Cold tongue upwelling / mixing Coupled equatorial feedbacks engage the global climate. The thermocline carries the essential ocean memory. Upwelling/mixing enables surface-thermocline communication. Remote consequences of these feedbacks are the main reason people outside the tropical Pacific care about what we do.

Upwelling controls property transports (BGC/nutrients/CO2). The vertical motion and exchanges produced by this circulation are now largely unconstrained by observations ... We must get this right. Walker Circulation El Niño precip teleconnections Up

Trade Easterlies Winds East New WarmWarm Water heated by the sun Guinea by the sun Pool UpwellingUpwelling South America

Cool lower water Thermocline

Cool lower water W.S.Kessler, NOAA/PMEL 1 Meridional circulation and processes z

y TIW x Green = Wind Brown = Fluxes Red = Solar Blue = Velocity = Mixing = Isotherms EUC

Upwelling is O(m/day) vertical mixing must be very strong Consider the modeling challenge ...

2 MARCH 2001 JOHNSON ET AL. 845 How well can we describe this circulation?

Observed v in the east-central Pacific: Two limitations MARCH 2001 JOHNSON ET AL. 845 Near-surface Ekman divergence drives upwelling, but ... Limitations: 1) Ship ADCP does not see upper 25m 2) TIW fluctuations

Need near-surface Red=Northward currents to Blue=Southward describe the structure of the (Johnson et al. 2001) divergence. (Scatterometer winds also need referencing over Mean meridional current in the east-central Pacific FIG.5.Vertical–meridionalsectionsbasedoncentered28 lat linear fitsstrong to data currents.) taken from 1708 to 958W, regardless of longitude (see 22 21 22 21 text). (a) MeridionalShipboard velocity, ADCPy (10 overms 170°W-95°W); CI 2, positive (northward) shaded. (b) Standard error of y,ey (10 ms ); CI 1, |y|. ey shaded. 10-year average (8400km) surfaceFIG.5.Vertical–meridionalsectionsbasedoncentered2 divergence and thermocline convergence,8 lat and linearsignal-to-noise fits to data taken from ratio 170 for8 yto. 95In8W, contrast regardless to the of longitudeu field, (see the y 22 21 y 22 21 y . text).little (a) cross-equatorial Meridional velocity, flow.(10 Meridionalms ); CI velocities 2, positive were (northward)y exceeded shaded. (b) their Standard standard error of errors,ey (10 (shadingms ); in CI Fig. 1, | | 5b)ey muchshaded. smaller than zonal velocities. Hence contour in- mainly near the surface3 and in the thermocline on either tervals for y were five times finer than for u. Poleward side of the equator. surfacevalues of divergencey were significant and thermocline to 40–50 convergence, m (278–258 andC). signal-to-noiseSince seawater ratio is nearlyfor y. In incompressible contrast to the andu field, vertical the littlePeak cross-equatorial poleward surface flow. speeds Meridional were 20.09 velocities6 0.02 were m yvelocity,exceededw, theiris negligible standard at errors the surface, (shading the in horizontal Fig. 5b) 21 21 muchs at smaller 4.28Sand0.13 than zonal6 0.03 velocities. m s Henceat 4.68 contourN. Surface in- mainlydivergence, nearu thex 1 surfacey y,canintheorybeintegrateddown- and in the thermocline on either tervalsdrifters for alsoy showwere five stronger times poleward finer than flow for inu. Poleward the north sideward of from the the equator. surface yielded estimates of the vertical valuesthan in the of y southwere from significant 1508 to 130 to8 40–50W(BaturinandNiiler m (278–258C). velocity,Since seawaterw. The horizontal is nearly incompressible divergence was and dominated vertical Peak1997), poleward despite the surface stronger speeds trades were in the2 Southern0.09 6 0.02 Hemi- m velocity,by the meridionalw, is negligible term, y aty,inmostplaces.However, the surface, the horizontal 21 21 ssphereat 4.2 (Hellerman8Sand0.13 and6 Rosenstein0.03 m s 1983).at 4.6 Equatorward8N. Surface divergence,the shoalingu ofx 1 they y EUC,canintheorybeintegrateddown- did result in a significant zonal drifterssubsurface also speeds show stronger were significant poleward flow in the in the Southern north wardterm, fromux,inlimitedregions. the surface yielded estimates of the vertical thanHemisphere in the south from from 60 150 m (248 to8C) 130 through8W(BaturinandNiiler 160 m (158C) velocity,The largestw. The feature horizontal in ux divergence(Fig. 6a) was was the dominated vertical

1997),and in despite the Northern the stronger Hemisphere trades in from the Southern 60 m (25 Hemi-8C) bydipole the centered meridional about term, they equatory,inmostplaces.However, with zonal divergence spherethrough (Hellerman 110 m (21 and8C). Rosenstein Interior equatorward 1983). Equatorward velocities theabove shoaling 105 m of and the convergence EUC did result from in 110 a significant to 270 m. zonal This 21 subsurfacepeaked at 0.05 speeds6 0.02 were m s significantat 1.48Sand in2 the0.04 Southern6 0.03 term,patternux was,inlimitedregions. due to the shoaling of the EUC. Equatorial 21 Hemispherems at 3.68 fromNat85m(near23 60 m (248C) through8C). 160 m (158C) zonalThe convergence largest feature below in u 270x (Fig. m was 6a) thewas result the vertical of the andMeridional in the Northern geostrophic Hemisphere velocities from estimated 60 m from (25 ver-8C) dipolegrowth centered of the westward-flowing about the equator with EIC zonal west divergence of 1408W throughtical integrals 110 m of (21 zonal8C). Interior density equatorward gradients (not velocities shown) above(Fig. 3a). 105 Zonal m and convergence convergence in from the NECC 110 to and 270 SEC m. This was peakedwere generally at 0.05 6 equatorward0.02 m s21 at over 1.48Sand the entire20.04 latitude6 0.03 patterndue to slight was due eastward to the weakening shoaling of of the the EUC. eastward-flow- Equatorial msrange.21 at They 3.68 wereNat85m(near23 also surface intensified8C). and stronger zonaling NECC convergence and slight below westward 270 m was strengthening the result of of the inMeridional the south. This geostrophic hemispheric velocities asymmetry estimated was from consis- ver- growthwestward of flowing the westward-flowing SEC, as expected EIC from west the of Sverdrup 1408W ticaltent with integrals the direct of zonal measurements density gradients of equatorward (not shown) trans- (Fig.balance. 3a). Zonal Zonal convergences convergence in centered the NECC near and 230 SEC m, 6 was38 wereports. Sincegenerally the equatorward equatorward geostrophic over the velocities entire latitude were duelatitude to slight were eastward tightly linked weakening to divergences of the eastward-flow- 60–80 m range.surface They intensified, were also the surface wind-driven intensified poleward and stronger Ekman ingshallower NECC and and about slight 28 westwardpoleward. strengtheningThis pattern was of due the incomponent the south. must This have hemispheric persisted asymmetry to at least was where consis- the westwardto the shoaling flowing and SEC, poleward as expected shift fromof the the NSCC Sverdrup and tentmaximum with the in direct equatorward measurements flow was of equatorward observed, roughly trans- balance.SSCC as Zonal they convergences move eastward centered (Johnson near 230 and m, Moore638 ports.85 m. Since This penetration the equatorward depth geostrophic was well below velocities the mixed were latitude1997). A were more tightly local estimate linked to of divergencesux from the slope 60–80 of m a surfacelayer depth intensified, (Figs. 2c the and wind-driven 2d). poleward Ekman shallowerthird-order and polynomial about 28 evaluatedpoleward. at This 1368 patternWsignificantly was due componentThe standard must errors have of persistedy (Fig. 5b)to at were least also where surface the toincreased the shoaling the magnitude and poleward of the shift dipole of the on the NSCC equator and maximumintensified in with equatorward off-equatorial flow maxima was observed, and dropped roughly be- SSCC(Fig. 7a, as dash–dotted they move line), eastward and other(Johnson features and off Moore the 21 21 85low m. 0.01 This m penetration s in places. depth The was maxima well below were 0.04 the mixed m s 1997).equator A (not more shown), local but estimate was not of appropriateux from the for slope a large- of a layerat 4.58 depthNand0.03ms (Figs. 2c and21 at 2d). 18S, and were likely related third-orderscale estimate polynomial of w. evaluated at 1368Wsignificantly toThe tropical standard instability errors wave of y activity(Fig. 5b) on were both also sides surface of the increasedSince uk they magnitudeand y y k u ofx, y they (Fig. dipole 6b) on was the contoured equator intensifiedequator, but with somewhat off-equatorial weaker maxima in the south and dropped (Chelton be- et (Fig.at five 7a, times dash–dotted coarser intervals line), and than otherux.Ontheequator features off the 21 21 lowal. 2000). 0.01 m While s in places. the mean The meridional maxima were currents 0.04 m were s equatorsurface meridional(not shown), divergence, but was noty appropriatey,wasstrong(Fig.7b) for a large- atnearly 4.58Nand0.03ms an order of magnitude21 at 18S, smaller and were than likely the related mean scaleat 60 ( estimate620) 3 of10w.28 s21.Meridionalconvergenceinthe 28 21 tozonal tropical currents, instability the y wavestandard activity errors on were both still sides half of the thermoclineSince uk reachedy and y 30y k (6u10)x, y y3(Fig.10 6b)s wason the contoured equator. equator,magnitude but of somewhat those for u. weakerThis combination in the south of (Chelton energetic et atThe five transition times coarser from meridional intervals than divergenceux.Ontheequator to conver- al.transient 2000). variability While the and mean a weak meridional mean resulted currents in a were low surfacegence occurred meridional at 50 divergence, m. Near-surfacey y,wasstrong(Fig.7b) meridional con- nearly an order of magnitude smaller than the mean at 60 (620) 3 1028 s21.Meridionalconvergenceinthe zonal currents, the y standard errors were still half the thermocline reached 30 (610) 3 1028 s21 on the equator. magnitude of those for u. This combination of energetic The transition from meridional divergence to conver- transient variability and a weak mean resulted in a low gence occurred at 50 m. Near-surface meridional con- 164 Y.-h. Tseng et al. / Ocean Modelling 104 (2016) 143–170

164 Y.-h. Tseng et al. / Ocean Modelling 104 (2016) 143–170 164 Y.-h. Tseng et al. / Ocean Modelling 104 (2016) 143–170

1 Fig. 18. Mean zonal velocity (color, in cm s − ) and potential density (black contours) along the equator from 1986 to 20 0 0. The white line denotes the core depth of the EUC (defined as the depth of the maximum eastward velocity). (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article).

Modern model zonal circulations (u)

The (observationally-constrained)1 EUC profiles are reasonable, but its shear ∂u/∂y is too weak. Fig. 18. Mean zonal velocity (color, in cm s − ) and potential density (black contours) along the equator from 1986 to 20 0 0. The white line denotes the core depth of the 1 EUC (defined asFig. the 18.depthMean of thezonal maximumvelocity eastward(color, in velocity).cm s− ) and(For potentialinterpretationdensity of the(black referencescontours) to alongcolourthe in thisequator figurefrom legend,1986 theto reader2000. Theis referredwhite lineto thedenotes web versionthe core depth of the

of this We article). have EUC (defined no constraintsas the depth of the maximum off theeastward equatorvelocity). (For (exceptinterpretation space-timeof the references to mean)colour in this ...figure modelslegend, the varyreader iswidely.referred to the web version of this article).

bservations OJohnson et al (2002)

Fig. 19. Mean zonal velocity in the central topical Pacific on a meridional section (140 °W) from 1986 to 20 0 0. Fig. 19. Mean zonal velocity in the central topical Pacific on a meridional section (140 °W) from 1986 to 20 0 0.

Fig. 19. Mean zonal velocity in(CORE-II hindcast simulaons. Tseng et al. 2016)the central topical Pacific on a meridional section (140°W) from 1986 to 2000. 4 Modern model CESM-POP with different horizontal resolutions (mean) meridional 1-degree 1/10-degree circulations u(eq,z) very similar: TAO equatorial current profiles “taught” models to get this right. But: v, w and T(y,z) quite different, especially near the surface: Colors = u Large implications for air-sea interaction. Vertical-meridional circulation unconstrained by observations. Short meridional scales (shear and properties) unsampled by TAO or Argo. Colors = w (Frank Bryan: CORE-II forcing, 30-year means) 5 Downward mixing on the equator

Diurnal turbulence extends downward into the upper EUC. Surface-thermocline transmission of heat and momentum. But ship-based experiments are necessarily sparse and Dissipation rate during 10 days of TIWE short-lived. (Lien and D’Asaro) They cannot describe the time evolution. 6 LETTER doi:10.1038/nature12363

Seasonal sea surface cooling in the equatorial Pacific cold tongue controlled by ocean mixing

James N. Moum1, Alexander Perlin1{, Jonathan D. Nash1 & Michael J. McPhaden2

Seasurface temperature(SST) is a critical control on the atmosphere1, Here we directly quantify variations in mixing on long time- and numerical models of atmosphere–ocean circulation emphasize scales using specialized instrumentation that we developed for use on its accurate prediction. Yet many models demonstrate large, system- oceanographic moorings18,19. Since late 2005, these sensors have pro- atic biases in simulated SST in the equatorial ‘cold tongues’ (expan- vided near-continuous measurements of mixing below the Tropical sive regions of net heat uptake from the atmosphere) of the Atlantic2 Atmosphere–Ocean (TAO) buoy maintained by the National Ocean and Pacific3 oceans, particularly with regard to a central but little- and Atmospheric Administration (NOAA) at 0, 140u W20. understood feature of tropical oceans: a strong seasonal cycle. The biases may be related to the inability of models to constrain turbulent 2 a New capabilitiesmixing realistically4, given that turbulent enable mixing, combined with La Niña El Niño

seasonal variations in atmospheric heating, determines SST. In tem- Niño 3.4 –2 perate oceans, the seasonal SST cycle is clearly related to varying solar b sampling5 mixing heating ; in the tropics, however, SSTs vary seasonally in the absence 25

of similar variations in solar inputs6. Turbulent mixing has long been SST (ºC) 22 )

–2 c a likely explanation, but firm, long-term observational evidence has –0.1 and beenshallow absent. Here we show thevelocity existence of a distinctive seasonal cycle (N m

w 0 of subsurface cooling via mixing in the equatorialPacific cold tongue, τ ) using multi-year measurements of turbulence in the ocean. In boreal –1 d s 0.3 2

spring, SST rises by 2 kelvin when heating of the upper ocean by the (m TIW KE 0

atmosphere exceeds cooling by mixing from below. In boreal sum- )

–2 e

mer, SST decreases because cooling from below exceeds heating from 20–60m 100 > q J above. When the effects of lateral advection are considered, the mag- (`W m 0 < 20 ↓ ↑↓ ↓ ↑↓ ↑↓ ↓↓↑ ↑ ↓↑ ↑↓ nitude of summer cooling via mixing (4 kelvin per month) is equi- f valent to that required to counter the heating terms. These results provide quantitative assessment of how mixing varies on timescales 30 longer than a few weeks, clearly showing its controlling influence on

seasonal cooling of SST in a critical oceanic regime. 40 Equatorial cold tongues in the Atlantic and Pacific are formed in part through diverging horizontal transport near the sea surface associated 50 with large-scale wind patterns, which bring cool waters towards the ocean LOST surface7,8.MaintainingcoolSSTsinthepresenceofintensesolarheating MOORING

requires a combination of subsurface mixing and vertical advection to Depth (m) 60 transport surface heat downward4,9–11.Limitedmeasurementsduringthe 2 log J (W m–2) passage of a tropical instability wave (TIW) at 0, 140u Wrevealedatenfold 70 10 q u (m s–1) 0 increase in subsurface turbulent heat flux12.Thiswassufficienttocool 0 1 2 3 –2 surface waters by 2 K per month, indicating that mixing alone has the potential to maintain the equatorial cold tongue, a conjecture supported 80 by climatological evidence that sea surface cooling is enhanced by TIWs12. Establishing general relationships that show how irreversible tur- 90 bulent mixing influences patterns in large-scale (10,000 km) ocean 2006 2007 2008 2009 2010 2011 phenomena has been notoriously difficult for two reasons. First, the Year delicate measurements required to quantify mixing have been limited to Figure 1 | Six-year record of mixing at the TAO mooring at 0, 1406 W. short shipboard campaigns (30 days) that target specific processes13–17. a, Nin˜o3.4 SST index, a measure of the relative strengths of El Nin˜o and La Nin˜a 28 Although these have led to fundamental advances in our knowledge events . b, SST. c, Wind stress, tw (7-day averages). d, Squared meridional of ocean dynamics at a few locations and on timescales of a few weeks, velocity at 40 m filtered at 12–33 days, taken as a proxy for kinetic energy in the such measurements are unable to determine whether changes in mix- tropical instability wave frequency band (TIW KE). e, Turbulent heat flux ing on seasonal and interannual timescales might influence large-scale averaged over depths 20–60 m, ,Jq.20-60m. f, Image plot of zonal currents (Kessler, Grissom, Cronin Enhanced realme TAO) (zonal velocity denoted u, eastward currents red, westward blue; see right inset dynamics. Second, diagnostics of mixing in numerical model solutions colour scale). Coloured bars show depths and durations of x-pod deployments; of ocean circulation are ambiguous, because models grossly oversimp- (OOMD-funded TPOS pilot) log10 Jq is indicated by the colour (10-day average, see left inset colour scale). lify their representation of mixing, which occurs on scales that are Arrows at top of f show deployments (red) and recoveries (black) of moorings. orders of magnitude smaller than the smallest spatial and shortest Grey bars show period of comparison experiment with shipboard turbulence 12,19 timescales resolved. profiling instrumentation . (Moum et al 2013) 1College of Earth, Ocean and Atmospheric Sciences, Oregon State University, Corvallis, Oregon 97331, USA. 2NOAA/Pacific Marine Environmental Laboratory, Seattle, Washington 98115, USA. {Deceased. 7

64 | NATURE | VOL 500 | 1 AUGUST 2013 ©2013 Macmillan Publishers Limited. All rights reserved Pacific Upwelling and Mixing Physics A Science and Implementation Plan The primary objectives of PUMP are: 1. To observe and understand the 3-D time evolution of the near-equatorial meridional circulation cell under varying winds, sufficiently well to serve (a) as background for the mixing observations in objective 2; (b) as a challenge to model representations. William S. Kessler James N. Moum 2. To observe and understand the mixing Meghan F. Cronin Paul S. Schopf mechanisms that determine (a) the depth of Daniel L. Rudnick LuAnne Thompson penetration of wind-input momentum and Revised January 2005 the factors that cause it to vary; (b) the transmission of surface heat fluxes into the upper thermocline and the maintenance of the thermal structure in the presence of meters per day upwelling.

3. To observe and understand the processes that allow and control exchange across the sharp SST front north of the cold tongue, including both small-scale frontal dynamics and the effects of tropical PUMP was too expensive in its day, instability waves. and our capabilities too weak, but new technology makes it tractable now. 8 Extra slides below The “classical” picture of equatorial Ekman divergence, but ... Easterly wind stress τ z y 2°N x

cold warm Easterlies spanning the equator drive Ekman transport that decreases with latitude. Downwelling is broad and symmetric. Snapshot (3 days in 1999)

But ... there is a sharp SST front near 2°N. How can the water cross the front? What mechanism would heat it up? How can equatorial Ekman divergence coexist with a sharp front?

Easterly wind stress : τ0 z “Stress” due to front : -τp y Ageostrophic tendencies : Ageostrophic flow Currents cold warm

(Cronin/Kessler 2009) At the center of the front, shear due to the front itself balances the Ekman flow (green arrow). → Convergence on the cold side, and divergence on the warm side. The cold diverging water slides under the front. Observed u at 140°W and 110°W

Johnson et al (shipboard ADCP mean) (White T contours) TAO ADCP at the Eq (~ monthly)

140°W 140°W

110°W 110°W

Neither the downward-looking shipboard nor the upward-looking moored ADCP samples the upper 25m Short space/time scales

Chelton et al. (2001)

Downwind scale of wind at front: (.1N/m2 over 250km => 1m/s over 50km = 3 hr) Surface boundary layer How does the thermocline communicate with the atmosphere?

The diurnal cycle Diurnal cycle composite at 2°N,140°W: Winds, current shear. is surprisingly 0 important ... 5 Can we teach this to models? 10

Depth (m) 15 20 Temperature 0 6 12 18 24 Local me

Cronin and Kessler (2009)

Much of the work of heat and momentum transmission Requirement to observe the to the thermocline is accomplished by the diurnal cycle. near-surface, including currents ... but how much is needed?

14 CESM w at 140°W, 50m

-10.2 -6.7 14.0 16.8