Can the Stratosphere Control the Extratropical Circulation Response to Surface Forcing? Chris Fletcher and Paul Kushner Atmospheric Physics Group University.

Can the Stratosphere Control the Can the Stratosphere Control the Extratropical Circulation Extratropical Circulation Response to Surface Forcing?Response to Surface Forcing?

Chris Fletcher and Paul KushnerAtmospheric Physics GroupUniversity of TorontoCanada

Judah CohenAER Inc., Lexington, MA, USA

AGU Chapman Conference, Santorini. September 27, 2007.

Thanks: Steven Hardiman (U Toronto), Michael Sigmond (UVic) and CFCAS funding

In this talk I will:

• Explain why we are still interested in Siberian snow cover

• Show results from a large ensemble of transient simulations using a low top model forced with anomalous Siberian snow extent

• Demonstrate the large variability in tropospheric responses to snow forcing

• Try to convince you that, to better predict how the troposphere will respond to snow forcing, you should look at the initial state of the stratosphere rather than the troposphere

• Conclude that it’s probably not just about snow

Why are we interested in Siberian snow?

• Snow forcing in autumn/fall is proposed as a source of seasonal predictability in the NH extratropics during winter

• Dynamical mechanism involves wave-mean flow interaction in the stratosphere

Cohen et al. [2007]

Results from NCEP Reanalysis

Cohen et al. [2007]Cohen et al. [2007]

• Correlations ~0.5 suggests it’s unreliable

• Stratospheric circulation anomalies can arise without a clear tropospheric precursor

Results from NCEP Reanalysis

Cohen et al. [2007]Cohen et al. [2007]

• Correlations ~0.5 suggests it’s unreliable

• Stratospheric circulation anomalies can arise without a clear tropospheric precursor

Results from ECHAM3 low top AGCM

Cohen et al. [2007] Gong et al [2003]

Motivation

Cohen et al. [2007]

Research Questions:1. Can snow really act as a

precursor to strat-trop interaction?

2. If so, are there conditions where interaction is more likely to occur?

• Reanalysis data are suggestive but not conclusive; r ~ 0.5• Previous modeling results are from a small ensemble, an old

AGCM and with no information about response variability

GFDL AM2 low-top AGCM

Full atmospheric GCM run in standard configuration for tropospheric seasonal/climate prediction:

– Finite-volume dynamical core: 2o x 2.5o (lat x lon)

– 24 vertical levels with lid at 1hPa; 4 above 100hPa

– Rayleigh drag in top level sponge layer

– Decorrelation timescales of NAM and geopotential height compare well with reanalyses

Experimental Design

i. Set of 100 independent Oct 1st initial conditions from long pre-industrial control run:

– Atmospheric composition = 1870 levels– Climatological SST / sea ice

Experimental Design

i. Set of 100 independent Oct 1st initial conditions from long pre-industrial control run:

– Atmospheric composition = 1870 levels– Climatological SST / sea ice

ii. From each initial condition we fix snow mass at Oct 1st levels then run two new simulations Oct 1st - Dec 31st:

(1) HIGH SNOW = Fixed Oct 1st snow + 40cm snow over Siberia(2) LOW SNOW = Fixed Oct 1st snow

Oct 1st . . . . . . . . . . . . . . . . . . . . . . .Dec 31st

HIGH SNOW k

LOW SNOW k

Initial Conditionk

Time:

40cm Siberia

Snow Mass for Ensemble Member k

Experimental Design

i. Set of 100 independent Oct 1st initial conditions from long pre-industrial control run:

– Atmospheric composition = 1870 levels– Climatological SST / sea ice

ii. From each initial condition we fix snow mass at Oct 1st levels then run two new simulations Oct 1st - Dec 31st:

(1) HIGH SNOW = Fixed Oct 1st snow + 40cm snow over Siberia(2) LOW SNOW = Fixed Oct 1st snow

iii. Response is defined as <HIGH> minus <LOW>

Oct 1st . . . . . . . . . . . . . . . . . . . . . . .Dec 31st

HIGH SNOW k

LOW SNOW k

Initial Conditionk

Time:

40cm Siberia

Snow Mass for Ensemble Member k

Surface Albedo Response

Albedo response ~0.4-0.5

Snow Perturbation Region (typical January extent)

Early Surface Response to Snow Forcing

Peak Cooling ~ 12 K in 2 weeks

SLP maximum ~ 6 hPa

Polar Cap Height Response

-25

15

70

Polar Cap Height Response

-25

15

70

Response highlyvariable in the troposphere:

Can we Better Predict the Response?

[Z] Initial Condition: 5 days before perturbation

48

-12

Following Reichler et al. [2005]

Strongest and most significant ‘predictor’ is

located in the lower stratosphere

Can we Better Predict the Response?

[Z] Initial Condition: 5 days before perturbation

Initially WARM

Initially COLD

115

50

25

Following Reichler et al. [2005]

Strongest and most significant ‘predictor’ is

located in the lower stratosphere

Strat-Trop Interaction Diagnostic

Snow forcing begins Oct 1st, but strat-trop interaction is associated with WAF pulses whose timing is difficult to predict:

- Find strongest WAF pulses then look at lagged SLP response - Does strat. initial condition influence interaction?

d1-50: Find day of max upward ∆WAF at 50 hPa

Wait 10 days, then record 30-day mean

∆SLP(e.g. Polvani and Waugh [2004])

30-day Mean ∆SLP Following WAF Pulses

Init. WARM: 18/51 Init. COLD: 20/49

(e.g. Polvani and Waugh [2004])

Annular Mode Response in SLP

Init WARM Init COLD

(e.g. Polvani and Waugh [2004])

Annular Mode Response in SLP

Init WARM Init COLD

Subtle difference in NAM probabilities, but focused in areas of low seasonal predictability (e.g. East Coast US, Northern Europe).

p(NAM < -1.0):WARM = 0.18COLD = 0.10

Conclusions

1. Can snow really act as a precursor to strat-trop interaction?- Response is highly variable but snow forcing does induce WAF

pulses, causing strat. and trop. warming responses2. Are there conditions where interaction is more likely to

occur?1. Initial condition in polar strat. provides useful predictor of trop.

response to snow forcing (better than trop. predictor)2. Similar freq of interaction in “Init WARM” and “Init COLD” cases3. “Init WARM”: interaction events produce more systematic

downward prop and negative NAM response- Is this really about the snow?

- No. Results should also apply to SST/sea ice forcing

1. Can snow really act as a precursor to strat-trop interaction?- Response is highly variable but snow forcing does induce WAF

pulses, causing strat. and trop. warming responses2. Are there conditions where interaction is more likely

to occur?1. Initial condition in polar strat. provides useful predictor of trop.

response to snow forcing (better than trop. predictor)2. Similar freq of interaction in “Init WARM” and “Init COLD” cases3. “Init WARM”: interaction events produce more systematic

downward prop and negative NAM response- Is this really about the snow?

1. No. Results should also apply to SST/sea ice forcing

- Results suggest a dynamical mechanism- Limitations: strat. resolution/drag scheme,

persistence?- Ongoing: do we observe the same behavior in a strat-

resolving model? Initial findings: confusing…

Conclusions

Fletcher, C. G. et al. (2007) Geophys. Res. Let., In Press.Preprint: http://www.atmosp.physics.utoronto.ca/people/cgf/

The end.

Northern Annular Mode in SLP

Reanalysis

Source: NCEP/CPC

AM2

Zonal mean climatologies

ERA-40 AM2-strat.AM2-trop.