1. 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

2. 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

3. Low-frequency circulation modulations Mesoscale eddy fluctuations

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

5. Stable yrs: 1993-94, 2002-04, 2010-2013Unstable yrs: 1996-2001, 2006-08 Altimeter-derived KE paths: Oscillations between stable and unstable states

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

7. Various dynamic properties representing the decadal KE variability

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

9. 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

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

11. 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

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

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

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

15. 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

16. 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

17. 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

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

19. 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

20. 850mb v’T’ and Q’ regressed to the KE dynamic state in 1977-2012 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

21. 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?

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

23. Regression between the KE index and the AVISO SSH anomaly field KE index: represented well by SSH anomalies in 31-36°N, 140-165°E Predicting KE index becomes equivalent to predicting SSH anomalies in this key box Implications:

24. 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

25. Examples of decadal KE index predictions and verifications

26. 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)

27. 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

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

29. 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

30. 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

31. Examples of decadal KE index predictions and verifications

32. 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

33. 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, 140-165°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.

34. 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, 140-165°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. 2013-2023 KE index forecast based on 2012 AVISO SSH data

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

36. 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

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