Decadal Variability, Impact, and Prediction of the Kuroshio Extension System B. Qiu 1, S. Chen 1, N. Schneider 1 and B. Taguchi 2 1. Dept of Oceanography, University of Hawaii, USA 2. Earth Simulator Center, JAMSTEC, Japan Climate Implications of Frontal Scale Air-Sea Interaction Workshop, Boulder, 5-7 August 2013

AVISO satellite altimeter data: 10/1992-present Observed decadal changes in the Kuroshio Extension system Forced vs. intrinsic KE variability and impact on SST front KE index and its prediction Outlines

Low-frequency circulation modulations Mesoscale eddy fluctuations

Altimeter-derived KE paths: Oscillations between stable and unstable states

Stable yrs: , , Unstable yrs: , Altimeter-derived KE paths: Oscillations between stable and unstable states

Yearly-averaged SSH field in the region surrounding the KE system

Various dynamic properties representing the decadal KE variability

Question 1: What causes the transitions between the stable and unstable dynamic states of the KE system?

Decadal KE variability lags the PDO index by 4 yrs ( r=0.67 ) Center of wind forcing is in eastern half of North Pacific basin + PDO generates negative local SSHAs through Ekman divergence, and vice versa

L H KE SSH field 145E165E155E135E L SSHA along 34°N PDO index center of PDO forcing H

L SSHA along 34°NPDO index/AL pressure center of PDO forcing SSH variability vs wind forcing Large-scale SSH changes are governing by linear vorticity dynamics: Given the wind forcing, SSH changes can be found along the Rossby wave paths: c R : observed value wind stress: NCEP reanalysis H H L

L Wind-forced SSHA along 34°NSSHA along 34°NPDO index/AL pressure center of PDO forcing H L H

KE strength of 142°-180°E (Taguchi et al. 2007) See also Nonaka et al. (2006), Pierini et al. (2009)

Question 2: What is the relative importance of the PDO-related wind forcing vs. the internal oceanic eddy forcing?

Quantifying wind- vs. eddy-forced KE variability in a 1.5-layer reduced-gravity North Pacific Ocean model Governing equations: where is velocity vector and is upper layer thickness. denotes the external wind forcing; given by interannually varying, or climatological, monthly NCEP data denotes the oceanic eddy forcing; inferred from the weekly AVISO anomalous SSH data North Pacific domain: 0°-50°N, 120°E-75°W Horizontal eddy viscosity: A h = 1,000 m 2 /s

Observed KE properties Modeled KE properties wind forcing only Modeled KE properties clim. wind + eddy forcing Wind forcing (only) case underestimates KE position/intensity changes and misses eddy signals Eddy + clim. wind forcing case fails to capture the decadal KE signals Wind forcing (only) case underestimates KE position/intensity changes and misses eddy signals Eddy + clim. wind forcing case fails to capture the decadal KE signals

Observed KE properties Modeled KE properties wind + eddy forcing Modeled KE properties clim. wind + eddy forcing Combined forcing by time-varying wind and eddy flux divergence generates the observed, decadal modulations of the KE system

KE dynamic state affects regional SST and cross-front SST gradient

JF rms 850mb v’T’ of stormtrack variability (Nakamura et al. 2004) JFM rms net surface heat flux variabilityRMS SSH variability (>2yrs) By carrying a significant amount of warm water from the tropics, KE is the region of intense air-sea interaction

850mb v’T’ and Q’ regressed to the KE dynamic state in When KE is stable, more heat release from O to A and a northward shift of stormtracks JF rms 850mb v’T’ of stormtrack variability (Nakamura et al. 2004) JFM rms net surface heat flux variability

Question 3: Can the decadally-modulating KE dynamic state be predicted in advance? Question 3: Can the decadally-modulating KE dynamic state be predicted in advance?

A composite index quantifying the KE dynamic state: the KE index KE index : average of the 4 dynamic properties (normalized)

Regression between the KE index and the AVISO SSH anomaly field KE index: represented well by SSH anomalies in 31-36°N, °E Predicting KE index becomes equivalent to predicting SSH anomalies in this key box Implications:

1. Prediction with Rossby wave dynamics h 1 (x, t) = h obs [ x+c R (t-t 0 ), t 0 ] where KE index and x-t SSHAs h obs (x, t 0 ) : initial SSHAs c R : Rossby wave speed The basis for long-term KE index prediction rests on 2 processes: 1. Oceanic adjustment in mid-latitude North Pacific is via slow, baroclinic Rossby waves The basis for long-term KE index prediction rests on 2 processes: 1. Oceanic adjustment in mid-latitude North Pacific is via slow, baroclinic Rossby waves

Examples of decadal KE index predictions and verifications

Mean square skill of the predicted KE index Compared to damped persistence, wave-carried SSHAs increase predictive skill with a 2~3 yr lead (Schneider and Miller 2001)

KEI>0 KE index-regressed curl field (tropical influence removed) stormtracks The basis for long-term KE index prediction rests on 2 processes: 2. The KE decadal variability affects the basin-scale wind stress curl field The basis for long-term KE index prediction rests on 2 processes: 2. The KE decadal variability affects the basin-scale wind stress curl field NCEP mean wind stress curl field

KE index-regressed curl field (tropical influence removed) Other studies of wind stress curl responses to KE variability

wind curl > 0 SSHA < 0 KEI>0 stormtracks wind curl > 0 SSHA < 0 SSHA < 0 RW propagation half of the oscillation cycle: ~5 yrs in the N Pacific basin KEI>0 The reason behind enhanced predictive skill with the 4~6 yr lead: delayed negative feedback mechanism

1. Prediction with Rossby wave dynamics h 1 (x, t) = h obs [ x+c R (t-t 0 ), t 0 ] 2. Prediction with Rossby wave dynamics + KE feedback to wind forcing h 2 (x, t) = h 1 (x, t) + ∫ b[ x+c R (t’-t 0 ) ] K(t’) dt’ t t0t0 where KE index and x-t SSHAs h obs (x, t 0 ) : initial SSHAs c R : Rossby wave speed where b(x) : feedback coefficient K(t) : forecast KE index

Examples of decadal KE index predictions and verifications

Mean square skill of the predicted KE index Compared to damped persistence, wave-carried SSHAs increase predictive skill with a 2~3 yr lead (Schneider and Miller 2001) Additional skill is gained with the 4~6 yr lead by considering the wind forcing due to KE feedback

Summary KE dynamic state (i.e. EKE level, path, and jet/RG strengths) is dominated by decadal variations. It impacts the broad-scale cross-front SST gradient. SSH anomaly signals in 31-36°N, °E provide a good proxy for the decadally-varying KE index. A positive KE index induces overlying-high and downstream- low pressure anomalies. This feedback favors a coupled mode with a ~10 yr timescale.  Oscillatory nature of this mode enhances predictability Compared to Rossby wave dynamics, inclusion of the KE- feedback wind forcing increases predictive skill with a lead time from 2~3 to 4~5 yrs. Due to the persistent negative PDO phase, a prolonged stable KE dynamic state (until 2017) is predicted.

KE dynamic state (i.e. EKE level, path, and jet/RG strengths) is dominated by decadal variations. It impacts the broad-scale cross-front SST gradient. SSH anomaly signals in 31-36°N, °E provide a good proxy for the decadally-varying KE index. A positive KE index induces overlying-high and downstream- low pressure anomalies. This feedback favors a coupled mode with a ~10 yr timescale.  Oscillatory nature of this mode enhances predictability Compared to Rossby wave dynamics, inclusion of the KE- feedback wind forcing increases predictive skill with a lead time from 2~3 to 4~5 yrs. Due to the persistent negative PDO phase, a prolonged stable KE dynamic state (until 2017) is predicted KE index forecast based on 2012 AVISO SSH data

Yearly SSH anomaly field in the North Pacific Ocean / : PDO index + -

Stable Dynamic State Unstable Dynamic State Reducing EKE level Strengthening RG and northerly KE jet Incoming positive SSHA Ekman conver- gence in the east – PDO + PDO Ekman divergence in the east Incoming negative SSHA Weakening RG and southerly KE jet Enhancing local EKE level Eddy-eddy interaction re- enhances the RG Schematic of the decadally-modulating Kuroshio Extension system N.B. Nonlinearity important; phase changes paced by external wind forcing

Pierini et al. (2009, JPO) Bimodal decadal KE variability can be purely nonlinearly driven