Radar imagery of Mercury’s putative polar ice: 1999–2005 Arecibo results John K. Harmona,*, Martin A. Sladeb, Melissa S. Ricec aNational Astronomy and Ionosphere Center, Arecibo Observatory, HC3 Box 53995, Arecibo, PR 00612, USA. Tel: (787)-878-2612; Fax: (787)-878-1861; Email: [email protected] bJet Propulsion Laboratory, California Institute of Technology, MS 238-420, 4800 Oak Grove Dr., Pasadena, CA 91109, USA. Tel: (818)-354-2765; Fax: (818)-354-6825; Email: [email protected] cDept. of Astronomy, Cornell University, Ithaca, NY 14853, USA. Tel: (607)-255-4709; Fax: (607)-255-6918; Email: [email protected] Submitted to Icarus: June 3, 2010 Revised: August 12, 2010 Accepted: August 18, 2010 36 manuscript pages 5 tables 7 figures Proposed running head: Mercury poles: Radar imagery Send correspondence and proofs to: John K. Harmon Arecibo Observatory HC3 Box 53995 Arecibo, PR 00612 Email: [email protected]; Tel: (787)-878-2612 x284; Fax: (787)-878-1861 2 Abstract We present an updated survey of Mercury’s putative polar ice deposits, based on high- resolution (1.5-km) imaging with the upgraded Arecibo S-band radar during 1999–2005. The north pole has now been imaged over a full range of longitude aspects, making it possible to distinguish ice-free areas from radar-shadowed areas and thus better map the distribution of radar-bright ice. The new imagery of the south pole, though derived from only a single pair of dates in 2005, improves on the pre-upgrade Arecibo imagery and reveals many additional ice features. Some medium-size craters located within three degrees of the north pole show near-complete ice coverage over their floors, central peaks, and southern interior rim walls and little or no ice on their northern rim walls, while one large (90 km) crater at 85°N shows a sharp ice-cutoff line running across its central floor. All of this is consistent with the estimated polar extent of permanent shading from direct sunlight. Some craters show ice in regions that, though permanently shaded, should be too warm to maintain unprotected surface ice owing to indirect heating by reflected and reradiated sunlight. However, the ice distribution in these craters is in good agreement with models invoking insulation by a thin dust mantle. Comparisons with Goldstone X-band radar imagery indicate a wavelength dependence that could be consistent with such a dust mantle. More than a dozen small ice features have been found at latitudes between 67° and 75°. All of this low-latitude ice is probably sheltered in or under steep pole-facing crater rim walls, although, since most is located in the Mariner-unimaged hemisphere, confirmation must await imaging by the MESSENGER orbiter. These low-latitude features are concentrated toward the “cold longitudes,” possibly indicating a thermal segregation effect governed by indirect heating. 3 The radar imagery places the corrected locations of the north and south poles at 7°W, 88.35°N and 90°W, 88.7°S, respectively, on the original Mariner-based maps. Keywords: Mercury; Mercury, surface; Radar observations; Ices 4 1. Introduction It has been nearly two decades since radar observations at Goldstone/VLA (Very Large Array) and Arecibo Observatory revealed radar-bright features at Mercury’s north and south poles that suggested the existence of polar ice deposits (Slade et al., 1992; Harmon and Slade, 1992; Paige et al., 1992; Butler et al., 1993; Butler, 1994). The early Arecibo delay-Doppler imagery of the poles, obtained at 15-km resolution, traced the radar features to the interiors of impact craters and thus provided strong support for the presence of frozen volatiles in shaded polar cold traps (Harmon et al., 1994). Accessing this ice with Earth- based radars is made possible by Mercury’s large (7°) orbital inclination to the ecliptic, which presents Earthward tilts of the poles of up to 12° and allows radar illumination of crater interiors that (because of Mercury’s near-zero rotational obliquity) are permanently shaded from direct sunlight. In the mid-1990s the Arecibo telescope underwent a major upgrading that substantially improved the sensitivity of the S-band (2380-MHz;λ 12.6-cm) radar. One of the early achievements with the upgraded radar was the re-imaging of Mercury’s north pole at a much finer (1.5–3 km) resolution. These 1998–1999 observations revealed many additional north polar “ice” features, including some at relatively low latitudes, and provided a more detailed picture of the ice distribution within individual craters (Harmon et al., 2001). In 2000 we commenced a multi-year program at Arecibo to do full-disk, dual-polarization radar imaging of Mercury. Unlike the 1998–1999 observations, which used standard (repeating-code) delay-Doppler, this program employed the so-called “long-code” delay-Doppler method (Harmon, 2002). This technique was specifically designed to eliminate the delay-Doppler aliasing that corrupts radar mapping of “overspread” targets. 5 Although the main benefit of using long-code was cleaner full-disk imaging of non-polar regions (Harmon et al., 2007), the same data were also suitable for imaging the poles. Furthermore, by adopting a 10-µs pulse width we were able to continue the imaging of the poles at the same 1.5-km resolution as for the 1999 imagery. In 2001 and 2002 we imaged the north pole over a range of longitude aspects but at less-than-optimal sub-Earth latitudes (= Earthward pole tilts) of 4.5–7.3°N. The observing aspect improved during 2004–2005, when we were able to make high-quality north polar images at sub-Earth latitudes in the range 8.8–11.8°N and from a longitude aspect roughly opposite that of 1999. Also, in March 2005 the sub-Earth latitude reached 7.5°S, enabling us to make our first post-upgrade images of the south pole. In this paper we present the Arecibo radar imagery of the Mercury poles based on the 1999–2005 observations. (For presentations of the non-polar imagery from this same period see Harmon et al. (2007) and Harmon (2007)). We combine long-code imagery from 2001–2005 with the 1999 imagery to give an essentially complete survey of the radar-bright “ice” deposits at the north pole. We also present and discuss the imagery of the south pole based on the March-2005 observations. We have two main objectives in presenting these results in this paper. The first is to provide as thorough an inventory as possible of the putative polar ice deposits and to point out certain results from the updated imagery that may have implications for the ice hypothesis. The second is to present a data base suitable for making comparisons with, and enhancing the scientific return from, the anticipated Mercury pole studies by the MESSENGER and BepiColombo orbiters. 6 2. Data and analysis The images used in this paper are derived from observations made on 23 dates organized in 13 groups of contiguous days. A listing of these “date groups” (in order of increasing sub- Earth longitude) is given in Table 1. Most of the observations were made when Mercury was Table 1 not far from inferior conjunction; the Earth-Mercury distance for the dates in Table 1 was in the range 0.56–0.82 AU. All of the observations were made using binary-phase-coded transmission with a 10-µs baud (synthesized pulse width), which gave a range resolution of 1.5 km. All but the 1999 dates used a long (240−1 element) code, which was effectively non-repeating and thus suitable for delay-Doppler imaging by the long-code method (Harmon, 2002; Harmon et al., 2007). A circularly polarized wave was transmitted and the echo was received in both (orthogonal) circulars. Transmitter power was in the range 780–900 kW for the 1999–2004 observations. In 2005 we ran at reduced power (400–440 kW for the north pole, 680 kW for the south pole) owing to transmitter problems. Details of the long-code analysis are as described in Harmon et al. (2007) and briefly reviewed here. The received echo time series was sampled once per baud and multiplied with a lagged replica of the code. This lagged-product time series was smoothed and decimated by a factor of 512 to reduce the Nyquist bandwidth to 195 Hz (twice Mercury’s total Doppler bandwidth), and then 8192-length blocks were Fourier transformed and squared to produce Doppler spectra with 0.0238-Hz resolution. The lags were shifted in one-baud steps to build up the delay-Doppler array. Successive delay-Doppler looks were summed over a full observing “run” (transmit/receive cycle), and from this array a flat noise spectral baseline was subtracted at each lag. The run-averaged array was normalized to units of radar cross 7 section, using the long-code calibration procedure discussed in Harmon (2002), and then an image was formed by doing a mapping from delay-Doppler space to planetary coordinates. The image pixels were normalized to a dimensionless reflectivity σ 0 (θ ), which is the radar cross section per unit surface area at incidence angle θ . Our spectral (Doppler) resolution of 0.0238 Hz gave a transverse mapping resolution of 1.1 km, which, when combined with the 10-µs delay resolution, gave a mapped image pixel measuring 1.5 ×1.1 km at the pole. (Since the planet rotated by 0.04° of longitude during a given run receive period, the images suffered an azimuthal smearing amounting to 0.6 km at 70° latitude and decreasing as the cosine of the latitude.) The final step in the processing involved summing the run-averaged images (typically 10 per day) over a data group or several date groups to produce the images shown in this paper.