Deep jets on gas-giant planets

Three-dimensional numerical simulations of the atmospheric flow on giant planets using the primitive equations show that shallow thermal forcing confined to pressures near the cloud tops can produce deep zonal winds from the tropopause all the way down to the bottom of the atmosphere. These deep winds can attain speeds comparable to the zonal jet speeds within the shallow, forced layer; they are pumped by Coriolis acceleration acting on a deep meridional circulation driven by the shallow-layer eddies. In the forced layer, the flow reaches an approximate steady state where east-west eddy accelerations balance Coriolis accelerations acting on the meridional flow. Under Jupiter-like conditions, our simulations produce 25 to 30 zonal jets, similar to the number of jets observed on Jupiter and Saturn. The simulated jet widths correspond to the Rhines scale; this suggests that, despite the three-dimensional nature of the dynamics, the baroclinic eddies energize a quasi-two-dimensional inverse cascade modified by the β effect (where β is the gradient of the Coriolis parameter). In agreement with Jupiter, the jets can violate the barotropic and Charney-Stern stability criteria, achieving curvatures ^∂2u/∂y² of the zonal wind u with northward distance y up to 2β. The simulations exhibit a tendency toward neutral stability with respect to Arnol'd's second stability theorem in the upper troposphere, as has been suggested for Jupiter, although deviations from neutrality exist. When the temperature varies strongly with latitude near the equator, our simulations can also reproduce the stable equatorial superrotation with wind speeds greater than . Diagnostics show that barotropic eddies at low latitudes drive the equatorial superrotation. The simulations also broadly explain the distribution of jet-pumping eddies observed on Jupiter and Saturn. While idealized, these simulations therefore capture many aspects of the cloud-level flows on Jupiter and Saturn.     

Interaction between eddies and mean flow in Jupiter's atmosphere: Analysis of Cassini imaging data

Beebe et al. [Beebe, R.F., et al., 1980. Geophys. Res. Lett. 17, 1-4] and Ingersoll et al. [Ingersoll, A.P., et al., 1981. J. Geophys. Res. 86, 8733-8743] used images from Voyagers 1 and 2 to analyze the interaction between zonal winds and eddies in Jupiter's atmosphere. They reported a high positive correlation between Jupiter's eddy momentum flux, , and the variation of zonal velocity with latitude, . This correlation implied a surprisingly high rate of conversion of energy from eddies to zonal flow: , a value more than 10% of Jupiter's thermal flux emission. However, Sromovsky et al. [Sromovsky, L.A., et al., 1982. J. Atmos. Sci. 39, 1413-1432] argued that possible biases in the analysis could have caused an artificially high correlation. In addition, significant differences in the derived eddy flux between datasets put into question the robustness of any one result. We return to this long-standing puzzle using images of Jupiter from the Cassini flyby of December 2000. Our method is similar to previous analyses, but utilizes an automatic feature tracker instead of the human eye. The number of velocity vectors used in this analysis is over 200,000, compared to the 14,000 vectors used by Ingersoll et al. We also find a positive correlation between and and derive a global average power per unit mass, , ranging from . Utilizing Ingersoll et al.'s estimate of the mass per unit area involved in the transport, this would imply a rate of energy conversion of . We discuss the implications of this result and employ several tests to demonstrate its robustness.     

Interaction of moist convection with zonal jets on Jupiter and Saturn

Observations suggest that moist convection plays an important role in the large-scale dynamics of Jupiter's and Saturn's atmospheres. Here we use a reduced-gravity quasigeostrophic model, with a parameterization of moist convection that is based on observations, to study the interaction between moist convection and zonal jets on Jupiter and Saturn. Stable jets with approximately the same width and strength as observations are generated in the model. The observed zonal jets violate the barotropic stability criterion but the modeled jets do so only if the flow in the deep underlying layer is westward. The model results suggest that a length scale and a velocity scale associated with moist convection control the width and strength of the jets. The length scale and velocity scale offer a possible explanation of why the jets of Saturn are stronger and wider than those of Jupiter.     

Waves in Jupiter's atmosphere observed by the Cassini ISS and CIRS instruments

The Cassini Imaging Science Subsystem (ISS) and Composite Infrared Spectrometer (CIRS) reported a North Equatorial Belt (NEB) wave in Jupiter's atmosphere from optical images [Porco, C.C., and 23 colleagues, 2003. Science 299, 1541-1547] and thermal maps [Flasar, F.M., and 39 colleagues, 2004. Nature 427, 132-135], respectively. The connection between the two waves remained uncertain because the two observations were not simultaneous. Here we report on simultaneous ISS images and CIRS thermal maps that confirm that the NEB wave shown in the ISS ultraviolet (UV1) and strong methane band (MT3) images is correlated with the thermal wave in the CIRS temperature maps, with low temperatures in the CIRS maps (upwelling) corresponding to dark regions in the UV1 images (UV-absorbing particles) and bright regions in the MT3 images (high clouds and haze). The long period of the NEB wave suggests that it is a planetary (Rossby) wave. The combined observations from the ISS and CIRS are utilized to discuss the vertical and meridional propagation of the NEB wave, which offers a possible explanation for why the NEB wave is confined to specific latitudes and altitudes. Further, the ISS UV1 images reveal a circumpolar wave centered at 48.5° S (planetocentric) and probably located in the stratosphere, as suggested by the ISS and CIRS observations. The simultaneous comparison between the ISS and CIRS also implies that the large dark oval in the polar stratosphere of Jupiter discovered in the ISS UV1 images [Porco, C.C., and 23 colleagues, 2003. Science 299, 1541-1547] is the same feature as the warm regions at high northern latitudes in the CIRS 1-mbar temperature maps [Flasar, F.M., and 39 colleagues, 2004. Nature 427, 132-135]. This comparison supports a previous suggestion that the dark oval in the ISS UV1 images is linked to auroral precipitation and heating [Porco, C.C., and 23 colleagues, 2003. Science 299, 1541-1547].     

Gravitational signature of rotationally distorted Jupiter caused by deep zonal winds

Both deep zonal winds, if they exist, and the basic rotational distortion of Jupiter contribute to its zonal gravity coefficients J_n for n ? 2. In order to capture the gravitational signature of Jupiter that is caused solely by its deep zonal winds, one must take into account the full effect of rotational distortion by computing the coefficients J_n in non-spherical geometry. This represents a difficult and challenging problem because the widely-used spherical-harmonic-expansion method becomes no longer suitable. Based on the model of a polytropic Jupiter with index unity, we compute Jupiter’s gravity coefficients J₂, J₄, J₆, … , J₁₂ taking into account the full effect of rotational distortion of the gaseous planet using a finite element method. For the model of deep zonal winds on cylinders parallel to the rotation axis, we also compute the variation of the gravity coefficients ΔJ₂, ΔJ₄, ΔJ₆, … , ΔJ₁₂ caused solely by the effect of the winds in non-spherical geometry. It is found that the effect of the zonal winds on lower-order coefficients is weak, ∣ΔJ_n/J_n∣ < 1%, for n = 2, 4, 6, but it is substantial for the high-degree coefficients with n ? 8.     

Jupiter's 24° N highest speed jet: Vertical structure deduced from nonlinear simulations of a large-amplitude natural disturbance

The evolution of a large-amplitude disturbance at cloud level in Jupiter's 24° N jet stream in 1990 is used to constrain the vertical structure of a realistic atmospheric model down to the 6 bar pressure level. We use the EPIC model (Dowling et al., 1998, The explicit planetary isentropic-coordinate (EPIC) atmospheric model, Icarus 132, 221-238) to perform long-term, three-dimensional, nonlinear simulations with a series of systematic variations in vertical structure and find that the details of the 1990 disturbance combine with the characteristics of the 24° N jet, the fastest on Jupiter, to yield a tight constraint on the solution space. The most important free parameters are the vertical dependence of the zonal-wind profile, and the thermal structure, below the cloud tops (p>0.7 bar) at the jet's central latitude. The temporal evolution of the disturbed cloud patterns, which spans more than 2 years, can be reproduced if the jet peak reaches ∼180 ms⁻¹ at the cloud level and increases to ∼210 ms⁻¹ at 1 bar and up to ∼240 ms⁻¹ at 6 bar; the observations were not reproduced for other configurations investigated. This trend is consistent with that measured by the Galileo Probe at 7° N; the implication is that this jovian jet extends well below the solar radiation penetration level situated near the 2 bar level.     

Wind variations in Jupiter's equatorial atmosphere: A QQO counterpart?

Jupiter's equatorial atmosphere, much like the Earth's, is known to show quasi-periodic variations in temperature, particularly in the stratosphere, but variations in other jovian atmospheric tracers have not been studied for any correlations to these oscillations. Data taken at NASA's Infrared Telescope Facility (IRTF) from 1979 to 2000 were used to obtain temperatures at two levels in the atmosphere, corresponding to the upper troposphere (250 mbar) and to the stratosphere (20 mbar). We find that the data show periodic signals at latitudes corresponding to the troposphere zonal wind jets, with periods ranging from 4.4 (stratosphere, 95% confidence at 4° S planetographic latitude) to 7.7 years (troposphere, 97% confidence at 6° N). We also discuss evidence that at some latitudes the troposphere temperature variations are out of phase from the stratosphere variations, even where no periodicity is evident. Hubble Space Telescope images were used, in conjunction with Voyager and Cassini data, to track small changes in the troposphere zonal winds from 20° N to 20° S latitude over the 1994-2000 time period. Oscillations with a period of 4.5 years are found near 7°-8° S, with 80-85% significance. Further, the strongest evidence for a QQO-induced tropospheric wind change tied to stratospheric temperature change occurs near these latitudes, though tropospheric temperatures show little periodicity here. Comparison of thermal winds and measured zonal winds for three dates indicate that cloud features at other latitudes are likely tracked at pressures that can vary by up to a few hundred millibar, but the cloud altitude change required is too large to explain the wind changes measured at 7° S.     

The jovian anticyclone BA: I. Motions and interaction with the GRS from observations and non-linear simulations

A study of the dynamics of the second largest anticyclone in Jupiter, Oval BA, and its red colour change that occurred in late 2005 is presented in a three part study. The first part, this paper, deals with its long-term kinematical and dynamical behaviour monitored since its formation in 2000 to September 2008 using ground-based observations archived at the public International Outer Planet Watch (IOPW) database. The vortex changed its zonal drift velocity from 1.8 m s⁻¹ in the period 2000-2002 to 0.8 m s⁻¹ in 2002-2003, and to 2.5 m s⁻¹ since late 2003. It also migrated southwards by 1.0 ± 0.5° in latitude between 2000 and 2004, remaining afterwards at an almost fixed latitude position. During the period 2000-2007, the oval also changed its triangular-like shape to a more symmetrical one. No latitudinal change was found in the months before the development of a red annulus in its interior. The colour change took place in less than 5 months in 2005-2006 and no red colour feature was observed to have been present or entrained by BA months before the annulus development. After detailed examination of the four encounters between BA and GRS that took place during this 9 year period, we did not detect any noticeable change in its drift rate or in apparent structure associated with the encounters at cloud level. Also, the area of BA did not significantly change in this period. Additionally, we found that BA displays a long-term oscillation of ∼160 days in its longitude position with peak to peak amplitude of 1.2°. Numerical experiments using the global circulation model EPIC reproduce accurately the shape, connecting it to its latitude migration, and morphology of the oval and confirm that no strong interaction between BA and the GRS is possible at least in the current situation.     

The cloud structure of the jovian atmosphere as seen by the Cassini/CIRS experiment

We analyze the thermal infrared spectra of Jupiter obtained by the Cassini-CIRS instrument during the 2000 flyby to infer temperature and cloud density in the jovian stratosphere and upper troposphere. We use an inversion technique to derive zonal mean vertical profiles of cloud absorption coefficient and optical thickness from a narrow spectral window centered at 1392 cm⁻¹ (7.18 μm). At this wavenumber atmospheric absorption due to ammonia gas is very weak and uncertainties in the ammonia abundance do not impact the cloud retrieval results. For cloud-free conditions the atmospheric transmission is limited by the absorption of molecular hydrogen and methane. The gaseous optical depth of the atmosphere is of order unity at about 1200 mbar. This allows us to probe the structure of the atmosphere through a layer where ammonia cloud formation is expected. The results are presented as height vs latitude cross-sections of the zonal mean cloud optical depth and cloud absorption coefficient. The cloud optical depth and the cloud base pressure exhibit a significant variability with latitude. In regions with thin cloud cover (cloud optical depth less than 2), the cloud absorption coefficient peaks at 1.1±0.05 bar, whereas in regions with thick clouds the peak cloud absorption coefficient occurs in the vicinity of 900±50 mbar. If the cloud optical depth is too large the location of the cloud peak cannot be identified. Based on theoretical expectations for the ammonia condensation pressure we conclude that the detected clouds are probably a system of two different cloud layers: a top ammonia ice layer at about 900 mbar covering only limited latitudes and a second, deeper layer at 1100 mbar, possibly made of ammonium hydrosulfide.     

Gravity waves in Jupiter's equatorial clouds observed by the Galileo orbiter

A reanalysis of Galileo images of Jupiter's Equatorial Zone obtained in 1999 and 2001 reveals the presence of trains of cloud bands, probably generated by gravity waves, similar to those seen in Voyager 1 and 2 (1979) and New Horizons (2007) images. No such waves have been found in a similar analysis of the Cassini flyby (2000) dataset. We present the properties of these waves and discuss them in the context of previously proposed models and observations.     

Life cycles of spots on Jupiter from Cassini images

Using the sequence of 70-day continuum-band (751 nm) images from the Cassini Imaging Science System (ISS), we record over 500 compact oval spots and study their relation to the large-scale motions. The ∼100 spots whose vorticity could be measured—the large spots in most cases—were all anticyclonic. We exclude cyclonic features (chaotic regions) because they do not have a compact oval shape, but we do record their interactions with spots. We distinguish probable convective storms from other spots because they appear suddenly, grow rapidly, and are much brighter than their surroundings. The distribution of lifetimes for spots that appeared and disappeared during the 70-day period follows a decaying exponential with time constant (mean lifetime) of 3.5 days for probable convective storms and 16.8 days for all other spots. Extrapolating the exponential beyond 70 days seriously underestimates the number of spots that existed for the entire 70-day period. This and other evidences (size, shape, distribution in latitude) suggest that these long-lived spots with lifetime larger than 70 days are from a separate population. The zonal wind profile obtained manually by tracking individual features (this study) agrees with that obtained automatically by correlating brightness variations in narrow latitude bands (Porco et al., 2003). Some westward jets have developed more curvature and some have developed less curvature since Voyager times, but the number of westward jets that violate the barotropic stability criterion is about the same. In the northern hemisphere the number of spots is greatest at the latitudes of the westward jets, which are the most unstable regions according to the barotropic stability criterion. During the 70-day observation period the Great Red Spot (GRS) absorbed nine westward-moving spots that originated in the South Equatorial Belt (SEB), where most of the probable convective storms originate. Although the probable convective storms do not directly transform themselves into westward-moving spots, their common origin in the SEB suggests that moist convection and the westward jet compose a system that has maintained the GRS over its long lifetime.     

EPIC simulations of the merger of Jupiter's White Ovals BE and FA: altitude-dependent behavior

Observations of the merger of Jupiter's White Ovals BE and FA show altitude-dependent behavior that we seek to capture in numerical simulations. In particular, it was observed that the upper portions of the vortices orbited each other before merging, but the lower portions translated into each other without orbiting, a phenomenon we term the pair-orbit vertical dichotomy. To reproduce this dichotomy in the EPIC model, it is sufficient to have (i) a decrease with altitude of the background zonal winds above the cloud level with a scale height of ∼2.4, (ii) a height scale of the winds inside the vortices that is the same or larger, and (iii) a maximum tangential velocity in each vortex of ∼100 m s⁻¹ or larger. Condition (i) is expected from thermal-wind analyses, (ii) is consistent with thermal-wind and Shoemaker-Levy 9 debris-trajectory analyses, and (iii) is consistent with cloud-top wind tracking. The model generally does not reproduce the dichotomy for vortices with smaller vertical extent or weaker circulations. Our simulated mergers correctly reproduce the observed ∼9° separation at which vortices start to orbit in the upper layers before they merge and the ∼70% area ratio of the final vortex BA to the sum of BE and FA.     

Velocity and vorticity measurements of Jupiter's Great Red Spot using automated cloud feature tracking

We have produced mosaics of the Great Red Spot (GRS) using images taken by the Galileo spacecraft in May 2000, and have measured the winds of the GRS using an automated algorithm that does not require manual cloud tracking. Our technique yields a high-density, regular grid of wind velocity vectors that is advantageous over a limited number of scattered wind vectors that result from manual cloud tracking. The high-velocity collar of the GRS is clearly seen from our velocity vector map, and highest wind velocities are measured to be around 170 m s⁻¹. The high resolution of the mosaics has also enabled us to map turbulent eddies inside the chaotic central region of the GRS, similar to those mapped by Sada et al. [Sada, P.V., Beebe, R.F., Conrath, B.J., 1996. Icarus 119, 311-335]. Using the wind velocity measurements, we computed particle trajectories around the GRS as well as maps of relative and absolute vorticities. We have discovered a narrow ring of cyclonic vorticity that surrounds the main anti-cyclonic high-velocity collar. This narrow ring appears to correspond to a ring surrounding the GRS that is bright in 5 μm [Terrile, R.J., Beebe, R.F., 1979. Science 204, 948-951]. It appears that this cyclonic ring is not a transient feature of the GRS, as we have discovered it in a re-analysis of Galileo data taken in 1996 first analyzed by Vasavada et al. [Vasavada, A.R., and 13 colleagues, 1998. Icarus 135, 265-275]. We also calculate how absolute vorticity changes as a function of latitude along a trajectory around the GRS and compare these measurements to similar ones performed by Dowling and Ingersoll [Dowling, T.E., Ingersoll, A.P., 1988. J. Atmos. Sci. 45, 1380-1396] using Voyager data. We show no dramatic evolution in the structure of the GRS since the Voyager era except for additional evidence for a counter-rotating GRS core, an increase in velocity in the main velocity collar, and an overall decrease in the length of the GRS.     

Merging circulations on Jupiter: Observed differences between cyclonic and anticyclonic mergers

The dynamics of mergers of large circulations in Jupiter's atmosphere may permit different models of the atmosphere to be tested. We report well-resolved observations of such events at visible wavelengths: three anticyclonic and three cyclonic events. A merger of anticyclonic white ovals in the South South Temperate domain (2002 March) is compared with the previously reported merger of ovals BE and FA in the South Temperate domain (2000 March). In each case, the two similar-sized ovals converged rapidly once they were separated by less than the sum of their diameters; they orbited around each other anticyclonically during the merger; the merged oval initially had the same rapid drift as the western parent; and, in an unexpected similarity, a cyclonic oval emerged westward from the point of merger. Evidence suggests that a merger of smaller ovals in the North North Temperate domain (2002 February) had similar dynamics. In contrast, mergers of cyclonic ovals in the North Equatorial Belt (‘barges’: 2001 November, 2005 May) proceeded in a different manner. The two parent barges showed no consistent acceleration towards each other as they converged; on contact there was no obvious sign of mutual circulation, and the low-albedo regions had almost passed each other before they finally merged; and the resulting barge had a drift rate intermediate between the two parents, and a length that was greater than either parent. Again, a third such event involving a smaller barge (2002 December) showed many of the same characteristics. These observations define different dynamical behaviour during anticyclonic and cyclonic mergers.     

Jupiter's White Oval turns red

Jupiter's remaining White Oval changed color in late 2005 and became noticeably red in early 2006, as reported by amateur observers. We present wind and color analyses from high spatial resolution images taken with the Hubble Space Telescope Advanced Camera for Surveys in April 2006. These images suggest that the recent color change was tied to a strengthening of this storm, as implied by increased vorticity, causing it to become more like the Great Red Spot. From a historical perspective, the current activity may be consistent with the generation of new anticyclones at this latitude in the coming months and years.     

On the HCN and CO2 abundance and distribution in Jupiter's stratosphere

Observations of Jupiter by Cassini/CIRS, acquired during the December 2000 flyby, provide the latitudinal distribution of HCN and CO₂ in Jupiter's stratosphere with unprecedented spatial resolution and coverage. Following up on a preliminary study by Kunde et al. [Kunde, V.G., and 41 colleagues, 2004. Science 305, 1582-1587], the analysis of these observations leads to two unexpected results (i) the total HCN mass in Jupiter's stratosphere in 2000 was (6.0±1.5)×¹⁰¹³ g, i.e., at least three times larger than measured immediately after the Shoemaker-Levy 9 (SL9) impacts in July 1994 and (ii) the latitudinal distributions of HCN and CO₂ are strikingly different: while HCN exhibits a maximum at 45° S and a sharp decrease towards high Southern latitudes, the CO₂ column densities peak over the South Pole. The total CO₂ mass is (2.9±1.2)×¹⁰¹³ g. A possible cause for the HCN mass increase is its production from the photolysis of NH₃, although a problem remains because, while millimeter-wave observations clearly indicate that HCN is currently restricted to submillibar (∼0.3 mbar) levels, immediate post-impact infrared observations have suggested that most of the ammonia was present in the lower stratosphere near 20 mbar. HCN appears to be a good atmospheric tracer, with negligible chemical losses. Based on 1-dimensional (latitude) transport models, the HCN distribution is best interpreted as resulting from the combination of a sharp decrease (over an order of magnitude in Kyy) of wave-induced eddy mixing poleward of 40° and an equatorward transport with velocity. The CO₂ distribution was investigated by coupling the transport model with an elementary chemical model, in which CO₂ is produced from the conversion of water originating either from SL9 or from auroral input. The auroral source does not appear adequate to reproduce the CO₂ peak over the South Pole, as required fluxes are unrealistically high and the shape of the CO₂ bulge is not properly matched. In contrast, the CO₂ distribution can be fit by invoking poleward transport with a velocity and vigorous eddy mixing (). While the vertical distribution of CO₂ is not measured, the combined HCN and CO₂ results imply that the two species reside at different stratospheric levels. Comparing with the circulation regimes predicted by earlier radiative-dynamical models of Jupiter's stratosphere, and with inferences from the ethane and acetylene stratospheric latitudinal distribution, we suggest that CO₂ lies in the middle stratosphere near or below the 5-mbar level.     

Cassini CIRS retrievals of ammonia in Jupiter's upper troposphere

The zonal mean ammonia abundance on Jupiter between the 400- and 500-mbar pressure levels is inferred as a function of latitude from Cassini Composite Infrared Spectrometer data. Near the Great Red Spot, the ammonia abundance is mapped as a function of latitude and longitude. The Equatorial Zone is rich in ammonia, with a relative humidity near unity. The North and South Equatorial Belts are depleted relative to the Equatorial Zone by an order of magnitude. The Great Red Spot shows a local maximum in the ammonia abundance. Ammonia abundance is highly correlated with temperature perturbations at the same altitude. Under the assumption that anomalies in ammonia and temperature are both perturbed from equilibrium by vertical motion, we find that the adjustment time constant for ammonia equilibration is about one third of the radiative time constant.     

Brightness power spectral distribution and waves in Jupiter's upper cloud and hazes

In this work we analyze the spatial structure of Jupiter's cloud reflectivity field in order to determine brightness periodicities and power spectra characteristics together with their relationship with Jupiter's dynamics and turbulence. The research is based on images obtained in the near-infrared (∼950 nm), blue (∼430 nm) and near-ultraviolet (∼260 nm) wavelengths with the Hubble Space Telescope in 1995 and the Cassini spacecraft Imaging Science Subsystem in 2000. Zonal reflectivity scans were analyzed by means of spatial periodograms and power spectra. The periodograms have been used to search for waves as a function of latitude. We present the values of the dominant wavenumbers for latitude bands between 32° N and 42° S. The brightness power spectra analysis has been performed in the meridional and zonal directions. The meridional analysis of albedo profiles are close to a k⁻⁵ law similarly to the wind profiles at blue and infrared wavelengths, although results differ from that in the ultraviolet. The zonal albedo analysis results in two distributions characterized by different slopes. In the near infrared and blue wavelengths, average spectral slopes are n₁=−1.3±0.4 for shorter wavenumbers (k<80), and n₂=−2.5±0.7 for greater wavenumbers, whereas for the ultraviolet n₁=−1.9±0.4 and n₂=−0.7±0.4, possibly showing a different dynamical regime. We find a turning point in the spectra between both regimes at wavenumber k∼80 (corresponding to L∼1000 km) for all wavelengths.     

Constraints on deep-seated zonal winds inside Jupiter and Saturn

The atmospheres of Jupiter and Saturn exhibit strong and stable zonal winds. How deep the winds penetrate unabated into each planet is unknown. Our investigation favors shallow winds. It consists of two parts. The first part makes use of an Ohmic constraint; Ohmic dissipation associated with the planet's magnetic field cannot exceed the planet's net luminosity. Application to Jupiter (J) and Saturn (S) shows that the observed zonal winds cannot penetrate below a depth at which the electrical conductivity is about six orders of magnitude smaller than its value at the molecular-metallic transition. Measured values of the electrical conductivity of molecular hydrogen yield radii of maximum penetration of 0.96RJ and 0.86RS, with uncertainties of a few percent of R. At these radii, the magnetic Reynolds number based on the zonal wind velocity and the scale height of the magnetic diffusivity is of order unity. These limits are insensitive to difficulties in modeling turbulent convection. They permit complete penetration along cylinders of the equatorial jets observed in the atmospheres of Jupiter and Saturn. The second part investigates how deep the observed zonal winds actually do penetrate. As it applies heuristic models of turbulent convection, its conclusions must be regarded as tentative. Truncation of the winds in the planet's convective envelope would involve breaking the Taylor-Proudman constraint on cylindrical flow. This would require a suitable nonpotential acceleration which none of the obvious candidates appears able to provide. Accelerations arising from entropy gradients, magnetic stresses, and Reynolds stresses appear to be much too weak. These considerations suggest that strong zonal winds are confined to shallow, stably stratified layers, with equatorial jets being the possible exception.