Styles and timing of volatile-driven activity in the eastern Hellas region of Mars

Transcription

1 JOURNAL OF GEOPHYSICAL RESEARCH, VOL. 110,, doi: /2005je002496, 2005 Styles and timing of volatile-driven activity in the eastern Hellas region of Mars David A. Crown and Leslie F. Bleamaster III Planetary Science Institute, Tucson, Arizona, USA Scott C. Mest Planetary Geodynamics Laboratory, NASA Goddard Space Flight Center, Greenbelt, Maryland, USA Received 19 May 2005; revised 22 September 2005; accepted 24 October 2005; published 24 December [1] Recent analyses of Mars Global Surveyor and Mars Odyssey data sets provide new insights into the geologic evolution of the eastern Hellas region of Mars, in particular, the role of volatiles. Here, we present results of our recent work and integrate these with previous studies by various investigators to provide a synthesis of the history of volatile-driven activity of the region. We utilize high-resolution images from the Mars Orbiter Camera and Thermal Emission Imaging System combined with Mars Orbiter Laser Altimeter digital elevation models and profiles to examine fluvial systems that dissect the circum-hellas highlands, to characterize stages in the development of the Dao, Niger, Harmakhis, and Reull Valles canyon systems, and to evaluate evidence for ancient lakes in Hellas Planitia. The occurrence of valley networks, dissected highland crater rims, and crater interior deposits such as layered plateaus suggests widespread ancient degradation of the circum-hellas highlands. Canyon development, which represents subsequent more localized activity, may have included an early fluvial phase followed by the collapse and sapping dominated stages that, along with recent wall erosion and floor resurfacing, produced the currently observed morphologies. The prominent role of collapse and sapping along the east rim of Hellas, along with the presence of numerous channels extending toward the basin and sequences of finely layered deposits along the basin rim, suggests a volatile-rich substrate across a broad depositional shelf. The east rim of the basin was an accumulation zone for atmospheric volatiles and/or the edge of volatile-rich deposits associated with the basin floor. This evidence combined with topographic data and cratered terrain preservation around the basin is consistent with a lacustrine period or periods in early Martian history. The style, magnitude, and spatial extent of volatile-driven activity in eastern Hellas have varied considerably with time, and these variations may represent a transition from a water- to an ice-dominated surface environment. Citation: Crown, D. A., L. F. Bleamaster III, and S. C. Mest (2005), Styles and timing of volatile-driven activity in the eastern Hellas region of Mars, J. Geophys. Res., 110,, doi: /2005je Introduction Copyright 2005 by the American Geophysical Union /05/2005JE [2] Hellas basin spans more than 2000 km in the southern cratered highlands of Mars and is the largest well-preserved impact structure on the Martian surface. Hellas is Mars deepest depositional sink, with elevations as low as 8200 m in western Hellas Planitia, and the Hellas region has long been recognized as a source for global dust storms [e.g., Greeley et al., 1992]. The basin and surrounding highlands exhibit landforms shaped by a diversity of geologic processes, preserve exposures of Noachian, Hesperian, and Amazonian units, and extend across a wide range in latitude and elevation, encompassing the Martian midlatitude zone where geomorphic indicators of ground ice are prominent [e.g., Greeley and Guest, 1987; Crown et al., 1992; Squyres et al., 1992; Leonard and Tanaka, 2001; Mest and Crown, 2001]. Geologically contemporaneous volcanism and volatiledriven activity in the circum-hellas highlands, particularly to the east, provide resources for potential Martian life [Farmer et al., 1993]. Hellas is a significant region for evaluating volatile abundance, distribution, and cycling as well as changes in surface conditions on Mars, given the nature and diversity of potential water- and ice-related landforms. Geologic studies using Mariner 9 and Viking Orbiter images have provided fundamental information on the geologic evolution of the region and interpretations of specific landforms (e.g., highland paterae, valley networks, and debris aprons). Mars Global Surveyor, Mars 1of19

2 Odyssey, and Mars Express data sets are now being used to test earlier hypotheses and provide new insights into the geology and climate history of this important part of the Martian surface. The purpose of this work is to synthesize new and recent work on (1) fluvial systems in the circum-hellas highlands, (2) the formation of the prominent canyons on Hellas east rim, and (3) the potential for ancient lakes on the basin floor, and, using this information, provide a summary of our understanding to date of the history of volatile-driven activity in the region. 2. Background [3] Hellas Planitia, which includes the basin interior and inner rim deposits, extends across 50 in longitude (from 46 to 95 E) and more than 20 in latitude (from 32 S to 55 S), using 4000 m elevation to roughly define the topographic depression of the basin (Figure 1). Elevations below 8000 m are located along the bases of scarps that define the northwest and southwest borders of Alpheus Colles. Hellas basin is surrounded by the cratered highland terrains of Noachis Terra to the west, Terra Sabaea to the northwest, Tyrrhena Terra to the northeast, and Promethei Terra to the east and southeast. Large expanses of ridged plains are found to the south in Malea Planum and to the northeast in Hesperia Planum; contained within or adjacent to these ridged plains are the highland volcanoes Tyrrhena and Hadriaca Paterae (to the northeast) and Amphitrites and Peneus Paterae (to the south). The Hellespontus Montes define the rugged west rim of the basin. The basin shows significant east-west asymmetry with more gradual slopes and less rugged topography to the east, along with the presence of Dao/Niger, Harmakhis, and Reull Valles and associated plains. [4] The Hellas region has been the subject of numerous geologic mapping studies at various scales, using different imaging data sets to examine the regional stratigraphy and provide constraints on the geologic history. Studies based on Mariner 9 images provided the first stratigraphic analyses of the region, putting the geology of the basin into a global context [Scott and Carr, 1978] and examining the volcanic history, basin interior deposits, and characteristics of the cratered highlands [Potter, 1976; Peterson, 1977; King, 1978]. Viking-based mapping includes the 1:15Mscale global series [Greeley and Guest, 1987; Tanaka and Scott, 1987] and a 1:5M-scale map focused on Hellas [Leonard and Tanaka, 2001; see also Tanaka and Leonard, 1995], as well as a series of detailed 1:500,000-scale studies covering parts of Tyrrhena and Hadriaca Patera [Gregg et al., 1998; D. A. Crown and R. Greeley, Geologic map of MTM and quadrangles, Hadriaca Patera region of Mars, submitted to U.S. Geological Survey Geologic Investigations Series, 2005 (hereinafter referred to as Crown and Greeley, submitted map, 2005)], Dao/ Niger, Harmakhis, and Reull Valles [Price, 1998; Mest and Crown, 2002, 2003], and Tyrrhena Terra [Mest and Crown, 2005a]. Syntheses of regional geology with maps of the eastern Hellas region are also given by Greeley and Crown [1990], Crown et al. [1992], and Mest and Crown [2001]. [5] Geomorphic studies of the region include investigations of explosive volcanism at the highland paterae [Peterson, 1978; Greeley and Spudis, 1981; Greeley and Crown, 1990; Crown and Greeley, 1993], emplacement of Tyrrhena Patera lava flows [Greeley and Crown, 1990; Crown et al., 1991], development of highland fluvial systems [Carr and Chuang, 1997; Mest and Crown, 2001; S. C. Mest et al., Watershed modeling in the Tyrrhena Terra region of Mars, submitted to Journal of Geophysical Research, 2005 (hereinafter referred to as Mest et al., submitted manuscript, 2005)]; formation of circum-hellas canyon systems and associated plains [Squyres et al., 1987; Price, 1998; Mest and Crown, 2001, 2002, 2003], and characteristics of lobate debris aprons [Squyres, 1979, 1989; Squyres and Carr, 1986; Pierce and Crown, 2003]. A series of recent studies have focused on impact crater degradational history, in particular for Millochau and Terby craters, which are both located north of Hellas [e.g., Ansan and Mangold, 2004; Berman et al., 2005; Mest and Crown, 2005b; Moore and Howard, 2005a; Wilson and Howard, 2005]. Recent climate modeling studies indicate that the Hellas basin has a major influence on southern hemisphere circulation and that, as a result, eastern Hellas is a zone of significant volatile accumulation [Colaprete et al., 2004a, 2005]. Earlier studies proposed that Hellas was the site of extensive lacustrine and glacial activity [Kargel and Strom, 1992; Moore and Wilhelms, 2001]. 3. Data Sets and Research Methods [6] In this investigation of the geology of the Hellas region of Mars we utilize imaging and topographic data sets from the Viking Orbiter (VO), Mars Global Surveyor (MGS), and Mars Odyssey (MO) missions. Imaging data sets include (1) individual VO frames with resolutions in the range from 10s to several hundred meters/pixel; (2) a VO MDIM (Mars Digital Image Mosaic) 2.1 photomosaic (256 pixel/deg or 236 m/pixel; gov); (3) MGS Mars Orbiter Camera (MOC) narrow-angle (2 12 m/pixel) and wide-angle (240 m/pixel) images ( and (4) MO Thermal Emission Imaging System (THEMIS) visible (VIS) wavelength images (20 m/pixel) and daytime and nighttime infrared (IR) images (100 m/pixel) ( Use of this suite of imaging data sets allows characterizations of the geomorphology of the surface at a range of geologically significant scales. To analyze the topography of the surface, we use MGS Mars Orbiter Laser Altimeter (MOLA) data, specifically (1) Mission Experiment Gridded Data Records (MEGDR; 128 pixel/deg or 462 m/pixel) and (2) individual Precision Experiment Data Record (PEDR) profiles, which have a footprint size of 160 m, an along-track spacing of 300 m, and an effective vertical resolution of 37 cm ( [7] Many of the data sets used in this investigation have been imported into ArcGIS 9 Geographic Information Systems (GIS) software for analysis. Prior to GIS import, MOC and THEMIS images were processed from raw to geometrically calibrated, noise-reduced, map-projected images using the Integrated Software for Imagers and Spectrometers (ISIS) software ( 2of19

3 Figure 1. Topography of the Hellas region. (a) MOLA 128 pixel/degree DEM in simple cylindrical projection showing locations of features and regions discussed in the text. Numbered black lines show locations of topographic profiles shown in Figure 1d. (b) MOLA 128 pixel/degree DEM of Hellas basin with contour intervals shown. Moore and Wilhelms [2001] discussed evidence for ice-covered lakes with stands or shorelines at 6900, 5800, and 3100 m in elevation. See text for discussion of geologic significance of 5800 m and 1800 m elevations. Box indicates location of Figure 1c. (c) MOLA 128 pixel/degree DEM merged with Viking MDIM 2.1 image mosaic of eastern Hellas Planitia showing locations of layered exposures identified in MOC and THEMIS images (yellow dots). (d) Profiles derived from MOLA 128 pixel/degree DEM (vertical exaggeration = 100) showing topographic characteristics of Hellas basin floor, rim, and surrounding highland terrains. Shaded region bounded by dashed lines lies between 5800 and 1800 m. See text for discussion. North is to the top in all figures unless otherwise indicated. 3of19

4 Figure 1. (continued) usgs.gov). In GIS, data are georeferenced to the global MOLA MEGDR data in simple cylindrical projection. 4. Regional Geology: Observations and Interpretations 4.1. Cratered Highlands [8] The Noachian Period included the formation of Hellas basin and numerous other large impact events in the southern highlands that contributed to the formation of cratered highland terrains and associated intercrater plains adjacent to the basin [e.g., Greeley and Guest, 1987; Schultz and Frey, 1990; Tanaka and Leonard, 1995; Leonard and Tanaka, 2001]. These highland terrains appear to have had extensive, complex histories of degradation resulting in widely variable preservation of the rugged highland topography. The rims of large impact craters and isolated and clustered knobs of mountainous material, some of which are likely remnants of eroded crater rims, rise to significant heights (typically 1 4 km) above younger plains. In the Hellas region, the transition from the Noachian to the Hesperian Period was marked by widespread volcanic activity to the south and northeast of Hellas [Plescia and Saunders, 1979]. For eastern Hellas, previous studies suggested that the explosive activity at Hadriaca and Tyrrhena Paterae that formed their now-eroded flanks was followed by voluminous effusive eruptions emplacing the ridged plains of Hesperia Planum in the Early Hesperian Epoch [Greeley and Spudis, 1981; Greeley and Crown, 1990]. General parallels for Amphitrites and Peneus Paterae and Malea Planum have been noted [e.g., Greeley and Spudis, 1981], with ridged plains emplacement in Malea Planum in the Late Noachian to Early Hesperian [Tanaka and Leonard, 1995; Leonard and Tanaka, 2001]. These ridged plains and the younger (Late Hesperian to Early Amazonian [e.g., Crown et al., 1992]) extensive lava flow field extending for more than 1000 km from the summit of Tyrrhena Patera to the southwest suggest a regional transition in eruptive style and a prolonged and voluminous magma source in the eastern Hellas region. Viking-based geomorphic and modeling studies showed that explosive eruptions driven by either magmatic volatiles or by interactions between rising magma and meteoric water could have provided sufficient energy to emplace pyroclastic flows comparable to the extents of the patera flanks [Greeley and Crown, 1990; Crown and Greeley, 1993]. [9] MOLA DEMs show significant and widespread variations in highland topography surrounding Hellas (Figure 1), suggesting the possibility of extensive modification in the Noachian Period, assuming highland topography was initially relatively uniform around the basin. Expanses of preserved rugged highlands are typically found at elevations between 2 and 4 km. Low-lying, relatively smooth intercrater plains within highland terrains occur at elevations between 500 m and 1.5 km; Hesperia Planum surfaces (away from Tyrrhena Patera) are typically 4of19

5 located at elevations between 1 and 1.5 km, and Malea Planum surfaces and smooth plains adjacent to Reull Vallis in Promethei Terra [see Mest and Crown, 2001, 2002, 2003] are found at elevations between 500 m and 500 m. The lowest area of the basin rim is the zone extending from Hellas Planitia to the northeast, which encompasses Dao and Harmakhis Valles and occurs at elevations between 4 and 1 km. In eastern Hellas, modification of original rugged highland terrain would have occurred prior to paterae formation and emplacement of the ridged plains in Hesperia Planum. Ridged plains partially fill a former low-lying region but remain at elevations significantly below adjacent highland surfaces. Kostama et al. [2005] note that the surface of Hesperia Planum is typically 450 m below the adjacent highlands and estimate that the ridged plains are between 250 and 500 m thick [see also Ivanov et al., 2005]. Significant modification of highland topography is supported by the noticeable lack of large highland craters on relatively smooth, low-lying surfaces adjacent to heavily cratered highland terrain, as well as the existence of both relatively continuous zones of rugged highlands and isolated and clustered mountainous remnants at similar radial distances from the basin center. [10] Alternatively, the circum-hellas highland terrains could have formed with significant initial topographic irregularities. Tanaka and Leonard [1995] suggested that the asymmetric distribution of features defining the uplifted basin rim were the result of a low-angle impact with a S60 E trajectory, and Wichman and Schultz [1989] utilized models of lithospheric flexure to assess the distribution of basin structures. Other studies have examined structures radial and concentric to the basin in order to infer positions of ring structures and to identify other impact basins surrounding Hellas that may have had persistent structural influences [e.g., Peterson, 1978; Schultz and Frey, 1990; Schultz, 1984]. Tanaka et al. [2002] attributed the missing highland terrain in parts of the circum-hellas highlands to catastrophic erosion of the rim of Hellas by a period of voluminous magmatic intrusion into volatile-rich rocks, causing massive destabilization of the surface. Although this interpretation might be applicable to the low-lying zones northeast and south of the basin where evidence for later volcanism also exists and the lack of highland topography is most prominent and areally extensive, it is not consistent with other circum-hellas regions in which similar vertical irregularities in highland terrains are apparent. For example, in the regions north of Hellas and west of Tyrrhena Terra, west of Hellas in Noachis Terra, and east of Hellas adjacent to Promethei Terra, there is no direct evidence of volcanism in the preserved surface materials and the magmatic erosion model is more difficult to support. Regardless of the nature of its formation, the existence of a topographic depression in the Hesperia Planum region during the Noachian Period is significant in that it may have provided connectivity between low-lying terrains in Hellas Planitia and those of the northern lowlands. If an extensive northern ocean did exist on early Mars [Parker et al., 1989, 1993], then it is likely that a southern extension into the cratered highlands through Hesperia Planum and into Hellas basin existed; correlations between the geologic records along the dichotomy boundary and in the Hellas region may provide important constraints on Martian climate history and the extent and timing of Martian lacustrine/marine activity. [11] The Hellas radial topographic profiles shown in Figure 1d, which are derived from the 128 pixel/degree MOLA DEM, provide additional constraints on the topographic characteristics of the basin and rim terrain. In particular, they allow qualitative assessments of roughness, in terms of magnitude and scale, and the profile pairs shown provide a clear indication of large-scale topographic irregularities. From examination of the profiles it is evident that the floor of the basin is relatively flat with irregular local topography, which appears to be a function of the deposits that have accumulated on the basin floor and their subsequent erosion and redistribution (for previous work on Hellas basin floor deposits, see section 4.4). The profiles show the east-west asymmetry in the basin rim and the missing highland terrain (more than 2 km in places) to the east (in particular, comparison of profiles 2, 3, and 4 with 8, 9, and 10). The south rim of Hellas exhibits evidence for modification of heavily cratered terrain and extensive resurfacing by volcanism in the form of Malea Planum and Amphitrites and Peneus Paterae. In a general sense this is similar to the ridged plains of Hesperia Planum and the volcano Tyrrhena Patera to the northeast of Hellas; however, comparisons of the topographic characteristics between the northeast and south circum-hellas regions indicate significant differences. In the northeast, the surfaces of Hesperia Planum are found at elevations significantly below the surrounding highlands; in the south, volcanism appears to have built up the rim (note slight deflection of color-coded topography in Figure 1a toward basin interior along south rim) rather than filled a depression. These younger volcanic surfaces are found at elevations comparable to degraded highlands elsewhere around the basin. Thus the relative magnitudes of highland terrain removal and volcanic resurfacing appear to have been different in the south and east, with potentially less extensive removal of rim materials to the south. On the basis of visual inspection of the profiles and the MOLA DEM, generally smoother surfaces at local scales occur between 5800 and 1800 m relative to above or below this elevation range Highland Fluvial Valleys [12] Cratered terrain and intercrater plains surfaces, the flanks of the highland paterae, and the rims of many impact craters exhibit valley systems interpreted to result from a combination of groundwater sapping and surface runoff [Pieri, 1980; Craddock and Maxwell, 1993; Carr, 1995; Carr and Chuang, 1997; Grant, 2000; Craddock and Howard, 2002]. The lower drainage densities determined for Martian basins relative to terrestrial basins dominated by surface runoff are in some localities attributed to a hybrid process in which precipitation provided surface runoff to form valleys as well as, due to high infiltration rates, recharged subsurface aquifers which caused modification of networks by sapping [e.g., Grant, 2000; Grant and Parker, 2002]. Snowmelt has also been considered a source for the fluids carving the valley networks [Carr and Head, 2004]. Integrated valley networks with sub-parallel, rectilinear, or dendritic patterns are found within Tyrrhena Terra 5of19

6 Figure 2. Fluvial valleys in the circum-hellas highlands. (a) MOLA 32 pixel/degree DEM of Mars Transverse Mercator quadrangles and and (b) part of Viking Orbiter image mosaic of MTM quadrangles, which are located north of Hellas basin in Tyrrhena Terra. Superimposed on the DEM are mapped valleys from Mest and Crown [2005a, 2005b] and Mest et al. (submitted manuscript, 2005). Vichada Valles is the large integrated valley network whose watershed dominates the region. (c) Viking Orbiter MDIM of part of Promethei Terra region east of Hellas showing fluvial valleys from Mest and Crown [2001] and Mest et al. (submitted manuscript, 2005) [see also Mest, 2004]. Parts of Reull Vallis are observed in the southern part of the area. (d) Part of THEMIS VIS image V (image width = 17.4 km, image center = S, E) showing detailed view of valleys dissecting plains which embay the highlands. and Promethei Terra [Carr and Chuang, 1997; Mest and Crown, 2001; Mest, 2004; Stepinski and Collier, 2004; Mest et al., submitted manuscript, 2005]. Drainage basins in Promethei Terra are significantly smaller in areal extent but more numerous than those in Tyrrhena Terra [Mest and Crown, 2004], in part reflecting differences in the topographic characteristics of these two highly cratered regions (Figure 2). Mapping studies suggest that Tyrrhena Terra fluvial systems are ancient [Mest and Crown, 2005a; Mest et al., submitted manuscript, 2005], whereas those in Promethei Terra are Hesperian in age and dissect an intermontane basin unit that accumulates in low-lying regions of the highlands [Mest and Crown, 2001, 2002]. The apparent age difference between valley networks in the two regions is consistent with the more pronounced and long-lived presence of volatile-driven activity near Promethei Terra, where canyon systems, sinuous fluvial channels, debris aprons, and gully systems are also observed. The wide distribution of valley networks in the circum-hellas highlands, and the existing age constraints that indicate predominantly ancient valley formation, suggest a more active hydrologic environment, with volatile release triggered or enhanced by impact events and/or volcanism. A more hydrologically active surface environment is consistent with several possible scenarios for early Martian conditions, including (1) more clement and relatively uniform (at least at similar latitudes) climate conditions in early Martian history across the Hellas region; (2) temporally, and 6of19

7 Figure 3. (a) Part of THEMIS VIS image V (image width = 17.4 km, image center = S, E) showing fluvial valleys on south flank of Hadriaca Patera that are truncated by main canyon of Dao Vallis. Sinuous, v-shaped interior valleys (v) are observed within larger, broad, radial troughs. Note fine-scale parallel troughs perpendicular to larger radial valleys that are located on mesas (m) of highstanding erosional remnants, as well as superposition of impact ejecta on valley floors. (b) Mars Express High-Resolution Stereo Camera (HRSC) image (orbit 528, image center = 93 E, 32 S) of Hadriaca Patera and Dao/Niger Vallis for context shows numerous valleys extending across flanks of volcano. potentially spatially, variable climatic conditions with a series of relatively short duration but intense episodes of erosion, either due to enhanced precipitation or release of volatiles from subsurface reservoirs. See Baker [2004], Clifford [2004], Colaprete et al. [2004b], Howard et al. [2004], and Moore and Howard [2004] for discussion of early Martian climate. If scenario #1 above is correct, this is in strong contrast to the apparent diversity in volatile abundance and/or local climatic conditions that are inferred from the nature and distribution of younger landforms. [13] The highland paterae exhibit radial valley systems dissecting their flank materials [Greeley and Spudis, 1981; Greeley and Crown, 1990; Gulick and Baker, 1990; Crown and Greeley, 1993; Plescia, 2004] (Figure 3). The amphitheater-headed valleys, layering exposed in valley walls, and erosional remnant mesas observed in Viking Orbiter images were attributed to the combined effects of surface runoff and groundwater sapping acting on a sequence of pyroclastic flow deposits. More deeply incised valleys within the wider, flatter-floored troughs along the flanks of Hadriaca Patera suggest a greater component of surface runoff than at Tyrrhena Patera and/or more friable materials within Hadriaca Patera. In THEMIS images, ejecta blankets from craters in the 5 15 km diameter range superpose both the upper patera surfaces and the floors of the erosional valleys, indicating that both formation and initial dissection of patera surfaces was ancient [Crown et al., 2005a; Williams et al., 2005]. THEMIS images show fine-scale parallel troughs perpendicular to the larger radial valleys, consistent with friable pyroclastic deposits and sustained erosion (fluvial or nival) of the highland paterae, at least in some localities (Figure 3a). The early erosion of the highland paterae may be the result of different climatic conditions regionally and/or to generation of localized microclimates due to eruption of volatile-rich magmas [e.g., Wilson and Mouginis-Mark, 1987]. [14] The interior crater rims of many large impact craters in the region exhibit evidence of dissection; in some cases numerous parallel troughs are observed (Figure 4) and in others alcove-gully apron systems as described by Malin and Edgett [2000a] are present. Some exterior rims also show erosional channels; these are less numerous and more sinuous, perhaps reflecting differences in volatile supply, slope and/or surface materials. Craters in general show a 7of19

8 to confidently comment on the age or range in ages of crater rim erosion. Figure 4. THEMIS VIS image V (image width = 17.4 km, image center = S, E) of impact crater east of Hellas with (a) numerous parallel troughs on its interior rim to the south, (b) a series of potential ice-rich debris lobes on its interior rim to the north, and (c) several sinuous valley segments extending from the rim through the ejecta blanket to the north. wide range of preservation states and contain an interesting array of interior deposits, some of which are significant accumulations of material [Ansan and Mangold, 2004; Ansan et al., 2005; Mest and Crown, 2005b; Moore and Howard, 2005a, 2005b; Wilson and Howard, 2005]. Current research is characterizing crater interior morphologies and addressing the nature of crater interior depositional environments to assess potential paleo-lacustrine activity and atmospheric and subsurface volatile inventories [see also Lahtela et al., 2005; Korteniemi et al., 2005a, 2005b]. At this time, not enough detailed analyses of impact crater morphologies across the Hellas region have been completed 4.3. Canyon Systems and Associated Plains [15] Dao, Niger, Harmakhis, and Reull Valles extend through the cratered highlands and a series of plains units in the eastern Hellas region toward Hellas Planitia (Figure 5). The three circum-hellas canyon systems represent a stage in regional geologic history following formation of valleys within highland terrains (including those on the paterae) in the Late Noachian and Early Hesperian Epochs and prior to formation of debris aprons and gullies in the Amazonian Period [Crown et al., 1992; Price, 1998; Leonard and Tanaka, 2001; Mest and Crown, 2001]. Mest and Crown [2001] mapped a dissected plains unit south of Tyrrhena Patera that was originally considered to be part of Hesperia Planum [Greeley and Guest, 1987]. Significant fluvial dissection of the surface is noted along with subdued wrinkle ridge morphology, and this may be a zone of modified ridged plains, consistent with the interpretation of a sedimentary component to the ridged plains near the upper reaches of Reull Valles [Mest and Crown, 2001] and the recognition of significant drainage from Hesperia Planum to the south [Crown and Mest, 2004; see also Gregg and Crown, 2005]. Dissection of plains units that cover low-lying regions of the highlands may mark a transition from widespread valley network formation to more localized volatile-driven erosion. Channeled and smooth plains in the eastern Hellas region are dissected by the canyon systems; along the east rim of Hellas basin, evidence for contemporaneous and pre-valles fluvial activity is evident in the form of numerous elongate, sinuous channels. These channels are contained within channeled and dissected rim units defined in earlier mapping studies [Greeley and Guest, 1987; Crown et al., 1992; Leonard and Tanaka, 2001; Mest and Crown, 2001], with fluvial activity inferred to be predominantly Hesperian in age. Some of these channels are not proximal to the canyon systems; in particular there is a concentration of channels to the northwest of Dao Vallis along the eastern rim of Hellas basin (Figure 6). Many others are observed adjacent to the canyon systems; some of these are truncated by canyon segments and others appear to form tributaries to canyon segments (Figure 5). On the basis of cross-cutting relationships, these channels appear to be mostly older than or contemporaneous with the formation of the main canyons, although continued widening of the canyons by wall collapse makes age relationships complicated to interpret, and some channels may be younger. [16] Dao (and its tributary Niger Vallis), Harmakhis, and Reull Valles have been classified as outflow channels, although they do not exhibit distinctive streamlined islands like those found within the circum-chryse outflow channels [Baker, 1982]. In addition, the hybrid nature (characteristics of both outflow and fretted channels) of Reull Vallis has been recognized [Carr, 1981; Mest and Crown, 2001]. The circum-hellas canyon systems have been studied as part of the global inventory of fluvial features on Mars [e.g., Sharp, 1973; Carr, 1981, 1996; Baker, 1982]; they have also been the subject of more detailed geomorphic analyses [Crown et al., 1992; Mest and Crown, 2001] and have been the focal points for a series 8of19

9 Figure 5. Viking Orbiter MDIM 2.1 image mosaic showing eastern Hellas region with geologic units and volatile-related features discussed in text. Arrows representing fluvial valleys on Hadriaca Patera (thin white arrows), elongate channels within various plains (thin black arrows), circum-hellas canyon segments (thick black arrows), and scoured region south of Reull Vallis are shown (white arrows). Numbers indicate approximate centers of other figures. of geological mapping studies [Price, 1998; Mest and Crown, 2002, 2003; Crown and Greeley, submitted map, 2005]. Because of its prominent central canyon and the recent identification of gullies on canyon walls [Malin and Edgett, 2000a], its proximity to Hadriaca Patera with the potential for volcano-ice interactions and hydrothermal systems, and its direct connection to Hellas Planitia, Dao Vallis has long been considered a landform of significance on Mars and remains one today. The nature, magnitude, and history of volatile-driven processes associated with Dao and the other circum-hellas canyon systems have important implications for geologic and climate studies of Mars. In this work, we synthesize the results of recent work that utilizes new Martian data sets to advance our understanding of the processes involved in canyon formation, their spatial and temporal variability, and their role in the volatile evolution of the Hellas region. [17] Dao/Niger, Harmakhis, and Reull Valles extend for 1200 km, 800 km, and 1500 km, respectively, through eastern Hellas, with individual canyon segments varying in width between 5 and 60 km [Crown and Mest, 1997; Crown et al., 1997; Mest and Crown, 2001]. Dao and Harmakhis Valles are parallel (NE-SW trending) canyon systems that traverse the east rim of the basin and extend into Hellas Planitia. On the basis of Viking Orbiter image analyses, Dao/Niger and Harmakhis Valles were interpreted to be collapsed regions of volcanic and sedimentary plains that were eroded by a combination of surface and subsurface flow [Baker, 1982, Squyres et al., 1987; Crown et al., 1992]. They are characterized by relatively steep-walled depressions, zones of subsided plains, and prominent central canyons whose walls, in high-resolution images, display gullies with associated depositional aprons that cover parts of the canyon floor (Figure 3). At Viking scales, evidence for surface flow consisted of scour zones and areas where lineations parallel to canyon walls were evident. MOC and THEMIS data reveal that the apparent scour zones are typically the result of small-scale collapse and that floor lineations occur in debris masses covering canyon floors; neither are fluvial in origin. A depositional lobe extending onto the basin floor from the terminus of Dao Vallis has also been noted [e.g., Crown et al., 1992]. [18] Reull Vallis is a morphologically diverse canyon system, consisting of three main segments (Figure 5). Reull 9of19

10 dominated by collapse and sapping [Bleamaster and Crown, 2004; Crown et al., 2004a] (Figure 8). Surface runoff may only have occurred locally or in stages now erased by subsequent collapse. MOLA DEMs clearly show significant subsidence (up to 500 m) of plains along the paths of Dao and Harmakhis Valles, confirming earlier interpretations of Viking data and supporting the existence of pathways in the subsurface (Figure 9). High-resolution images suggest that sapping is a significant process at a range of scales, with the development of sapping networks from fractured surface materials, scarp retreat, and capture of local topography (Figure 8). Initial classifications of the circum-hellas canyons as outflow channels imply, in a general way, a contribution of flooding in their development. Current analyses of topography and small-scale surface morphology are not consistent with sustained surface runoff over significant lengths, but cannot rule out catastrophic flooding in earlier stages of canyon formation. Viking, THEMIS, and MOC images show small channels that may delineate local scouring from surface flow, which may be promoted by the Figure 6. MOC wide-angle image R (image width = km; image center = S, E) showing a group of parallel, elongate, sinuous channels on the rim of Hellas basin northwest of Dao Vallis. These channels are typical of features extending across plains in the eastern Hellas region that define a zone of significant drainage into the basin. Vallis extends from the ridged plains of Hesperia Planum (N-S orientation) through resurfaced highlands in Promethei Terra (NE-SW orientation) and then through smooth and channeled plains (SE-NW orientation) where it ends abruptly, adjacent to the upper reaches of Harmakhis Vallis [Crown and Mest, 1997; Mest and Crown, 2001, 2002, 2003]. Reull and Harmakhis Valles may have been connected at one time, but a large debris apron obscures their intersection. Subsequent development of the canyons may have also altered an earlier surface or subsurface connection, as their elevations in adjacent segments near the intersection point are currently offset by 2 km. The second and third segments of Reull Vallis exhibit numerous morphologic similarities to Dao and Harmakhis Vallis, but its upper reaches are quite different, exhibiting streamlined islands, scour marks, narrow, sinuous channels, and lateral terraces or benches (Figure 7). These are all evidence for a fluvial stage or zone, which does not occur or is not preserved at Dao or Harmakhis Vallis. The emplacement and erosion of a sequence of plains adjacent to Reull Vallis (i.e., smooth and channeled plains in Figure 5) may be directly tied to flooding associated with this canyon system [Mest and Crown, 2001, 2002, 2003]. Exhumed craters to the south of Reull Vallis, identified by their morphology and almost completely infilled interiors, suggest several hundred meters of material has been removed [Crown and Mest, 2001]. [19] Recent studies using MOC and THEMIS images in combination with MOLA topographic data suggest that the lateral and vertical growth of Dao and Harmakhis Valles are Figure 7. Evidence for fluvial processes along Reull Vallis. (a) Viking Orbiter MDIM 2.1 image mosaic of part of Martian 1:5M quadrangle MC-28 showing upper part of middle segment of Reull Vallis with sinuous path, potential scour marks (s), and lateral terraces or benches (t) as the canyon cuts through an impact crater. Figure is 65 km across. (b) Part of THEMIS VIS image V (image width = 17.4 km, image center = S, E) showing part of upper segment of Reull Vallis exhibiting sinuous path, evidence for deflection of flow around small crater at center, and layering or terracing in canyon walls. Downstream direction is toward image bottom in both cases. 10 of 19

11 Figure 8 11 of 19

12 Figure 9. Viking Orbiter MDIM 2.1 image mosaic of Hadriaca Patera and the upper reaches of the Dao/Niger Vallis system (center); lines and labels mark locations of topographic profiles derived from MOLA DEM (128 pixels/degree), which are shown at bottom of figure. Parts of THEMIS IR images (a) I (image center = S, E) and (b) I (image center = S, E) (image widths = 30.7 km) provide detailed views of plains in early stages of modification as part of canyon development; arrows show approximate local boundaries of Dao/Niger Vallis. Topographic profiles show high degree of variability in canyon segment widths and depths and clearly indicate significant subsidence in plains along path of canyon system. irregular topography resulting from collapse and the presence of subsurface water or ice. [20] Geomorphic and topographic analyses suggest that a generalized sequence of events has occurred along various segments of Dao and Harmakhis Valles at different scales and at different times [Crown et al., 2004a, 2004b] (see Figures 3b, 8, and 9). Although likely the result of a complex formational history involving different processes and subject to local controls such as subsurface stratigraphy and geologic structure, the currently preserved morphology of the canyons suggests a general evolution consisting of the following idealized stages: (1) withdrawal of underlying Figure 8. High-resolution images show the effects of collapse and sapping in the development of Dao and Harmakhis Valles at a range of scales. (a) Part of THEMIS IR image I (image width = 30.4 km, image center = S, E) showing collapsed and slumped plains along Dao Vallis; note fractures in plains parallel to the main canyon walls. (b) Part of THEMIS VIS image V (image width = 17.4 km, image center = 37.5 S, E) of curvilinear chain of pits which forms a tributary to Dao Vallis as well as aligned, tear-drop-shaped pits to the south. (c) Part of THEMIS IR image I (image width = 30.7 km, image center = S, E) of Niger Vallis near junction with Dao Vallis showing plains in various stages of disruption; note development of sapping networks from fractures in plains, an example of which is shown in detail on inset image (part of MOC M , image width = 2.85 km, image center = S, E). (d) Part of THEMIS IR image I (image width = 30.4 km, image center = S, 92.5 E) showing hilly remnant of collapsed plains in Niger Vallis and fractures and pits in surrounding highstanding plains surfaces. 12 of 19

13 materials, and (5) resurfacing of canyon floors, resulting in various combinations of remnant hills and mounds of collapsed plains, mass-wasted deposits extending from canyon walls, and lineated valley fill deposits extending down-canyon. The morphologies currently observed and their distribution suggests several possible scenarios for overall canyon development, including a regional episode of subsidence or temporally distinct episodes of initial subsidence and collapse in different places. Either of these would be followed by variable progression of subsequent morphologic stages at different times in different places depending on the local environment. The morphologic differences between Dao and Harmakhis Valles relative to Reull Vallis can be attributed to different stages of maturity and/or preservation, which may be partly controlled by formation in different surface materials with potentially different volatile abundances and surface environments due to distance from Hellas. The canyon systems may have evolved from elongate channels similar to those observed on plains in eastern Hellas, with a limited contribution of catastrophic flooding associated with Dao and Harmakhis Valles. Figure 10. Part of MOC image M (image width = 2.80 km, image center = S, E) of deepest part of Hellas basin in western Hellas Planitia showing polygonal depressions in layered sediments. Moore and Wilhelms [2001] referred to this as honeycomb material, which they interpreted to form as a result of material being deformed as ice blocks settled on a muddy substrate. support causing subsidence of plains, (2) continued collapse and surface fracturing of subsided plains resulting in the formation of numerous subsided blocks, (3) resultant irregular topography facilitates sapping, with growth of sapping valleys along fractures and over subsurface voids, (4) collapse of canyon walls and localized erosion of wall 4.4. Hellas Planitia [21] The deposits of Hellas Planitia which cover the floor of Hellas basin, as well as those that form the channeled plains to the east, have been attributed generally in previous studies to a combination (typically unspecified) of volcanic, sedimentary, and aeolian materials dissected and redistributed by fluvial activity and sculpted by the wind [Potter, 1976; Greeley and Guest, 1987; Crown et al., 1992; Moore and Edgett, 1993; Tanaka and Leonard, 1995; Price, 1998; Leonard and Tanaka, 2001]. Additional diverse interpretations of the geology of the basin interior include widespread glaciation [Kargel and Strom, 1992; see also Baker et al., 1991] and, most recently, evidence in high-resolution images for ice-covered lakes [Moore and Wilhelms, 2001]. The eastern part of Hellas Planitia is thought to be Hesperian in age and post-dates formation of the central plateau; braided channels in eastern Hellas Planitia are associated with Dao and Harmakhis Valles. In Viking Orbiter images, wrinkle ridges, mesas, and channels are evident in the basin interior. Layering is expressed in local scarps and benches [Moore and Wilhelms, 2001]. Recent work suggests layered, indurated finegrained materials are abundant in the basin interior; Tanaka and Leonard [1995] favor an aeolian dominated environment, whereas Moore and Wilhelms [2001] favor a sequence of ice-covered lakes. [22] The possibility of past lacustrine activity in the Hellas basin interior is significant, given interest in the potential development of life on Mars, the identification of layered deposits in various localities on the Martian surface [Malin and Edgett, 2000b], and debates regarding the possibility of oceans and paleolakes covering the northern lowlands and other parts of the planet [e.g., Parker et al., 1989, 1993; Scott et al., 1995; Malin and Edgett, 1999]. Moore and Wilhelms [2001] suggested that Hellas basin contained ice-covered lakes from observations of MOC images, primarily of the deepest parts (below 6900 m) of western Hellas Planitia. Here, layered sediments and polygonal cavities were attributed to a lacustrine environ- 13 of 19

14 Figure 11. Parts of MOC images (a) E (image width = 2.98 km, image center = S, E) and (b) E (image width = 3.07 km, image center = S, E) show intricate curvilinear patterns in finely layered deposits that are concentrated along the scarp defining the eastern edge of Hellas basin near Dao and Harmakhis Valles (see Figure 1c). Layered outcrops are exposed in locally highstanding mesas, knobs, and surfaces, typically near 5700 m in elevation. ment (Figure 10). Viking and MOLA data were used to propose additional shorelines at 5800 m and 3100 m [Moore and Wilhelms, 2001]. Moore and Wilhelms [2001] noted a consistency of landforms and deposit contacts around the basin at 5800 m in elevation as well as a change in the appearance of Dao and Harmakhis Valles, which, below 5800 m, they state resemble terrestrial marine channels associated with large rivers. Kargel and Strom [1992] attributed this morphologic change to a channel flowing into a lake. Along the 3100 m contact, Moore and Wilhelms [2001] described a change from more rugged highland terrain to smoother surfaces with rimless or low-rimmed craters. They mapped a layered mantling material in an annulus around much of the basin periphery extending from approximately 3100 m inward to approximately 5800 m, which they attributed to episodic deposition of potentially lacustrine sediment due to variations in climate. Tanaka and Leonard [1995] noted dissection of Hellas rim materials by numerous small channels in this elevation zone. [23] After identifying intricate curvilinear patterns in finely layered deposits in the plains near where Dao Vallis enters Hellas Planitia (Figure 11), Crown et al. [2004b, 2005b] systematically searched MOC images to examine the distribution of these layered outcrops in eastern Hellas. Layered outcrops are concentrated along the scarp that defines the eastern edge of the basin near Dao and Harmakhis Valles (Figure 1c) and are exposed in locally highstanding mesas, knobs, and surfaces, typically near 5700 m in elevation. MOLA topographic contours are deflected toward the basin interior in precisely the area exhibiting the layered outcrops, delineating a substantial depositional shelf in eastern Hellas (Figure 1). This shelf could be a preferential accumulation zone for atmospheric volatiles and/or the edge of a volatile-rich deposit associated with the basin interior, such as at the margin of an ice-covered lake. [24] The numerous, typically singular channels extending toward Hellas Planitia suggest significant drainage into Hellas from the north and northeast along the depositional shelf (Figure 5). We suggest that Dao and Harmakhis Valles once may have been similar to these narrow, elongate channels and grew by collapse of volatile-rich shelf sediments. Morphologic evidence that collapse has dominated at least the later stages of canyon formation and that large canyon segments may form from collapse around small linear features support this hypothesis (see section 4.3). The upper reaches of Dao and Harmakhis Valles as well as the middle and distal segments of Reull Vallis (i.e., the nonfluvial segments) all appear bound by the 1800 m elevation contour. Over the 1800 to 5800 m interval, the topographic expression of the highlands changes significantly (Figure 1): (1) Below 5800 m large impact craters are not apparent and the topography is dominated by the erosional morphology of the basin interior deposits. Comparison of radial profiles (Figure 1d) shows that the basin floor has relatively uniform characteristics. (2) Above 1800 m typical cratered terrain and intercrater plains surround the basin. Profiles show that preserved highland remnants and large craters dominate local topography. Smooth zones are present in areas of significant resurfacing (e.g., the south rim volcanics in profile 6, intercrater plains in profiles 8 and 9). Comparison of profiles 2 with 8 and 3 with 9 show the topographic asymmetry of the basin over this elevation interval, with smooth and channeled plains (profiles 2 and 3) at elevations 2 km lower in eastern Hellas adjacent to the canyon systems. (3) Between 5800 and 1800 m large craters are present (and can dominate local topography), but they are fewer in number and appear degraded and/or buried (Figure 1b). The prominent canyons of Dao and Reull Valles show more than 2 km of relief relative to adjacent plains. Similarities in profiles 1 and 7 and 5 and 11 coupled with significant topographic offset and local variability in profiles 2, 3, and 4 relative to 8, 9, and 10 illustrate that the large-scale basin asymmetry is correlated with the shelf in eastern Hellas. [25] Topographic data, cratered terrain preservation, canyon morphology, and the distribution of layered outcrops are all consistent with ancient lakes in Hellas, perhaps to a larger extent than envisioned previously. Our observations as summarized here build on and advance the analyses of Moore and Wilhelms [2001], who evaluated potential volcanic, aeolian, and glacial contributions in formulating their lacustrine hypothesis. Our identification of finely layered exposures along and above the scarp at 5800 m supports the morphologic changes noted at the termini of Dao and Harmakhis Valles and the deposit contacts mapped by 14 of 19

15 Figure 12. (a) Part of THEMIS IR image I (image width = 30.7 km, image center = S, E) showing lobate debris aprons extending from highland massifs north of Reull Vallis; note (at top of image) debris appears to flow through valley in massif and spread at massif base. (b) MOC image E (image width = 2.87 km, image center = S, E) showing wall and floor of Dao Vallis. Canyon floor is covered by lineated valley fill with lineations parallel to canyon walls. Canyon walls show Martian gully systems with wellincised channels. Moore and Wilhelms [2001]. The 5800 m elevation clearly marks the boundary between the basin interior deposits and the rim and is a likely paleolake margin. Our observations suggest a zone on the rim between 5800 and 1800 m that is distinct morphologically from the basin interior deposits and the surrounding highland terrains. This is based on evidence for collapse along the full extents of Dao and Harmakhis Vallis, the change in morphology along Reull Vallis, preservation of highland craters, and the channeled shelf in eastern Hellas. We do not see clear evidence for shorelines within the 5800 to 1800 m zone; however, given the preservation of highland craters in particular, the 1800 m elevation would appear to be a maximum highstand for a Hellas paleolake, effectively extending (from 3100 m) Moore and Wilhelms largest Hellas possible paleolake further into the highlands. Although obscured by Hesperian and Amazonian units, an even larger fluvial/lacustrine system may have once connected Hellas basin with the ancient depression in Hesperia Planum and perhaps the northern lowlands as well. This may explain the location of Reull Vallis upper reaches in Hesperia Planum as well as other recently identified channels extending south from Hesperia Planum [Crown and Mest, 2004]. Ages for possible Hellas paleolakes are poorly constrained at present given the degree of erosion of the deposits, which hinders the traditional crater counting approach. Given the primarily Hesperian ages of units in the basin interior and along the rim and the smoothing of highland topography associated with Noachian cratered terrains surrounding the basin, any large paleolakes would have presumably been Noachian in age and followed creation of the cratered highlands, although more localized lakes on the basin floor in more recent time periods cannot be ruled out. We do not offer a specific interpretation here regarding the origin of the finely layered exposures observed within the depositional shelf in eastern Hellas and consider their mode of deposition (subaerial versus subaqueous) an open question. Their layered nature and distribution suggests they are related to a Hellas-centered phenomenon; likely candidates include (1) lacustrine sediments deposited by a Hellas paleolake or series of paleolakes, (2) fluvial sediments deposited by subaerial and/or subaqueous drainage into Hellas from the east, and/or (3) sequences of aeolian sediments deposited in eastern Hellas, perhaps related to cycles of deposition and removal of ice-rich deposits on the basin floor. It is also likely that complex sequences of volatile and sediment redistribution occurred within the Hellas region, with prominent effects on the east rim. Further work is needed to address the specific nature, duration, and extent of lacustrine activity in the Hellas region. This study shows the significance of specific evaluations of the rim geology of Hellas, both within the 5800 to 1800 m zone and in the surrounding highlands, where crater interiors appear to have recorded episodes of erosion and deposition [e.g., Mest and Crown, 2005b; Moore and Howard, 2005a, 2005b]. It is clear that the nature of volatile-driven activity in this region is a critical parameter in understanding Martian climates and their spatial and temporal variability Recent Ice-Related Features [26] The regional geology of eastern Hellas is overprinted with a suite of more recent features indicative of contained ice, melting of ice, or its release to the atmosphere (Figure 12). Lobate debris aprons with lineated surfaces showing viscous flow are observed extending from canyon walls, highland massifs, and crater rims [Squyres, 1979; Pierce and Crown, 2003; Baratoux et al., 2002; van Gasselt et al., 2005]. These features may be analogous to rock glaciers or debris-covered glaciers, although their precise emplacement mechanisms remain unknown. Lineated valley fill deposits are observed in various segments of the Dao, Harmakhis, and Reull Valles systems with regions of both down-canyon and acrosscanyon flow recognized [Bleamaster and Crown, 2004; Crown et al., 2004a]. A variety of smaller ice-rich flow features and their inferred remnants have also been recognized in similar settings, with interpretations including glacial flow and ice-rich mass-wasting [Arfstrom, 2003; Howard, 2003; Milliken et al., 2003; Crown et al., 2004a; Arfstrom and Hartmann, 2005; Berman et al., 2005]. These are often found in association with the ice-cemented mantling deposits characteristic of the Martian midlatitudes [Mustard et al., 2001] and prominent gully systems [Malin and Edgett, 2000a]. Recent work demonstrates that within Dao and Harmakhis Valles, the morphology of wall mantling deposits and exposed gullies, and their spatial 15 of 19