Page 1
Stratosphere-Troposphere Dynamical Coupling and its relationship to
Annular Variability of the Troposphere
Michael Blackburn(1), Joanna D. Haigh(2), Isla Simpson(2,3), Sarah Sparrow(1,2)
(1) NCAS-Climate, Department of Meteorology, University of Reading, UK
(2) Space and Atmospheric Physics, Imperial College London, UK
(3) Department of Physics, University of Toronto, Canada.
NCAS Staff Meeting 11 November 2009
Page 2
Outline
• Motivation – response to climate forcings
• Idealised stratospheric heating experiments
equilibrium response
spin-up ensembles – mechanisms
unforced annular variability
• Dependence on tropospheric climatological basic state
Page 3
Climate Change: annular response
Lorenz & DeWeaver (2007)
IPCC AR4 models
2080-2099 minus 1980-1999
A2 scenario (“business as usual”)
Zonal mean zonal wind 850hPa zonal wind
Temperature change
Yin (2005); Miller et al (2006);
Page 4
© Imperial College LondonPage 4
Polar cap T trend (2001-2050) DJF u trend (2001-2050)
Comparing IPCC models with and without ozone recovery:
With recovery With recoveryWithout recovery Without recovery
21st century Ozone Recovery
Son et al (2008), Science
Page 5
Solar index regressions using reanalysis data
Frame & Gray (2009)
ECMWF reanalyses 1979-2001 (ERA-40)
Observed stratospheric temperature signal
solar max - solar min
Crooks & Gray (2005);
Page 6
Circulation changes over the 11-year cycle
• Weakening and poleward shift of the mid-latitude jets• Weakening and expansion of the Hadley cells• Poleward shift of the Ferrell cells
Haigh and Blackburn (2006)
Multiple regression analysis of NCEP/NCAR reanalysis, DJF, 1979-2002
Page 7
Simplified GCM - “dynamical core” model
Control run zonal wind
Control run temperature
Relaxation Temperature
Based on Hoskins & Simmons (1975) primitive equation model• Spectral dynamics: T42 L15
• Newtonian relaxation (Held-Suarez)
• Boundary layer friction (Rayleigh drag, σ > 0.7)
• No orography / forcing of planetary waves
Page 8
Idealised stratospheric heating
• Heating perturbations can be applied to the stratosphere by changing the relaxation temperature profile
P10 Polar heating (10K)
5K0K
5K 0K
E5 Equatorial heating (5K) U5 Uniform heating (5K)
10K
• Applied 3 different
heating perturbations
Haigh et al (2005)
Page 9
Equilibrium ResponseZonal mean Temperature
Zonal mean zonal wind
Control zonal wind
E5 U5 P10
E5 U5 P10
E5 case gives a similar response in the troposphere to that seen over the solar cycle
Page 10
• Poleward or equatorward shift of tropospheric jet dependent on stratospheric heating distribution
• Coherent displacement of the jet and storm-track
• How does this arise?
• Spin-up ensemble for the equatorial heating case:
– 200, 50-day runs
Ensemble spin-up Experiments
5K 0K4.5K0.5K
Simpson et al (2009)First recall storm-track diagnostics....
Page 11
• Flux of wave activity in latitude-height plane
• Conserved following eddy group velocity (assumptions)
• Components proportional to eddy heat + momentum fluxes
• E-P flux divergence quantifies eddy forcing of mean state
Eliassen-Palm flux
Eliassen & Palm (1961); Charney & Drazin (1961); Andrews & McIntyre (1976); Edmon et al (1980)
Page 12
Eddy-feedback processes
Ensemble spin-up response to stratospheric heating distributions in the idealised model (Simpson et al, 2009)
Tropopause [qy] trigger
|][|
][~2
cu
qn y
Refraction feedback amplifies tropospheric anomalies
Baroclinicity feedback moves wave source
t
uF
.
E-P Flux, days 0 to 9 E-P Flux, days 20 to 29 E-P Flux, days 40 to 49
u, days 20 to 29 u, days 40 to 49Heating: δT_ref
zFz
u
Page 13
Response to forcing projects onto leading annular modes
(2D phase space projection)
EOF 1 (51.25%) EOF 2 (18.62%)Control Run
Latitude (equator to pole) →
Hei
ght
→Leading Modes of Variability
Sparrow et al (2009)
Po
lew
ard
Eq
ua
torw
ard
Narrower, Stronger
Broader, Weaker
PC1 Amplitude
PC
2 A
mpl
itude
E5U5
C
Page 14
• Poleward jet migration
• Driven by upper-level momentum / EP-fluxes
Low frequency variability - dynamics
Sparrow et al (2009)
Low frequency phase-space circulation
• Positive eddy feedback
• Leads to long timescales of variability
• Similarity to forced response
High-PC1 composite
Page 15
E5 dependence on tropospheric basic state
• Equilibrium experiments with modified tropospheric reference temperature
• Stronger response to stratospheric forcing for lower latitude jets
• Indicative of stronger eddy feedback (despite weaker eddies in control)
TR1 TR2 TR3 TR4
Decreasing baroclinicity Increasing baroclinicity
TR5
TR
u
E5δu
NOTE: THERE IS 1 BLANK BOX
HIDING TEXT ON THE RIGHT
Page 16
Control climatology E5 response
Page 17
Possible causes of sensitivity
• Under investigation….
• Not due to proximity of stratospheric heating to jet (equatorial and polar heating responses scale similarly)
• Timing of spin-up response should indicate refraction versus baroclinic mechanism:
- apparently conflicting evidence
- projection of eddy forcing onto wind response varies
- responses differ only after ~50 days
• Are different responses related to unforced variability?
Page 18
Relationship to unforced internal variability
• Find strongest response to forcing for lower latitude jets
• How is this related to the unforced internal variability?
• Fluctuation-Dissipation Theorem (FDT) predicts a stronger response for longer timescales of internal variability
• Due to stronger internal (eddy) feedbacks, maintaining the leading mode(s) of variability against dampingNOTE: THERE IS 1
BLANK BOX HIDING PLOTS ON
THE RIGHT FDT references: Leith (1971); Ring & Plumb (2008) etc
Page 19
© Imperial College LondonPage 21
Timescales of variability
• 1-point correlation maps of zonal wind anomalies wrt peak easterly response at 200hPa
• Mid-latitude jets: short timescale; propagating
• Low latitude jets: long timescale; stationary
Page 20
Annular variability in TR3 control
• Evidence for 2 types of natural variability:
poleward propagating anomalies – short timescale
persistent stationary anomalies – long timescale
• Persistent behaviour dominates for lower latitude jets
• Propagating behaviour dominates for higher latitude jets
• Need to separate and characterise these distinct “modes” of variability
Page 21
Conclusions
• Mechanism identified by which stratospheric change displaces the tropospheric jet and storm-track.
• Relevant to the tropospheric response to all stratospheric climate forcings.
• Dynamical mechanism is related to the slowest modes of annular variability.
• Forced response and variability are both driven by storm-track transient eddy feedback.
• Strength of eddy feedback depends on the latitude / width of the jet:
- GCMs need realistic variability for correct forced response (FDT)
- 2 types of annular variability in sGCM - under investigation
Page 22
- Questions? -
NCAS Staff Meeting 11 November 2009
Page 24
Idealised GCM: annular response
Lorenz & DeWeaver (2007)
Zonal wind response to localised heating 150hPa deep, 20° wide latitude
Page 26
Phase Space View of Momentum Budget
• Eddies change behaviour at high and low frequencies and jet migration changes direction.
• At low frequencies it is unclear what drives the poleward migration.
0
1 sp
ZONAL EDDY Su dp C Cg t
PC1 →
PC
2 →
PC1 →
PC
2 →
Low Pass High Pass
Page 27
Empirical Mode Decomposition: Phase SpaceMode 1 Mode 2
Mode 4
Mode 3
Mode 6Mode 5
Tc = 4.96 ± 0.05 days Tc = 8.0 ± 0.3 days Tc = 20.3 ± 0.8 days
Tc = 39 ± 2 days Tc = 78 ± 5 days Tc = 198 ± 19 days
Page 28
Transformed Eulerian Mean Momentum Budget
High Frequencies: • Eddies drive equatorward
migration.• Eddies out of phase with
winds near the surface.
Intermediate Frequencies:• Eddies drive poleward
migration.• Residual circulation drives
jet migration at lower levels.
• Eddies in phase with the winds near the surface.
][][
][cos][cos
][][ *
**
Fp
uwu
a
vvf
Fcos
1][
adt
ud––+ ω
Page 29
Eddy feedback processes
Refractive Index determined by wind anomalies
|][|
][~2
cu
qn y
Eddies propagate towards high refractive index
Resulting EP-flux divergence drives zonal wind changes (phase offset)
Eddy source lags baroclinicity (zonal wind anomalies) by 2-4 days
Latitude Latitude Latitude Latitude
Hei
ght
Hei
ght
Hei
ght
Hei
ght
Latitude
Hei
ght
Latitude
Hei
ght
Latitude
Hei
ght
LatitudeH
eigh
t
Hig
h F
requ
ency
Low
Fre
quen
cy
Page 30
Conclusions
• Annular variability at different timescales in a Newtonian forced AGCM:
– Equatorward migration of anomalies at high frequencies
– Poleward migration at low frequencies
• For all timescales the jet migration is driven by the eddies at upper levels and conveyed to lower levels by the residual circulation.
• Evidence for two feedback processes:
• Eddy source responds to low-level baroclinicity, with lag 2-4 days:
– High frequency flow is so strongly eddy driven that wind anomalies almost out of phase with wave source.
– Low frequency wind anomalies and eddy source are almost in phase.
• Wind anomalies dominate refractive index, leading to positive eddy feedback via EP-flux divergence.
• Direction of propagation from relative phases of wave source/sink and wave refraction.