3.1. Rotation-averaged images
We first present longitude-averaged images at 13 and 22 cm, for which we incorporated all 10 days of observing, approximately 12 rotations of Jupiter. These images were built essentially by assuming that the radiation belts are circularly symmetric about a tilted magnetic axis. We use an axis tilted by from the rotation axis towards . To account for the wobble introduced by the tilt, we applied a time-dependent rotation to the coordinates (and, for linear polarization, to Stokes Q and U visibilities by double the angle). Other than this, we used standard synthesis imaging and deconvolution techniques. It is assumed that there were no large temporal changes of the intensity of the radiation belts during the observations; excepting the impact of Comet SL-9, during many years of observation no short-term variations have been confirmed, although long-term variations have been established by Klein, Thompson & Bolton (1989) and by Bolton et al. (1989).
Fig. 1 shows the resulting images. They have the magnetic axis vertical. Longitude-dependent features are smeared out. In addition, the up-and-down rocking of the radiation belt in front of the disk adds a smeared-out component of synchrotron radiation to the brightness of the thermal emission.
Images of Jupiter at 13 cm are rare, the only previous ones being those of Kenderdine (1980). Fig. 1 shows features with unprecedented clarity. At 13 cm the thermal radiation of the disk produces the bright central region, with the synchrotron radiation producing the extensions. Thermal radiation, which is unpolarized, is absent in the linearly-polarized image. At 22 cm the thermal disk appears much less bright than at 13 cm, whereas in reality its brightness temperature is slightly higher; the synchrotron radiation is much brighter (by a factor of approximately ).
The images clearly show the radiation from the two populations of energetic electrons, one with large pitch angles producing radiation concentrated at the magnetic equator, and the other one with smaller pitch angles producing radiation up to high latitudes. Even though these are averaged maps, the east side of the belt is brighter than the west one. This is not unexpected; de Pater & Jaffe's (1984) observations made in 1981 also show the belt on the east to be brighter than the west for most CML. We will develop an explanation for this asymmetry in Paper II.
Fig. 2 shows one-dimensional brightness scans as a function of position along the magnetic equator. The scans show that the equatorial belts extend to 4 at both 13 and 22 cm; at that distance the signal is still above the noise, estimated to less than K. We note that the slope of brightness of the belt is very regular from the peak to the noise level: neither at 2.5 nor at 3 (positions of the orbits of Amalthea and Thebe) is there a distinct change of slope, showing that they produce no large change in the synchrotron-emitting electron population. The peak to peak separation is 2.9 at both wavelengths. De Pater & Klein (1989) and de Pater (1991) summarized previous measurements of this separation, which is not always the same. However, the most recent measurements, those of 1989 with the VLA, were close to our value of 2.9 . We note that the separation we measure is the average over 12 rotations. The distance of the peaks from the center of Jupiter changes with CML, as is described below and documented in Paper II.
3.2. Images at different longitudes
For aperture synthesis, a significant portion of the plane must be sampled before a Fourier transform can produce a reliable image. For an east-west array, such as the ATCA, only a very limited coverage at a particular longitude is obtained on any one day. However, on successive days different parts are covered. With our 10 days of observations, (5 days, a 5 day gap, and 5 more days), the resulting coverage for a given longitude is quite reasonable, provided that we consider 40- bins of longitude.
We have formed images at 18 nominal values of CML (, , ..., ), 9 of which are independent, by using data within of the CML. We make images in the four Stokes parameters, CLEAN them, and construct images of the total and linearly polarized intensity. Within the of rotation for each image, we apply corrections (as in the previous section) to remove some of the effects of the magnetic field wobble.
Fig. 3 shows maps at 13 and 22 cm for two longitudes apart. At the east side of the belt is dimmer than the west, while at the east side is brighter than the west. The color table was chosen to span the full range of brightnesses, 100 K to 1320 K at 22 cm and 50 K to 530 K at 13 cm. The non-uniformity of the brightness of the disk, especially notable at 13 cm, is the result of synchrotron radiation of the equatorial belt in front of it. In the images at where the magnetic dipole points away from the observer, the belt is north of disk center, and conversely for the images at .
3.3. Reconstruction of the belts in 3 dimensions
Studying a series of two-dimensional projections as above is not the optimal way of understanding three-dimensional objects, so we also have produced cubes of Jupiter with three spatial axes. Only a brief description of the technique used to produce these will be given here - full details can be found in Sault et al. (1996b).
Consider imaging a rotating object where the source is optically thin, and where the emission is radiated isotropically. In this case, an interferometer measures an instantaneous sample in a three-dimensional Fourier domain. With many such measurements combined, the use of a three-dimensional Fourier transform will produce a cube with three spatial axes. This cube can then be deconvolved using a three-dimensional algorithm.
We have implemented such an imaging and deconvolution scheme, incorporating a "MX"-like CLEAN algorithm. In this form of CLEAN, source components are subtracted from the visibility data (rather than the images) during so-called major cycles. Using this characteristic, we are able to relax the optically thin assumption by determining when a particular component is shadowed behind the disk of Jupiter, and not subtracting it if it is. As we are not interested in imaging the disk, we subtract a constant temperature elliptical disk model from the data before imaging the total intensity data (we have used temperatures of 280 and 350 K at 13 and 22 cm respectively; see de Pater et al. 1995).
The assumption that is not readily relaxed is that the emission is radiated isotropically. For the synchrotron radiation, where beaming is involved, this is not satisfied. With the isotropic assumption, the resultant cubes represent the average emission at each 3-D position. Asymmetries in the emission of a given longitude between east and west limb passage are averaged out. However, as will be seen later these asymmetries are of order 20 to 30% and do not obscure certain important features of the radiation belts.
3.3.1. Total intensity
Fig. 4 (top row) shows sample views of the cube from three different angles in total intensity. The thick disk coincides with Jupiter's magnetic equator and is a result of emission from the electron population with large pitch angles. The inner extensions to higher latitudes are due to the population with smaller pitch angles. In this presentation the disk of Jupiter has been removed.
A very important feature seen in Fig. 4 is that the equatorial belt is not cylindrically symmetric or planar, but is warped, with some portions being tilted relative to the "average" magnetic equator by or more. In addition, some portions are at larger radii than others. Further discussion of these features is given in Paper II.
3.3.2. Linear polarization
The same procedure was used to build 3-D images of the linearly polarized emission (Fig. 4 bottom). The most striking feature is that the low pitch angle emission is seen at high latitudes, north and south, like two little rings, but is weaker (absent in the visualization) at medium latitudes. The linearly polarized emission at high latitudes corresponds to mirror regions where the electrons are reflected when the magnetic field strength reaches the critical value; it is there that the electrons spend most time, that the emission is the most intense. We see also the warped, asymmetric shape of the two little rings, as in the main belt.
De Pater & Jaffe (1984) first noted maxima at high latitudes in VLA images in total intensity. As seen in Fig. 1, they exist in our 2-D images in total intensity with low contrast, and in linearly polarized radiation with somewhat higher contrast simply because the total dynamic range is smaller. At 13 cm we find that the degree of linear polarization is constant at over most of the image and the brightness temperature is 15-20% higher at the high latitude peaks than at intermediate latitudes, both in total intensity and in linear polarization. The high contrast in the 3-D reconstructions of linear polarization in Fig. 4 arises because effects of different regions along the line of sight are largely eliminated.
© European Southern Observatory (ESO) 1997
Online publication: July 3, 1998