The Effect of Removing a Well-Resolved The Effect of Removing a Well-Resolved Stratosphere on the Simulation of the Stratosphere on the Simulation of the Tropospheric Climate, and Climate Tropospheric Climate, and Climate ChangeChange

Michael Sigmond (University of Victoria) Michael Sigmond (University of Victoria) Paul J. Kushner (University of Toronto)Paul J. Kushner (University of Toronto) John F. Scinocca (University of Victoria, John F. Scinocca (University of Victoria, CCCma)CCCma)

The Effect of Removing a Well-Resolved The Effect of Removing a Well-Resolved Stratosphere on the Simulation of the Stratosphere on the Simulation of the Tropospheric Climate, and Tropospheric Climate, and Climate Climate ChangeChange

Michael Sigmond (University of Victoria) Michael Sigmond (University of Victoria) Paul J. Kushner (University of Toronto)Paul J. Kushner (University of Toronto) John F. Scinocca (University of Victoria, John F. Scinocca (University of Victoria, CCCma)CCCma)

Motivation:Motivation:

Sigmond et al, 2007, (JGR, in press): Investigated Sigmond et al, 2007, (JGR, in press): Investigated robustness of the simulated response to climate change robustness of the simulated response to climate change

We forced 2 AGCM with a generic SST perturbation, varied horizontal resolution, and a single tuning parameter, and compared the responses

Here: investigate robustness of response to climate change to changing model top height

Or: compare the global warming responses in ‘high-top’ with a ‘low-top’ model

Do we need a well-resolved stratosphere to realistically model the future tropospheric climate? (Shindell et al 1998, Fyfe et al. 1999, Gillet et al. 2002)

Method:Method:

Use different versions of the Canadian AGCM (T63 resolution)Use different versions of the Canadian AGCM (T63 resolution) forcing:forcing: 1) double atmospheric CO1) double atmospheric CO22 concentration concentration 2) Forcing with 2) Forcing with ‘best-guess’ SST increase in 2xCO2 world (repeating annual cycle)(repeating annual cycle)

Ensemble average SST response of 17 AR4 models in A1B scenario (2090-2100 minus 2000-1990)

equilibrium runs

All plots DJF

How to compare high-top with How to compare high-top with low-top models?low-top models?

1) Take ‘best-tuned’ low-top model and compare it to ‘best-tuned’ 1) Take ‘best-tuned’ low-top model and compare it to ‘best-tuned’ high-top modelhigh-top model

HIGH:HIGH: - CMAM: state-of-the-art stratosphere resolving GCM - CMAM: state-of-the-art stratosphere resolving GCM - 71 levels with top at 0.001 hPa- 71 levels with top at 0.001 hPa - Used in several studies with interactive chemistry for - Used in several studies with interactive chemistry for stratospheric O stratospheric O33 predictions predictions

- Here: dynamical part (no coupling to chemistry) - Here: dynamical part (no coupling to chemistry)

LOW:LOW: - GCM3: standard Canadian ‘tropospheric’ model - GCM3: standard Canadian ‘tropospheric’ model - 31 levels with top at 1 hPa- 31 levels with top at 1 hPa - Used for climate prediction, e.g. in IPCC AR4 report- Used for climate prediction, e.g. in IPCC AR4 report

Problem:

Model versions have different settings (vertical resolution, tuning, timestep) and physics

differences can be caused by more than just the model lid height

How to compare high-top with How to compare high-top with low-top models? (2)low-top models? (2)

2) Take low-top model and add layers2) Take low-top model and add layers

Problems: - we need to add physics Problems: - we need to add physics (radiation, non-orographic gravity wave drag)(radiation, non-orographic gravity wave drag) - we need to decrease time step- we need to decrease time step

How to compare high-top with How to compare high-top with low-top models? (2)low-top models? (2)

2) Take low-top model and add layers2) Take low-top model and add layers

LOWERED:LOWERED: - lowered version of ‘HIGH’: 41 levels with top at 10 hPa, - lowered version of ‘HIGH’: 41 levels with top at 10 hPa, with physics and dynamics as similar to standard CMAMwith physics and dynamics as similar to standard CMAM

- Not trivial to construct (radiation, sponge layer)- Not trivial to construct (radiation, sponge layer)

Problems: - we need to add physics Problems: - we need to add physics (radiation, non-orographic gravity wave drag)(radiation, non-orographic gravity wave drag) - we need to decrease time step- we need to decrease time step

3) Take the high-top model and remove layers above a certain 3) Take the high-top model and remove layers above a certain heightheight

LOWERED (only removing levels above 10 hPa) LOWERED (only removing levels above 10 hPa) (5y, control)(5y, control)

U

T

HIGH LOWERED LOWERED-HIGH

HIGH HIGH LOWERED LOWERED

(- Removing all layers above 10 hPa)(- Removing all layers above 10 hPa)

- Removing (non-zonal) sponge layer- Removing (non-zonal) sponge layer

- Remove non-LTE LW radiation module- Remove non-LTE LW radiation module

Was not ‘tuned’ for model with 10 hPa top, not needed below 10 hPaWas not ‘tuned’ for model with 10 hPa top, not needed below 10 hPa

- conserve angular momentum in column- conserve angular momentum in column

Instead of letting momentum of gravity waves escape to space, deposit in Instead of letting momentum of gravity waves escape to space, deposit in

uppermost layer (see Shaw et al. poster)uppermost layer (see Shaw et al. poster)

HIGH vs LOWERED (5y, control)HIGH vs LOWERED (5y, control)

U

T

HIGH LOWERED LOWERED-HIGH

RESPONSE to Climate RESPONSE to Climate changechange

(40 year equilibrium runs)(40 year equilibrium runs)

HIGH LOW

AO+ AO+

LOWERED LOW-G

∆ ∆ SLP = SLPSLP = SLP2xCO22xCO2 - SLP - SLPcontrolcontrol

AO+

(0.001 hPa top) (1 hPa top)

(10 hPa top)

Model lid height?

No! Just lowering model lid height does NOT change pattern of response

(amplitude ~50%)

HIGH LOW

LOWERED (10 hPa top) LOW-G

∆∆u = uu = u2xCO22xCO2 - u - ucontrolcontrol(0.001 hPa top) (1 hPa top)

-HIGH and LOWERED responses similar, but LOW response is different

-Anomalous LOW response must be caused by difference in physics/model settings in LOW compared to LOWERED/HIGH

Which model setting in LOW compared to Which model setting in LOW compared to LOWERED causes the response to be so LOWERED causes the response to be so different?different?

LOW LOW-G LOWERED

# levels 32 32 41

top: 1 hPa 1 hPa 10 hPa

Vert res (tropopause)

~2 km ~2 km ~1.2 km

Sponge layer: YES YES NO

non-oro gravity waves:

NO NO YES

Gphil (~oro gravity waves; Scinocca and

McFarlane 2000)

1.0 0.65 0.65

Make setting in LOW equal to that in LOWERED/HIGH and check if responses become more similar

HIGH LOW

LOWERED LOW-G

∆ ∆ SLP SLP

AO+

(Gphil=1.0)

(Gphil=0.65) (Gphil=0.65)

(Gphil=0.65)

AO+ AO+

HIGH LOW

LOWERED (Gphil=0.65) LOW-G

∆ ∆ UU (Gphil=0.65) (Gphil=1.0)

(Gphil=0.65)

Response is more dependent on Gphil than on model lid height!!

UUcontrolcontrol

HIGH LOW LOW-G

Close to observations

Too weak (waveguide to narrow)

Closer to observations

gphil=0.65gphil=0.65 gphil=1.0

ConclusionsConclusions

Assessing the benefit of including a well-resolved stratosphere on Assessing the benefit of including a well-resolved stratosphere on the simulation of climate (change) is the simulation of climate (change) is not straightforwardnot straightforward

Response in Response in standard ‘low-top’standard ‘low-top’ model (no AO response) is model (no AO response) is differentdifferent from that in standard from that in standard ‘high-top’‘high-top’ model (AO+) (for this model (AO+) (for this model)model)

When When only lowering model lid heightonly lowering model lid height, the responses , the responses do not do not changechange very much very much

By making the By making the orographic gravity waveorographic gravity wave settings in the standard settings in the standard low-top model consistent with that in the standard high-top model, low-top model consistent with that in the standard high-top model, we can get a very similar response as in the high-top model we can get a very similar response as in the high-top model

The strength of orographic gravity waves appears crucial for The strength of orographic gravity waves appears crucial for response to climate change, response to climate change, more so than the model lid height more so than the model lid height (in this model) (in this model) (pretty scary, isn’t it?)(pretty scary, isn’t it?)

Michael.sigmond@ec.gc.ca

LOW vs LOW-G LOW vs LOW-G LOW LOW-G LOW-G minus LOW

Closer to observations

gphil=0.65

gphil=0.65 gphil=1.0

CONTROL

2xCO2 Climate

∆∆T T

LOWERED LOW-G

HIGH LOW

HIGH LOW

LOWERED (10 hPa top) LOW-G

∆∆u (40 years)u (40 years)(0.001 hPa top) (1 hPa top)

HIGH LOW

∆∆u (40 years)u (40 years)(0.001 hPa top) (1 hPa top)

LOWERED (10 hPa top) LOW-G

HIGH LOW

∆∆u (year 1-20)u (year 1-20)(0.001 hPa top) (1 hPa top)

LOWERED (10 hPa top) LOW-G

HIGH LOW

∆∆u (year 21-40)u (year 21-40)(0.001 hPa top) (1 hPa top)

LOWERED (10 hPa top) LOW-G

‘‘Construction’ of LOWEREDConstruction’ of LOWEREDstep 1: removing all layers above 10 hPastep 1: removing all layers above 10 hPa

U

T

HIGH LOWERED LOWERED-HIGH

‘‘Construction’ of LOWEREDConstruction’ of LOWEREDstep 2: removing (non-zonal) sponge layer step 2: removing (non-zonal) sponge layer

U

T

HIGH LOWERED LOWERED-HIGH

‘‘Construction’ of LOWEREDConstruction’ of LOWEREDstep 3: Remove non-LTE LW radiation step 3: Remove non-LTE LW radiation

modulemodule

U

T

HIGH LOWERED LOWERED-HIGH

‘‘Construction’ of LOWEREDConstruction’ of LOWEREDstep 4: conserve angular momentum in step 4: conserve angular momentum in

columncolumn

U

T

HIGH LOWERED LOWERED-HIGH