How the ocean`s mean structure conspires to generate - eur
Transcription
How the ocean`s mean structure conspires to generate - eur
How the ocean's mean state generates submesoscale turbulence EUR-Oceans Hot Topics Conference Gran Canaria 6-8 November 2013 Shafer Smith CAOS/Courant/NYU Monday, November 25, 2013 Outline • Eddies, baroclinic instability and the oceanic energy cycle • The structure of the eddy field: geostrophic turbulence and SQG • How the eddies stir tracers Monday, November 25, 2013 Sea Surface Height anomaly ... eddies everywhere Chelton, Schlax & Samelson (2011) SSH on 28 August 1996 constructed from TOPEX/Poseidon (T/P) data only (top) and from the m Monday, November 25, 2013 What’s the source? Monday, November 25, 2013 Energy cycle Ferrari & Wunsch ’10 Blue: Geostrophic Green: Ageostrophic Transfers in TW = 106 W Resevoirs in EJ = 1018 J Monday, November 25, 2013 Energy cycle Wind work on interior ocean: 0.8TW Ferrari & Wunsch ’10 Blue: Geostrophic Green: Ageostrophic Transfers in TW = 106 W Resevoirs in EJ = 1018 J Monday, November 25, 2013 Energy cycle Wind work on interior ocean: 0.8TW Baroclinic instability APE to EKE (eddies): 0.9TW Ferrari & Wunsch ’10 Blue: Geostrophic Green: Ageostrophic Transfers in TW = 106 W Resevoirs in EJ = 1018 J Monday, November 25, 2013 The stratified ocean Monday, November 25, 2013 The stratified ocean Gill, Green and Simmons (1974): APE ~ (Basin scale/Deformation scale)2 ~ 1000 2 ~ (3000 km/100 km) ...enormous resevoir of energy available for conversion... Monday, November 25, 2013 Thermal wind balance with two density layers 94 Lectures on Geophysical Fluid Dynamics Figure 2.12 The mean westerly flow in a system healed near the equator (where the upper, warmer layer is thickest) and cooled near the pole. isopycnal displacements that contain a relatively large amount of available potential energy. In the Earth's atmosphere, the sun creates available potential energy by net heating in the tropics and net cooling in the polar regions. If the Earth Monday, November 25, 2013 Thermal wind balance with two density layers 94 Lectures on Geophysical Fluid Dynamics Figure 2.12 The mean westerly flow in a system healed near the equator (where the upper, warmer layer is thickest) and cooled near the pole. Sloping isopycnals store APE. When steep enough, baroclinic instability occurs, converting APE eddy KE large amount of available potenisopycnal displacements thatto contain a relatively tial energy. In the Earth's atmosphere, the sun creates available potential energy by net heating in the tropics and net cooling in the polar regions. If the Earth Monday, November 25, 2013 The classic eddy cycle Wind work on gyres Large scale Small scale Deformation scale of energy transfer in a two-layer fluid. The gure 6.6 Schematic diagram upper leve presents baroclinic energy (vertical wave number 1) and the lower level represent arotropic energy (vertical wave number 0). Horizontal wave number increases to th ght. Solid arrows indicate the direction of energy transfer; dashed arrows indicate poten al enstrophy transfer. Salmon (1980) Monday, November 25, 2013 The classic eddy cycle Wind work on gyres Large scale Small scale Deformation scale gure 6.6 Schematic diagram of energy transfer in a two-layer fluid. The upper leve presents baroclinic energy eddies (vertical wave 1) and the lower Classic picture: grownumber in horizontal and level represent arotropic energy (vertical wave number 0). Horizontal wave number increases to th vertical scale >>> big, smooth structures, ght. Solid arrows indicate the direction of energy transfer; dashed arrows indicate poten al enstrophy transfer. with ‘blurry’ SSH image consistent Salmon (1980) Monday, November 25, 2013 Idealized evolution of BCI Simulation by J. Taylor Ro = 0.125 Boussinesq Vertical vorticity (color), Isopycnal surface (white) 10km 10km 1km Deformation scale: ld = N H/2πf � 320km Monday, November 25, 2013 Idealized evolution of BCI Simulation by J. Taylor Ro = 0.125 Boussinesq Vertical vorticity (color), Isopycnal surface (white) 10km 10km 1km Deformation scale: ld = N H/2πf � 320km Monday, November 25, 2013 APE in the ocean JOURNAL OF PHYSICAL OCEANOGRAPHY 1/2 APE (Eady time scale) Eddies form mostly in high APE areas, consistent with BCI Tulloch et al. (2011) Monday, November 25, 2013 Eddy kinetic energy 60°N 0°N 60°N 120°E (cm/s)2 180°E 25 50 240°E 75 100 125 300°E 150 175 200 360°E 225 250 275 60°E 300 325 350 120°E 375 KE × sin2 φ Eddies form mostly in high APE areas, consistent with BCI Stammer & Wunsch (1998) Figure 4 Estimate of the geostrophic kinetic energy (KE) (cm s−1 )2 of oceanic variability at the sea surface, here multiplied by sin2 φ, where φ is the latitude, to avoid the equatorial singularity in noisy data. Note the very large spatial changes of kinetic energy. Figure taken from Wunsch & Stammer 1998. Monday, November 25, 2013 intense currents such as the Gulf Stream along the Eastern U.S. coast or the Agulhas Current What about submesoscales? (e.g. turbulence below the deformation scale) Monday, November 25, 2013 Sea Surface Height anomaly ... eddies everywhere Blow up this region Chelton, Schlax & Samelson (2011) SSH on 28 August 1996 constructed from TOPEX/Poseidon (T/P) data only (top) and from the m Monday, November 25, 2013 Sea Surface Temperature SST provides a ‘blow up’ of a region, exposing finer detail of eddies. Monday, November 25, 2013 Sea Surface Temperature SST provides a ‘blow up’ of a region, exposing finer detail of eddies. Energetic vortices and fronts range down to resolution of instrument Monday, November 25, 2013 (Sea Surface Roughness contrast .. .glitter analysis shows even finer scales) 10 km Thanks to B. Chapron Monday, November 25, 2013 Monday, November 25, 2013 Shcherbina et al (2013) eW su at pp er ed or re is ting gio d m in n i on ar fo n th M ked rm win e s od in ati te ra urv era Fi on r. (a di ey te gu fo ) en , R re r m Se Ad ts sh es 1 o a m at a ow olu a. ( del sur f in is re ing tio c) de ac od NA BI ER ET .: AL OS CA LE ES BM SU UL E TU RB NC E ST AT IS TI CS Simulation provides further blow up... Vorticity: shows continuous range of finer scale turbulence Velocity spectrum SHCHERBINA ET AL.: SUBMESOSCALE TUR (a) shallower than -3 slope: energetic submesoscales Monday, November 25, 2013 Shcherbina et al (2013) (c) AUGUST 2008 Submesoscales in BCI sim 1757 KLEIN ET AL. 1754 JOURNAL OF PHYSICAL OCEANOGRAPHY KE z=0 VOLUME 38 KE z=-150 KE from U,V KE from SSH KE z=-400 JOURNAL OF PHYSICAL OCEANOGRAPHY VOLUME 38 Surface vorticity Upper Ocean Turbulence from High-Resolution 1748 3D Simulations PATRICE KLEIN AND JOURNAL OF PHYSICAL OCEANOGRAPHY KE from U,V KE from BT BACH LIEN HUA Laboratoire de Physique des Océans, IFREMER/CNRS, Brest, France GUILLAUME LAPEYRE Laboratoire de Météorologie Dynamique/IPSL, Ecole Normale Supérieure/CNRS, Paris, France XAVIER CAPET Institute of Geophysics and Planetary Physics, University of California, Los Angleles, Los Angeles, California SYLVIE LE GENTIL Laboratoire de Physique des Océans, IFREMER/CNRS, Brest, France 1748 HIDEHARU SASAKI Earth Simulator Center, JAMSTEC, Yokohama, Japan (Manuscript received 29 January 2007, in final form 17 December 2007) Monday, November 25, 2013 KE z=-800 FIG. 5. Velocity spectra estimated from u and ! (red curve) at (a) z " #2 m, (b) #150 m, (c) #400 m, and (d) #800 m. The black curve in (a) and (b) is the surface velocity spectrum estimated from the SSH. The blue curve in (c) and (d) is the velocity spectrum estimated from the barotropic velocity. The green curve in each panel is the density spectrum. Surface density spectrum is scaled by a factor of g2/(N 20&20) " 7.1 using N0 /f0 " 37 to match the surface velocity spectrum (see text). Other density spectra are scaled to match the velocity spectra at k " 15. The horizontal axis displays the nondimensional wavenumber k. The value k " 10 corresponds to a wavelength of 300 km. Units on the vertical axis are m3 s#2. Upper Ocean Turbulence from High-Resolution 3D Simulations J O U R N A L O F PPHATRICE YSICA L OC E AB NACH O GL R A PH HUA Y KLEIN spectralIEN range. This confirms locity spectrum, suggesting that the energy there AND is whole the geostrophic equilibrium mentioned examining captured by this mode. Aprincipally UGUST 2008 K L before E Brest, I Nwhen EFrance T A L . the graLaboratoire de Physique des Océans, IFREMER/CNRS, To quantify the degree of ageostrophy of the flow in dient wind balance, that is, the velocity field near the the upper layers (where the Rossby number is large) surface almost equilibrates the pressure field. Then we V KLEIN ET AL.: VERTICAL Submesoscales in VELOCITIES BCI simFROM Note vertical velocities associated with fine-scale vorticity features: submesoscale fronts and filaments are sites of intense upwelling (Raf’s talk) Monday, November 25, 2013 How can BCI generate submesoscales at surface? Monday, November 25, 2013 Conditions for BCI ng-force to zonal PV perturbations (Pedlosky 1987; Walker and Pedlosky 2002). M nic instabilities generated by criteria non-zonal Charney Stern Pedlosky formean BCI - flows need:are very effective at generatin in the (Phillips • Q y changes urbulence (Spall 2000;sign Arbic andinterior Flierl 2004), duemodel) to a strong nonlinear feedback • Uz has the same sign at the upper and lower boundaries (Eady eneration problem) and eddy scale and anisotropy (Smith 2007a). That said, such instabiliti • near Uz atmeridional the lowercontinental boundary has the sameand signhere as Q the interior y in rimarily boundaries, we focus on flows that (original Charney problem) onal, such ACC, boundary current extensions and theofsubtropical • Uasatthe the upper boundary has the opposite sign Q in the counter cu z y interior (‘oceanic Charney problem’) e large-scale meridional QGPV gradient, generalized as in Bretherton (1966) to For ocean-relevant cases need sign function change ofsheets, singleisfunction (a la nd lower boundary conditions via delta Bretherton ’66): ∂s f 2 dU ∂ Q̃ =β−f + 2 δupper , where s=By/N2 ∂y ∂z N dz Monday, November 25, 2013 erturbations (Pedlosky 1987; Walker and Pedlosky 2002). Moreover, Mean PV structure from hydrography ated by non-zonal mean flows are very effective at generating baroZonally-averaged Qy/β from OCCA atlas Uz ; Arbic and Flierl 2004), due to a strong nonlinear feedback between > 1.5# a) (50m) 0 ale and anisotropy (Smith 2007a). !n=27That said, such instabilities# likely 1km Depth nal continental boundaries, n =27.5 we ! and here .5# focus on flows that are rela0 Qy C, 2km boundary current extensions and the subtropical counter currents. ".5# nal QGPV gradient, generalizedn as in Bretherton (1966) to include ! =28 3km "# nditions via delta function sheets, is −60 −40 −20 0 Latitude ∂s f 2 dU ∂ Q̃ = β − f + δ , upper 2 b) ∂y ∂z N dz 20 40 60 < "1.5# 2 where s(z) = By/N2 = -fUz/N(1) Tulloch, Marshall, Hill and Smith (JPO 2011) gradient, s = −by /N 2 the meridional isopycnal slope, f the Coriolis Monday, November 25, 2013 erturbations 3km n ! =28and Pedlosky 2002). (Pedlosky 1987; Walker "# Moreover, Mean PV structure from hydrography < "1.5# ated by −60 non-zonal flows are 0very effective at40generating −40mean −20 20 60 baroLatitude Schematic QGPV gradient Uz ; Arbic and Flierl 2004), due to a strong nonlinear feedback between (50m) b) ale and anisotropy (Smith 2007a). That said, such instabilities likely nal continental boundaries, and here we focus on flows that are rela- Qy/β C, boundary current extensions and the subtropical counter currents. nal QGPV gradient, generalized as in Bretherton (1966) to include nditions via delta function sheets, is ∂s f 2 dU ∂ Q̃ β − f cross-section + 2 δof upper ) Zonally= averaged the, meridional QGPV gradient (in (1) units of β), from ∂y ∂z N dz tlas. The upper surface gradient Uz is shown aboveMarshall, the dashed at a2011) depth Tulloch, Hillline andevaluated Smith (JPO 2 Monday, Novembers 25, = 2013 −by /N gradient, the meridional isopycnal slope, f the Coriolis erturbations (Pedlosky 1987; Walker and Pedlosky 2002). Moreover, Mean PV structure from hydrography ated by non-zonal mean flows are very effective at generating baroMoreover, baroclinic instabilities generated by nonzonal Hydrography... mean flows are very effective at generating baroclinic tur; Arbic and Flierl 2004), due to a strong nonlinear feedback between ulence (Spall 2000; Arbic and Flierl 2004) because of strong nonlinear feedback between eddy generation and ale and anisotropy (Smith 2007a). That said, such instabilities likely ddy scale and anisotropy (Smith 2007a). Here we focus on ows that are relatively zonal, such as the ACC, boundary nal continental boundaries, and here we focus on flows that are relaurrent extensions, and subtropical return flows. large-scale gradient, generalC,The boundary currentmeridional extensions QGPV and the subtropical counter currents. zed as in Bretherton (1966) to include upper and lower oundary via delta function sheets, is (1966) to include nal QGPVconditions gradient, generalized as in Bretherton 2 ~ › Q ›s f dU nditions via delta function sheets, 5b ! f 1 2 is dupper , ›y ›z N dz (1) igure 3: Schematic of typical2zonal velocity profiles in westerly, mixed and easterly sheared me ∂s f dU ∂ Q̃ 2 = β − f + δ , (1) where b is the planetary PV gradient, s 5 2by/N is the upper 2 ∂y ∂z UN(z) dz ows. The curve indicates the Coriolis horizontalparameter, dashed line indicates approximately the heig meridional isopycnal slope, fand is the N2 Tulloch, Marshall, Hill and Smith (JPO 2011) s the stratification, and dupper is a delta function at the 2 is zero assuming negligible planetary PV gradient. The surface she t which thesQGPV gradient, = −bgradient /N the meridional isopycnal slope, f the Coriolis y Monday, November 25, 2013 erturbations (Pedlosky 1987; Walker and Pedlosky 2002). Moreover, Mean PV structure from hydrography ated by non-zonal mean flows are very effective at generating baroMoreover, baroclinic instabilities generated by nonzonal Hydrography... mean flows are very effective at generating baroclinic tur; Arbic and Flierl 2004), due to a strong nonlinear feedback between ulence (Spall 2000; Arbic and Flierl 2004) because of strong nonlinear feedback between eddy generation and ale and anisotropy (Smith 2007a). That said, such instabilities likely ddy scale and anisotropy (Smith 2007a). Here we focus on ows that are relatively zonal, such as the ACC, boundary nal continental boundaries, and here we focus on flows that are relaurrent extensions, and subtropical return flows. large-scale gradient, generalC,The boundary currentmeridional extensions QGPV and the subtropical counter currents. zed as in Bretherton (1966) to include upper and lower oundary via delta function sheets, is (1966) to include nal QGPVconditions gradient, generalized as in Bretherton 2 ~ › Q ›s f dU These sheets, types efficiently generate surface b... nditions via delta function is (1) 5b ! f 1 2 dupper , ›y ›z N dz igure 3: Schematic of typical2zonal velocity profiles in westerly, mixed and easterly sheared me ∂s f dU ∂ Q̃ 2 = β − f + δ , (1) where b is the planetary PV gradient, s 5 2by/N is the upper 2 ∂y ∂z UN(z) dz ows. The curve indicates the Coriolis horizontalparameter, dashed line indicates approximately the heig meridional isopycnal slope, fand is the N2 Tulloch, Marshall, Hill and Smith (JPO 2011) s the stratification, and dupper is a delta function at the 2 is zero assuming negligible planetary PV gradient. The surface she t which thesQGPV gradient, = −bgradient /N the meridional isopycnal slope, f the Coriolis y Monday, November 25, 2013 erturbations (Pedlosky 1987; Walker and Pedlosky 2002). Moreover, Mean PV structure from hydrography ated by non-zonal mean flows are very effective at generating baroMoreover, baroclinic instabilities generated by nonzonal Hydrography... mean flows are very effective at generating baroclinic tur; Arbic and Flierl 2004), due to a strong nonlinear feedback between ulence (Spall 2000; Arbic and Flierl 2004) because of strong nonlinear feedback between eddy generation and ale and anisotropy (Smith 2007a). That said, such instabilities likely ddy scale and anisotropy (Smith 2007a). Here we focus on ows that are relatively zonal, such as the ACC, boundary nal continental boundaries, and here we focus on flows that are relaurrent extensions, and subtropical return flows. large-scale gradient, generalC,The boundary currentmeridional extensions QGPV and the subtropical counter currents. zed as in Bretherton (1966) to include upper and lower oundary via delta function sheets, is (1966) to include nal QGPVconditions gradient, generalized as in Bretherton 2 ~ › Q ›s f dU These sheets, types efficiently generate surface b... nditions via delta function is (1) 5b ! f 1 2 dupper , ›y ›z N dz Why and igure 3: Schematic of typical2zonal velocity profiles in westerly, mixed andhow? easterly sheared me ∂s f dU ∂ Q̃ 2 = β − f + δ , (1) where b is the planetary PV gradient, s 5 2by/N is the upper 2 ∂y ∂z UN(z) dz ows. The curve indicates the Coriolis horizontalparameter, dashed line indicates approximately the heig meridional isopycnal slope, fand is the N2 Tulloch, Marshall, Hill and Smith (JPO 2011) s the stratification, and dupper is a delta function at the 2 is zero assuming negligible planetary PV gradient. The surface she t which thesQGPV gradient, = −bgradient /N the meridional isopycnal slope, f the Coriolis y Monday, November 25, 2013 The spectra Regimes characterization Eddy PV from Phillips vs. Charney BCI Charney (w/ Bsurf gradient) Oceanic tracer Phillips stirred by a turbu (no Bsurf gradient) Roullet, McWilliams, Capet & Molemaker (JPO 2012) Monday, November 25, 2013 The spectra Regimes characterization Eddy PV from Phillips vs. Charney BCI Charney (w/ Bsurf gradient) Oceanic tracer Phillips stirred by a turbu (no Bsurf gradient) Note small eddies and filaments Roullet, McWilliams, Capet & Molemaker (JPO 2012) Monday, November 25, 2013 Simple model for Charney BCI Consider simple model for the oceanic version of Charney-type baroclinic instability that Shane showed earlier... N = N0 U(z) = Γ(z + z2/(2H) - H/3) so By = -f Uz(0) = -fΓ Uzz(z) = Γ/H and Qy = β - Γf2/HN2 = constant Have BCI when Qy and By have same sign. If Γ>0 possible to have By with no BCI... Monday, November 25, 2013 Γ =-1 H=1 Linear instability of simple Charney Growth rates Amplitudes Note: 1. Growth at all scales, as in original problem 2. Growth focused at upper surface -- instability excites buoyancy anomalies at surface => SQG behavior Monday, November 25, 2013 Baroclinic instability & SQG 1. Surface ∇B opp sign from interior ∇Q => Charney instability, with amplitude sharply peaked at surface Monday, November 25, 2013 Baroclinic instability & SQG 1. Surface ∇B opp sign from interior ∇Q => Charney instability, with amplitude sharply peaked at surface Monday, November 25, 2013 2. Eddy velocity u grows fastest at surface Baroclinic instability & SQG 1. Surface ∇B opp sign from interior ∇Q => Charney instability, with amplitude sharply peaked at surface 2. Eddy velocity u grows fastest at surface 3. Surface buoyancy conservation (because w=0 at z=0) is ∂tb + (u+U)⋅∇(b+B) = 0 so variance is generated: d/dt〈½b2〉=〈-bu〉⋅∇B > 0 for down-gradient flux. Monday, November 25, 2013 Baroclinic instability & SQG 1. Surface ∇B opp sign from interior ∇Q => Charney instability, with amplitude sharply peaked at surface 2. Eddy velocity u grows fastest at surface 3. Surface buoyancy conservation (because w=0 at z=0) is ∂tb + (u+U)⋅∇(b+B) = 0 so variance is generated: d/dt〈½b2〉=〈-bu〉⋅∇B > 0 for down-gradient flux. 4. Eddy buoyancy b dominates interior PV q, so inversion to get streamfunction is SQG-like Monday, November 25, 2013 Nonlinear Charney: Horizontal slices hC/H = -1 Thanks to Shane Keating Surface PV PV at z=-.3 Monday, November 25, 2013 Nonlinear Charney: PV(x,z) hC/H = -1 hC/H = -1/3 hC/H = -1/7 hC/H = -1/11 Monday, November 25, 2013 Submesoscale structures from interior stirring Monday, November 25, 2013 NATRE and SFTRE experiments North Atlantic Tracer Release Experiment Monday, November 25, 2013 NATRE data set NATRE North Atlantic Tracer Release Experiment data set • High Resolution Profiler provided 127 vertical profiles of T, S, and shear (0-4000m depth) • Moored current meters provided U, V velocities for one year (200 ! 3500m depth) Monday, November 25, 2013 T-S Profiles in NATRE T-S relationship is tight in thermocline At MS level, T-S relationship characterized by 0.2 psu thermohaline fluctuations S (psu) Monday, November 25, 2013 T-S Profiles in NATRE T-S relationship is tight in thermocline Vertical mixing by internal wave breaking or double diffusion? At MS level, T-S relationship characterized by 0.2 psu thermohaline fluctuations S (psu) Monday, November 25, 2013 T-S Profiles in NATRE T-S relationship is tight in thermocline At MS level, T-S relationship characterized by 0.2 psu thermohaline fluctuations S (psu) Monday, November 25, 2013 T-S Profiles in NATRE T-S relationship is tight in thermocline or stirring along isopycnals? At MS level, T-S relationship characterized by 0.2 psu thermohaline fluctuations S (psu) Monday, November 25, 2013 T-S Profiles in NATRE T-S relationship is tight in thermocline At MS level, T-S relationship characterized by 0.2 psu thermohaline fluctuations S (psu) Monday, November 25, 2013 T-S Profiles in NATRE T-S relationship is tight in thermocline At MS level, T-S relationship characterized by 0.2 psu thermohaline fluctuations S (psu) Thermohaline fluctations have little signature on density: compensated fronts of 1-5km in horizontal, 10-100m in vertical. => Stirring along isopycnals will effectively stir tracers inclined to isopycnals, not density (McVean and Woods1980) Monday, November 25, 2013 Lateral structure in simulation ‣ T, S passive tracers => filamentation compensated in effect on density ‣ Interior: little density gradient, ample tracer variance along isopycnals Monday, November 25, 2013 K 0 Spice 10 PSD ‣ Spectra: Tracer ~ K-1, density ~ K-5 K-1 −1 2 10 −2 Density 10 K-5 −5 K −4 10 Dens & Spice spectra −6 10 0 10 1 10 K (nondimensional) 2 10 Vertical structure in simulation ‣ 3D cascade => ample strain and shear at submesoscales ‣ Shear/Strain ~ N/f (independent of scale) ‣ Tracer (T & S) filaments are 3D, with 2490 ratios followingJ O URNAL OF PHYSICAL aspect shear/strain ~N/f z (N/f)2 (S aspect ratio)2 Monday, November 25, 2013 OCEANOGRAPHY FIG. 10. (left) Horizontal and (right) vertical slice central simulation. The dashed black line in each p panel. The white contours in the vertical slice are th and the white line is a reference line with slope f/N, is centered. The line is situated next to a filamen filaments are easily found in any snapshot slice. Slope f/N the right panel follows a 21 power law (the thin solid line has a 21 slope for comparison), also consistent with isotropy. Notice that the spectra become very noisy at the largest vertical wavenumbers. This hap- b Conclusions • Oceanic APE resevoir generates ample eddies through BCI • At surface, eddy stirring of buoyancy gradients readily generates submesoscale motions • At depth, eddy stirring leads to compensated fine-scale T/S fronts Monday, November 25, 2013 But .. don’t really need Charney BCI Consider two runs with N constant and U1(z) = U0 cos(πz) U2(z) = U0 cos(πz) + Γ(z2/2 + z - 1/3), Γ>0, hC>0 Qy(z) U2(z) U1(z) Surface component does not contribute to BCI (no Charneytype instability), yet have surface buoyancy gradient Monday, November 25, 2013 But .. don’t really need Charney BCI Linear BCI is nearly the same in both cases Growth rates Monday, November 25, 2013 Amplitude (max) But .. don’t really need Charney BCI Slices of PV U1(z) U2(z) Monday, November 25, 2013 But .. don’t really need Charney BCI Patches of surface buoyancy field U1(z) U2(z) Note secondary vortex formation Monday, November 25, 2013 But .. don’t really need Charney BCI Surface KE spectra U2(z) U1(z) -2 -3 Monday, November 25, 2013