The atmospheric response to an Oyashio SST front shift in an

atmospheric GCM

Dima Smirnov, Matt Newman, Mike Alexander, Young-Oh Kwon & Claude

Frankignoul

August 6, 2013Workshop on SST Fronts

Boulder, Colorado

Impact of SST fronts on mean state Significant impact has now been shown

Minobe et al., 2008

Nakamura et al., 2008

Front (solid)No front (dash)

SST anomalies in front/no-front experiments approach 10°C

75%300%

Impact on variability

Is the response mainly in the boundary layer? Locally confined? Is the atmosphere sensitive enough to respond to

realistic SST front variability?

30% ~12% w/ obs SST (solid)smoothed (dash)

ΔSST

Taguchi et al., 2009

Experimental design

SST anomaly based on the Oyashio Extension Index

(1982-2008)

Outside of the frontal region (dSST/dy < 1.5 °C 100 km-1),

SST anomalies are masked

dSST/dy (°C 100 km-1)

WARM

COLD

OEI from Frankignoul et al., 2011

Model information

NCAR’s Community Atmosphere Model (CAM), version 5

25 warm/cold ensembles with different atmospheric initial

states from control run (taken a year apart)

Two simulations:

1. High-resolution (HR): Uses 0.25° CAM5.

2. Low-resolution (LR): Uses 1° CAM5.

Identical initial land, sea-ice and atmospheric initial

conditions

Compare the Ensemble mean difference (WARM – COLD)

between the HR and LR model responses

Model Experiments

Horizontal circulation

Turbulent heat flux is 10-20% stronger in LR

LR response is seasonally dependent

Both models imply a ~6-month persistence time for a 150-m mixed layer

HR LR

L L

Mean Nov-Mar difference: SLP (contour), turbulent heat flux (color), 2-m wind (arrow)

NCEP

L

SST (thin contour), SLP (thick contour)

Vertical circulationω (contour, 1.5x10-3 Pa s-1) div (color, s-1)

latitude

ERA-Int

HR

LR

What is the cause of the stronger circulation in the HR model?

+50%

Vertical circulation: forcingDecompose ω using the generalized ω equation:

thermal advection vorticity advectiondiabatic heatingHR: Model

OutputHR: All forcing

Re-constructed (left) not perfect, but still useful to compare contribution of individual terms.

Vertical circulation: forcingDiabatic heating:

Vorticity advection:

Δω (contour)

ΔQDIAB (color)

HR LR

Δω (contour)Δ(HR-LR) (color)

HR LR

Role of eddies : high-pass v’T’

NCEP

HR

LR

Eddies in HR show a much greater sensitivity to the SST frontal shift

850mb v’T’ (mean: contour, diff: color)

Cross-section across the front

2

-2K m s-1

Thermodynamic budget: 950mb

HR LR

°C day-1

<5%

Thermodynamic budget: 700mbHR

°C day-1

LR

Conclusions A high resolution model (<1°) is required to capture

the atmospheric response to the Oyashio SST front shift

For CAM5, movement of heat from the warm side of the SST front is strongly resolution dependent: In HR, a strong upward heat flux maintains a vertical

circulation through the depth of the troposphere In LR, heat is removed largely by horizontal eddy

fluxes, causing a shallower vertical circulation Unlike the LR, the HR develops a robust shift in the

storm track Collectively, what does this mean for the large scale

response?

Remote response

HR LRNDJ NDJ

HR JFM

Sea-level pressure

LR JFM

Looking ahead Can the difference in the HR and LR responses be

explained with a simpler model? Is the difference related to differences in the mean state? Employ a simplified GCM forced by diabatic heating.

How much of the difference in the HR and LR responses is actually due to a better resolved SST front, versus a higher-resolution atmosphere. A “smooth” HR simulation (1° SST with a 0.25° GCM)

appears to suggest that atmospheric resolution plays a larger role than SST front strength.

Additional Slides

Precipitation

Role of eddies

EHFC – low pass

Smoothed HRExperiment (0.25° CAM5)