CHARACTERIZATION OF ROCK POPULATIONS ON PLANETARY SURFACES : TECHNIQUES AND A PRELIMINARY ANALYSIS OF MARS AND VENUS J. B. GARVIN, P. J. MOUGINIS-MARK,andJ. W. HEAD Department of Geological Sciences, Brown University, Providence, Rhode Island, U.S.A. (Received 4 November, 1980) Abstract. Characteristics of rock populations on the surfaces of Mars and Venus can be derived from analyses of rock morphology and morphometry data. We present measurements of rock sizes and sphericities made from Viking lander images using an interactive digital image display system. The rocks considered are in the gravel size range (16- 256 mm in diameter). Mean sphericities, form ratios, and roundness factors are found to be very similar for both Viking lander sites. Size distributions, however, demonstrate differences between the sites; there are significantly more cobble size fragments at VL-2 than at VL-1. A model calling for aphanitic basalts emplaced as ejecta or lava flows at the Viking sites is supported by the rock shape, size, and roundness data. Morphologic features pertaining to the modification history of a rock are considered for Mars and Venus. A multi-parameter clustering algorithm is utilized to objectively categorize martian and venusian rocks in terms of various criteria. Erosional markings such as flutes are demonstrated to be most important in separating VL-1 rock morphologic groups, while rock form (i.e., shape) represents the primary separator of subpopulations at VL-2 and the Venera landing sites. Fillets are common around VL-1 and Venera 10 fragments. Obstacle scours occur frequently only at VL-1. Cavities in rocks are ubiquitous at all lander sites except Venera 9. Eolian processes, possibly assisted by local solution weathering, are a strong candidate for the origin of cavities and flutes in martian rocks. 0. Introduction Rock populations on Earth typically contain morphological information pertinent to their compositions, modes of origin, emplacement styles, and subsequent weathering histories (Oilier, 1969; Folk, 1974). Terrestrial analogues for the Martian environment (Morris et al. , 1972, McCauley et al. , 1979), permit some inferences to be made about the evolution of the Martian surface. Previous studies of block fields and fine particles on Mars (Moore eta!., 1977, Evans and Adams, 1979; Strickland, 1979), and Venus (Florensky et al., 1977; Keldysh, 1979), have identified rock subpopulations within the fields of view of the Viking and Venera landers, but have lacked the large-scale data base required for multiple-parameter morphological analysis. In this report, we provide an overview of a data collection and analysis scheme that has been developed for the interpretation of rock morphology from lander images (Garvin et al. , 1980). Emphasis is placed here on our approach to solving the problem of how to best characterize rock populations on planetary surfaces. It involves the collection of quantitative data such as rock size and sphericity, as well as qualitative information regarding morphological features. Full descriptions of the morphological attributes chosen with specific rock examples are presented. Data analysis techniques are also Th e Moon and the Planets 24 (1981) 355-387. 0165--{)807 /81/0243--{)355$04.95 . Copyright © 1981 by D. R eidel Publishing Co., Dordrecht, Holland, and Boston, U.S.A. 356 J. B. GARVIN ET AL. discussed in the context of identifying key characteristics of a rock that place it in a single category with similar rocks. Actual rock characteristics observed from Viking and Venera lander imagery are summarized. Finally, we present some speculations regarding the block fields on Mars and Venus, in an attempt to answer key questions such as their overall rock type, mode of emplacement, and modification history. These speculations arise from the analysis of planetary rock morphology and morphometry data sets and their comparison to terrestrial data. 1. Technique 1.1. THE PROBLEM To characterize planetary surfaces and the processes acting on them from a small number of lander sites is especially difficult when digital images are all that are available. For Mars and Venus this is the case. The two Viking landers on the martian surface have returned thousands of images of their vicinities with resolutions of up to 1 mm per picture element or pixel (Tucker, 1978). The Venera 9 and 10 spacecraft each returned a 180° panorama of their landing sites with an optimal resolution of about 10 mm per pixel (Keldysh, 1979). These resolutions are not sufficient to distinguish most mineral grains in rock, nor are they a replacement for the kinds of detailed information field and laboratory studies provide for terrestrial rock populations. However, a systematic qualitative and quantitative analysis of planetary surface images such as those returned from the Viking and Venera spacecraft provide information suitable for comparison with terrestrial data sets created by far more detailed studies. 1 .2. THE APPROACH In order to characterize the blocks visible in planetary surface images in a rigorous fashion, a scheme for data collection must be derived. This standardization permits more rapid collection of data and forms a basis for comparison between populations. 1.1 .1. Data Gathering A. The data. An interactive digital image processing system (Garvin et al., 1980) was employed to display mosaics of VL-1 and VL-2 images covering a camera elevation band between - 20° and - 40° (0° representing the horizon) around the spacecraft. Distortion of objects below - 40° (less than 2m from the camera) is extreme (Mutch, 1978), and rocks in the - 40° to - 60° range (1.5 to 2.0 m from camera) can be elongated by as much as a factor of two. This is because the images are cylindrical mercator projections with the least distortion at 0° elevation, and maximum distortion at the pole (- 60° elevation). The field of view above - 20° (> 3.8 m from the camera) was not imaged with sufficient resolution (< 3 mm/pixel) for the detailed evaluation of cavities and flutes on rock surfaces. Only high resolution images taken with BB2 and BB3 black and white diodes were considered. This choice of camera diodes means that rocks approxi- CHARACTERIZATION OF ROCK POPULATIONS ON PLANETARY SURFACES 357 mately 3m from the cameras(- 20° to - 40° camera inclination angles) are in the best focus, and that only rocks from ~ 2.0 to 4.0 m from the cameras were studied. Lander images were also chosen so that they gave the most uniform lighting conditions in terms of solar elevation for the azimuth ranges of each mosaic. In general, high sun images were used (> 50° solar elevation), while lower sun images (20° to 30° solar elevation) were also utilized for rock morphology determination. Table I lists the mosaic images for each lander and camera on Mars. It is important to note that these mosaics form the basis for the rock size and sphericity (i.e., shape) determinations, but that multiple images were considered when the overall morphology was evaluated, at which time both high and low sun images were essential. Stereo pairs were used when available, to aid in esti­ mations of parameters such as rock form. TABLE I List of camera events included in the mosaics created as the basis for VL-1 and VL-2 rock morphology data collection. All elevations are - 20° top, - 40° bottom. Note that along with the images included here, others were used to provide multiple sun angle and stereo coverage of the areas. Lander/Camera Camera Event Label Start Azimuth Stop Azimuth (degrees) (degrees) VL -1/Camera 1 11B058/036 180.0 220.0 11A114/019 220.0 320.0 VL-1/Camera 2 12A140/024 20.0 132.5 12A152/026 132.5 192.5 VL-2/Camera 1 21A115/016 152.5 227.5 21A057 /008 227.5 235.0 21G095/524 235 .0 240.0 21B022/033 240.0 310.0 VL-2/Camera 2 22H008/593 80.0 90.0 22A251/030 90.0 95.0 22H037 /597 95.0 102.5 22G237/590 102.5 115.0 22H025/595 115.0 172.5 22B012/032 172.5 195.0 The Venera 9 and 10 sites were measured from an 11" by 14" photograph using a digitizing table. Venera camera angular resolution is approximately an order of magnitude less than Viking (0.33° /pixel vs 0.04° /pixel; see Keldysh, 1979), so only the rocks within 2.5 m of the Venera spacecraft were measured for maximum compatibility of rock morphology data sets between planets. B. Techniques for data gathering. The interactive digital image processing system was used in the collection of data from the Viking lander mosaics. Using this system, sections of the mosaics are displayed on a volatile display screen, contrast stretched, and optionally 358 J. B. GAR YIN ET AL. enlarged. An operator then selects a rock by moving a cursor ( crosshairs on the screen) about its perimeter. This process of outlining a rock triggers the computer to generate the major or length (L) and minor or width (W) axes for the rock based on the outline. By monoscopically ranging to the lowest elevation point on the rock, the absolute lengths of the axes are automatically computed as well as the least projection elongation or two­ dimensional sphericity (W/L; see Folk, 1974). The average rock diameter is then calcu­ lated by averaging the axes: (W + L)/2. After outlining a rock, the operator must qualitatively assess its morphologic features based on the 'menu' in Figure 1. A computer-driven digitizing table was used for this purpose in a manner similar to that employed for the interpretation of planetary impact craters (Arvidson et al. , 1974; Cintala et al., 1976a; Cintala et al., 1976b ). In this study, 8" by 10" and 11" by 14" photographic prints of areas included in the mosaics, but at high and low sun angles, were used for rock morphology determination.