Radar Altimetry of Mercury: A Preliminary Analysis - Brown ...

JOURNAL OF GEOPHYSICAL RESEARCH, VOL. 91, NO. B1, PAGES 385-401, JANUARY 10, 1986Radar Altimetry of Mercury' A Preliminary AnalysisJ. K. HARMON AND D. B. CAMPBELLNational Astronomy and Ionosphere Center, Arecibo, Puerto RicoD. L. BINDSCHADLER AND J. W. HEADDepartment of Geological Sciences, Brown University, Providence, Rhode IslandI. I. SHAPIROHarvard-Smithsonian Center for Astrophysics, Cambridge, MassachusettsMeasurements of Mercurian topography based on Arecibo radar observations are presented. The data,which were obtained from 1978 to 1984, cover much of the equatorial zone of Mercury between 12øNand 5øS latitude. Over thirty continuous altitude profiles were obtained, each spanning from 20 to 90degrees of longitude at a resolution of 0.15 ø (longitude) by 2.5 ø (latitude). Radar depths for large craterssupport previous indications from imagery that Mercurian craters are shallower than lunar craters of thesame size. One very large (800 km) impact basin shows some distinct topographic structure, although itsrelative shallownes suggests postimpact modification by isostatic relaxation or volcanic filling. Theplains of Tir Planitia appear topographically smooth to the radar. These plains extend well into thehemisphere not imaged by Mariner 10, possibly forming part of a large annulus of smooth plains aroundCaloris Basin. The circum-Caloris smooth plains are strongly down-bowed, indicating subsidence undera load. This and other similarities to lunar maria suggest a volcanic origin for these plains. Additionalareas of topographically smooth terrain have been found in both the imaged and unimaged hemispheres.Several ridges, scarps, and fault zones have been identified in the altimetry. Three mapped arcuate scarpsshow heights of about 700 m and cross-sectional widths of about 70 km. One of these features is clearly aridge, while the other two scarps have a more ridge-like appearance than is suggested by images. Onelarge-scale topographic drop of 3 km correlates well with a mapped system of faults and intracraterscarps. The equatorial zone of Mercury shows 7 km of maximum relief, although the typical elevationdifference between highlands and lowlands is closer to 3 km. Three major highland areas are found, thelargest two of which are roughly antipodal and aligned within about 10 ø of the "hot poles" of Mercury.The unimaged hemisphere, possessing both large craters and topographically smooth areas, does notappear to be markedly different in its topography from the imaged hemisphere. No evidence has beenfound for another Caloris-type impact structure in the unimaged hemisphere.1. INTRODUCTIONA program of 2380 MHz radar observations of Mercury hasbeen conducted at Arecibo Observatory since 1978. These observationshave yielded an extensive set of altitude measurementsover the equatorial zone of the planet. In this paper wepresent the altimetry results and discuss some of the possibleimplications for Mercurian geology.The earliest information on Mercurian topography was providedby radar-ranging observations [Smith et al., 1970;Ingalls and Rainville, 1972]. These measurements showed thatthe range of altitudes on Mercury was somewhat less than hadbeen measured for the equatorial zones of Venus and Mars.The most detailed of the Mercury radar measurements, andthe last to be published, were by Zohar and Goldstein [1974].Their data showed "hills and valleys" with 1-2 km relief andprovided the first evidence of craters.With the Mariner 10 spacecraft encounters of 1974-1975came the first detailed look at the surface of Mercury. TheMariner 10 photographic images, which covered 45% of theplanet at 0.1-4 km resolution, revealed a heavily cratered,lunar-like surface [Murray et al., 1974; Trask and Guest,1975]. The images suggested tectonics dominated by compressiveforces as manifested in the unique Mercurian scarpsystem. While these results were consistent with the relativelyCopyright 1986 by the American Geophysical Union.Paper number 5B5488.0148-0227/86/005 B- 5488 $05.00385subdued topography measured by earth-based radar, Mariner10 carried no altimeter and quantitative altimetry was limitedto shadow measurements of high-relief features such as cratersand scarps [Gault et al., 1975; Strom et al., 1975] and a fewphotoclinometry results [e.g., Hapke et al., 1975]. In addition,many of the Mariner 10 images were obtained at unfavorableillumination angles and one entire hemisphere (the dark sideat the times of encounter) was not imaged at all.It was clear, then, that useful earth-based radar work onMercury remained to be done and a regular program of radarrangingobservations of the planet was undertaken at Arecibo.In the following section we present a brief discussion of theobservations and the data reduction to altitude profiles. Thisdiscussion is followed by a display of the altitude profiles irl aform which is intended to be convenient for the reader wishingto make comparisons with the U.S. Geological Survey shadedreliefand geologic maps and the Mariner 10 images. In thelatter half of the paper we point out some of the more interest-ing features of the Arecibo altimetry, and discuss them in thecontext of the geological interpretations which have accumulatedin the decade since the Mariner 10 encounters.2. DATA ACQUISITION AND REDUCTIONObservations and Planet CoverageThe observations were made with the 2380 MHz (12.6 cmwavelength) radar on the 305-m-diameter telescope at the AreciboObservatory in Puerto Rico. The specifications of the

386 HARMON ET AL.' RADAR ALTIMETRY OF MERCURYTABLE 1. Radar System CharacteristicsTABLE 2. Locations of Mercury Radar Profiles' 1978-1984ParameterValue or TypeNumber Longitude LatitudeFrequencyTransmit powerAntenna gainSystem temperatureTransmit waveformPolarizationSampling intervalFrequency resolutionIntegration time2380 MHz (12.6 cm wavelength)400 kW71 dB35ø-90øKphase-coded CW (4/•s baud-length)right circular (transmit)left circular (receive)20.061 or 0.122 Hz10-20 minobserving system and some relevant observational parametersare listed in Table 1. The measurements were made in theso-called "delay-Doppler" mode. Delay (altitude) discriminationwas achieved by phase-coding the transmitted signal,and Doppler (longitude) discrimination was obtained fromFourier analysis of the coherently detected echo. The delay-Doppler data were analyzed in such a way as to yield a profileof altitude along the subradar track.Mercury's 7 ø orbital inclination to the ecliptic renders a_+ 12 ø equatorial band accessible to earth-based radar altimetry.Achieving extensive coverage over even this very restrictedlatitude band requires making observationsa largenumber of orbital aspects which, in turn, requires that observationsbe spread out over several years. The data presentedin this paper were derived from approximately 150 days ofobservations spread over the six-year period 1978-1984. Thelocation tracks of the Arecibo altitude profiles on Mercury'ssurface are mapped in Figure 1.1 4ø-27øW 10.7ø-9.9øN2 5ø-42øW 3.8ø-4.1 øN3 6ø-54øW 2.0ø-3.2øN4 21ø-61øW 0.5øS-0.4øN5 24ø-69øW 10.3ø-8.3øN6 26ø-56øW 1.8ø-1.4øS7 37ø-69øW 9.4ø-7.7øN8 43 ø-71 øw 5.0ø-4.3 øS9 66ø-85øW 3.5ø-4.1 øN10 71ø-97øW 7.2ø-6.5øN11 97ø-129øW 6.5ø-5.7øN12 104ø-126øW 5.5ø-5.3øN13 106ø-120 øw 6.00-5.3 ON14 109ø-138øW 3.6ø-4.8øN15 112ø-148øW 5.1ø-6.5øN16 113ø-141 øW 0.1 øS-0.7øN17 122ø-140øW 4.5ø-4.5øS18 127ø-169øW 5.9ø-5.3øN19 137ø-224øW 2.0øS-0.2øN20 150ø-197øW 5.3ø-7.4øN21 158ø-241øW 6.9ø-10.3øN22 174ø-209øW 4.4ø-5.9øN23 182ø-218øW 2.0ø-3.4øN24 216ø-259øW 4.3ø-5.3øN25 246ø-269øW 0.7ø-0.1 øS26 261 ø-330øW 7.9ø-8.9øN27 261ø-291øW 5.7ø-6.2øN28 279ø-338øW 10.4ø-10.8øN29 283ø-308øW 2.1ø-2.7øN30 287ø-309øW 0.3ø-0.8øN31 291ø-319øW 11.6ø-11.8øN32 320ø-339øW 1.0ø-l.3øN33 328ø-360øW 0.1ø-l.2øN34 338ø-359øW 11.4ø-10.9øNAltitudeMeasurementsThe first step in the data analysis was to produce an alti- templates were then fit to the run-averaged delay-Dopplertude profile for each observing "run." One to three observing array to determine the time delay to the leading edge of theruns were made on a given day; a run consisted of a transmit planet at each Doppler frequency. From these delays wereand a receive period, each equal in duration to the round-trip subtracted the computed time delays to a reference sphere oflight-travel time (10-20 min). The decoded echo was Fourier radius 2439 km. The resultant residual delays were then exanalyzedto yield a delay-Doppler array for each 8.4 or 16.8 s pressed as altitudes relative to the reference sphere. Using thecoherence interval, giving a frequency resolution of 0.06 or Doppler frequencies to provide an effective longitude discrimi-0.12 Hz or, equivalently, longitude resolutions of about 0.07 ø- nation, the final result for each run was an altitude profile0.15 ø . These arrays were then summed incoherently over the along a roughly east-west linear track extending approxifullreceive period to give a single delay-Doppler array for mately 7 ø of longitude to either side of the subradar point.each run. Numerically computed echo power-versus-delay This is the same analysis technique used by Ingalls and Rainville[1972] and Shapiro et al. [1972], and the reader is referredto these papers for a more detailed explanation.By making several runs on the same day and observing oncontiguous days (Mercury rotates by about 5 ø per day), weachieved extensive overlap among the individual profiles. The8final step in the analysis consisted of collating overlappingprofiles and averaging them over 0.15 ø longitude bins to pro-4duce a single "composite" profile. Each composite profilespans from 20 ø to 90 ø of longitude and consists of data taken0 •from as many as 11 days of observing. A list of the compositeMercury profiles from 1978-1984 is given in Table 2.The surface resolution ("radar footprint") of the altitudemeasurements is approximately 0.15 ø x 2.5 ø (6.4 x 106 km).560 270 ' ' ' 180 ' ' ' 9'0 ' ' 0The 0.15 ø longitude bin size is roughly equal to the coarsestresolution of the individual profiles and is larger than theLongitude (øw)maximum longitude smear from planet rotation over an ob-Fig. 1. Location tracks on Mercury for the Arecibo radar alti- serving run. The latitude resolution is much coarser, as eachmetry profiles (1978-1984). The radar resolution cell actually extendsover 2.5 ø of latitude centered on the indicated track (see text). Note altitude data point can be influenced by echoes arising fromthat the latitude scale is exaggerated; the Arecibo coverage is confined anywhere within roughly a degree to the north or south of theto a near-equatorial band.subradar track. Interpretation problems resulting from scat-

_390 HARMON ET AL.' RADAR ALTIMETRY OF MERCURYi i i• I 10.4_5.7 5.5I I I9.310.3 •••280 270 260 250 240 230 220,Longitude (øw)0.2•'•Fig. 2d. Altitude profiles for the H-9 quadrangle (216ø-288øWlongitude). The display format follows that of Figure 2a (see caption).This entire quadrangle lies in the unimaged hemisphere.crater rim heights if the rim crest fills only a small fraction ofthe area of the radar resolution cell. There may also be acorresponding tendency to underestimate crater floor depths,although the intrinsically high radar cross sections of craterfloors should mitigate this effect. Finally, a bias can be introducedif the crater floor is offset to the north or south of thesubradar track, in which case the measure depth will includea contribution from the planet's spherical figure. This biasamounts to 0.37 km for a 1 ø latitude offset, and increases asthe square of the offset. The most striking example of thiseffect is the anomalous 6 km drop seen at 143øW, 1.8øS(Figure 2b). This is apparently a signature of the crater Theophanes,which lies 2.5 ø south of the nominal subradar track.This is an extreme case, however, and in most cases we wouldexpect the upward depth bias due to planet curvature to beabout 10% or less. We have elected not to correct for thecurvature bias because of (1) the potential errors in subradarlatitude associated with uncertainties in the true pole position,and (2) the likelihood that this bias will be offset to someextent by underestimation of the rim height.In Table 3 are listed the radar-derived depth estimates for23 craters ranging from 40 to 260 km in diameter. Only cratersin the imaged hemisphere of Mercury are included.4.5NameMozartHandelChaikovskijLysippusZeamiRudakiThakurTyagarajaYeatsLu HsunAsvaghosaTABLE 3. Mercury Crater Depth EstimatesDiameter, Depth,Position km km Class190.6øW, 8.0øN 260 2.9 134.0øW, 3.9øN 155 3.1 150.5øW, 8.0øN 145 2.4 2133.0øW, 1.3øN 130 1.8 2147.6øW, 2.5øS 123 2.4 154.5øW, 3.5øS 115 1.8 263.8øW, 2.7øS 105 2.3 2148.9øW, 4.3øN 104 4.1 134.9øW, 9.7øN 100 2.0 182.1øW, 7.3øN 100 1.4 159.9øW, 3.5øS 99 2.5 224.0øW, 0.1 oS 98 3.1 121.2øW, 10.7øN 86 2.0 1129.9øW, 0.1øN 75 2.1 1132.8øW, 6.3øN 72 1.8 255.8øW, 3.7øS 65 2.7 1121.9øW, 5.4øN 58 1.8 2117.8øW, 6.0øN 55 1.8 1119.0øW, 0.2øN 51 2.0 2132.7øW, 4.0øN 50 2.5 2115.7øW, 5.2øN 48 1.8 2110.5øW, 5.5øN 41 1.2 2113.5øW, 6.5øN 40 1.9 2Degradational states for these craters were obtained from thegeologic maps of the H-6 and H-8 quadrangles •Schaber andMcCauley, 1980; De Honet al., 1981] and were estimated forthe H-7 quadrangle from Mariner 10 images and the USGSshaded-relief map of the quadrangle. Following the treatmentof Malin and Dzurisin F1977], we have grouped the cratersinto "fresh" or class 1 (USGS classes C3-C5) and "degraded"or class 2 (USGS classes C•-C2) and have plotted their depthsas a function of diameter (Figure 3). Included in Figure 3 arethe Mariner 10 shadow-derivedepth data of Malin and Dzurisin[1977]; their power-law relation for fresh Mercurian cratersis plotted along with an envelope encompassing theirrange of values for both fresh and degraded craters. Alsoshown in Figure 3 is the power-law depth/diameter relation10.95608.9 h,. • FI•O9 1.5 • •,1.0I I I I550 540 530 520Longitude (øw)310 300 290Fig. 2e. Altitude profiles for the H-10 quadrangle (288ø-360øWlongitude). The display format follows that of Figure 2a (see caption).This entire quadrangle lies in the unimaged hemisphere.Diameter (km)i i • i !Fig. 3. Depth versus diameter for fresh class 1 (triangles) and degradedclass 2 (crosses) Mercurian craters as measured by radar altimetry(Table 3). The dotted line shows the approximate range ofshadow-derivedepth values for fresh and degraded Mercurian cratersl,Malin and Dzurisin, 1977]. The straight lines are the fittedpower-law c•epth/diameterelations for fresh craters on the moonl,Pike, 1974] and Mercury rMalin and Dzurisin, 1977].iooo

::HARMON ET AL.: RADAR ALTIMETRY OF MERCURY391..•-•'EL)1" ' Basin •'• .................... :,:.Homer:'.o 4 /' --....-4..--•.- - '' 'ß E]ecta1-,5045 40 35 30 25:i Longitude (øw)Fig. 4. Altitude profiles selected from the H-6 quadrangle (see Figure 2a) showing topography across Homer Basinand an unnamed basin to the west. Vertical bars indicate the + 1 standard deviation altitude errors. Broken lines indicatethe approximate locations of the basin rims as discerned from Mariner 10 images and USGS maps. Arrows denote theapproximate locations of the inner (I) and outer (O) basin rings of Homer. The location of the rim/ejecta from TitianCrater in the floor of the western basin is shown. The subradar tracks for these profiles are shown on the USGSshaded-relief map at the top.for fresh lunar craters [Pike, 1974]. The radar-derived depthsfall well within the range of depths found by Malin and Dzurisin[1977], although our average depth for fresh craters is 17%less than their 2.9 km depth for 100-km-diameter fresh craters.Unfortunately, the two sets of depth estimates (radar andshadow) are disjoint; no craters were measured by both methodsand thus it is impossible to establish reliably whetherthere is a systematic disagreement between the two techniques.It is clear, however, that most of the radar depths of freshMercurian craters are significantly lower than the 4.2 kmdepth typical of 100-km-diameter fresh lunar craters, whichlends support to the assertion that large craters on Mercuryare shallower than their lunar counterparts.Radar profiles in the unimaged hemisphere of Mercury(Figures 2d, 2e) show several features that are obviously ofimpactorigin. Examples include structures at 306øW, 11.7øNand at 355øW, 11.0øN, as well as a possible peak-ring basinlocated at 279øW, 8.3øN. The depths of these craters are in the2-3 km range, which is consistent with the depths of craters inthe imaged hemisphere.Craterand Basin StructureDespite the subduing effect of the finite radar resolution cell,most of the crater profiles exhibit some evidence of upraisedrims. Rim heights of 0.5-1.0 km are typical, while the freshcraters Tyagaraja and Zeami display asymmetric rims risingas much as 1.5 km above the surrounding terrain (see Figure2c). Mozart, the largest fresh crater within our radar coverage,

392 HARMON ET AL.' RADAR ALTIMETRY OF MERCURYhas an asymmetric rim structure which is probably influencedstrongly by the underlying large-scale topography. There is,however, a distinct break in slope in the radar profiles southand east of Mozart (see Figure 6) which correlates well withthe edge of Mozart's continuous ejecta blanket as mapped by$chaber and McCauley [1980]. Here the radar profiles indicatea thickness of about 1 km for the rim/ejecta deposits at adistance of 1.2 crater radii from the rim crest.Interior structure can be discerned for some of the largercraters and basins. For example, the craters Asvaghosa andYeats both show hints of central peak topography (Figure 2a).One of the profiles across Mozart (Figures 2c and 6) showssome interior structure which may be indicative of an innerring, the existence of which is also hinted at by the USGS H-8shaded-relief map. Mozart appears to have a somewhat bowl-shaped floor (Figures 2c and 6, uppermost profile), in contrastto the usual assumption that larger craters and basins areflat-floored structures. The large crater in the unimaged hemisphereat 279øW, 8.5øN (Figure 2d), though comparable in sizeto Mozart, shows shallower walls broken by terrace featuresthis is a result of' either a lower basin production rate or ahigher degree of isostatic relaxation than has operated on themoon. Others have discussed the lack of identifiable multiringstructures on Mercury compared to the moon or Mars andhave emphasized the role that crustal characteristics may haveplayed in the formation of very large impact structures on theterrestrial planets [Wood and Head, 1976; Strom, 1979]. Morerecently, Spudis and Strobell [1984] claim to have identified anumber of degraded multiring structures based on carefulphotogeologic mapping of massifs and massif chains, arcuatesegments of ridges and scarps, rejuvenated crater rims, andregions of anomalously high topography. Topographic informationprovided by radar allows us to examine and evaluatethree large impact structures discussed in these works.The one large basin identified in both the Schaber et al.[1977] and Spudis and Strobell [1984] surveys for which wehave extensive radar coverage is that which is bisected by theequator in the western portion of the H-7 quadrangle. Schaberet al. [1977] list it as an 839-km-diameter, single-rimmedbasin centered on 130øW, 1.8øN. It is the second largest Mercurianbasin in their survey after Caloris. Spudis and Strobell[1984] identify this same basin (which they name "Mena-Theophanes") as a multiring structure centered on 129øW, IøS,with concentric rings of diameters 260, 475, 770, and 1200 km.In Figure 5 we show four of the altitude profiles across thebasin on an expanded scale (see also Figure 2b) along withmarkers indicating the approximate basin edges as given bythe USGS shaded-relief map and Schaber et al. [1977]. Thesuggestive of an inner ring. This structure is likely to be apeak-ring basin of the same class as the basins Renoir andRodin.The peak-ring structure within Homer Basin (Figure 4)shows up only as rather subdued breaks in slope on the easternside of the basin and as a 600-m-high interior peak in thetopography on the western side. Homer shares its western rimwith a larger, older basin to the west. This unnamed basin iscentered at 44øW, 2.1øS and was included along with Mozartand Homer in a survey of large craters and basins by Schaberet al. [1977]. Superposition of ejecta from Homer as well as itshighly degraded appearance imply that this basin is quite old.The basin is 2.2 km deep at its lowest elevation (measuredprofiles show that the topography across this basin is complexand strongly latitude-dependent. The basin floor has been significantlyaltered by postbasin impacts; several have left cratersmore than 70 km in diameter. Other portions of the basinfloor appear topographically smooth, possibly indicating afrom the rim crest). This is shallower than the 2.8 km depth smooth plains fill. The southernmost profile at 4.5øS is themeasured for Homer, although it is likely that the apparent simplest of the four profiles in Figure 5. It shows 1.2 km ofrather smooth, down-bowed relief, an upraised rim in the east,depth of Homer is enhanced by the fact that it lies astride aand a western rim which corresponds to a scarp visible inregional west-facing downslope. Unlike Homer, the basin atMariner 10 imagery. Another profile at 2øS (Figure 2b; profile44øW shows no obvious topographic or morphological ring.not shown in Figure 5) shows a rise-up from the basin floorIn addition to the superposition of ejecta from Homer, theinterior of this basin has been modified by the impact thatjust north of Theophanes Crater which could correspond toformed Titian Crater (see Figure 4) and by extensive smooth the outer basin ring of Spudis and Strobell [1984]. The mostplains formation [De Honet al., 1981]. Radar profiles place northerly profile in Figure 5 shows a very prominent basinthe topographic rim of the unnamed basin near the narrow rim in the NW along with two smooth, down-bowed sectionstrough which De Hon et al. [1981] map as a crater chain. The of basin floor in the NW and NE. The two northern profiles intopographic rise from the basin floor begins well to the east of Figure 5 show very little topographic expression across thethis feature, however, and the trough itself is apparently tooshallow or too narrow to yield an identifiable radar signature.There are several cases where crater profiles are influencedor distorted by underlying topography. For example, theNE rim of the basin, whereas the eastern portion of the thirdprofile (that nearest the equator) does show some structurewhich may be basin-related. The radar data show that, overall,the interior of the basin is not significantly lower than theprominent rim of the crater ^svaghosa (Figure 2a and 10) is level of the adjacent terrain. This suggests that the basin hasbeen severely modified by postimpact processes such as isoprobablyaccentuated by the fact that it was excavated from astatic relaxation, impact cratering, and volcanic filling. In adtopographicallyhigh region formed by the combined ejecta ofdition, the interior of the basin may have experienced someseveral large impact craters [De Hon et al., 1981] and lieslocalized subsidence due to the emplacement of smooth plains.adjacent to two topographically lower plains regions to theeast and west. Craters Handel and Yeats and Homer Basin are Although there are some apparent correspondences betweentilted, apparently due to a significant west-facing regionaltopography and the locations of concentric basin rings identislope(see section 6), as are the craters Rudaki and Zeami fied by Spudis and Strobell [1984], the radar data do not offer(Figures 2a and 2c). In addition, the interior of the crater unambiguous support for the argument that this basin is aYeats displays a topographic high that may be the signature multiring structure.of an intracrater scarp, although there may also be a contributionfrom the central peak.Two other basins identified by Spudis and Strobell [1984]are also covered by altimetry profiles. The first of these is afour-ring, 1250-km-diameter structure centered on 168øW,Large BasinsThe apparent dearth of large impact structures (diametergreater than 400 km) on Mercury relative to lunar abundances6øN; the second is a five-ring, 1500-km-diameter structurecentered on 4øW, 10øN [Spudis and Strobe!l, 1984]. In neithercase do we find any long-wavelength topographic signature inhas been discussed by Schaber et al. [1977]. They suggesthat the radar data indicative of a basin. While some smaller-scale

HARMON ET AL..' RADAR ALTIMETRY OF MERCURY 393It I I It0.7-4.5 _150i 1140 130 t20 I0Longitude (øw)Fig. 5. Altitude profiles selected from the H-7 quadrangle (see Figure 2b) showing topography across the large basincentered at 130øW, 1.8øN [Schaber et al., 1977]. Vertical bars indicate the q- 1 standard deviation altitude errors. Bracketsdenote the approximate locations of the basin rim as given by the USGS shaded-relief map and Schaber et al. [1977]. Seetext for discussion. The subradar tracks for these profiles are shown on the USGS shaded-relief map at the top.features in the altimetry may correspond to remnant ring provide strong evidence for a large expanse of downwarpedstructures associated with these basins, no such identifications smooth plains extending well into the unimaged hemisphere;could be made unambiguously. We stress, however, that our this may, in fact, be part of an irregular annulus of smoothradar data, with its limited coverage and resolution, is inferior plains entirely surrounding Caloris Basin (see section 5). Theto images for the identification of remnant ring structures as- radar data also show that the region south of Caloris is one ofsociated with highly degraded basins.the highest regions in the equatorial zone of Mercury in termsThe largest prominent impact structure on Mercury is Ca- of absolute altitude (see section 7). The photoclinometry relorisBasin. The southern edge of the basin (Caloris Montes) sults of Hapke et al. [1975] suggest that the center of Calorisextends down to 14øN latitude, 5 ø north of the nearest subra- is at least 7 km below the eastern edge of the basin. Thus it isdar track. Although the Arecibo profiles do not cross the likely that the inner edge of the smooth plains annulus isbasin proper, the profiles in the southern environs of Caloris higher than the center of Caloris. Clearly, the distinctive radarare very distinctive and probably reflect topographic structure signature of a basin like Caloris would be easily recognized ifrelated to the event that formed Caloris and to postimpact such a basin lay within roughly 45 ø of the equator in theprocesses. Radar profiles in this region (see Figures 2c and 6) unimaged portion of the planet.

394 HARMON ET AL.' RADAR ALTIMETRY OF MERCURYT _ Mozart8.5While such scattering studies are obviously of interest, we willrestrict the remainder of our discussion to inferences ofœ/•ta i Smooth PIo,ns I', CalørloSndFl ecta_Mozart/., •;,.• • Smooth Pla,ns smoothness based solely on the altimetry data.The rms roughness values in Figure 7 range from about 0.1to 1 km. The heaviest concentration of smooth profiles extendsfrom 160 ø to 240øW longitude. The smooth area between160øW and 180øW coincides with the smooth plains of'! ITir Planitia (Figure 6). Figure 7 shows a local rise in theroughness at 190øW due to the crater Mozart, and then anotherlarge expanse of relatively smooth terrain extending.... I I Plainswest of Mozart and well into the unimaged hemisphere. Thesesmooth plains are probably continuous with Tir Planitia toithe south of Mozart [see $chaber and McCauley, 1980]; to-TIgether, these two regions probably constitute a single enormousexpanse of smooth plains.•9o •ao •o •o These results lend support to the notion that there exists anLongitude (øw)annular region of smooth plains surrounding Caloris [Stromet al., 1975; Dzurisin, 1976, 1978; Melosh and Dzurisin, 1978a;Fig. 6. Altitude profiles from the H-8 quadrangle (see Figure 2c) McKinnon, 1979]. Additional support is provided by theshowing topography in the environs of Tir Planitia. Vertical barsindicate the + 1 standard deviation altitude errors. Broken lineslarge-scale topography of the region south of Caloris. Thedenote the approximate boundaries of major geological terrain types profiles across Tir Planitia have a distinct concave appearanceas given by the geologic map of Schaber and McCauley [1980]. The (Figures 2c and 6) and show that Tir Planitia has a floorvertical broken line (T-T) indicates the location of the Mariner 10 approximately 1 km below the terrain to the east. The conwesternterminator.cavity of profiles west of Mozart is even more striking; herethe floor is as much as 2.5 km below the surrounding terrain.The Tir Planitia depression has its lowest altitude at 178øW,5. SMOOTH PLAINSwhereas the western depression is lowest at 208øW. ThisSmooth plains constitute the youngest of the major geologic would place the longitudinal bisector of the plains annulus atunits on the surface of Mercury [Trask and Guest, 1975]. 193øW longitude, close to the apparent longitudinal bisectorAlthough the smooth plains of Mercury bear some resem- of Caloris Basin. In Figure 8 we show the mappe distributionblance to lunar maria, their mode of emplacement remains of smooth plains around Caloris and have indicated the apuncertain.It has been suggested that the smooth plains are proximat extent of the smooth plains identified by radar inimpact melt and/or ejecta from large basins [Wilhelms, 1976; the unimaged part of the planet. Although the distribution isOberbeck et al., 1977•, although there appears to be somewhat irregular, there is a clear tendency for regions of smooth plainsmore evidence supporting a volcanic origin [Strom et al., to fall within approximately one basin diameter of the rim of1975; Trask and Strom, 1976; Cintala et al., 1977].Caloris. A final piece of evidence for the continuation of theComparison of the Arecibo radar profiles with geologic smooth plains into the unimaged hemisphere is provided inmaps of Mercury shows a strong correspondence between Figure 9. This albedo map of Mercury [Murray et al., 1972],areas mapped as smooth plains on the basis of Mariner 10 which was constructed from telescopic observations, showsimagery [Trask and Guest, 1975; $chaber and McCauley,semiannular features around Caloris which correspond well1980; De Honet al., 1981] and those areas which have a with the smooth plains in the imaged hemisphere as well assmooth appearance in the radar altimetry. The largest mappedwith the radar-inferred smooth plains on the unimaged side.expanses of smooth plains near the equator occur in the H-8 Additional photographic imaging of the region west of thequadrangle [$chaber and McCauley, 1980], and our profiles in Mariner 10 terminator is required, however, to establishthis quadrangle (Figure 2c) have a generally smoother ap- whether the annular plains do, in fact, completely surroundCaloris.pearance than do profiles in the H-6 and H-7 quadrangles(Figures 2a and 2b).We obtained a quantitative measure of the "roughness" atany given point along a radar profile by fitting a least squares12line to the 5 ø longitude segment (210 km E-W) centered onthat point, and then assigning the rms altitude variation aboutthat line as the "roughness" at that center point. The movinglinear fits effectively high-pass filter the profiles, providing ameasure of "topographic roughness" characteristic of structureon the scale of large craters. The combined roughness estimatesare shown on the map in Figure 7. From comparisons-4of Figure 7 with geologic maps and images, we find that areas• I I I I I [ I [ I I CI •of low roughness values correlate well with mapped expanses360 270 180 90 0of smooth plains or with intercrater plains areas with lowdensities of large craters. We also find that smooth plainsLongitude (øw)areas tend to yield the narrowest radar scattering functions, Fig. 7. Profile location map of Mercury (see Figure 1) with vertiwhichindicates that the relative smoothness also holds at the cal shading added to indicate topographic roughness in km r.m.s. (seesmaller scales (less than 100 m) dominating the quasispecular text for the roughness definition). The trace at the bottom gives theroughness as a function of longitude, averaged over all latitudes. Notescattering characteristics. This has been corroborated by the large expanse of low roughness covering the centralongitudes. Asrecent quasispecular scattering analyses by Clark et al. r1984]. in Figure 1, the latitude scale is exaggerated.

HARMON ET AL.' RADAR ALTIMETRY OF MERCURY 395Fig. 8. Distribution of smooth plains (shaded) in and around Caloris Basin, adapted from a figure by Trask and Guest[1975]. The crosshatching indicates the probable extension of the smooth plains terrain into the unimaged hemisphere asindicated by the radar altimetry (see text). Also indicated are the locations of Tir Planitia, Budh Planitia, and the Mozartcrater and ejecta deposits (labelled Mo).The most plausible explanation for the distinctive downbowedtopography over Tir Planitia and the plains west ofMozart is that there has been subsidence in these areas inresponse to an emplaced load. The down-bowed section ofprofiles on Tir Planitia corresponds closely to the location ofthe undivided smooth plains region mapped by Schaber andMcCauley [1980], whereas the higher, slightly rougher area tothe east was mapped as intercrater plains and mixed smoothplains/Caloris ejecta terrain (see Figure 6). The distinctive concaveshape of the topography and the abundance of mare-likeridges within Tir Planitia noted by Strom et al. [1975] andDzurisin [1978] are consistent with a lithospheric loading andflexure process analogous to the emplacement and deformationof lunar maria [Solomon and Head, 1979, 1980].The radar altimetry results appear to strengthen the analogybetween the Mercurian smooth plains and lunar mariaand, hence, to provide indirect support for a volcanic originfor the smooth plains. Dzurisin [1976], who supports the volcanictheory, suggested that Tir Planitia may have been filledwith magma withdrawn from beneath Caloris and that thissapping process may have resulted in the subsidence and ridgeformation in the Caloris Basin floor. It has been further suggested[McKinnon, 1979] that the subsequent subsidence ofthe emplaced plains outside of Caloris may have induced thefinal episode of uplift and graben formation in the floor ofCaloris.Although the plains region southeast of Caloris is quitesmooth, some distinct topographic features can be identified inradar profiles (see Figure 6). Some of these features have beenidentified as ridges on the basis of comparisons of altimetrywith Mariner 10 images and maps. The smallest and mostnumerous of these are mare-like ridges [Strom et al., 1975],100 to 200 m high and located in the low plains. Two largeasymmetric ridges bound the low plains of Tir; the first rises350 m above the mixed smooth plains/Caloris ejecta to theeast and runs roughly north-south along 174øW, while thesecond rises about 1 km above the edge of the Mozart ejectato the west at about 183øW. Both of these boundary ridgesrun approximately north-south over a distance of at least 400km. Just to the east of the low plains, two topographic highsappear in the northernmost profile in Figures 2c and 6. Locatedat approximately 170øW and 172øW at 7.2øN, thesefeatures correspond to mapped deposits of Caloris ejecta[Schaber and McCauley, 1980], perhaps implying embaymentof ejecta by smooth plains material.The expanse of smooth plains west of Mozart containstopographic highs that are similar to those seen in profiles ofTir Planitia. The most prominent of these is a 1.1-km-highfeature at 209øW, 9øN, which appears to extend at least as farsouth as 5.9øN (Figure 2c).A relatively smooth section of radar profile extends from220 ø to 240øW at 10øN (Figures 2d and 7), west of the pro-

396 HARMON ET AL.: RADAR ALTIMETRY OF MERCURY360270:.+ '30:+ 30360 270 S 90 0Fi•. 9. Albedo map oœ Mercury œ•om telescopic observations. Note the scmiannula• œ½atu•½s ccm½•½d on 190øW thatclosely •½s½mb1½ the distribution oœ smooth plainsu•oundin8 the Calofis Basin. Fi8u•c œ•om M•rray ½• aL [1972].Rcpfimcd by permission oœ Academicposed circum-Caloris annulus. This section may be smooth in height [Dzurisin, 1978], although a height of 3 km wasplains or, by analogy with the terrain southeast of Tir Planitia,it could be a region of intercrater plains bounding the smoothreported for a portion of Discovery Scarp [Strom et al., 1975].Three of the arcuate scarps in the tectonic maps of Dzurisinplains to the east. The flat section of plains from 230ø-240øW [1976, 1978] and Strom et al. [1975] have been identified inat 10øN (Figure 2d) is of interest in that it is known to be a radar profiles. All three scarps are located in the same regionregion of high thermal inertia [Chase et al., 1976], a possible of the H-6 quadrangle (Figure 2a); we display them on anindicator that a high percentage of the surface is covered by expanded scale in Figure 10. These scarps are delineated byrock or rock outcroppings. A swath of thermal data extending shadow and, as they are near the Mariner 10 eastern terminafrom200øW, 18øN to 245øW, 7øN (much of which presumably tor (illumination from the west), they have been mapped asincludes the smooth plains annulus) showsignificantly higher east-facing dips. The first of these scarps is Santa Mariathermal inertias than does the terrain further west and south(250øW to 340øW), suggesting a soil of higher density, a great-Rupes, which runs south of the crater Asvaghosa and iscrossed by a radar profile at 19øW, 3.9øN (Figure 10). Theer exposure of consolidated rock, or both [Chase et al., 1976]. profile shows an asymmetric ridge-like feature 700 m high andAlthough the largest expanse of smooth terrain is in the 70 km wide in cross section. The maximum slope of the ridgeregion south of Caloris, radar profiles and Figure 7 indicate faces east and corresponds to the mapped position of theseveral other areas of comparably low roughness. Two of scarp [De Hon et al., 1981]. The central cleft is probably athese areas are in the unimaged hemisphere at 285ø-297øW, small 25-km-diameter impact crater.8ø-11øN, and 330ø-352øW, liøN (Figures 2d and 2e). A few The second of these features runs to the NW of the cratersmaller smooth regions can be seen in the imaged hemisphere. Donne. It is crossed by a radar profile at 15øW in the southernOne area east of Asvaghosa Crater actually consists of twoplains regions separated by a ridge (Figures 2a and 10). Theseplains have been mapped as smooth plains by De Hon et al.[1981]. Smooth, concave areas are also seen at the northernprofile of Figure 10, about 4 ø east of Santa Maria Rupes. Itconsists of a complex double-peaked ridge with its maximumslope facing east and corresponding to the location of themapped scarp [De Hon et al., 1981]. The western peak isand southern portions of a large basin in the H-7 quadrangle lower and broader than the eastern peak, but its west-facing(see Figure 5). The concave shape of these areas could be the slope does appear highlighted in Mariner 10 imagery. Themaximum height and cross-sectional width of this doubleresultof basin excavation, but it is more likely to be the resultpeaked feature are comparable with those of Santa Mariaof subsidence (see section 4).Rupes (see Figure 10).6. SCARPS AND FAULT ZONESThe third "scarp" is located in the upper profile in Figure 10at 13øW, 10.3øN, due east of Asvaghosa Crater. Roughly 750Scarps are the most distinctive and widespread tectonic featureson the surface of Mercury. As defined by Dzurisin [1976,1978-1, planimetrically arcuate scarps have been generally acceptedas being thrust faults caused by crustal shortening andm high and 70 km wide, it shows a more symmetric androunded profile than the other two features. The shadowthrown by this feature is very prominent due to its proximityto the Mariner 10 terminator and, again, coincides with thecompression associated with core shrinkage i-Strom et al., eastern edge of the radar feature. The radar profile of this1975l, lithosphericooling [Solomon, 1976], and/or tidal de- feature is much more characteristic of a "ridge" than a "scarp,"spinning [Melosh and Dzurisin, 1978bl. Shadow measurements casting doubt as to whether it is, in fact, a thrust fault. Theof these features indicate that they range from 500 to 1000 m picture is less clear for the other two "scarps"(in the lower

HARMON ET AL..' RADAR ALTIMETRY OF MERCURY 397l I tlJ .......... t111 .... '1 I I ti1t ' I ttll ........ ,J,,I I t . I I1Asvaghos 'at0.74.03.8Santa Maria Rupes /30 2520 15 I0 5 0Longitude (øw)Fig. 10. Altitude profiles selected from the H-6 quadrangle (see Figure 2a) showing topography across the threemapped arcuate scarps discussed in the text. Vertical bars indicate the_ 1 standard deviation altitude errors. Arrows (S)denote the locations of the downthrown (shadow) sides of the scarps as determined from the USGS geologic andshaded-relief maps and Mariner 10 images. The subradar tracks for the two profiles are shown on the Mariner 10 image atthe top. The shorter dark line on the right side of the image is an image flaw.profile in Figure 10); both show significant west-facing downslopesbut are asymmetric, with maximum dip on the eastern(shadow) sides. In any event, the radar results suggesthatcare must be taken in the identification of scarps when usingimages obtained at only one illumination angle.The most distinctive large-scale topographic feature on theH-6 quadrangle (Figure 2a) is a 3 km drop in mean elevationthat occurs between 30øW and 40øW longitude. The mostimpressive part of this drop is at 37.5øW longitude in the twoprofiles at 2.7øN and 4.0øN latitude, just west of HandelCrater. Here most of the 3 km drop occurs within 70 km (1.5 øof longitude), although some of this west-facing slope is takenup across Handel itself. To the north, a somewhat shallowerslope occurs across and to the west of Yeats Crater (10øN). Tothe south, the high eastern rim and asymmetric profile ofHomer Basin (Figures 2a and 4) suggest that the basin alsostraddles this west-facing slope. Inspection of Mariner 10images and the geologic map of the H-6 quadrangle [-De Honet al., 1981] shows that this regional slope occurs in an areawith several west-facing intracrater fault scarps. These mappedscarps [De Honet al., 1981] transect the craters Yeats andSinan, the crater just NW of Handel, and a crater NW ofYeats. The crater NW of Handel does, in fact, show a radarsignature in the form of two narrow topographic lows near thewest rim of Handel at 37øW and 38øW, 4øN, centered on theregion of maximum slope (Figure 2a). Mariner 10 images

....398 HARMON ET AL.' RADAR ALTIMETRY OF MERCURY20 /• Crater r,m! Fault scarp Mercator Project•on.• RidgeAlttmetry 0 200.... groundtrack k•lometerscontour,,elev(•tionin kmSinan10-....I•/Yea•t,•• 0 / . ./9 . e... • .......................-I06O50 40 30 20Longitude (Ow)Fig. ll. Schematic map of the central portion of the H-6 quadrangle covering the region where radar altimetryindicates a large west-facing downslope. The topographi contours are shown based on radar altimetric profiles (subradartracks shown as dotted lines). The indicated ridges and intracrater scarps are from the geologic map of De Honet al.[1981]. Note that although no expression of faulting (e.g., scarps or ridges) is seen south of Handel, the trend of thetopography remains the same. Note also the narrow plateau east of Rudaki Crater. Abbreviations are of crater names; Ru,Rudaki; Ch, Chaikovskij; Ti, Titian.show some hints of continuation of the faults outside the craters.The indistinct nature of the presumed faults outside thecraters may reflect a difference in material properties betweenthe floor of a crater and the plains and ejecta external to acrater.The major topographic features in the area of the 3 kmdrop are shown in Figure 11. As can be seen in this schematic,it appears that the scarps in this region are part of a singlesystem of faults that trends north-south in the northern half ofthe H-6 quadrangle and is associated with the 3 km drop inmean elevation seen in radar altimetry. This system of topographicfeatures does not correlate well with any of the basinrings of Spudis and Strobell [1984] and any curvature of thisstructural trend is very slight. Thus, it seems unlikely that thisfault system is related to any extremely large impact structure.Another west-facing slope with a 2 km drop can be seen intwo radar profiles at 8øW longitude, 2.0øN and 3.8øN latitude(Figure 2a). The high side of this slope is the highest absolutealtitude that we have measured for the planet (Figure 12).Unfortunately, it lies too close to the Mariner 10 eastern terminatorto allow for comparison with images and appearsonly at the very edge of the radar profiles.A 1.5-2 km west-facing slope occurs at approximately 52øWat latitudes extending from 4.8øS to the equator. It apparentlydefines the western side of a rather narrow plateau of intercraterhighlands [De Hon et al., 1981] separating the basin centeredat 44øW, 2.1øS from the lowlands west of Rudaki Crater(see Figure 11). There is no obvious connection between thisfeature and the large fault zone to the northeast. No obviousstructural or topographic features can be seen in the altimetryor Mariner l0 derived images or maps that might show aconnection between the two. Detailed comparisons withimages are difficult in this area, however, due to the highMariner 10 illumination angles at this longitude.An escarpment can be seen in the radar profile acrossZeami Crater on the eastern edge of the H-8 quadrangle(Figure 2c). The eastern wall of Zeami drops 3.3 km to thecrater floor, whereas the west rim climbs only 2 km beforedropping back to an altitude 3 km below the level of theterrain to the east of the crater. The geologic map of $chaber-2--•--• f I I I I I I I I I.560 270 180 90 0Longitude (øw)Fig. 12. Combined plot of all Arecibo radar profiles of Mercuryfrom 1978-1982, showing absolute altitudes relative to the 2439.0-kmradiusreference sphere (denoted by the central horizontal line). Latitudesof profiles on this figure range from 11.8øN to 5.0øS.

__HARMON ET AL.' RADAR ALTIMETRY OF MERCURY 399and McCauley [1980] shows extensive faulting on the easternwall of Zeami; similarities have been noted between Zeamiand certain "floor-fractured" craters on the moon [Schultz,1977' Head et al., 1981]. It seems likely that the interior ofZeami has been influenced by the underlying topography ofthe area and that a large escarpment or steep slope was presentbefore the event that excavated Zeami. It is possible thatthe Zeami escarpment/slope may be an extension of thesystem of faults and lineaments which trend NE of TolstojBasin [Schaber and McCauley, 1980]. Given the proximity ofZeami to the Mena-Theophanes Basin (see section 4), however,basin rim structure must be considered a possible contributingor alternative explanation for the Zeami escarpment.el_>_7. EQUATORIAL GLOBAL TOPOGRAPHYTo illustrate the equatorial topography on the global scale,we have plotted all of the composite radar profiles from 1978-1982 on a 0-360 ø longitude scale in Figure 12. The altitudescale is absolute; the zero-altitude datum is defined by the2439.0-km-radius reference sphere (see section 2). Figure 13shows a histogram giving the distribution of altitudes fromFigure 12.The mean of the altitudes in Figure 12 is+0.7 km (2439.7km mean radius), which is consistent with the 2439 ___ 1 kmmean equatorial radius measured by Ash et al. [1971] fromplanetary radar observations. Also, there is good spot agreementwith the 2439.5 _+ 1 km radius measured at 295.3øW,1.1øN from radio occultations [Fjeldbo et al., 1976]. The zeroaltitudedatum corresponds to the typical elevation of Mercurianlowlands (see Figure 12 and Figures 2a-2e) and is close tothe "most probable" altitude (+0.3 km) as given by the peakof the histogram (Figure 13).The extreme range of Mercurian altitudes is 7 km (Figures12 and 13) as measured from the lowest crater floors to thehigh plateau near the Mariner 10 eastern terminator. The elevationdifference between Mercurian highlands and lowlandsis typically about 3 km, which corresponds to the approximateequivalent width of the altitude distribution in Figure 13. Forcomparison, the moon has approximately 10 km of peak-topeakrelief as measured by Apollo laser altimetry [Kaula et al.,1974]. Some of this relief is due to the roughly 2.5 km offsetbetween the moon's center-of-mass and center-of-figure, sincethe topographic datum used by Kaula et al. [1974] was basedon a sphere about the center-of-mass. Altimetry results fromthe Lunar Sounder Experiment show about 7 km of relief ifcrater floors are excluded and if the topographic datum isbased on a 1738 km sphere about the center-of-figure [Brownet al., 1974].Radar altimetry for Mercury shows two major topographichighs in the equatorial zone of the planet. The first extendsacross the eastern terminator in Mariner 10 images from350øW to 35øW and resembles a plateau with an abrupt dropoffon its western side. This drop in elevation is associatedwith an extensive system of faults (see section 6). The secondmajor high area covers a broad region south of Caloris Basinbetween 160øW and 240øW. Caloris itself is approximatelybisected by the Mariner 10 western terminator at 190øW longitude.This high area contains two local topographic lowscentered on 180øW and 210øW which correspond to areas ofsmooth plains. A third, less extensive highlands area can beseen in the more northerly profiles near 310øW in the unimagedhemisphere (Figures 2e and 12).The fact that Mercury is in a precise 3'2 spin-orbit resonanceindicates that the long axis of the planet's dynamicalfigure is aligned with the perihelion subsolar points at 0øW[ 1-5 0 5,,qlt•tude (kin)Fig. 13. Histogram of the Mercurian altitudes shown in Figure12. The histogram is normalized by the number of altitude datapoints within each altitude bin, not by area. The zero-altitude datumcorresponds to a 2439.0-km-radius reference sphere. The highest andlowest altitudes measured are indicated by arrows.and 180øW longitude. The Arecibo results (Figure 12) showthat Mercury's topographic figure is roughly aligned with itsdynamical figure, although the two large bulges appear toalign somewhat better along a 10ø-190øW longitude axis.Goldreich and Peale [1966] have shown that a difference betweenthe equatorial moments-of-inertia of only about 0.01%for an ellipsoidal figure would likely ensure a high probabilityof capture into the 3:2 resonance state. This corresponds tovariations in an ellipsoidal dynamical figure of about 100 m. Itis then conceivable that the dynamical figure of Mercurycould be dominated by a long-wavelength component of uncompensatedtopography associated with the observed bulges.An alternative explantion for the 3:2 spin-orbit resonance isthat there is a lunar-like mascon associated with the smoothplains in or around Caloris. An early suggestion was made byMurray et al. [1974] that the mascon was located in the interiorplains of Caloris itself. This was later challenged [Dzurisin,1976; Melosh and Dzurisin, 1978a] on the basis that amascon is inconsistent with evidence that the final episode inthe tectonic history of Caloris was uplift of the basin floor.Strom [1979] points out, however, that this argument againsta Caloris mascon presumes that the final uplift was isostatic.Melosh and Dzurisin [1978a], offering an alternative to a Calorismascon, argued that a positive gravity anomaly associatedwith 400 m of uncompensated material in a 1300-kmwidecircum-Caloris smooth plains annulus would suffice tocontrol the planet's dynamical figure. McKinnon [1979]claims that such an annular ring-load would be "substantiallycompensated" due to its large size, but that some stresseswould be set up which would inhibit complete compensation.The apparent similarity between the circum-Caloris smoothplains and lunar maria, strengthened somewhat by the radarevidence for subsidence, establishes a circum-Caloris masconas a plausible hypothesis. However, the available data (radarand imaging) are insufficiento determine how much of thesmooth plains material remains uncompensated.8. CONCLUSIONSThe Arecibo radar observations of Mercury provide informationon the morphology of surface features with horizontaldimensions ranging from the 50 km scales characteristic of

400 HARMON ET AL.' RADAR ALTIMETRY OF MERCURYscarps, ridges and moderate-sized craters, to the 1000-2000km scales of large impact basins and major upland regions.When used in tandem with Mariner 10-derived images andmaps, radar altimetry has allowed us to better constrain andmore closely examine previous interpretations of the geologyof the planet.Crater depth measurements appear to confirm earlier findingsfrom shadow measurements [Gault et al., 1975; Malin andDzurisin, 1977], although the slightly shallower radar depthswarrant further study. Both radar and shadow measurementssuggest that Mercurian craters are shallower, on the average,than lunar craters. Only one large (800 km) basin has beenfound to show a radar altimetric signature, the overall shallownessof which suggests significant modification by isostaticrelaxation or volcanic filling.Altimetry over the southern environs of Caloris Basin indicatesthat the circum-Caloris smooth plains extend well intothe unimaged hemisphere. Both imaged and unimaged sectionsof these plains are strongly down-bowed, suggesting thatthere has been subsidence under the load of the smooth plainsfill. This and other similarities to lunar maria offer evidence,albeit circumstantial, in favor of a volcanic origin for thesmooth plains around Caloris. Although the Arecibo data areconsistent with the hypothesis that there is an irregular annulusof smooth plains surrounding Caloris Basin, new photographicimages of the region to the west of the Mariner 10terminator will be required to establish this definitively.Some of the most important results of this study bear uponthe tectonic evolution of Mercury. We have obtained altimetricprofiles across three features mapped as arcuate scarps.One of these features appears to be a ridge rather than a scarpwhile the other two features, though showing some crosssectionalasymmetry consistent with thrust-faulting, have profileswhich are more ridge-like than is suggested by images.These results should be factored into models that account forsuch structures as compressional tectonic features. Altimetryhas allowed us to document the existence of at least one largefault zone associated with 3 km of reliefi This fault zone alsoappears to be intimately connected with one of the two broadtopographic bulges in the equatorial zone of Mercury. Thesebulges rise roughly 3 km above the mean equatorial radiusand are aligned along an axis within about 10 ø of the so-called"hot poles" of Mercury. These bulges present an interestingalternative to the hypothesis that the smooth plains in oraround Caloris control the dynamical figure of Mercury. Thequestion of whether their origin is due to tidal stresses or tosome endogenic process is crucial to understanding the tectonichistory of the planet.Radar observations of the equatorial zone of the unimagedhemisphere reveal uplands and lowlands, a number of largecraters, and some topographically smooth areas. The topographyof this hemisphere shows no marked differences with respectto the imaged hemisphere and no evidence has beenfound for the existence of another Caloris-type impact struc-ture.The discussion we have included in this paper is intended tobe descriptive and preliminary. There are several areas whichwarrant more intensive study. Further geophysical modellingof the Caloris region should include the radar topography as aconstraint. It is clear, however, that a reasonably completeunderstanding of this complex and interesting region of Mercurywill require an orbiting spacecraft with altimetric, gravimetric,and imaging capabilities. The Arecibo altimetry hasprovided some new information on the topography of ridges,scarps and faults; more detailed interpretation of these find-ings is in order. Finally, we note that the possibilities for studyof the topography of Mercury with earth-based radar havenot been exhausted, and continued observations should bemade of the unimaged hemisphere and of areas in the imagedhemisphere deemed interesting on the basis of existing data.APPENDIX' THE MERCURIAN COORDINATE GRIDSince the spatial resolution of the Arecibo radar altimetry iscomparable with uncertainties or inconsistencies in the cartographicgrid systems, an assessment of the grid systems is inorder.Longitude CoordinateThe USGS shaded-relief map of the H-6 quadrangle adoptsthe longitude reference used in the Mariner 10 cartography[Davies and Batson, 1975], in which the 20 ø longitude meridianis defined as passing through the small reference craterHun Kal. Davies and Batson [1975] state that the IAU 0 ølongitude meridian (defined by referencing to the subsolarmeridian on January 1, 1950) is at 359.5 ø ___ 0.5 • on the Mariner10 grid. Comparisons of Arecibo altimetry features withthe USGS H-6 quadrangle showed the Arecibo longitudes tobe systematically larger by approximately 0.4 ø . This differenceimplies that the 0 ø longitude reference used in the Areciboaltimetry reduction programs is at approximately 359.6øW inthe Mariner 10 (Hun Kal) system, which is consistent with theDavies and Batson [1975] estimate of the position of the IAUmeridian. Comparisons of Arecibo longitudes with longitudestabulated in the Astronomical Almanac [U.S. Naval Observatory,1984], which uses the 1979 IAU system, shows preciseagreement to within 0.01 øA position check of the longitudes of altimetry features withtheir locations on the USGS H-7 and H-8 shaded-relief mapsshowed no systematic discrepancies such as that found incomparisons with the USGS H-6 map. This contrast impliesthat there is an inherent misregistration of the USGS H-7 andH-8 maps relative to H-6. A check of the common boundaryof the H-6 and H-7 relief maps does, in fact, show such amismatch. These findings were confirmed by spot checks betweenthe USGS relief maps and the Mercury Control Net[Davies and Katayama, 1976]. This check showed preciseagreement between USGS and Control Net longitudes for theH-6 quadrangle, but that the USGS longitudes were approximately0.5 ø larger than Control Net longitudes for the H-7and H-8 quadrangles.Since the USGS shaded-relief maps provide the best meansof interpreting the radar altimetry, it was decided to adjust theArecibo profiles to conform to the USGS longitudes. Accordingly,the Arecibo profiles on the H-6 quadrangle (Figure 2a)were shifted east by 0.4 ø, whereas no shift was made to profileson the other quandrangles.LatitudeCoordinateBoth the Arecibo and USGS coordinate systems follow theIAU convention placing the Mercurian spin axis normal tothe orbital plane. The best estimate of the pole position fromMariner 10 data is offset by 2 ø from the IAU pole, with anerror oval encompassing the IAU pole [Klaasen, 1976]. Invokingsome dynamical studies by Peale [1969], Klaasen concludedthat the spin axis was most likely aligned within approximately1 ø of the orbit normal. This, then, admits thepossibility of errors in subradar latitude of the order of adegree.

HARMON ET AL.: RADAR ALTIMETRY OF MERCURY 401Acknowledgments. We are grateful to the staff of Arecibo Observatoryfor their assistance in various aspects of this project. In particular,we would like to thank R. Velez, T. Crespo, A Hine, and G. Giles.We would also like to thank T. Forni and J. Chandler for theirassistance in the ephemeris generation. One of us (I.I.S.) gratefullyacknowledges support from National Science Foundation grantPHY-8243330. D.L.B. wishes to thank the staff and administration ofArecibo Observatory for their help and for support during his stay. Aportion of this work was supported by NASA grant NGR-40-002-116from the Planetary Geology Program to J.W.H. The NationalAstronomy and Ionosphere Center (Arecibo Observatory) is operatedby Cornell University under contract with the National ScienceFoundation and with support from the National Aeronautics andSpace Administration.REFERENCESAsh, M. E., I. I. Shapiro, and W. B. Smith, The system of planetarymasses, Science, 174, 551-556, 1971.Brown, W. E., G. F. Adams, R. E. Eggleton, P. Jackson, R. Jordan,M. Kobrick, W. S. Peeples, R. J. Phillips, L. J. Procello, G. Schaber,W. R. Sill, T. W. Thompson, S. H. Ward, and J. S. Zelenka, Elevationprofiles of the moon, Proc. Lunar Planet. Sci. Conf. 5th, 3037-3048, 1974.Chase, S.C., E. D. Miner, D. Morrison, G. Munch, and G. Neugebauer,Mariner 10 infrared radiometer results: Temperatures andthermal properties of the surface of Mercury, Icarus, 28, 564-578,1976.Cintala, M. J., C. A. Wood, and J. W. Head, The effect of targetcharacteristics on fresh crater morphology: Preliminary results forthe moon and Mercury, Proc. Lunar Planet. Sci. Conf. 8th, 3409-3425, 1977.Clark, P. E., M. E. Strobell, G. G. Schaber, and R. F. Jurgens, Usingnew radar reflectivity maps to characterize features on Mercury,Lunar Planet. Sci., 15, 166-167, 1984.Davies, M. E., and R. M. Batson, Surface coordinates and cartographyof Mercury, J. Geophys. Res., 80, 2417-2430, 1975.Davies, M. E., and F. Y. Katayama, The control net of Mercury:November 1976, Rep. R-2091-NASA, Rand Corp., Santa Monica,Calif., 1976.Davies, M. E., S. E. Dwornik, D. E. Gault, and R. G. Strom, Atlas ofMercury, NASA SP-423, National Aeronautics and Space Administration,Washington, D.C., 1978.De Hon, R. A., D. H. Scott, and J. R. Underwood, Geologic map ofthe Kuiper quadrangle of Mercury, U.S. Geol. Surv., Reston, Va.,1981.Dzurisin, D., Scarps, ridges, troughs and other lineaments on Mercury,Ph.D. thesis, Calif. Inst. of Technol., Pasadena, 1976.Dzurisin, D., The tectonic and volcanic history of Mercury as inferredfrom studies of scarps, ridges, troughs, and other lineaments, d.Geophys. Res., 83, 4883-4906, 1978.Fjeldbo, G., A. Kliore, D. Sweetnam, P. Esposito, B. Seidel, and T.Howard, The occultation of Mariner 10 by Mercury, Icarus, 29,439-444, 1976.Gault, D. E., J. E. Guest, J. B. Murray, D. Dzurisin, and M. C. Malin,Some comparisons of impact craters on Mercury and the moon, d.Geophys. Res., 80, 2444-2460, 1975.Goldreich, P., and S. Peale, Spin-orbit coupling in the solar system,Astron. d., 71,425-438, 1966.Hapke, B., G. E. Danielson, K. Klaasen, and L. Wilson, Photometricobservations of Mercury from Mariner 10, J. Geophys. Res., 80,2431-2443, 1975.Head, J. W., W. B. Bryan, R. Greeley, J. E. Guest, P. H. Schultz, R. S.J. Sparks, G. P. L. Walker, J. L. Whitford-Stark, C. A. Wood, andM. H. Carr, Distribution and morphology of basalt deposits onplanets, in Basaltic Volcanism on the Terrestrial Planets, BasalticVolcanism Study Project, Pergamon, New York, 1981.Ingalls, R. P., and L. P. Rainville, Radar measurements of Mercury:Topography and scattering characteristics at 3.8 cm, Astron. J., 77,185-190, 1972.Kaula, W. M., G. Schubert, R. E. Lingenfelter, W. L. Sjogren, and W.R. Wollenhaupt, Apollo laser altimetry and inferences as to lunarstructure, Proc. Lunar Planet. Sci. Conf. 5th, 3049-3058, 1974.Klaasen, K. P., Mercury's rotation axis and period, Icarus, 28, 469-478, 1976.Malin, M. C., and D. Dzurisin, Landform degradation on Mercury,the moon, and Mars: Evidence from crater depth/diameter re-lationships, d. Geophys. Res., 82, 376-388, 1977.Malin, M. C., and D. Dzurisin, Modification of fresh crater land-forms: Evidence from the moon and Mercury, J. Geophys. Res., 83,233-243, 1978.McKinnon, W. B., Caloris: Ring load on an elastic lithosphere, EosTrans. AGU, 60, 871, 1979.Melosh, H. J., and D. Dzurisin, Tectonic implications for the gravitystructure of the Caloris Basin, Mercury, Icarus, 33, 141-144, 1978a.Melosh, H. J., and D. Dzurisin, Mercurian global tectonics: A conse-quence of tidal despinning?, Icarus, 35, 227-236, 1978b.Murray, B.C., M. J. S. Belton, G. E. Danielson, M. E. Davies, D. E.Gault, B. Hapke, B. O'Leary, R. G. Strom, ¾. Suomi, and N. Trask,Mercury's surface: Preliminary description and interpretation fromMariner 10 pictures, Science, 185, 169-179, 1974.Murray, J. B., A. Dollfus, and B. Smith, Cartography of the surfacemarkings of Mercury, Icarus, 17, 576-584, 1972.Oberbeck, ¾. R., W. L. Quaide, R. E. Arvidson, and H. R. Aggarwal,Comparative studies of lunar, Martian, and Mercurian craters andplains, J. Geophys. Res., 82, 1681-1698, 1977.Peale, S. J., Generalized Cassini's laws, Astron. J., 74, 483-489, 1969.Pike, R. J., Depth/diameter relations of fresh lunar craters: Revisionfrom spacecraft data, Geophys. Res. Lett., 1, 291-294, 1974.Schaber, G. G., and J. F. McCauley, Geologic map of the Tolstojquadrangle of Mercury, U.S. Geol. Surv., Reston, Va., t980.Schaber, G. G., J. M. Boyce, and N.J. Trask, Moon-Mercury: Largeimpact structures, isostasy and average crustal viscosity, Phys.Earth Planet. Inter., 15, 189-201, 1977.Schultz, P. H., Endogenic. modification of impact craters on Mercury,Phys. Earth Planet. Inter., 15, 202-219, 1977.Shapiro, I. I., S. H. Zisk, A. E. E. Rogers, M. A. Slade, and T. W.Thompson, Lunar topography: Global determination by radar,Science, 178, 939-948, 1972.Smith, W. B., R. P. Ingalls, I. I. Shapiro, and M. E. Ash, Surfaceheightvariations on Venus and Mercury, Radio Sci., 5, 411-423,1970.Solomon, S.C., Some aspects of core formation in Mercury, Icarus,28, 509-522, 1976.Solomon, S.C., and J. W. Head, Vertical movement in mare basins:Relation to mare emplacement, basin tectonics, and lunar thermalhistory, J. Geophys. Res., 84, 1667-1682, 1979.Solomon, S.C., and J. W. Head, Lunar mascon basins: Lava filling,tectonics, and evolution of the lithosphere, Rev. Geophys., 18, 107-141, 1980.Spudis, P. D., and M. E. Strobell, New identifications of ancientmulti-ring basins on Mercury and implications for geologic evolution,Lunar Planet. Sci, 15, 814-815, 1984.Strom, R. G., Mercury: A post-Mariner 10 assessment, Space Sci.Rev., 24, 3-70, 1979.Strom, R. G., N.J. Trask, and J. E. Guest, Tectonism and volcanismon Mercury, J. Geophys. Res., 80, 2478-2507, 1975.Trask, N.J., and J. E. Guest, Preliminary geologic terrain map ofMercury, d. Geophys. Res., 80, 2461-2477, 1975.Trask, N.J., and R. G. Strom, Additional evidence of Mercurianvolcanism, Icarus, 28, 559-563, 1976.U.S. Geological Survey, Kuiper quadrangle, Map 1-960, H-6,1:5,000,000 ser., Reston, Va., 1976.U.S. Geological Survey, Beethoven quadrangle, Map 1-1029, H-7,1:5,000,000 ser., Reston, Va., 1977a.U.S. Geological Survey, Tir (Tolstoj) quadrangle, Map 1-993, H-8,1:5,000,000 ser., Reston, Va., 1977b.U.S. Naval Observatory, The Astronomical Almanac, Nautical AlmanacOffice, U.S. Government Printing Office, Washington, D.C.,1984.Wilhelms, D. E., Mercurian volcanism questioned, Icarus, 28, 551-558, 1976.Wood, C. A., and J. W. Head, Comparison of impact craters onMercury, Mars and the moon, Proc. Lunar Planet. Sci. Conf. 7th,3629-3651, 1976.Zohar, S., and R. M. Goldstein, Surface features on Mercury, Astron.J., 79, 85-91, 1974.D. L. Bindschadler and J. W. Head, Department of GeologicalSciences, Brown University, Providence, RI 02912.D. B. Campbell and J. K. Harmon, National Astronomy and IonosphereCenter, Arecibo Observatory, P.O. Box 995, Arecibo, PR00613.I. I. Shapiro, Harvard-Smithsonian Center for Astrophysics, Cambridge,MA 02138.(Received February 20, 1985;revised August 7, 1985;accepted August 20, 1985.)