained the mesospheric cooling observed above stratospheric sudden warmings as due to a GW feedback– Filtering of GW momentum fluxes leads to

a positive wave drag anomaly

• Same reasoning applies to climate perturbations, e.g. to the ozone hole

• How robust is this effect?

Shaw & Shepherd (JAS, in press)

Response of downwelling over SH polar cap to combined effects of climate change and ozone depletion• Solid line shows October, dashed shows November• Left is total downwelling, right only from resolved EPFD

From Manzini et al. (2003 JGR)

• There is a strong constraint from (angular) momentum conservation

• In the steady limit, downwelling is constrained by “downward control” (Haynes et al. 1991 JAS) [F is force/unit mass]

• For GWs, this simplifies to

€

w*

= −1

aρ cosφ

∂

∂φ

ρF cosφ

2Ωsinφdz

z

∞

∫

€

w*

= −1

acosφ

∂

∂φ

cosφu'w'

2Ωsinφ

⎛

⎝ ⎜

⎞

⎠ ⎟

(assuming no flux of momentum to space)

• Thus the downwelling at a given height is independent of exactly where the waves break above that height– What goes up must come down

• But what happens at the model lid?

• If any momentum flux remaining at the model lid is thrown away, then

which now depends on model lid height

€

w*

= −1

acosφ

∂

∂φ

cosφu'w'

2Ωsinφ

⎛

⎝ ⎜

⎞

⎠ ⎟z

−cosφu'w'

2Ωsinφ

⎛

⎝ ⎜

⎞

⎠ ⎟ztop

⎡

⎣

⎢ ⎢

⎤

⎦

⎥ ⎥

• To conserve momentum, any remaining momentum flux at the model lid must be deposited as a drag, e.g. in the top few levels of the model– This maintains the integrity of the

downward control relation– Throwing away momentum flux is

equivalent to imposing an opposite drag above the model lid

• Also, there must be no Rayleigh drag or zonal mean sponge layer (Shepherd et al. 1996 JGR)

GW feedbacks to a radiative perturbation

Physicallyconsistent

Rayleigh drag (violates momentum constraint)

From Shepherd & Shaw (2004 JAS)

• Difference between 80-km and 96-km lids with Hines GWD in Met Office UM (letting momentum flux at model lid escape to space)

• Influence extends to low altitudes

From Lawrence (1997 JGR)

• Top: ratio of downwelling in 96-km model from below 80 km to below 96 km

• Bottom: ratio of downwelling in 80-km model to that in 96-km model

From Lawrence (1997 JGR)

• The effect of a background jet on an anti-symmetric source spectrum is to create a dipole of negative drag above positive drag, hence polar downwelling

• Imposing a polar cooling shifts each part of the drag dipole, so the difference drag is composed of two dipoles, driving two circulation cells (left)

Rayleigh drag gives a single-signed response (unphysical)

• Enforcing momentum conservation can improve the robustness of GWD feedback to polar cooling

Non-MC AD99

MC AD99+ RD

MC AD99Low lid

AD99

Circulation response to polar cooling at ~15 km

• Dashed is 80°N, solid is 85°N

• a=control– cf. Garcia & Boville (JAS,

1994)

• c=MC AD99– Physical response is

significant

• e=non-MC AD99 (50 km lid)– Spurious response is also

significant

Vertical profile of downwelling in response to polar cooling around 15 km, with AD99 GWD scheme

• Sensitivity of AD99 and H97 induced downwelling to model lid height

MC

Inferred from downward control (dashed)

non-MC (solid)

Downwelling at 25 km, 85°N

Tropospheric circulation

Actual (solid)

Anti-symmetric Asymmetric Sensitivity to the source spectrum

Resting state

With polar cooling

Difference

• Conclusion: GW induced warming above an imposed polar cooling is robust to– Model lid height– Source spectrum– Breaking criterion– Background flow

if any only if momentum is conserved

Zonal mean wind at SH midlatitudes in CMAM and in observations

• GW drag doesn’t just slow the mesospheric jet, it reverses it above about 90 km altitude (so isn’t really a “drag”)• Requires non-zero GW phase speeds

From Beagley et al. (2000 GRL)

• Doubled CO2 simulations with the CMAM (note no heterogeneous chemistry in these runs)

• We separate the effect of doubled CO2 from that of the associated change in SSTs (taking SSTs from CCCma coupled atmosphere-ocean run)– The combined response is surprisingly linear

• Figure shows temperature change in January(blue is 99% significant, purple 90%)

Fomichev et al. (JC, 2006)

Total response From 2xCO2 From SSTs

• There is a robust dynamical temperature response at the summer mesopause

• Tropospherically induced dynamical changes negate the CO2-induced cooling

• From gravity-wave drag• Consistent with the lack of

a cooling trend in observations

Fomichev et al. (JC, 2006)

Summary

• There are some robust aspects to the GW response to climate change (assuming fixed source spectra)– Based on filtering of GW fluxes

• Robustness depends on enforcing momentum conservation

• Without momentum conservation, model intercomparisons will be ill-posed

• However there is a robust response in the lower tropical stratosphere

• Tropospherically induced changes now augment the CO2-induced cooling

• Increased upwelling from stratospheric wave drag (in both NH and SH)

Fomichev et al. (JC, in revision)

• The annual cycle of tropical and extratropical 50 hPa temperature (global mean is subtracted) points to a strengthened diabatic circulation

Fomichev et al. (JC, in revision)

Control

2xCO2+SST

2xCO2+SST

Extratropics

Tropics

• Changes to tropical upwelling at 70 hPa– Black from resolved EPFD, gray total– Half these models use Rayleigh drag

From Butchart et al. (CD, in press)