Regional coupled modeling of eddy-wind interaction in the California - PowerPoint PPT Presentation

Regional coupled modeling of eddy-wind interaction in the California Current System — Eddy kinetic energy and Ekman pumping Hyodae Seo Woods Hole Oceanographic Institution Art Miller & Joel Norris Scripps Institution of Oceanography PICES-2014 Annual Meeting Yeosu, Korea, October 21, 2014

Eddy-wind interaction: wind stress τ = ρ C D (U a − U o ) |U a − U o |

Eddy-wind interaction: wind stress τ = ρ C D (U a − U o ) |U a − U o | 10m wind Increased wind over warm SST U a = U ab + U aSST Wallace et al (1998) Correlation (SST and wind speed): high-passed Xie 2004

Eddy-wind interaction: wind stress τ = ρ C D (U a − U o ) |U a − U o | 10m wind Increased wind over warm SST U a = U ab + U aSST Wallace et al (1998) Uniform eastward wind over an anticyclonic eddy in the Southern Ocean (Chelton 2013) a b Surface temperature with contour interval = 0.5 cm da Dipole Ekman velocity SST and SSH and height surface temperature Correlation (SST and wind speed): high-passed 2 2 0.5 6 d D ⊖ 1 1 0.25 3 s τ 0 0 0 0 U ⊕ –0.25 –3 –1 –1 –0.5 –6 –2 –2 –2 –1 0 1 2 –2 –1 0 1 2 2 Ekman pumping anomaly 90° out of phase with Xie 2004 SSH → propagation of an eddy

Eddy-wind interaction: wind stress τ = ρ C D (U a − U o ) |U a − U o | surface current U o =U ob + U oe (U ob ≪ U oe ) W e = τ /[ ρ (f+ ζ )] a b Surface temperature SST and SSH with contour interval = 0.5 cm da Monopole Ekman velocity and height 2 2 τ 6 0.5 d 1 1 3 0.25 s U ⊕ 0 0 0 0 –3 –0.25 –1 –1 τ –6 –0.5 –2 –2 –2 –1 0 1 2 2 –2 –1 0 1 2 Upwelling at the center of an anti- surface current cyclonic eddy: damping of an eddy

Eddy-wind interaction: wind stress τ = ρ C D (U a − U o ) |U a − U o | surface current U o =U ob + U oe (U ob ≪ U oe ) W e = τ /[ ρ (f+ ζ )] a b Surface temperature SST and SSH with contour interval = 0.5 cm da with contour interval = 0.5 cm da Dipole Ekman velocity Monopole Ekman velocity surface temperature and height 2 2 2 τ 6 6 0.5 d D ⊖ 1 1 1 3 3 0.25 s U ⊕ τ 0 0 0 0 0 0 U ⊕ –3 –3 –0.25 –1 –1 –1 τ –6 –6 –0.5 –2 –2 –2 –2 –1 0 1 2 2 –2 –1 0 1 2 2 –2 –1 0 1 2 Upwelling at the center of an anti- Feedback to ocean surface current cyclonic eddy: damping of an eddy would be different!

Eddy-wind interaction: wind stress τ = ρ C D (U a − U o ) |U a − U o | 10m wind U a = U ab + U aSST surface current U o =U ob + U oe (U ob ≪ U oe ) resulting wind stress τ ≈ τ b + τ SST + τ oe

Eddy-wind interaction: wind stress τ = ρ C D (U a − U o ) |U a − U o | 10m wind U a = U ab + U aSST surface current U o =U ob + U oe (U ob ≪ U oe ) resulting wind stress τ ≈ τ b + τ SST + τ oe Effects of τ SST and τ cur on the ocean? EKE and Ekman pumping

Result from previous studies and the goal of this study • Previous studies considered either SST or U o in τ formulation in ocean-only models and saw weakened eddy variability. uncoupled SST - τ coupled SST SST uncoupled EKE U o - τ coupled EKE - τ coupling: Jin et al. (2009) U o - τ coupling: Eden and Dietze (2009) SST • This study examines the relative importance of SST and u sfc in a fully coupled regional model.

Regional coupled model Scripps Coupled Ocean-Atmosphere Regional Model • Seo et al. 2007, 2014 Ocean Atmosphere • An input-output based WRF or bulk physics τ (Q & FW) coupler; portable & flexible WRF ROMS 6-h coupling SST & U sfc • 7 km O-A resolutions & matching mask 6-h NCEP FNL monthly SODA • 6-yr integration (2005-2010) T tot T b T e Smoothing of mesoscale SST and U o (Putrasahan et al. 2013) 5° loess smoothing U e U tot U b (~3° boxcar smoothing)

Experiments τ = ρ C D (U a -U o )|U a -U o | T tot = T b + T e U tot = U b + U e 5° loess filtering ( ≈ 3° boxcar smoothing) Experiments τ formulation τ ation includes es CTL T b T e U b U e noT e T b T e U b U e noU e T b T e U b U e noT e U e T b T e U b U e noU tot T b T e U b U e

Summer surface eddy kinetic energy EKE time-series CTL noT e NoT e U e cm 2 s -2 — CTL = 171 25-30% — noT e = 174 — noU e = 231 EKE difference — noT e U e = 230 — noU tot = 247 noU e noT e U e noU tot JAS 2005-2010 • T e no impact • 25% weaker EKE with U e • 30% weaker EKE with U b +U e

Eddy kinetic energy budget advection by mean and eddy current (offshore) ! ! ! ! ! ! ! ! Ke t + U ⋅ Ke + # u ⋅ Ke + ∇⋅ ( # u # p ) = ∇ ∇ ! ! u ⋅ ! " ! ! ! − g " w + ρ o ( − " u ⋅ ( " u ⋅ U )) + " " ρ ∇ τ + ε P e → K e K m → K e wind work (P) if positive barotropic baroclinic (eddy drag if negative) conversion conversion (BT) (BC) Upper 100 m average H~fL/N, where f=10 -4 , L=10 4 m, N=10 -2 → H=10 2 m

EKE budget: CTL BT BC P Significant difference in only P v ′τ y ′ u ′τ x ′ u ′ τ y ′ v ′ • P a primary source of EKE. τ x ′ - Wind work from v ′τ y ′ - Eddy damping by u ′τ x ′ 150 m average

EKE budget: CTL BT BC P Significant difference in only P v ′τ y ′ u ′τ x ′ u ′ τ y ′ v ′ • P a primary source of EKE. τ x ′ - Wind work from v ′τ y ′ - Eddy damping by u ′τ x ′ 150 m average

Cross-shore distribution of EKE and P cross-shore distance (km) 50 [cm 2 s -2 ] 50 EKE 77 — CTL • P and BC — noT e maximum near the — noU e coast (20-30 km). • noU e ➞ CTL: P • P decreases by [10 -5 kgs -1 m -3 ] 1.26 20% 1.33 1.57 cross-shore distance (km)

Eddy drag and wind work u ′τ x ′ = eddy drag eddy drag 42% CTL= - 0.47 stronger [10 -5 kgs -1 m -3 ] noT e = - 0.53 eddy noU e= - 0.33 drag v ′τ y ′ = wind work wind work 16% CTL=1.74 weaker [10 -5 kgs -1 m -3 ] noT e =1.86 wind noU e= 1.90 work Ue: increases the eddy drag and weakens the wind work

Ekman pumping velocity Stern 1965; Gaube et al. (2014) ✓ ◆ 1 τ τ τ W tot = r ⇥ ( f + ζ ) ρ o  r ⇥ τ τ ⇥ r ˜ background wind stress = W cur + W SST W tot ✓ ◆ τ x r ⇥ ˜ τ τ τ 1 τ y ∂ζ τ x ∂ζ β ˜ + r ⇥ τ τ τ 0 SST = ˜ ∂ x � ˜ + . � (10) ρ o ( f + ζ ) 2 ρ o ( f + ζ ) 2 ρ o ( f + ζ ) ρ o ( f + ζ ) ∂ y | {z } | {z } | {z } | {z } ˜ W SST ˜ ˜ W lin W c W SST W ζ W β W ζ W β Curl-induced linear Ekman pumping Vorticity gradient-induced nonlinear Ekman pumping β Ekman pumping (negligible) SST induced Ekman pumping (Chelton et al. 2004) ≈ α c ∇ c SST W SST = ∇× # τ SST ( ) ( ) ρ o f + ζ ρ o f + ζ

Wind stress curl and cross-wind SST gradient ≈ α c ∇ c SST W SST = ∇× # τ SST ( ) ( ) ρ o f + ζ ρ o f + ζ OBS CTL Wind stress curl α c =0.6 [Nm -2 per 10 7 m] α c =0.8 noU e noT e α c =0.1 α c =0.6 Cross-wind SST gradient [°C per 100km] JAS 2005-2009; QuikSCAT wind stress and TRMM SST

Ekman pumping velocity JAS climatology OBS W sst W lin W tot W ζ CTL W sst W lin W tot W ζ JAS 2005-2009 m/day

Ekman pumping velocity JAS climatology noT e W sst W lin W tot W ζ noU e W sst W lin W tot W ζ JAS 2005-2009 m/day

Long-term effect of SST and vorticity on Ekman pumping velocity Wek: CTL-noTe Wek from CTL Wek vs crosswind SST gradient W ctl -W noTe Wek [mday -1 ] r=-0.06 crosswind SST gradient [°C per 100km] Wek: CTL-NoUe Wek vs surface vorticity • SST and vorticity induce the Wek response of comparable W ctl -W noUe magnitudes but of different Wek [mday -1 ] spatial pattern. r=-0.3 ‣ indicative of different feedback processes JAS 2005-2009 surface vorticity [ day -1 ]

Summary • Examined the relative importance of τ SST vs τ cur in EKE and Ekman pumping velocity in the CCS using a regional coupled model. • Surface EKE is weakened almost entirely due to mesoscale current. - SST has no impact. • EKE budget: enhanced eddy drag and reduced wind work. • W SST reflects the crosswind SST gradient, while W ζ surface vorticity - Associated patterns of change imply different feedback processes. - Further investigation on the mechanisms for feedback is underway.

Thanks!

Summertime climatology: coastal upwelling • CTL yields reasonable representation of the observed summertime upwelling condition in CCS. JAS 2005-2010

Change SST and surface current CTL-NoU e CTL CTL-NoT e Change in SST pattern reflects the change in surface current: advection by CTL-NoU e CTL CTL-NoT e mean and eddies. CTL-NoU e

Cross-shore vs depth EKE CTL-noT e CTL EKE CTL-noU e CTL-noT e U e 100-150m cm 2 s 2 CTL-noU tot 41N 34N alongshore averages

Change in JAS SST NOAA OI SST CTL CTL-NoT e CTL-NoT e U e CTL-NoU tot CTL-NoU e

Change JAS Surface current CTL CTL-NoT e Overlaid with contours for SST difference Surface currents show both alongshore and offshore component (Ekman current). Change in offshore (onshore) temperature advection by mean current mainly responsible for the change in SST CTL-NoT e U e CTL-NoU tot CTL-NoU e

wind speed (and also stress) is ENHANCED (REDUCED) over warm (cold) SST. It is a response to change in SST, damping the SST anomaly.