Peering through Jupiter’s clouds with radio spectral imaging

See allHide authors and affiliations

Science  03 Jun 2016:
Vol. 352, Issue 6290, pp. 1198-1201
DOI: 10.1126/science.aaf2210

A radio view into Jupiter's atmosphere

Jupiter's atmosphere is a complex system of belts, layers, storms, and cloud systems. de Pater et al. used Earth-bound radio observations to peer beneath its surface. Previous radio studies have been limited to average properties at each latitude, but the new observations allow a full two-dimensional view. This can be related to features (such as storms) seen in visible or infrared images. The results aid our understanding of gas giant atmospheres and will provide important context for the Juno spacecraft that arrives at Jupiter in July 2016.

Science, this issue p. 1198


Radio wavelengths can probe altitudes in Jupiter’s atmosphere below its visible cloud layers. We used the Very Large Array to map this unexplored region down to ~8 bar, ~100 kilometers below the visible clouds. Our maps reveal a dynamically active planet at pressures less than 2 to 3 bar. A radio-hot belt exists, consisting of relatively transparent regions (a low ammonia concentration, NH3 being the dominant source of opacity) probing depths to over ~8 bar; these regions probably coincide with 5-micrometer hot spots. Just to the south we distinguish an equatorial wave, bringing up ammonia gas from Jupiter’s deep atmosphere. This wave has been theorized to produce the 5-micrometer hot spots; we observed the predicted radio counterpart of such hot spots.

Despite the fact that Jupiter has been observed for decades from the ground and in situ by spacecraft, questions remain about its bulk composition and global atmospheric dynamics. Much information on Jupiter's deep atmospheric composition was obtained, however, by the probe on the Galileo spacecraft, even though it descended in a 5-μm hot spot; i.e., a region that is “dry”: devoid of clouds and condensable gases. The deep (>8 bar) mole fractions of NH3 and H2S were found to be enriched by a factor of 3 to 5 over the solar composition (1). This was several times higher than NH3 mole fractions derived from radio spectra globally at higher altitudes [pressure (P) < 4 bar] and locally over the North Equatorial Belt (NEB) and Equatorial Zone (EZ) (2). With NH3 being the dominant source of opacity at radio wavelengths, this difference could not be reconciled. A longitude-resolved global radio map of Jupiter (3), constructed from data at a single frequency of 15 GHz, showed regions similar to 5-μm hot spots with depleted NH3 down to P > 5 bar but no regions with NH3 mole fractions equal to the Galileo-derived deep abundance. A two-layer model of Jupiter's cloud-layer circulation was developed (4), explaining that NH3 could be globally depleted (at 0.5 to 4 bar) if upward transport occurred primarily in precipitating thunderstorms. Our observations show that a two-layer circulation pattern is not needed to explain the NH3 concentrations, at least in the equatorial region.

We used the upgraded Karl G. Jansky Very Large Array (VLA) (Table 1) to observe Jupiter over the full frequency range from 4 to 18 GHz (1.7 to 7 cm) (5). Our new maps have a sensitivity up to an order of magnitude higher than the previously obtained longitude-resolved 15-GHz map, and achieve a 2 to 5 times higher spatial resolution (~1300 km at disk center in the new 15-GHz map versus ~5000 km in the previous map). These maps probe depths in Jupiter’s atmosphere between ~0.5 and ~10 bar, a vertical range of over 100 km (2, 6), covering the altitude range where Jupiter's main cloud layers are formed (Fig. 1) (7).

Table 1 VLA data and products.

Column 1: Observations were conducted: A configuration: 2014/05/04; B configuration: 2013/12/23 and 2014/01/09; C configuration: 2014/12/27; D configuration: 2014/08/16. Column 2: VLA frequency band. Column 3: Center frequency of each 1-GHz-wide map. Column 4: Jupiter's effective equatorial radius (Reff) for the data from the combined arrays. The C-band longitude-resolved map was constructed from only A-configuration data, with Reff = 17.487′′ Column 5: Disk-averaged Tb (final value) (5). Columns 6 and 7: Resolution or beam size (HPBW, full width at half power) in the final longitude-smeared images. Column 8: HPBW near the center of the disk in the final longitude-resolved images. 1° in latitude or longitude at Jupiter's disk center corresponds to ~1200 km.

View this table:
Fig. 1 Contribution functions.

(A and B) Normalized contribution functions are shown for two different NH3 profiles, as marked. The dotted line is the temperature-pressure curve in Jupiter’s atmosphere. Cloud layers that are expected to form under chemical equilibrium are indicated as shaded regions in (A) (7).

Figure 2, A to D, shows radio maps averaged over 12 to 18 GHz, 8 to 12 GHz, and 4 to 8 GHz, after subtraction of a best-fit uniform limb-darkened disk (5). Panels E to H show Jupiter’s almost-simultaneous appearance at visible wavelengths for comparison (8). A wealth of structure is visible in each radio map, which is easily distinguished from some “ripples” and large-scale dark and light areas that are instrumental artefacts. In panel D, some of these large-scale patterns are caused by Jupiter’s synchrotron radiation (2, 5). Brightness temperatures (Tbs) in excess of the best-fit subtracted disk show up as bright (light-colored) regions, and lower temperatures are dark. A higher Tb indicates probing deeper levels in the atmosphere and hence suggests a low atmospheric opacity (i.e., low NH3 concentration), and dark regions correspond to a high opacity. Our observations thus provide the three-dimensional distribution of NH3 gas, which we use to trace dynamics, assuming the following: In parcels of rising air, NH3 is enriched above solar levels, or within and above the NH3 ice layer, NH3 is saturated. Descending flows that originate at or above the NH3 cloud base carry depleted ammonia with respect to the deep abundance.

Fig. 2 Longitude-resolved VLA maps of Jupiter.

(A to D) Longitude-resolved radio maps of Jupiter, after subtraction of a uniform limb-darkened disk (5). Each map has been integrated over the full ~4- to 6-GHz bandwidth. (E to H) Visible-light maps taken close in time with the VLA maps, as compiled by the amateur community (8). The axes indicate west longitude (in System III, prime meridian rotation angle of the magnetic field) and planetographic latitude. In addition to the GRS and Oval BA, several features discussed in the text are highlighted by arrows on both a 12- to 18-GHz and visible wavelength map. The red arrows in Fig. 2C indicate the plumes used to derive their phase velocity through comparison with the 8- to 12-GHz map from 9 January 2014 (5). Tb min and Tb max give the differences in temperature for the minimum and maximum observed deviations from zero (i.e., with respect to the uniform limb-darkened disk that was subtracted).

Historically, the zone-belt structures at visible, 5-μm, and radio wavelengths were thought to be well correlated. Brownish belts were thought to correspond to bands that are bright at 5 μm and in the radio, and whiter zones to bands that were dark at 5 μm and in the radio (9). A detailed comparison of the images in Fig. 2, however, reveals important refinements to this view. A radio-hot belt is visible right at the interface between the NEB and EZ, on the north side of the 7.8°N eastward jet. The northern and southern edges of the South Equatorial Belt (SEB) are radio-bright, whereas the darkest reddish interior is radio-dark. Several radio-bright bands are seen between 40° and 60°S, which do not necessarily seem to be connected to optical counterparts. The radio-hot belt at ~8°N contains structure at the limits of our resolution, as well as bright regions elongated in longitude. Elongated radio-bright regions coincide with the dark-gray-bluish regions in Fig. 2, E to H, at the same latitude (yellow arrows in Fig. 2, B and F) and are usually interpreted as “holes” in the cloud deck. These radio-bright, dark-gray-bluish regions are most likely related 5-μm hot spots (3, 10, 11).

All of these radio-bright bands are also visible at 4 to 8 GHz, although those at 40° to 60°S just barely. The radio-hot belt at ~8°N shows less detailed structure overall than at 8 to 12 GHz, despite the higher spatial resolution in this band. To determine how deep the low NH3 concentration in this radio-hot belt extends, we made longitude-smeared and longitude-resolved maps integrated over 1-GHz-wide frequency bands (Table 1), and we modeled these maps with radiative transfer (RT) calculations (5, 6, 12). In our calculations, we assume that the NH3 concentration (mole fraction) is equal to 5.7 × 10−4 at P > 8 bar, the average value detected by the Galileo probe at these depths (1), which is enhanced by a factor of ~4.5 over the solar N/H ratio (13). We assume that all constituents (C, S, N, and O) are enhanced by this same factor. Ammonia vertical profiles in Fig. 3A include an equilibrium case (profile a), where ammonia gas is decreased at higher altitudes due to solution in the water cloud (~7.3 bar), the formation of the NH4SH cloud (~2.5 bar), and the NH3 ice cloud (~0.8 bar). NH3 follows the saturated vapor pressure curve within and above the NH3 ice cloud. Several other NH3 profiles used in our calculations (Fig. 3, B to F) are indicated. For simplicity, we adopt step functions at different altitudes, as shown. For some profiles we adopted a relative humidity (RH) of ~1 to 10% above the NH3 ice cloud. All model atmospheres have a temperature of 165 K at 1 bar to match the Voyager radio occultation profile (14). By changing the NH3 profile, the contribution functions in an atmosphere may change substantially (Fig. 1), which helps to understand differences in the model spectra.

Fig. 3 Frequency dependence of specific features.

(A) Altitude profiles of the NH3 concentration used in the various models. The black-dotted line shows the temperature-pressure profile. Profile a follows the saturated vapor pressure curve for NH3 within and above the NH3 ice cloud (RH = 100%), whereas for the horizontal lines at a pressure of 0.6 bar, RH = 1%. (B) Longitude-averaged brightness temperature as a function of frequency for the NEB and EZ, with RT models superimposed; Tb maxima (NEB) and minima (EZ) from fig. S3 are plotted (5). Error bars are the standard deviation along the constant-latitude line. (C to F) Tb versus frequency for the GRS, Oval BA, hot spots, and plumes. Models are superimposed. The Tb for the GRS, Oval BA, and four plumes are averaged over the central regions; the values for the four plumes were averaged. The error bars give the standard deviation. The Tb for hot spots are local maxima; the error bars indicate the spread in the measured Tb.

Figure 3B shows spectra of the NEB, at the location of the longitude-smeared radio-hot belt (peak NEB values from fig. S3), and the EZ (minimum values) (5), with several models based on the profiles in Fig. 3A superimposed. Clearly, the radio-hot belt is well matched by profile e: NH3 is depleted with respect to the deeper atmosphere by a factor of 5 to 6 down to P ~ 8 bar, or likely deeper as shown by profile e′. In contrast, the EZ is consistent with profile a, contradicting prior analyses that the EZ is depleted in NH3 above the ~4-bar level (2). The advance in our understanding is due to the upgrade in VLA sensitivity (15), higher spatial resolution in longitude-averaged images than obtained previously (2, 6), and full coverage over the entire 4- to 18-GHz frequency range.

The radio-hot belt contains numerous hot spots that contribute to the average NEB value. Spectra of these (at ~10°N; Fig. 3C) were extracted from 1-GHz-wide longitude-resolved maps, where maxima in Tb were recorded. We determined values for four different hot spots on each day and averaged these. The true Tb may be slightly higher, because our maps do not resolve structure within the hot spots. A comparison with models shows that hot spots are characterized by a low NH3 concentration, less than ~10−4 over the 1- to 8-bar pressure level (profile e). A reasonable match to the data is given by profile f, with a concentration ~10−5 at the highest altitudes (P < 1 bar), gradually increasing at deeper levels.

Just south of the radio-hot belt, striking oval-shaped dark regions are visible at 4 to 8 GHz (Fig. 2D); these are also quite prominent at 8 to 12 GHz (Fig. 2C) and only barely visible at higher frequencies (Fig. 2, A and B). Spectra of these regions (~4°N), referred to as plumes, are shown in Fig. 3D. A comparison with models shows that the plumes bring NH3 gas from deep (P > 8 bar) atmospheric levels where the ammonia concentration is 5.7 × 10−4, up to high altitudes (P ~ 1.5 bar), contrary to previous findings that there are no locations where the full deep NH3 concentration is brought up to the 2-bar level (4). A two-layer circulation pattern with thunderstorm-dominated upwelling (4) is therefore inconsistent with NH3 concentrations in equatorial regions.

These plumes (Fig. 2, C and D) display a regular wave pattern with wavelength 30° and are most pronounced over a 180° range in longitude. A comparison of the location of five plumes on the 8- to 12-GHz map on 23 December 2013 (indicated by red arrows in Fig. 2C) with that on 9 January 2014 (5) reveals an eastward phase velocity of 102.2 ± 1.4 m/s relative to the System III coordinate system, which is ~20 to 25% slower than the 7.8°N eastward jet at 130 m/s (16). These characteristics are very similar to those displayed by 5-μm hot spots (17). These plumes could be the deep signature of the equatorially trapped Rossby wave thought to be responsible for forming the 5-μm hot spots. Numerical simulations of such Rossby waves showed regions with a high NH3 concentration in between (and slightly south of) the simulated hot spots (10, 11). At higher altitudes, the ammonia gas in these plumes will condense out, and as such could be responsible for the spectroscopically identified fresh ammonia ice clouds detected by the Galileo spacecraft at these latitudes (18).

Other prominent features in the radio maps are anticyclones, in particular the Great Red Spot (GRS) and Oval BA, but also the series of white ovals at 40°S (orange arrows) and numerous other small ovals. Anticyclones look relatively dark, often surrounded by a radio-bright ring, in contrast to cyclonic features, which are radio-bright (white arrows). These features are most apparent in the 8- to 12- and 12- to 18-GHz maps (i.e., at pressure levels <2 to 3 bar) (Fig. 1). The GRS is partially embedded in the SEB, which is relatively quiescent east of the GRS and extremely turbulent to the northwest. This turbulence produces the same fine-scale inhomogeneity in the radio maps as well as the visible clouds. Oval BA is also characterized by a turbulent, relatively radio-bright, region to the west (resembling a cyclonic feature) and a quiescent region to the east. Figure 3, E to F, shows spectra of the GRS and Oval BA. The NH3 concentration over the GRS itself looks similar to that of the EZ, except that it is depleted by a factor of ~2 down to ~1.5 bar (profile b) as compared to the EZ. The NH3 concentration in the bright ring is depleted twice as much, down to ~2 to 2.5 bar (profile d). Although one would expect Oval BA and the GRS to have similar NH3 profiles, given their similar secondary circulation and sensitivity to environmental stratification (19), the VLA spectra reveal differences that might be related to entrainment from surrounding belts and zones with different NH3 concentrations.

Our VLA maps at 8 to 18 GHz reveal structure at every resolvable length scale, tracing dynamical flows primarily at pressures ≲2 to 3 bar, where ammonia is strongly modulated by cloud condensation. At all observed frequencies, we find a radio-hot belt near the 7.8°N eastward jet, which contains the radio counterpart of the 5-μm hot spots, characterized by a low NH3 concentration down to ~8 bar. The ascending branch of the equatorially trapped Rossby wave that produces the 5-μm hot spots carries the deep atmospheric NH3 concentration up to the visible cloud layers, producing dark plumes in the VLA maps.

Supplementary Materials

Materials and Methods

Figs. S1 to S5

Tables S1 and S2

References (2026)

References and Notes

  1. See the supplementary materials.
Acknowledgments: This research was supported by NASA Planetary Astronomy (PAST) award NNX14AJ43G and NASA Outer Planets Research Program award NNX11AM55G. The National Radio Astronomy Observatory (NRAO) is a facility of NSF operated under cooperative agreement by Associated Universities, Inc. VLA data used in this report, associated with project code 13B-064, are available from the NRAO Science Data Archive at Ground-based optical Jupiter maps (8) are available from the Planetary Virtual Observatory & Laboratory (PVOL) website at We thank three anonymous reviewers for their thoughtful comments on this paper.
View Abstract

Stay Connected to Science

Navigate This Article