V. TOLLS,1 Z. WANG,1 G. WINNEwISSER,9 AND Y. F. ZHANG1. Received 2000 ... from Jupiter and Saturn by the Submillimeter Wave Astronomy. Satellite ...
The Astrophysical Journal, 539:L147–L150, 2000 August 20 q 2000. The American Astronomical Society. All rights reserved. Printed in U.S.A.
SUBMILLIMETER WAVE ASTRONOMY SATELLITE OBSERVATIONS OF JUPITER AND SATURN: DETECTION OF 557 GHz WATER EMISSION FROM THE UPPER ATMOSPHERE E. A. Bergin,1 E. Lellouch,2 M. Harwit,3 M. A. Gurwell,1 G. J. Melnick,1 M. L. N. Ashby,1 G. Chin,4 N. R. Erickson,4 P. F. Goldsmith,5 J. E. Howe,6 S. C. Kleiner,1 D. G. Koch,7 D. A. Neufeld,8 B. M. Patten,1 R. Plume,1 R. Schieder,9 R. L. Snell,6 J. R. Stauffer,1 V. Tolls,1 Z. Wang,1 G. Winnewisser,9 and Y. F. Zhang1 Received 2000 February 7; accepted 2000 June 23; published 2000 August 16
ABSTRACT We have used the Submillimeter Wave Astronomy Satellite to carry out observations on Jupiter and Saturn in two bands centered at 489 and 553 GHz. We detect spectrally resolved 557 GHz H2O emission on both planets, constraining for the first time the residence levels of external water vapor in Jupiter’s and Saturn’s stratosphere. For both planets, the line appears to be formed at maximum pressures of about 5 mbar. For Jupiter, the data further show that water is not uniformly mixed but increases with altitude above the condensation level. In each planet, the amount of water implied by the data is 1.5–2.5 times larger than inferred from Infrared Space Observatory data. In addition, our observations provide new whole-disk brightness measurements of Jupiter and Saturn near 489 and 553 GHz. Subject headings: planets and satellites: individual (Jupiter, Saturn) — radio continuum: solar system — radio lines: solar system — submillimeter measurements of the whole-disk brightness temperature of Jupiter and Saturn near 489 and 553 GHz.
1. INTRODUCTION
Recent observations with the Infrared Space Observatory (ISO) have shown that all four giant planets and Titan contain significant amounts of water vapor in their upper atmospheres and that CO2 is also present in Jupiter, Saturn, and Neptune (Feuchtgruber et al. 1997, 1999; Lellouch et al. 1997a; Coustenis et al. 1998). These oxygen compounds are external in origin and believed to result from meteoritic infall and subsequent chemical processing. Infall rates have comparable values of 10 5–10 7 H2O molecules cm22 s21 for the five planets. Remarkably similar values of the external fluxes at Titan and Saturn, (1–4) # 10 6 cm22 s21 (Coustenis et al. 1998; Moses et al. 2000a), suggest that interplanetary dust, as opposed to planetary rings, is the dominant source of exogenous oxygenbearing material to the planets. While the ISO observations have definitely proved the existence of such an external supply, they do not constrain the residence levels of water and CO2 in the planetary stratospheres. We here report the detection of water emission at 557 GHz from Jupiter and Saturn by the Submillimeter Wave Astronomy Satellite (SWAS). These spectrally resolved observations provide information, for the first time, on the stratospheric levels where water is located and in Jupiter’s case on the vertical distribution of H2O above its condensation altitude. The observations also provide new (space-borne for the first time)
2. OBSERVATIONS AND PLANETARY DISK BRIGHTNESS
We observed Jupiter in the two periods 1999 September 1–6 (epoch 1) and September 18–23 (epoch 2). Within each of these epochs, the solid angle of Jupiter (QJ) changed by ∼2%–3%. However, QJ changed by nearly 10% between epochs. For our disk brightness calculations we have treated each epoch separately and then averaged the results. Saturn was observed in 1999 September and November. All observations were carried out in the spacecraft nod mode (Melnick et al. 2000). Table 1 presents relevant physical data along with continuum measurements. The data are corrected for the double-sideband (DSB) response but uncorrected for antenna efficiency (i.e., on the TA∗ scale). The spectra in the 557 GHz band are shown in Figure 1. The continuum flux can be derived from TA∗ using S(l) p 2kTA∗(l)/eA A p, where eA p 0.66 5 0.03 is the aperture efficiency, derived from observations of Mars and the Moon (Gurwell et al. 2000; Melnick et al. 2000), and Ap is the physical area of the SWAS antenna. The Rayleigh-Jeans temperature (TRJ) of Jupiter was calculated using the Planck equation in the Rayleigh-Jeans limit. The total observed flux and values of TRJ are given in Table 1. At 489 GHz, our temperature (TRJ p 157 5 5 K, equivalent to a brightness temperature TB p 168.5 5 5 K) is much larger than that reported by Goldin et al. (1997; TRJ ∼ 136 5 2 K). Models predict that Jupiter’s TB decreases by about 10 K from 392 to 489 GHz and by about 4 K from 452 to 489 GHz. Therefore, our measurement would roughly extrapolate to TB p 178.5 5 7 K at 392 GHz, i.e., higher than the photometric determination of Griffin et al. (1986; TB p 162.5 5 2 K). In contrast, it would lead to TB p 172.5 5 7 K at 452 GHz, in reasonable agreement with Hildebrand et al. (1985; TB p 167 5 2 K). We note that the SWAS bandwidth of ∼350 MHz is much smaller than any of the previous measurements. The determination of Saturn’s disk brightness temperature
1 Harvard-Smithsonian Center for Astrophysics, 60 Garden Street, Cambridge, MA 02138. 2 De´partement de Recherche Spatiale, Observatoire de Paris, F-92195 Meudon, France. 3 511 H Street SW, Washington, DC 20024-2725; also Cornell University. 4 NASA Goddard Space Flight Center, Greenbelt, MD 20771. 5 National Astronomy and Ionosphere Center, Department of Astronomy, Cornell University, Space Sciences Building, Ithaca, NY 14853-6801. 6 Department of Astronomy, University of Massachusetts, Amherst, MA 02133. 7 Ames Research Center, Moffett Field, CA 94035. 8 Department of Physics and Astronomy, Johns Hopkins University, 3400 North Charles Street, Baltimore, MD 21218. 9 I. Physikalisches Institut, Universita¨t zu Ko¨ln, Zu¨lpicher Strasse 77, D-50937 Ko¨ln, Germany.
L147
L148
SWAS OBSERVATIONS OF JUPITER AND SATURN
Vol. 539
TABLE 1 Observations: Total Continuum Flux and Whole-Disk Brightness TA∗ Source Jupiter . . . . . . . . . . Averagea . . . . . . Saturn . . . . . . . . . .
Dates (1999) Sep 1–6 Sep 18–23 Sep 1 Novb
Flux
TRJ
v (arcsec)
tint (hr)
489 GHz (K)
553 GHz (K)
489 GHz (Jy)
553 GHz (Jy)
45.95 # 42.97 47.97 # 44.87 … 19.56 # 17.89
8.2 7.6 15.8 8.4
2.96 5 0.03 3.13 5 0.03 … 0.51 5 0.03
2.95 5 0.03 3.18 5 0.03 … 0.59 5 0.03
41891 5 1932 44297 5 2038 … 7217 5 425
41750 5 1926 45005 5 2038 … 8350 5 491
489 GHz (K) 159 154 157 121
5 5 5 5
7 7 5 11
553 GHz (K) 124 123 124 100
5 5 5 5
6 6 4 11
a
Average values are presented only for TRJ (see § 2). The elliptical SWAS beam was oriented with its long axis aligned with the major axis of the Saturnian disk and ring system. The rings subtended an angle of 440. 40 # 150. 79. b
is complicated by the inclusion of the Saturnian rings in the SWAS field of view. In 1999 September, the rings were inclined with respect to the Earth by an angle b p 217. Thus, the rings block some of the flux from the planet. They also emit appreciable radiation and conversely are partly eclipsed by the disk. Ring emission and obscuration effects must thus be accounted for to calculate Saturn’s disk brightness temperature. To estimate these effects, we used the formalism described in Melnick et al. (1983). Following Grossman (1990) and de Pater & Dickel (1991), we adopted ring brightness temperatures of 66 K at 489 GHz and 70 K at 553 GHz, assigning a 1 j
uncertainty of 8 K on these values. Using the expressions provided in Melnick et al. (1983), we found that the rings contribute ∼29% of the observed flux at 489 GHz and 32% at 553 GHz and that ∼10% of the flux from the disk is absorbed by the ring system. The ring-corrected Rayleigh-Jeans temperatures of Saturn are given in Table 1. The 489 GHz value (TRJ p 121 5 11 K, i.e., TB p 132 5 11 K) is again larger than that of Goldin et al. (1997; TRJ ∼ 111 5 2 K). In contrast, after applying a 4 K correction from 489 to 452 GHz, as indicated by models, it appears to be in very good agreement with Hildebrand et al. (1985; TB p 137 5 4 K at 452 GHz).
3. WATER VAPOR IN JUPITER’S AND SATURN’S STRATOSPHERES
Fig. 1.—Observed spectra of Jupiter and Saturn near 557 GHz showing emission in the 110–101 transition of water vapor (n0 p 556,936.044 MHz). The intensity scale is in kelvins, and the spectra shown here are uncorrected for the DSB response of the SWAS instrument. The broad wings (5100 km s21) seen in the spectrum of Jupiter have been demonstrated not to be real through observations at various local oscillator settings. We are unaware of any atmospheric feature that could provide the general slope seen in the spectra of Jupiter and Saturn; however, SWAS measurements alone cannot rule out the possibility.
The spectrum in the 553 GHz channel (Fig. 1) clearly shows a detection of 556.936 GHz water vapor emission in both Jupiter and Saturn. The lines exhibit an FWHM of 24 5 2 km s21 for Jupiter and 20 5 4 km s21 for Saturn. The line-tocontinuum ratio is about 8% for Jupiter and 9% for Saturn. We analyzed these features with a standard radiative transfer model (see, e.g., Lellouch, Encrenaz, & Combes 1984; Lellouch et al.1997b). The H2O emissions are formed at stratospheric levels warmer than the tropospheric continuum. Continuum opacity in the vicinity of the H2O line is due to collisionally induced absorption of H2 and He and to the far wings of the rotational transitions of ammonia and phosphine, particularly the nearby NH3 line due to the J p 1–0 transition at 572.5 GHz and the PH3 J p 2–1 doublet at 533.8 GHz. For Jupiter, ammonia cloud opacity was also considered. To account for pressure and thermal broadening, we assumed all molecular lines to be Voigt-shaped. For the H2O line strength, we used S p 5.27 # 10220 cm molecule21 at 296 K, corresponding to an ortho-para ratio of 3 and a T1.5 dependence of the partition function. An important parameter is the collisional line width of H2O for H2 and He broadening. Following Dutta et al. (1993), we used half-widths of 3.2 and 0.9 MHz torr21 at 300 K for H2 and He broadening, respectively, with temperature exponents of 0.9 and 0.5. A complication is that the disk-averaged spectra are affected by rotational smearing due to the projected diurnal velocity (12.6 km s21 limb equatorial velocity at Jupiter; 9.9 km s21 at Saturn). This effect, clearly visible in our ∼1 km s21 resolution data, is included by partitioning the planetary disks into a square grid of regular spacing (40 elements along a diameter) and calculating a synthetic spectrum at each point.10 These elementary synthetic spectra were then shifted in frequency to 10 In practice this was done by spline interpolation in air mass on spectra precalculated at eight discrete air masses at the planet (1, 1.1, 1.3, 1.8, 2.5, 4, 8, and 20).
No. 2, 2000
BERGIN ET AL.
Fig. 2.—(a) Observed spectrum of Jupiter (histograms) compared to three models using the vertical distributions of water shown in (b) using the same line designations. The continuum level in the models is rescaled to the observed value. (b) Three vertical distribution profiles of water in Jupiter’s atmosphere. Short-dashed lines: Uniform q(H2O) p 2 # 1029 above the condensation level. Long-dashed lines: Profile calculated assuming an external flux of 2 # 106 cm22 s21, the eddy diffusion coefficient (K) profile of Gladstone et al. (1996), and the assumption that all water is removed by vertical transport. Solid lines: A similar profile but with a steeper vertical slope, which could result from the combination of a larger deposition rate, a smaller eddy K, and the inclusion of chemical losses in the lower stratosphere.
account for the local projected diurnal velocity and finally summed with appropriate weights. 3.1. Jupiter For Jupiter we used the temperature profile of Fouchet et al. (2000a). This profile combines Galileo probe measurements in the troposphere and at pressures less than 1 mbar, with information on the stratospheric temperature from CH4 and H2 emissions observed with the ISO short-wavelength spectrometer (SWS). The temperature is 150 K at 10 mbar, smoothly increasing to 165 K at 1 and 0.1 mbar. We took NH3 and PH3 abundances and vertical distributions from Fouchet et al. (2000b). Ammonia has a deep mixing ratio of 2.0 # 1024, sharply decreasing from the 0.9 bar level to 2.0 # 1025 at 0.55 bar and 2 # 1028 at 0.1 bar. For phosphine, we assumed a 6 # 1027 mixing ratio below the tropopause and a photolytical cutoff above. Without any cloud attenuation, our continuum model has disk-averaged brightness temperatures (TB) of 180 K at 300 GHz, 169 K at 392 GHz, and 155 K at 675 GHz, somewhat overpredicting the Griffin et al. (1986) measurements (respectively, about 170, 162.5, and 148.5 K). Better agreement is reached if a cloud with base at 0.68 bar and transmission T p 0.8 is added. This is qualitatively consistent with the suggestion by Griffin et al. that NH3 ice particles may attenuate Jupiter’s submillimeter radiation. With T p 0.8 , we calculate a Rayleigh-Jeans temperature TRJ p 118 K at 553 GHz. This is a tolerable 1 j, 5% mismatch with the SWAS measurement, and in what follows, the model is multiplied by a constant factor to match the observed continuum; i.e., the H2O emission is analyzed in the line-to-continuum ratio.11 11 We note, however, a substantial discrepancy at 489 GHz, where the model with T p 0.8 predicts TRJ p 138 K or TB p 149.5 K. The SWAS temperature at 489 GHz cannot be reproduced by the model, which even for T p 1 yields TRJ p 144 K (TB p 155 K). With T p 0.8, the model prediction is consistent both with the Goldin et al. (1997) measurement and with the model of Weisstein & Serabyn (1996), who obtain TB ∼ 151 K at 489 GHz. On the other hand, the SWAS value agrees reasonably well with Hildebrand et al. (1985). Future observations may clarify this issue.
L149
The observed line width (FWHM p 24 5 2 km s21 or 45 5 4 MHz) gives a direct indication of the line formation level. Attributing all of the line width to pressure broadening gives an upper limit to the probed pressure level. For a Jovian composition and at 150 K, the Lorentz FWHM is 8150 MHz atm21. Thus, the bulk of the water vapor resides at pressures less than 5.5 mbar. This is an upper limit, because most of the observed FWHM is actually due to rotational smearing.12 For a typical H2O mixing ratio in Jupiter’s middle stratosphere, q p 1.5 # 1029, from the ISO/SWS spectra (Lellouch et al. 1997a; Feuchtgruber et al. 1999), condensation occurs at P p 14 mbar. Thus, the upper limit of 5.5 mbar is consistent with removal of water vapor by condensation around this level. The ISO data could be fitted either with a vertically uniform distribution above the condensation point or with a more physical H2O profile resulting from the deposition of water by micrometeoritic impact in the microbar region and downward transport to the condensation sink. Assuming no other losses (such as photolysis or chemical reactions) besides condensation and the eddy diffusion (K) profile of Gladstone, Allen, & Yung (1996), a deposition flux of (1.2 5 0.8) # 10 6 cm22 s21 was found to match the ISO spectra (Lellouch et al. 1997a; Feuchtgruber et al. 1999). A vertically uniform profile (with q p 2 # 1029 ) above the condensation level satisfies the observed 557 GHz line contrast but produces too broad (FWHM ∼ 35 km s21) emission wings (Fig. 2). The observed line is more peaked than the model, implying that q increases with increasing altitude. Calculation of contribution functions (not shown) indicates that line wings sound levels just above the condensation point, while at infinite spectral resolution, the line core probes a broad layer centered near 0.3 mbar and extending from 0.05 to 10 mbar. In spite of considerable rotational (and hence vertical) smearing, the high signal-to-noise ratio (S/N) of the spectrum permits the retrieval of the vertical distribution of water at 1–10 mbar. An H2O distribution calculated as described above from the eddy K profile of Gladstone et al. (1996), but with an external flux 1.7 times larger (i.e., 2 # 10 6 cm22 s21), produces an emission line with adequate contrast but that is marginally too broad (Fig. 2). This profile has an average mixing ratio slope over 1–10 mbar, defined as 2d(log q)/d(log p), of 0.8, which appears to be a lower limit to the actual H2O slope (Fig. 2). Increasing the slope to ∼1.3 improves the fit, although further discrimination is precluded by S/N limitations. Our preferred H2O profile (Fig. 2, solid line) has a slope of 1.3 and a total H2O column density of 2.8 # 10 15 cm22. 3.2. Saturn A similar analysis, performed for Saturn, turned out to be less instructive as a result of the lower S/N of the data and the increased importance of condensation. We first constructed a continuum model as for Jupiter, using a PH3 tropospheric mixing ratio of 4.5 # 1026 and a deep ammonia abundance of 1.75 # 1024 (de Graauw et al. 1997). We obtained a diskaveraged TB of 131 and 118.5 K (i.e., TRJ p 119 and 106 K) at 489 and 553 GHz, fully consistent with the Weisstein & Serabyn (1996) model and with the SWAS measured values. At 140 K, and for a Saturnian composition, the Lorentz FWHM of the 557 GHz line is 9400 MHz atm21. If entirely 12 Indeed, if the local H2O emission on Jupiter were spatially constant and infinitely narrow, Jupiter’s rotation would produce a line with FWHM of Î3veq p 22 km s21, almost equal to that observed.
L150
SWAS OBSERVATIONS OF JUPITER AND SATURN
attributed to collisional broadening, the observed FWHM (20 5 4 km s21 p 37 5 7 MHz) would correspond to a pressure of 4 mbar. However, this maximum pressure is implausible because the condensation pressure of water in Saturn’s stratosphere is ∼0.5 mbar. Thus, unlike in Jupiter, pressure broadening contributes a negligible fraction (∼1/10) of the observed FWHM. The latter is consistent with pure rotational smearing, which typically induces a 17 km s21 FWHM. Therefore, no further information can be obtained on the H2O distribution. From the detection of 8 H2O lines from ISO/SWS, Feuchtgruber et al. (1997) and Moses et al. (2000a) inferred a mean mixing ratio of 7 # 1029, assuming a constant value at P ! 0.5 mbar. A good fit to the SWAS spectrum is obtained for a slightly larger value, q p 1.1 # 1028, giving a stratospheric column density of 1.9 # 10 15 cm22. As on Jupiter, the unresolved ISO/SWS lines did not provide the vertical profile of H2O. However, based on detailed photochemical modeling, Moses et al. (2000a) favored a vertically increasing mixing profile with a total column density of 2.7 # 10 15 cm22. We find (Fig. 3) that their preferred H2O profile must be scaled by a factor (2 5 0.5) to fit the SWAS data.13 4. DISCUSSION
On both planets a vertical concentration slope is expected for H2O that has a high-altitude production region and a low-altitude condensation sink. This situation is observed for many minor species in outer planet atmospheres, particularly for Jupiter’s and Saturn’s hydrocarbons (Be´zard et al. 1995; Fouchet et al. 2000a; Moses et al. 2000b) and Titan’s nitriles (e.g., Hidayat et al. 1997). On Jupiter, Fouchet et al. 2000a inferred a slope of 0.6 5 0.2 for ethane and 0.8 5 0.1 for acetylene. The slope of the mixing ratio we infer for water is higher than that of C2H6 and similar to or higher than that of C2H2. This is readily understood, because, as pointed out by Gladstone et al. (1996), the chemical losses of C2H6 are small and its photolysis, shielded by methane absorption, is negligible. Thus, the vertical profile of C2H6 essentially reflects its longevity in vertical transport. In contrast, ˚ , are efC2H2 and H2O, being good absorbers at 1500–2000 A ficiently dissociated in the middle stratosphere, and their mixing ratio must significantly decrease downward. This conclusion is supported by the recent Saturn photochemical model of Moses et al. (2000a, 2000b). The photolysis rates of C2H6, C2H2, and H2O at 1 mbar in Saturn’s atmosphere are 6.8 # 10210, 4.6 # 1029, and 2.5 # 1028 s21, respectively. The corresponding mix13 Note that this gives as good a fit to the data as the constant q p 1.1 # 1028 mixing ratio at P ! 0.5 mbar, illustrating that the vertical profile of water in Saturn’s atmosphere remains observationally undetermined.
Vol. 539
Fig. 3.—Saturn’s observed spectrum (histograms) compared to models based on several profiles of H2O in Saturn’s stratosphere. Dashed line: q p 1.1 # 1028, uniform at P ! 0.5 mbar. Solid lines: H2O profiles equal to 1, 2, and 3 times ( from bottom to top) the nominal H2O profile of Moses et al. (2000a), a detailed photochemical model, and a comparison with the ISO/SWS data. The continuum level in the models is rescaled to the observed value.
ing ratio slopes between 0.05 and 0.5 mbar are about 0.3, 0.8, and 0.8, respectively. This qualitatively confirms the above picture. However, a direct correlation between the slope and the photolysis rate is not to be expected, as the slope results from the combined effect of the deposition/production rate and altitude (which differs for different species), the transport rate, and the photolytical, chemical, and condensation losses in the lower stratosphere. Therefore, a more detailed discussion for Jupiter must await the development of a complete Jupiter model including the oxygen chemistry and its testing against the ISO and SWAS data. Finally, although the SWAS data tend to suggest larger amounts of water (and therefore larger input fluxes) than inferred from ISO, it is hard to say whether this indicates a real variability, as could result from an increase of small cometary impacts. On this issue, a monitoring of the H2O lines by SWAS (and ODIN and FIRST in the future) will be very valuable. We thank Martin Burgdorf, Gordon Bjoraker, Gary Davis, Eli Dwek, Dale J. Fixsen, Peter J. Gierasch, Alexey B. Goldin, Leo Metcalfe, Alberto Salama, and Edward L. (Ned) Wright for stimulating and informative discussions and communications. We are grateful to the referee The´re`se Encrenaz for valuable comments and suggestions. This work was supported by NASA’s SWAS grant NAS5-30702. R. Schieder and G. Winnewisser would like to acknowledge the generous support provided by the DLR through grants 50 0090 090 and 50 0099 011.
REFERENCES Be´zard, B., et al. 1995, Icarus, 118, 384 Coustenis, A., et al. 1998, A&A, 336, L85 de Graauw, Th., et al. 1997, A&A, 321, L13 de Pater, I., & Dickel, J. R. 1991, Icarus, 94, 474 Dutta, J. M., Jones, C. R., Goyette, T. M., & de Lucia, F. C. 1993, Icarus, 102, 232 Feuchtgruber, H., et al. 1997, Nature, 389, 159 ———. 1999, in The Universe as Seen by ISO, ed. P. Cox & M. F. Kessler (ESA SP-427; Noordwijk: ESA/ESTEC), 133 Fouchet, T., et al. 2000a, A&A, 355, L13 ———. 2000b, Icarus, 143, 223 Gladstone, G. R., Allen, M., & Yung, Y. L. 1996, Icarus, 119, 1 Goldin, A. B., et al. 1997, ApJ, 488, L161 Griffin, M. J., et al. 1986, Icarus, 65, 244 Grossman, A. W. 1990, Ph.D. thesis, Caltech
Gurwell, M., et al. 2000, ApJ, 539, L143 Hidayat, T., et al. 1997, Icarus, 126, 170 Hildebrand, R. H., et al. 1985, Icarus, 64, 64 Lellouch, E., Encrenaz, T., & Combes, M. 1984, A&A, 140, 405 Lellouch, E., et al. 1997a, in Proc. First ISO Workshop on Analytical Spectroscopy, ed. A. M. Heras, K. Leech, N. R. Trams, & M. Perry (ESA SP419; Noordwijk: ESA), 131 ———. 1997b, Planet. Space Sci., 45, 1203 Melnick, G., Russel, R. W., Gosnell, T. R., & Harwit, M. 1983, Icarus, 53, 310 Melnick, G., et al. 2000, ApJ, 539, L77 Moses, J. I., et al. 2000a, Icarus, 145, 166 ———. 2000b, Icarus, 143, 244 Weisstein, E. W., & Serabyn, E. 1996, Icarus, 123, 23