Recent intensification of wind-driven circulation in the Pacific and the

ongoing warming hiatus

England, M. H., S. McGregor, P. Spence, G. A. Meehl, A. Timmermann, W. Cai, A. Sen Gupta, M. J McPhaden, A.

Purich, A. SantosoNature Clim. Change 4, 222–227 (2014)

Yu Kosaka, Journal Club (Apr 24, 2014)

The global warming hiatus

• The surface global warming appears to have stopped for this century• CMIP5 historical simulation (until 2005) + RCP scenario (after 2006)

does not reproduce thus hiatus as ensemble mean

Hiatus is attracting wide interests

• Mar 2014 issue of Nature Clim Change was a special issue on hiatuso Editorialo 1 of Correspondenceo 5 Commentarieso News Featureo 1 Article (England et al.)

• Mar 2014 issue of Nature Geoscio Editorialo 2 Commentarieso 1 of Letters

http://www.nature.com/nclimate/focus/slowdown-global-warm/index.html

• News Feature on Jan 15 issue of Nature

What is causing this hiatus?

Changes in radiative forcing• Solar activity minimum ~2009 (Kaufmann et al. 2011 PNAS)

• Anthropogenic aerosol increase (Kaufmann et al. 2011 PNAS)

• Small volcanic eruptions (Solomon et al. 2011 Science, Santer et al. 2014 Nat Geosci)

• Decline in methane emissions (Estrada et al. 2013 Nat Geosci)

• Stratospheric water vapor increase (Solomon et al. 2010 Science)

(Probably) internal variability • La Niña-like decadal trend (or PDO, IPO) (Meehl et al. 2011 Nat CC, 2013 J Clim, Kosaka & Xie 2013 Nature)

Energy budget• Deeper ocean heat uptake (Watanabe et al. 2013 GRL,

Levitus et al. 2012 GRL, Balmaseda et al. 2013 GRL)

– But is this internal or forced?

Obs. global-mean temperature and

trade winds• Observed hiatus and

accelerated GW coincide with IPO+ and IPO–, respectively

• Recent IPO– consistent with negative wind stress in the Eq Pacific

• Note large uncertainty in reanalysis before 1980s (Tokinaga et al. 2013 Nature)

o The recent trend is not necessarily “unprecedented” in WASWind data (Tokinaga and Xie 2011)

ANN GISTEMP anom and 5-yr running mean

Wind stress anom and 20-yr trends[6ºS-6ºN, 180º-150ºW]

Observed (& CMIP5) trends for 1992-

2011

• Acceleration of Walker circulation (against slowdown due to GW)

• Negative phase of IPO (La Niña-like + extratropical anom)

SLP & sfc winds (ERAi)

SSH (AVISO)

SST (HadISST)

SAT (GISTEMP)

SAT (CMIP5 historical + RCP4.5)

Emphasis on the tropical Pacific

• IPO is defined as EOF3 of 13-yr low-pass filtered global SST

• Observed recent sfc wind trend is stronger than regressed trend onto the IPO index

• Suggesting that tropical processes induced the recent IPO trend

Ocean circulation trend (1992-2008)

• Surface Ekman divergence• Thermocline convergence advecting warm

subtropical water to the equatorial subsurface• Acceleration of Equatorial undercurrent

Surface u Surface v

100-200m u 100-200m v

Based on SODA

Trends in ocean-atmos circulation &

ocean temp

Stronger trade winds

Shallower thermocline in east Pacific

Surface cooling in central & east Pacific

Ocean temperature based on reanalyses

Deeper thermocline in west Pacific

Subsurface warming in west Pacific

Global SAT change

x-z perspective

Trends in ocean-atmos circulation &

ocean temp

Stronger trade winds

Equatorial upwelling

Surface cooling

Ocean temperature based on reanalyses

East-west SSH gradient

Subsurface warming

Equatorward subsurface current

(subtropical cell)

Global SAT change

y-z perspective

Model experiment I: OGCM• OGCM + atmos EBM + sea ice model + land model

o Atmos. heat and moisture are coupled with ocean/ice/lando Pacific surface wind (momentum + scalar wind) is prescribed

Global SAT anom Global SAT change(?)

Radiative forcing

historical (no volcano) RCP6

Sfc wind

control climatology –

Pacific wind-forced

climatology

climatology + obs linear

trend(45ºS-45ºN

Pac)

(1)Trend reverses at 2012(2)Stabilizes at 2012 level(3)Trend continues until

20202011/20121991/1992 time

OGCM trend for 1992-2011

• Negative IPO trend• Intensification of

subtropical cell (sfc div, thermocline conv)

• Sfc cooling, subsfc warming

• Eq Pacific cooling is far stronger (> 2x) than obs

SST & ocean circulation Pacific zonal-mean temp & circulation

• Thermocline warming is consistent with obs• Net ocean heat gain of 1.2 x 1022 J at 2012 due to sfc wind• Explains ~50% of slowdown in global SAT rise until 2012

Ocean reanalysis trend

OGCM trend for 1992-2011

• Accelerated subtropical cell circulation

• Intensified subtropical gyre

v and (u,v) trends

SST and column integrated circulation trends

Model experiment II: CGCM• CGCM (CSIRO-Mk3L)• Pacific surface wind (momentum flux only) to ocean

is prescribed

Radiative forcing 20C3M SRES A1B

Sfc wind

control climatology

Pacific wind-forced

climatology

climatology + obs linear trend(45ºS-45ºN Pac)

(1)Trend reverses at 2012

(2)Stabilizes at 2012 level

(3)Trend continues until 2020

2000/2001

2011/20121991/1992 time

• 12-member ensemble each

• Controls model IPO to follow observed trend while ocean heat budget is kept closed

(cf. Kosaka and Xie (2013) restored tropical eastern Pacific SST → Ocean heat budget was not closed)

Global SAT

• CMIP3 (20C3M + ?) + CMIP5 (historical + RCP4.5 or RCP8.5)

• 5-yr running mean, changes from 2000 is added to obs

• Adjusted CMIP result with (Pacific wind-forced – control) of O/CGCM

• CGCM (w/ atmos circulation) explains the current hiatus

Spatial structure of SAT trends for

1992-2011

• Negative IPO pattern• Pacific cooling is too strong, cooling expands to the entire tropics

CGCM control

CGCM Pacific wind trend-forced

Tropical Pacific wind stress vs global

SAT

• 20-yr tropical Pacific τ trend vs. 10-yr global SAT trend

• Positive correlation (r = 0.3) b/w tropical Pacific τ and global SAT trends in CMIP5

• Obs: some outside the CMIP5 cluster; global SAT less sensitive to τ

c.f. 20-yr tropical Pacific SST trend vs. 20-yr global SAT trend by Fyfe & Gillett (2014)

from Fyfe & Gillett (2014 Nature CC)1993-2012 trend

r = 0.63

CMIP5 captures the current hiatus?

• The recent τ trends is far outside the CMIP5 ensemble spread

Very low probability of the current hiatus in CMIP5 (Fyfe & Gillett 2014)

• Amplitude (or phase transition speed) of IPO is underestimated

or • The forced equatorial Pacific τ

trend (Walker circulation slowdown) is overestimated

Ob

s 1992

-2011 20-yr Pacific trade wind trends

(48 members)

Summary & Discussion• Prescribing Pacific sfc wind trends to OGCM/CGCM reproduces the current

hiatus in global SAT changes (50% by ocean processes, 50% by atmos. feedback)

• Intensification of Pacific trade winds due to negative IPO redistribute heat vertically in the ocean, through accelerating the subtropical cell circulationo Cooling at the surface → lowers global SATo Warming in the subsurface (below ~125m)

• The recent intensification of Pacific trade winds is “unprecedented” in 20CR (but considering uncertainty in reanalysis, this is not necessary; Tokinaga et al. 2013)

• The recent trade wind acceleration is outside the CMIP5 ensemble spread

CMIP5 predicts very low chance of the current hiatus in global SAT (Fyfe & Gillett 2014)

• What is inducing this acceleration?o CMIP models project deceleration associated with the global warming

o Rapid Indian Ocean warming (Luo et al. 2012 PNAS)

o Phase transition of AMO in the mid-1990s (Chikamoto et al. 2012 GRL)

o Pacific internal process (after 1997/98 El Niño) (Trenberth & Fasullo 2013 Earth’s Future)