Notice: Wiley Online Library will be unavailable on Saturday 27th February from 09:00-14:00 GMT / 04:00-09:00 EST / 17:00-22:00 SGT for essential maintenance. Apologies for the inconvenience.
 Geologic mapping of the northern plains of Mars, based on Mars Orbiter Laser Altimeter topography and Viking and Mars Orbiter Camera images, reveals new insights into geologic processes and events in this region during the Hesperian and Amazonian Periods. We propose four successive stages of lowland resurfacing likely related to the activity of near-surface volatiles commencing at the highland-lowland boundary (HLB) and progressing to lower topographic levels as follows (highest elevations indicated): Stage 1, upper boundary plains, Early Hesperian, <−2.0 to −2.9 km; Stage 2, lower boundary plains and outflow channel dissection, Late Hesperian, <−2.7 to −4.0 km; Stage 3, Vastitas Borealis Formation (VBF) surface, Late Hesperian to Early Amazonian, <−3.1 to −4.1 km; and Stage 4, local chaos zones, Early Amazonian, <−3.8 to −5.0 km. At Acidalia Mensa, Stage 2 and 3 levels may be lower (<−4.4 and −4.8 km, respectively). Contractional ridges form the dominant structure in the plains and developed from near the end of the Early Hesperian to the Early Amazonian. Geomorphic evidence for a northern-plains-filling ocean during Stage 2 is absent because one did not form or its evidence was destroyed by Stage 3 resurfacing. Remnants of possible Amazonian dust mantles occur on top of the VBF. The north polar layered deposits appear to be made up of an up to kilometer-thick lower sequence of sandy layers Early to Middle Amazonian in age overlain by Late Amazonian ice-rich dust layers; both units appear to have outliers, suggesting that they once were more extensive.
If you can't find a tool you're looking for, please click the link at the top of the page to "Go to old article view". Alternatively, view our Knowledge Base articles for additional help. Your feedback is important to us, so please let us know if you have comments or ideas for improvement.
 The surface of Mars broadly consists of an ancient highland-lowland topographic dichotomy, punctuated by huge impact scars and volcanic regions. The largest region of lowlands (below −2000 to −3000 m elevation) on Mars roughly centers on the north pole and covers one third of the planet. Origin of the northern lowlands has been attributed to impacts or tectonism [e.g., Wise et al., 1979; Wilhelms and Squyres, 1984; Frey and Schultz, 1988], but subsequent resurfacing appears to have removed any clear morphologic evidence for its earliest origin. Greatly improved topographic mapping of Mars by the Mars Orbiter Laser Altimeter (MOLA) [Smith et al., 1999] reveals that three broad, enclosed basins occur within the lowlands—the Borealis, Utopia, and Isidis basins (Figure 1).
 In this study, we report on our initial results obtained from geologic mapping of the northern plains using Mars Global Surveyor data. We propose a major shift in interpretation of the geologic character and history of the northern plains from plains-wide volcanism and/or sedimentary infilling such as resulting from deposition within an ocean, to a history dominated by broad, episodic erosion and modification of volatile-rich material and local to regional sedimentary and volcanic infilling. This new perspective has many implications regarding the nature of northern plains materials, evolution of regional to global liquid volatile (H2O and perhaps CO2) tables, and development and preservation of tectonic structure. We also find that contractional ridge structures have a younger age than their Late Noachian to Early Hesperian highland counterparts [Tanaka et al., 1991; Watters, 1993; Dohm et al., 2001b]. Finally, we observe numerous potential polar outliers documenting possible former extents of dust mantles and polar layered deposits.
2. Geology of the Northern Plains
 Our geologic map of the northern plains of Mars (Figure 2) relies mostly on a MOLA digital elevation model (DEM) at 1/128° resolution (about 500 m/pixel) compiled from hundreds of millions of nadir-pointing laser altimeter shots (above ∼70°N, we use a 1/256° resolution DEM). We also consulted derivative shaded relief, slope, and aspect (slope direction) maps. These data are of sufficient resolution for fairly consistent topographic rendering of landforms without the complexities of atmospheric haze and albedo variations inherent in spacecraft images. MOLA data density and thus DEM accuracy increases with latitude, except above 87°N., where only limited, off-nadir data were collected due to the orbital path of the spacecraft. Locally, MOC context (∼250 m/pixel) and narrow angle images (>1.4 m/pixel) and Viking images (tens to >100 m/pixel) provided additional albedo and morphologic information used in the mapping (e.g., residual polar ice was mapped strictly by its high albedo).
 Our units are shown in the geologic map (Figure 2), and summaries of their physical characteristics, stratigraphic information, and crater densities and distributions are provided in Tables 1 and 2 and Figure 3. The units are grouped geographically into highland, lobate, plains, and polar materials (Figure 2b). The following stratigraphy (from oldest to youngest units) includes some significant modifications to previous mapping [Scott et al., 1987].
Table 1. Characteristics of Map Units in the Northern Plains of Marsa
Marginal member of the Vastitas Borealis Formation
47 ± 10
Hummocky member of the Vastitas Borealis Formation
73 ± 2
Boundary plains unit 2
103 ± 5
99 ± 29
Older chaos material
126 ± 23
88 ± 6
Older lobate material
104 ± 4
Boundary plains unit 1
177 ± 9
223 ± 9
Highland material, undivided
393 ± 7
 The crater-density data in Table 2 must be applied with caution. The more significant considerations include the following: (1) The craters were located using Viking 1:2,000,000-scale photomosaics produced both digitally and by hand, resulting in misregistrations with our MOLA-based geologic map commonly on order of 10 to 20 km. Thus craters occurring near contacts may be included in the incorrect unit. This will produce the largest errors for units having small areas, irregular contacts, and low crater densities. In particular, crater counts for the polar layered deposits (units Apl1–2) were most sensitive to this error, and so we performed our own counts for those units. (2) Embayed craters are included in counts of their surrounding units, making the count too high for those units. Thinner units having lower superposed crater densities overlying more densely cratered surfaces have the greatest error. Units most affected by this include smooth lobate unit 2 (unit Als2), whose count is perhaps entirely made up of embayed craters, coarse lobate material (unit Alc), apron material (unit Aa), and younger chaos material (unit Act). (3) A unit's crater retention age may not reflect the unit's emplacement age because of various possible resurfacing scenarios including crater obliteration and deep burial and later exhumation.
Highland material, undivided (unit HNu; Figures 4 and 5) forms the rugged ancient highland terrain surrounding much of the northern lowlands, as well as part of Acidalia Mensa. The unit is marked in places by large craters, hills, valley networks, isolated depressions, and linear and sinuous ridges, scarps, and troughs. Interpretation: Mixture of volcanic and sedimentary material, gardened by impacts. A long history of resurfacing and deformation accounts for secondary erosional, depositional, and tectonic structures.
 The knobby unit (unit HNk; Figure 4) consists of knobs and mesas of highland rocks and interposed slope and plains materials. The unit forms much of the highland-lowland boundary (HLB), which underwent degradation starting at the end of the Noachian and extending into the Early Hesperian [Tanaka, 1986; Frey et al., 1988; Maxwell and McGill, 1988; McGill, 2000]. Interpretation: Highland rocks fractured and collapsed due to basal sapping of volatiles and mass-wasted debris.
 The oldest exposed northern plains unit, boundary plains unit 1 (unit Hb1, Figure 4), occurs adjacent to older highland and plateau materials including the knobby unit along the HLB of Utopia, Amazonis, Chryse, and Acidalia Planitiae and Arabia Terra. The unit appears to be missing along some parts of the HLB including northwestern Arabia Terra (Figure 5), Tempe Terra, and Isidis Planitia. The unit slopes gently downward away from the HLB. Irregular depressions and scarps many tens of kilometers in length and 200–300 m in relief locally give the unit a broadly scalloped appearance (Figure 4). The superposed crater density (Table 2) indicates an Early Hesperian age. Interpretation: Because this unit mostly occurs just below the knobby unit (unit HNk) and highland material (unit HNu), the unit likely results in most places from collapse, comminution, erosion, transport, and deposition of highland clastic material, perhaps by volatile-assisted slope processes. Previously, plains materials in the northern lowlands have been generally viewed as volcanic plains [e.g., Greeley and Spudis, 1981; Scott et al., 1987]; also, the VBF has been interpreted to be sediments underlain by Early Hesperian volcanic ridged plains [Head et al., 2001]. However, we find no clear evidence for volcanic vents or flows in boundary plains unit 1.
Older lobate material (unit AHl) consists of extensive tongue-shaped flows and forms the gently sloping shields of the Elysium and Tharsis rises and Syrtis Major Planum (Figure 1). Dissected outcrops of the unit exhibit layered sequences. The flows predate the Vastitas Borealis Formation and both overlie and underlie boundary plains unit 2 (unit Hb2). Locally, the unit is marked by various vent and tectonic structures. The unit has a Late Hesperian mean crater age (Table 2). Interpretation: Lava flows and possibly other volcanic materials. Ages of flows range from Early Hesperian to Early Amazonian [Scott et al., 1987].
Channel material (unit Hch) covers the floors of outflow channels tens to hundreds of kilometers wide and hundreds of kilometers long crossing highland and boundary plains units in Xanthe Terra and Chryse Planitia (Figure 6). Channel dissection ceased near the end of the Hesperian (Table 2). The northern extent of the channel material has been modified by the VBF, and in southern Chryse Planitia, some channel areas appear partly disintegrated and transitional with older chaos material (unit Hct). Interpretation: Fluvial sediments and debris flows sourced by catastrophic discharges from older chaos material (unit Hct) [e.g., Baker and Milton, 1974; Nummedal and Prior, 1981]; exposed deposits generally Late Hesperian, but some older, buried deposits likely [e.g., Rotto and Tanaka, 1995; Tanaka, 1997; Nelson and Greeley, 1999]. Chryse outflow-channel sediments may have been restricted to Borealis basin, given (1) the paucity of ghost craters (which would have been buried by the younger sediments) in the channel-ward part of the basin (Figure 3), and (2) the lack of an obvious outflow-channel connection in the saddle between Borealis and Utopia basins (Figure 7).
Older chaos material (unit Hct) forms irregular depressions tens to hundreds of kilometers across and hundreds of meters to more than a kilometer deep filled with irregular knobs and mesas at the heads of Chryse outflow channels (unit Hch). Outcrops of the unit postdate adjacent highland material (unit HNu) and both predate and postdate various outflow channels [Rotto and Tanaka, 1995]. Some of the older chaos material modifies the youngest outflow channels, Simud and Tius Valles. Interpretation: Highland and plains materials disrupted and eroded by volatile discharge and subsequent collapse in association with outflow-channel dissection [e.g., Sharp, 1973; Carr, 1979; Hoffman, 2000].
 Outcropping along the northwest periphery of the Alba Patera shield, the Scandia unit (unit Hs, Figure 8) forms the discontinuous plateaus and knobs of Scandia Colles that range from 20 to 200 m in height. Some northern outcrops of the unit are broken up by depressions filled with younger chaos material (unit Act). In one case, a plateau made up of the Scandia unit may be armored by superposed crater ejecta (Figure 8). In Viking images, broad, thin lobate flows form an apron surrounding the southern margin of the unit. Interpretation: Because of the modification of the unit, its relative age is highly uncertain. Previously, the unit was interpreted to be remnants of ancient Noachian material because of degraded crater forms within the unit [Scott et al., 1987]. However, we suggest that the unit is Hesperian, because it doesn't have the more rugged relief and high crater density typical of the Noachian knobby unit. The degraded, planar nature of the unit suggests that it may have originated as either polar dust deposits and/or aeolian deposits of fines produced by Alba Patera volcanism [cf. Tanaka, 2000] or by erosion of Noachian highland material (and thus the Scandia unit might be a fine-grained facies of the boundary plains units (units Hb1 and Hb2)). Mass wasting of the unit may account for its partial lobate apron.
Boundary plains unit 2 (unit Hb2, Figures 4–7) is chiefly identified as smooth plains material marked by wrinkle ridges that occurs at the base of the lower, rugged part of boundary plains unit 1 (unit Hb1) in Utopia, Chryse, Acidalia, and Amazonis Planitiae. Elsewhere, the unit embays highland material (unit HNu) in northwestern Arabia Terra and southern Isidis Planitia and the older lobate material (unit AHl) of Alba Patera. The coalesced, fretted trough floors of Deuteronilus Mensae, while partly buried by apron material (unit Aa), forms a continuous surface with the unit Hb2 surface in northwestern Arabia Terra (Figure 5) and has a density of craters, both partly buried and superposed, similar to that of unit Hb2 (Table 2); thus the fretted floors may consist of unit Hb2. In turn, the unit is overlain by flows from Elysium and Tharsis (units AHl and Als1–2) in Arcadia and Amazonis Planitiae, cut by outflow channels in Chryse Planitia (unit Hch; Figure 6), and modified by the VBF (Figures 4–6). The unit's crater density indicates a Late Hesperian age (Table 2). The unit slopes gently away from the highland margin and appears relatively smooth at subkilometer scale [Kreslavsky and Head, 2000] but somewhat rougher at longer length scales (Figure 4). The unit also may form the smooth, lower tier of Acidalia Mensa, embaying highland material and rising tens of meters above the adjacent plains (Figure 9). Ghost craters are common in some outcrops of the unit (Figure 3). Interpretation: Where the unit occurs adjacent to boundary plains unit 1, it likely results from reworking of that member by collapse, erosion, transport, and deposition of clastic material from scalloped terrain. Elsewhere, the unit also may be made up of sediments eroded from highland materials (units HNu and HNk) and perhaps some volcanic flows.
 The Vastitas Borealis Formation (VBF) covers the majority of the Borealis, Utopia, and Isidis basin floors (Figure 2a). Previously, the VBF was divided into grooved, ridged, mottled, and knobby members [Scott et al., 1987]. These members largely reflect differences in surface character rather than distinctions in lithology or age. Parker et al. [1989, 1993] mapped a plains unit (outlined by their Contact 2) that in places coincides with the VBF contact. Because the more widespread, subtler morphologic features characterizing the VBF commonly are more detectable with high-resolution (1/128°) MOLA gridded topography than with Viking images, we have been able to map VBF “contacts” more completely than previously possible. We substitute “margin” for “contact” in describing the edge of the VBF, because we interpret that the unit is not stratigraphic but results from in situ reworking of preexisting materials that has destroyed pre-VBF landforms including craters and depositional, erosional, and tectonic landforms. Parker , who believes that the same feature represents a possible shoreline and not necessarily a material-unit boundary, also recognized a semantics problem with using “contact.”
 The hummocky member (unit AHvh) accounts for the majority of the VBF. This unit is marked by numerous low hillocks and arcuate ridges (in places organized into thumbprint terrain); by patches of grooves hundreds of meters wide forming networks of polygons several kilometers across; by local systems of wrinkle-ridge-like, mostly asymmetric, low ridges; by dozens of unevenly scattered circular depressions (ghost craters) up to tens of kilometers in diameter (Figure 3); and by many superposed, relatively high-albedo impact craters. In the MOLA data, the hummocky morphology appears to be pervasive along the unit margin (e.g., Figure 4), whereas the grooved and mottled appearances delineated by Scott et al.  and others are additional or dominant features within the interior of the unit. MOC images show that thumbprint ridges generally are made up of continuous and discontinuous chains of pitted domes; also, the domes locally appear unorganized (Figure 10). The unit has an overall Late Hesperian to Early Amazonian crater density (Table 2). The unit is surrounded by Hesperian boundary plains units and channel material and overlain by Amazonian polar and lobate materials. In the Deuteronilus region, the unit's margin occurs along a subtle scarp (Figure 5). Interpretation: We propose that the VBF is a mixture of undifferentiated Noachian to Hesperian materials and local outflow-channel sediments that have been highly and pervasively altered by in situ permafrost-related processes such as cryoturbation, desiccation, and thermokarst involving near-surface deformation and mass-movements aided by subsurface volatiles as previous workers have suggested for northern plains landforms [e.g., Carr and Schaber, 1977; Lucchitta, 1981; Rossbacher and Judson, 1981; McGill and Hills, 1992; Kargel et al., 1995]. The various surface textures and bright crater ejecta may be more an indication of locally varying physical properties, volatile content, and/or subsequent modification than lithology or pervasive structural character, so we refrain from mapping additional units where these features occur. Previously, Tanaka  assigned the VBF as the Late Hesperian stratigraphic referent, from which crater-density boundaries were derived to define chronologic boundaries for the end of the Hesperian. However, we feel that crater densities provide a less ambiguous means of defining relative age, considering that many regional geologic units on Mars including the VBF may be significantly time transgressive. Thus, we rely on the N(5) = 67 crater density to define the Hesperian/Amazonian boundary [Tanaka, 1986], which, considering unit AHvh's crater density (Table 2) and that development of the unit may have been lengthy in time, places this unit within the Late Hesperian and possibly the Early Amazonian.
 The marginal member (unit AHvm) of the VBF forms low plateaus dissected by linear to sinuous, 2- to 5-km-wide troughs that extend for tens of kilometers and resemble the burrows of ant colonies (hence, we informally call this “ant-farm terrain”; see Figure 11). Locally, the troughs form rectilinear or irregular networks and some terminate at and/or connect to large knobs. The plateaus are bounded by linear, lobate, or sinuous scarps similar in trace to the troughs within the plateaus. In places, the troughs have angular corners. In higher resolution Viking and MOC images, low, narrow medial ridges occur within the trough floors (Figures 11 and 12). Also, the troughs appear to dissect ridges of boundary plains unit 2 (Figure 11) and streamlined islands in channel material (Figure 12). The troughs of the marginal member dissect boundary plains units 1 and 2 and channel material. Interpretation: The unit's unusual morphologic character defies comparison with common terrestrial analogs such as glacial, periglacial, lacustrine, or tectonic terrains, but shows evidence for both brittle and viscous deformation that may be related to the presence of near-surface volatiles [e.g., Scott, 1982; Tanaka, 1997]. The unit's crater density (Table 2) places it in the Early Amazonian, but its gradational appearance with the hummocky member of the VBF (unit AHvh; Figure 11) suggests that it formed contemporaneously with unit AHvh, perhaps as a late-stage, incipient form of degradation at the margin of the VBF.
 The polar layered deposits 1 (unit Apl1) form the base of part of Planum Boreum and possibly highstanding knobs and mesas south of Chasma Boreale and underlie the evenly bedded polar layered deposits 2 (unit Apl2; Figure 13). Unit Apl1 may be up to a kilometer thick along the margins of Chasma Boreale and thins out away from there; it presumably underlies much of Planum Boreum, particularly near Olympia Planitia. MOC images and MOLA data reveal that this unit has irregular bedding, locally steep scarps, a dark color, and some large, superposed craters (Figure 13) [Edgett and Malin, 2000; Kolb and Tanaka, 2001; Byrne and Murray, 2002]. Previously we thought that some of the layered deposits on the margin of Planum Boreum opposite Chasma Boreale were underlain by the ejecta of a 24-km-diameter crater and thus consisted of unit Apl1 (Figure 14, inset), but a higher-resolution DEM (Figure 14) shows the opposite relation and thus the deposits are more likely the upper unit (unit Apl2). Interpretation: Possibly an eroded sand sea [Byrne and Murray, 2002] or modified polar layered deposits. Formerly, unit Apl1 likely was more extensive in places, perhaps accounting for some of the material underlying the dunes of Olympia Planitia [Zuber et al., 1998; Fishbaugh and Head, 2000] and the pitted and flat-topped knobs in the plains surrounding Planum Boreum that rise to similar elevations of unit Apl1 at the margin of Planum Boreum (Figure 13). Alternatively, those knobs have been interpreted to be volcanic cones [Hodges and Moore, 1994; Sakimoto et al., 2001].
Younger chaos material (unit Act) occurs within depressions tens to hundreds of kilometers across and tens to a few hundreds of meters deep in the Acidalia Mensa, Cydonia, Isidis, and Scandia regions (Figures 9, 15, and 16). The unit locally includes polygonal fractures, knobs, and irregular scarps. Younger chaos material postdates channel material (unit Hch) and the VBF (unit AHvh), and may be coeval with tholi material (unit At). Interpretation: Mostly VBF and perhaps other materials that have been disrupted and eroded by volatile discharge and subsequent collapse.
Tholi material (unit At) forms circular to irregular domical hills and rugged complexes tens to a few hundred kilometers across and tens to hundreds of meters high south of Olympia Planitia (Figure 16). The larger complexes have interior and surrounding depressions of collapsed terrain (unit Act) tens to hundreds of meters deep and a few narrow sinuous ridges a couple kilometers wide and tens of kilometers long. Many of the hills are bounded by shallow moats. Interpretation: Lava, mud, or ice extrusions and/or pingo-like intrusions formed the domes and perhaps phreatic or cryoclastic eruptions and discharge-related collapse produced the depressions and moats. Alternatively, the deposits and associated depressions have been interpreted as glacial kames and kettles by Fishbaugh and Head [2000, 2001].
 The Medusae Fossae Formation (unit Am; MFF) forms sequences hundreds to more than a kilometer thick extending discontinuously for thousands of kilometers along the highland-lowland boundary between the Elysium rise and Olympus Mons. In many places, the MFF displays systems of linear ridges and grooves. Interpretation: Various origins for the MFF have been proposed, including poorly indurated volcanic ash or aeolian sediment, carved into yardangs by the wind [e.g., Scott and Tanaka, 1982; Scott et al., 1987].
Coarse lobate material (unit Alc) forms flows in Amazonis and Utopia Planitiae, in association with smooth lobate unit 1 (unit Als1, Figure 17). The proximal parts of flows in Amazonis Planitia appear to be buried by smooth lobate unit 1 and Olympus Mons aureole deposits (mapped as unit Aa). In Utopia Planitia, the unit emanates from mostly northwest-trending fissures and troughs in the northwest margin of the Elysium rise. The unit includes hummocky, pitted, and channeled surfaces. The flows typically are several tens of meters thick; in Utopia, the flows appear to embay and dissect ejecta blankets of large craters. Interpretation: Volcaniclastic flows formed from magma/volatile interactions; subsequent modification due to volatile escape and perhaps heating and activity of volatile-rich substrate [Christiansen, 1989; Skinner and Tanaka, 2001].
Smooth lobate unit 1 (unit Als1) consists of Early to Middle Amazonian tongue-shaped flows tens of kilometers wide that extend for hundreds of kilometers in Amazonis and Utopia Planitiae (Figures 17 and 18). The unit buries the VBF and other units of Noachian and Hesperian age. In eastern and southern Amazonis, the flows mostly appear to underlie the Olympus Mons aureole (unit Aa) and the Medusae Fossae Formation (unit Am). The flows are sparsely cratered in the MOLA DEM. The unit generally exhibits well-preserved flow morphologies. Interpretation: Lavas flows [Scott et al., 1987]; high effusion rates may account for their great extent across flat plains.
Apron material (unit Aa) includes smooth, sloping deposits along the base of highstanding scarps of Deuteronilus Mensae as well as the extremely rugged aureole materials of Olympus Mons. Interpretation: Deposits derived from ice-lubricated mass-wasting deformation processes [e.g., Tanaka, 1985; Mangold and Allemand, 2001].
Smooth lobate unit 2 (unit Als2) consists of Late Amazonian tongue-shaped flows tens of kilometers wide that extend for hundreds of kilometers across Elysium Planitia. The unit buries smooth lobate unit 1 (unit Als1) and other adjacent units. The unit locally embays and encircles dozens of impact craters a few kilometers and larger in diameter, indicating that the unit is probably tens to a few hundred meters thick in these areas. The flows are sparsely cratered in MOC images and may be less than 10 million years old [Hartmann and Neukum, 2001]. The unit mostly displays crisp, platy flow morphologies; in southwestern Amazonis Planitia, the flows contain sinuous channels bounded by rounded scarps. Interpretation: Mostly lava flows [Plescia, 1990]; high effusion rates may account for their great extent across flat plains. Near Cerberus Rupes, the unit may include young channel floors and deposits [Burr et al., 2002]. Distal material in Amazonis may also be debris flows [e.g., Rice et al., 2002], perhaps co-mingled with fluvial deposits [Scott et al., 1987; Fuller and Head, 2002].
Hummocky material (unit Ah) occurs in Borealis basin south of Chasma Boreale (Figure 13). The unit may be tens of meters thick, sufficient to bury and obscure the hummocky topography of the underlying VBF (unit AHvh). The unit may abut the ejecta ramparts of impact craters as seen in MOLA DEM's and thus may be uncratered at diameters of several kilometers and larger. Interpretation: Poorly indurated aeolian dust deposits, perhaps including reworked polar layered deposits or other polar region materials.
Polar layered deposits 2 (unit Apl2) form the upper part of Planum Boreum (Figure 13) and show fine layering, unconformities, and minor deformation in MOC images [e.g., Edgett and Malin, 2000; Kolb and Tanaka, 2001; Malin and Edgett, 2001]. On part of the margin of Planum Boreum, the unit appears to be draped over underlying topography (Figure 14). These deposits display no impact craters larger than a few hundred meters across in Viking images and MOLA DEM's [Herkenhoff and Plaut, 2000]. Interpretation: Likely made up of water ice and dust; alternating layers may result from varying proportions of dust and ice.
Dune material (unit Ad) forms broad, pebbly surfaced patches in MOLA DEM's covering Olympia Planitia (Figure 16) and other areas south of Chasma Boreale (Figure 13) and appears to be made up mostly of individual and chains of low-albedo barchans as seen in higher resolution Viking and MOC images [e.g., Breed et al., 1979]. Interpretation: Vast seas of wind-blown, sand-size particles, recently active and eroding from the dark polar layered deposits 1 (unit Apl1).
Polar residual ice (unit Api) has high albedo and appears to be largely uncratered in Viking and MOC images and MOLA DEM's [Herkenhoff and Plaut, 2000]. It forms a veneer on plateau surfaces of Planum Boreale. The mapping of this unit shown in Figure 2 is based on the Viking Mars Digital Image Mosaic and MOC wide-angle images, which show different extents of ice in places. Occurrences of the unit in the Scandia region are not mapped in order to show the underlying geology evident in the MOLA DEM. Interpretation: Recently deposited carbon dioxide ice precipitated from the atmosphere.
2.2. Tectonic Structure
 Most of the extensive systems of subtle, linear ridges throughout the northern lowlands ring Tharsis rise and Utopia basin; also ridges oriented radial to Utopia occur within that basin (Figures 2, 4, and 7). Most ridges have asymmetric cross-sectional topographic profiles. Ridge heights are similar among the plains units (∼100 to 300 m), which also are similar to ridge heights in highland and Tharsis volcanic rocks elsewhere on Mars [Golombek et al., 2001]. The maximum slopes on the ridges derived from the MOLA DEM are ∼0.5°.
 Cross-cutting relations indicate a lengthy history of ridge development in the northern plains, beginning in the Early Hesperian in boundary plains unit 1 (unit Hb1). The areal density of wrinkle ridges locally appears to decrease with decreasing age of the boundary plains and VBF units (units Hb1–2, AHvh). Ridges are generally absent in outcrops of older lobate material bordering the lowlands, where tectonic structures are largely absent or extensional faulting tends to dominate. Younger Amazonian lobate units (units Alc, Als1–2) that extend into the plains mostly bury ridge structures in the VBF and the boundary plains units, with local exceptions (e.g., Figure 18). In southern Utopia, the unit AHvh margin does not deflect across the ridges (Figure 4), indicating that either the ridges postdate unit AHvh or the local development of unit AHvh was not surface elevation dependent. The marginal member of the VBF (unit AHvm) appears to destroy ridges in Arcadia Planitia (Figure 11). In Chryse and Acidalia, ridges formed both prior to and following Late Hesperian channel dissection. Within the VBF, ridges are locally absent, perhaps where buried by outflow channel sedimentation in Borealis basin and where obscured by tholi and younger chaos materials (units At and Act) and related resurfacing features. We therefore see the ridge systems as a partly preserved record of contractional strain primarily from the later part of the Early Hesperian into the initial part of the Early Amazonian. The nearly ubiquitous Hesperian and Amazonian resurfacing apparently has destroyed or subdued the record of possible earlier ridge development in the northern plains. Thus the dominant signature of plains contractional deformation extends to younger times than the Late Noachian to Early Hesperian ages assigned to most highland and Tharsis-related contractional ridges [e.g., Scott et al., 1987; Tanaka et al., 1991; Watters, 1993; Dohm et al., 2001b].
 The apparent subtle nature of the lowland ridges, compared with those observed in highland rocks in Viking images, has been attributed to burial beneath a blanket of ocean sediment [Kreslavsky and Head, 2001; Thomson and Head, 2001]. However, the ridges retain relief similar to that of their highland counterparts, suggesting that their generally poor detectability in Viking images may have resulted instead from (1) the greater atmospheric scattering in lowland images, and (2) the lack of steep, readily perceived crenulations on the lowland ridges. Wrinkle (crenulated) ridges on Mars typically form in lava sequences, but uncrenulated scarps are more common in highland rocks [Watters, 1993] that may include sediments and highly mechanically weathered material. Okubo and Schultz  indicate that crenulations result from secondary synthetic and antithetic faulting within mechanically heterogeneous material; thus the northern plains ridges may reflect formation in mechanically homogenous material materials that inhibit such secondary faulting., whereas stronger lava sequences, possibly interbedded with weak materials, may favor the development of pronounced secondary faults. Notably, a small medial ridge in unit AHvm in Chryse Planitia has crenulated shoulders (Figure 12), perhaps indicative of a shallow, mechanically heterogeneous stratigraphy.
 Most ridges appear to be controlled by the regional topography, with major systems concentric to the north Tharsis region and Utopia basin (Figures 2a and 7). Within Utopia basin, many of the ridges trend radial to the basin center. Ridges in Isidis Planitia follow NW, E, and NE trends largely independent of the basin structure but perhaps concentric and radial to the HLB, which coincides with the intersection of Isidis and Utopia basin rims on the NE margin of Isidis basin. Compressional stresses responsible for the ridges probably resulted from planetary contraction and Tharsis loading [e.g., Tanaka et al., 1991], with ridges orientations locally controlled by Utopia basin [Thomson and Head, 2001] and other buried structure.
 A few narrow grabens of Tantalus Fossae cut northeastern Alba Patera and adjacent plains materials (Figure 2). Many are buried by the VBF but a few cut this unit, indicating that Tantalus Fossae development likely extended from the Hesperian into the Amazonian, in agreement with previous results [Tanaka, 1990]. The tensional stresses imposed by rifting associated with Alba Patera that produced these grabens apparently also precluded the development of ridges in this region.
 In southern Amazonis Planitia, wrinkle ridges trending ∼N50°W (concentric to Olympus Mons) deform an older flow sequence of smooth lobate unit 1 (unit Als1) and other older units (units AHl and Hb2) and are buried by a lightly cratered flow sequence of unit Als1 (Figure 18). These relations place the deformation in about the Middle Amazonian. The ridges are crosscut by linear, locally en echelon scarps trending N60°E (Figure 18); vertical offsets across the scarp alternate between northwest and northeast facing. Possibly these scarps formed by strike-slip faulting given that (1) they trend oblique to the wrinkle ridges, (2) one scarp forms at the termination of a wrinkle ridge, (3) the geometry of the ridges and scarps are consistent with a possible maximum compressive principal stress radial to Olympus Mons, and, (4) in contrast, wrinkle-ridge interactions generally result in curved ridge traces rather than linear sets [cf. Schultz, 1989; Watters, 1993].
 Tectonic deformation may also involve broad zones of uplift and lowering. Tanaka et al. , for example, proposed deformation along the VBF contact northeast of Alba Patera, in the outflow channels, and across Isidis Planitia. In the present study, with significantly higher resolution MOLA DEM data, such deformation is not resolved. Northeast of Alba Patera, the VBF/Hb2 contact exhibits ∼800 m of elevation range, some of which could be due to tectonic warping. However, the older Hb2/AHl contact that borders Alba Patera shows <300 m of relief; this observation seems to contradict significant warping. The southward-sloping outflow-channel floors stretching across hundreds of kilometers (shorter ones may be controlled by local structures) appear to be restricted to the region of chaos development within southern Chryse Planitia and northern Xanthe Terra. Since some of the chaos within the outflow channels appear to postdate some of the channel system that coursed into Borealis basin, development of the chaos terrains may be responsible for altering drainage direction. While the southwest-trending dip of the Isidis basin floor could have a tectonic origin, it might instead be related to volcanic loading of Syrtis Major Planum and local collapse due to magmatic activity rather than to plains-wide tectonic deformation. Finally, the Chryse outflow channels and young flow sequences (units Als1–2 and Alc) in the northern plains consistently trend in the direction of regional slopes. Overall, we find no unambiguous evidence for regional tectonic tilting of the surface.
2.3. Crater Modification
 Impact craters scattered throughout the northern plains provide a basis for analyzing the temporal and spatial history of resurfacing at vertical scales generally ranging from tens to hundreds of meters using MOLA data. Given that fresh impact craters generally have pronounced rims, distinct ejecta lobes, and deep, bowl-shaped floors, subdued crater rims and floors and missing ejecta features provide evidence for various sorts of resurfacing. Because the age and initial morphology of each crater are not precisely known, only qualitative and general statistical assessments of crater modification history can be made.
 Shallow circular depressions that have subtle, raised rims and lack distinct ejecta (“ghost craters”) are scattered throughout the northern plains (Figure 3) and occur within the VBF (units AHvh and AHvm) and boundary plains units (units Hb1–2). Those that we identified range from 13 to 188 km in rim diameter and have crater floors ∼105 ± 69 m deep (from our data for 295 craters) below the surrounding plains. When the density of ghost craters is added to that superposed on the VBF, an Early Hesperian crater density results [Kreslavsky and Head, 2001]. Furthermore, additional larger, but subtler “quasi-circular depressions” recognized in fine-scale topographic contouring indicate an Early Noachian density [Frey et al., 2002]. Some of the ghost craters on the boundary plains units have greater depth and rim height than those common in the VBF.
 Ghost craters (Figure 3) are unevenly distributed in the northern plains and likely were obliterated in places by Late Hesperian and Amazonian resurfacing. Such regions include: (1) central Utopia Planitia, where young lava and volcaniclastic flow deposits from Elysium Mons (units Als1 and Alc) have raised the plains floor by tens of meters (sufficient to bury ghost crater rims and floors) [Skinner and Tanaka, 2001]; (2) southern Utopia Planitia, where ring fractures in polygonal troughs occur rather than ghost craters, indicating that ring fractures may result from burial of ghost craters and subsequent contraction of the overlying sediments [McGill, 1989; McGill and Hills, 1992]; (3) Isidis Planitia, where young alluvium from Libya Montes and perhaps volcanic materials from Syrtis Major Planum may have been deposited [Crumpler et al., 2002]; (4) the lower part of the Acidalia side of Borealis basin, where sediments from the Chryse outflow channels must have collected (mapped as unit AHvh but probably originally unit Hch); (5) the Scandia region, where complex local resurfacing activity resulted in the Scandia, tholi, younger chaos material and polar units and features; and (6) Amazonis and Elysium Planitiae, where lavas, degraded flows, and local fluvial deposits from southeastern Elysium Mons, the western flank of the Tharsis rise, and northwestern Olympus Mons (units Als1–2 and Alc) have buried much of the plains [Scott et al., 1987; Fuller and Head, 2002].
 How were fresh craters with deep, bowl-shaped floors and prominent rims apparently transformed into ghost craters with subdued rims and flat floors of remarkably uniform depth? One hypothesis is that ghost craters represent Early Hesperian craters that have been buried by materials comprising the VBF [Kreslavsky and Head, 2001; Head et al., 2001]. However, we find that ghost craters also occur in the boundary plains units. Thus, processes other than burial may have contributed to the formation of the northern plains ghost craters. For example, at latitudes >30°N, many sub-kilometer and larger plains craters in Viking and MOC images appear heavily modified by terrain softening and perhaps aeolian infilling, in which the crater rim appears relaxed and the crater floor is nearly filled with layered and/or ridged debris [Squyres and Carr, 1986; Zimbelman et al., 1989]. Ghost craters are similar to virtually rimless, flat-floored craters prevalent in highland regions of Mars that have greater ranges in floor depth and rim height than their lowland counterparts and have been attributed to fluvial erosion and infilling, mass wasting, and perhaps other processes [e.g., Craddock et al., 1997]. In addition, other effects such as fluidization and smoothing of fluid-saturated crater-floor material due to shaking from impacts and other seismic events [Clifford, 1993] may also have contributed to ghost-crater development. Thus the actual dominant modificational process(es) that produced the lowland ghost craters remains uncertain and subject to further analysis.
 Other modified crater forms include (1) pedestal craters, which commonly have distinct rims and moderately deep floors, lack ejecta ramparts, and are perched on low plateaus tens of meters or more high and from ∼2 to 10 crater diameters wide [Mouginis-Mark, 1979], and (2) craters with distinct rims and ejecta but having shallow floors at or above the level of the adjacent surface in northern Arcadia Planitia and near Planum Boreum. Pedestal craters are locally abundant in the plains north of Alba Patera, in southern Amazonis Planitia, and especially on the northern rim of Utopia basin (Figure 3). The ejecta of pedestal craters seem to have armored a blanket of unconsolidated, perhaps ice-rich material that was later stripped away [e.g., Mouginis-Mark, 1979; Schultz and Lutz, 1988]. A particularly extreme example of the second crater type at 45.2°N, 184.6°E in Vastitas Borealis has a moderately to well-preserved rim 23 km in diameter and is infilled to an elevation 200 m above the surrounding plains. Large craters south of Olympia Planitia and Planum Boreum (e.g., Figures 14 and 16) appear partly filled by asymmetric, dome-shaped polar layered deposits 2 (unit Apl2) hundreds of meters thick. This second class of modified craters may have been filled by dust and ice entrapped only within the craters and/or by a broader mantle (or mantles) that was later stripped away except for within the craters. The deposit forms may reflect preferential erosion due to both local down-slope winds along the crater walls, producing the moats surrounding the domes, as well as regional prevailing winds, resulting in the off-centered dome crests. Collectively, the pedestal and infilled craters indicate that unconsolidated deposits had formed extensive blankets and isolated deposits within topographic traps in the northern plains; the blankets were preserved long enough to be impacted before eroding away. The deposits may have consisted of dust and ice related to volcanic and impact activity or the erosion and redistribution of material from elsewhere [cf. Schultz and Lutz, 1988; Tanaka, 2000; Malin and Edgett, 2001; Mustard et al., 2001].
2.4. Lowland Knobs
 Along the highland boundary, many knobs and mesas as much as several kilometers or more across result from deep dissection, fracturing, and mass wasting of the HLB, particularly along eastern Arabia Terra, eastern and southern Utopia Planitia, western Elysium Planitia, northern Tempe Terra, eastern Chryse and Acidalia Planitiae, and Phlegra Montes [e.g., Maxwell and McGill, 1988; McGill, 1989]. The larger, denser occurrences of these features are mapped as the knobby unit (unit HNk). Fields of smaller knobs mostly up to a few kilometers across extend as much as a few hundred kilometers from highland material (unit HNu) and the knobby unit and appear to be remnants of highland material. Similarly, groups of knobs surround Scandia Colles and most are likely remnants of the Scandia unit (unit Hs). In Viking and MOC images of Cydonia Colles, some knobs have scalloped margins and outer, lower platforms more than 1 km across that have been attributed to wave-cut erosion [Parker et al., 1993] or collapse of surrounding plains [Tanaka, 1997]. These features are particularly well developed in areas where depressions occur as seen in MOLA data (e.g., Figure 15 and MOC image SP2-49603; 34.7°N, 346.1°E), lending support to a collapse origin for the circum-knob platforms.
 Knobs of other types and occurrences in the northern plains may have other sources and origins. In some regions, such as on northern slopes and plains north of Alba Patera, many knobs form pitted domes having kilometer- and sub-kilometer-size summit craters. The size and spatial distribution of the craters suggest that they are likely impact craters; their morphology is consistent with pedestal craters. Their high density but mostly smaller sizes compared to pedestal craters elsewhere in the plains (Figure 3) may reflect a relatively short-lived yet less thoroughly effaced mantle.
2.5. VBF Structure
 Most of the northern plains are covered by the Vastitas Borealis Formation (18 × 106 km2, or 1/8 of Mars' total surface area), which in this work is suggested to result from pervasive, in situ modification of preexisting materials in the northern lowlands in the Late Hesperian and Early Amazonian below elevations ranging mostly from −3100 to −4100 m (Figure 19). Our mapping of the VBF largely coincides with that of Scott et al. ; however, we now have a much higher resolution topographic dataset to help map, measure, and interpret VBF structure more accurately and precisely.
2.5.1. Thumbprint “Ridges” and Unorganized Hummocks and Knobs
 These structures characterize much of the marginal parts of the hummocky member of the VBF (unit AHvh) [cf. Kargel et al., 1995], distinguishing the unit from adjacent plains materials (including units Hch, AHvm, and Hb1–2). Thumbprint ridges and unorganized hummocks are coarsely resolved in a slope map derived from the 1/128° MOLA DEM by a moderate yet perceptibly greater density of slope pixels >0.3° (Figure 4). Locally along the southern margin of the unit, thumbprint ridges form whorled patterns and have been described, mapped, and interpreted previously using Viking images [e.g., Carr and Schaber, 1977; Lucchitta, 1981; Rossbacher and Judson, 1981; Grizzaffi and Schultz, 1989]. Many of these ridge patterns are too small to resolve with MOLA DEM's. However, exceptions include a well-resolved pattern of thumbprint ridges in northern Arcadia Planitia (55–57°N, 183–187°E) and some of the ridges in Isidis Planitia. Thumbprint ridges seen in high-resolution Viking (<∼50 m/pixel) and MOC NA images appear to be both continuous and discontinuous and made up of relatively bright, pitted domes. Those within the northwestern fretted troughs of Deuteronilus Mensae and southwestern Utopia Planitia have lower relief, greater brightness, and greater discontinuity than ones in Arcadia and Isidis Planitiae (Figure 10).
 Based on their appearance and distribution in Viking images, the pitted domes making up thumbprint ridges have been interpreted as possible phreatic craters [Frey and Jarosewich, 1982], recessional glacial moraines [Grizzaffi and Schultz, 1989], and mud volcanoes [Tanaka, 1997]. In MOC NA images, the thumbprint ridges in Isidis Planitia are among the tallest and appear as arcuate, continuous to widely spaced chains of pitted domes up to 100 m high as measured in Viking images using photoclinometry [Davis and Tanaka, 1995]; farther north in other plains areas, these ridges appear relatively subdued (Figure 10). As such, they appear to be eruptive features; a general lack of associated flows suggests they formed by explosive vs. effusive discharge of subsurface, volatile-rich material. At present Martian temperatures, such discharge may involve CO2, whereas H2O explosive discharge would likely require volcanic heating [Hoffman et al., 2001]. The arcuate chains of pitted domes may result from eruptions occurring along periglacial or other deformation structures in near-surface material as opposed to along deep-seated faults [Tanaka, 1997]. In eastern Acidalia Planitia and Deuteronilus Mensae, the hummocky member (unit AHvh) includes dark lobate forms with raised margins and pitted mounds (e.g., MOC image M0302228), suggesting inflation and overflow of a fluidized mass of the substrate followed by deflation and local gas discharge. In some areas, sequences of flow deposits and pitted mounds suggest multiple episodes of fluidization and discharge (e.g., MOC SP248003). Such flows may explain other lobate margins of the VBF as observed in Viking and MOC images [Parker et al., 1993; Tanaka, 1997; Malin and Edgett, 1999].
2.5.2. Polygonal Troughs
 Networks of shallow (meters to a few tens of meters deep and up to a kilometer wide), intersecting linear troughs form polygons ranging from hundreds of meters to several kilometers or more in width. Polygonal troughs occur prominently in unit AHvh peripheral to the center of Utopia basin [McGill, 1989] where they are buried by coarse lobate material (unit Alc) sourced from Elysium Fossae, in Cydonia Labyrinthus (Figure 15) mapped as younger chaos material (unit Act), and in some of the low plains of Borealis basin surrounding Planum Boreum where it appears buried by various younger materials and structures (Figures 13 and 16). Polygonal troughs contribute to the apparent short-wavelength slope roughness of unit AHvh in the MOLA 1/128° resolution DEM. The full extent of the distribution of this structure may be underappreciated given data limitations, particularly at latitudes <60°N where MOLA shot point density decreases dramatically. Suggested origins include contraction of thick, volatile-rich sediments and tectonic rebound, perhaps from unloading of a local standing-body of water or ice [e.g., Pechmann, 1980; McGill and Hills, 1992; Hiesinger and Head, 2000; Thomson and Head, 2001]. The distinctive polygonal terrain of Cydonia Labyrinthus occurring within a depression appears to be most consistent with the contraction of volatile-rich material, whereas other occurrences in lower parts of the Utopia and Borealis basins could conceivably relate to either of the aforementioned mechanisms.
2.5.3. Marginal Troughs
 A particularly enigmatic set of landforms characterizes the marginal member of the VBF (unit AHvm). The member consists of networks of intersecting, linear to irregular troughs (generally 2–5 km wide, 30–70 m deep) with medial ridges (100–500 m wide, <20 m high) extending for tens of kilometers within smooth plains (Figures 11 and 12). In some cases, the trough areas appear elevated by meters to tens of meters above adjacent plains by scarps similar in trace to nearby troughs. Troughs in Arcadia Planitia broaden to 20 km across and incise wrinkle ridges in unit Hb2 in places. MOC images indicate that in some cases the trough has a scalloped floor that may reflect karst-like collapse, and the medial ridge may be continuous or discontinuous (MOC images 0303674, 0304711, 0402506, 0904678, 2300231). Some troughs end at low mesas, and in one MOC image (2301055) the central ridge tapers in width and height as it approaches the mesa. This ridge nearly connects with a subtle ridge on the mesa. In southern Acidalia Planitia, outflow channel bars and floors appear to be incised by the troughs. In MOLA shaded relief views (Figure 12a), the troughs are readily observed but the medial ridges are too narrow to be resolved, whereas in high-resolution MOC and Viking images, the troughs are commonly too subtle to see, but the central ridges are pronounced (Figures 12b and 12c). One flat-topped medial ridge clearly developed deforms a small crater (Figure 12c).
 Based on their associations in Viking images, interpretations for the features include tunnel valleys and eskers [Kargel et al., 1995], coastal spits [Parker et al., 1993], and periglacial deformation features [Scott, 1982; Tanaka, 1997]. The new MGS data favor an origin by contractional deformation to form the central ridge (Figure 12) and perhaps synclinal folding or graben faulting and karst collapse in places to form the troughs. Because the deformation originates in many cases at knobs and mesas that are also deformed, the deformation may extend below any thin sedimentary units covering the channel floors into underlying rocks; thus, the knobs may form sites of nucleation for the deformation. The nature of such deformation is uncertain. Possibly, the marginal member has undergone thin-skinned deformation in mechanically stratified material [Okubo and Schultz, 2002] dictated by the near-surface vertical distribution of volatiles and their ice and liquid phases as they occur within poorly consolidated and sorted deposits, including former patches of channel material and boundary plains units [e.g., Tanaka, 1997].
2.6. Discharge/Collapse Complexes
2.6.1. Xanthe Terra Chaos
 The chaotic terrains made up of older chaos material (unit Hct) in Xanthe Terra from which the Chryse outflow channels originate have been described, mapped, and interpreted extensively from Mariner 9 and Viking imaging data [e.g., Sharp, 1973; Carr, 1979; Nummedal and Prior, 1981; Witbeck et al., 1991; Rotto and Tanaka, 1995; Tanaka, 1997, 1999]. The perplexing aspects evident from Earth-based radar topography [Lucchitta and Ferguson, 1983] and confirmed by the MOLA DEM include: (1) the chaotic terrain floors form depressions hundreds of meters below the floor elevations of the outflow channels to the north that drain into Borealis basin, and (2) Simud/Tiu Valles within the chaotic complex slope downward to the south. The reversed channel-floor gradient might be explained by ponded flows [Lucchitta and Ferguson, 1983] or subglacial floods whose northward scour was controlled by a northward-sloping surface of an ice sheet [Chapman and Tanaka, 2002]. Alternatively, volatile discharge from higher chaotic terrains may have flowed down into nearby, lower ones (as floods, density flows, or glaciers). In Simud and Tiu Valles, both highland rocks and material beneath the channel floors have collapsed, resulting in net volume losses. Channels connecting many of these chaotic zones to possible sedimentary repositories generally appear to be lacking (except on the very margins of the chaos)—that is, the missing materials appear to have disintegrated in place and removed via the atmosphere or were transported through subsurface conduits. This may have been accomplished by eruption and possibly subsurface flow of fines and volatiles, which could have dispersed without a trace. Smooth highland surfaces surrounding the chaos may be covered by dust or ash blankets generated by phreatomagmatic eruptions [Chapman and Tanaka, 2002]. Outflow channel dissection also may have led to unsealing of confined aquifers, perhaps charged with H2O and/or CO2, that led to the formation of the chaos terrain [e.g., Carr, 1979; Nummedal and Prior, 1981; MacKinnon and Tanaka, 1989; Hoffman, 2000; Tanaka et al., 2001].
2.6.2. Acidalia Mensa and Cydonia
 These knobby and chaotic terrains (made up of units HNk, Hb2, and Act) occur along the highland margin and in the northern plains (Figures 9 and 15). They are characterized by zones hundreds of kilometers across that generally include: (1) remnants of highland material tens to a couple hundred kilometers across that stand hundreds of meters above surrounding plains, (2) fields of knobs within plains surrounding the highland remnants, and (3) depressions up to a few hundred meters deep enveloping or scattered through the knobby areas and locally including polygonal troughs. Highland rocks associated with these terrains show evidence for faulting, tilting, and collapse producing relief of hundreds of meters. Acidalia Mensa includes a low, relatively smooth platform ∼250 km across that embays a large highland remnant. A shallow trough borders the south edge of the plateau and extends 180 km southeast. The depressions and troughs occur in part within and thus postdate the VBF. Also, other smaller, local depressions and, in some cases, associated knobs occur in the plains east of Acidalia Mensa. Cydonia Mensae and adjacent Cydonia Labyrinthus and Colles show similar morphologic relations.
 Collectively, the features seem to reflect central uplift (resulting in the highstanding and tilted highland remnants) surrounded by collapse and perhaps reconsolidation of sediment (producing the surrounding depressions, knobs, and polygonal troughs). If so, they may be produced by the upwelling of volatiles beneath the area of uplift that were drawn from surrounding rocks. Such localized volatile upwellings could be caused by magmatic intrusions, but this seems exceptionally fortuitous that in multiple localities such intrusion would occur chiefly beneath large highland inliers. A more plausible interpretation could be analogous to pingo formation in which thickening of the cryosphere possibly due to atmospheric cooling or a decrease in heat flow increases hydrostatic flow of water into the region beneath the inliers where the underlying cryosphere would be at relatively higher elevation [Clifford, 1993]. As water was drawn away from adjacent areas, collapse would ensue to form the adjacent moats, knobs and polygonal fractures; meanwhile, the areas above the upwellings could be raised if superhydrostatic pressures were achieved.
2.6.3. Western Isidis Planitia
 A ∼300 × 800 km swath of fractured and knobby terrain including some shallow depressions several tens of meters deep mapped as younger chaos material (unit Act) forms the lowest, western part of Isidis Planitia and the foot of the eastern slope of Syrtis Major Planum. The planum forms a broad, low plateau made of lava flows emanating from calderas and other vents [Scott et al., 1987]. The chaos structures may have formed by fracturing and collapse due to a late-stage of magmatic heating beneath Syrtis Major Planum that led to discharge of volatiles from nearby sediments within Isidis or perhaps by the over pressurization of subsurface volatiles due to structural and topographic controls.
2.6.4. Scandia Tholi and Cavi
 These features (Figure 16) include low, round, broad mounds that vary in shape, size, and relief scattered across a 400 × 1000 km region between Scandia Colles and Olympia Planitia. Some have pancake shapes and may occur in chains. Relief varies from tens to hundreds of meters. Two rugged mountainous terrains extend for about 200 and 300 km respectively and include the deep, irregular depressions of Scandia Cavi. Some of the tholi are surrounded by shallow moats tens of meters deep. We envision that shallow slurries were expelled onto the surface from below to produce the tholi; the evacuated material led to collapse to form the moats and cavi. For the larger tholi, the interplay between buildup and collapse features appears to have been more complex. Alternatively, Fishbaugh and Head [2000, 2001] have suggested that the features form glacial kame and kettle topography due to retreat of polar layered deposits.
3. Volatile-Driven Resurfacing of the Northern Plains
3.1. Progressive Lowering of Volatile Table
 Based on our geologic mapping and landform analysis, we suggest that the northern plains of Mars underwent four stages of resurfacing involving volatile-driven processes. This activity obliterated or heavily modified earlier landforms, resulting in HLB fretted and knobby terrains, ghost craters, chaotic terrain, outflow channels, collapse depressions, VBF-associated thumbprint and polygonal and marginal trough terrains, and polar tholi and cavi terrains. Such resurfacing likely involves erosion, ductile deformation, collapse, and effusive and violent eruptions of volatile-charged material. The elevation dependence of the reworking (Figure 19) may be related to gradual lowering of the threshold for near-surface volatile activity, as governed by the composition (H2O and possibly CO2) and distribution of subsurface volatiles and by the geothermal gradient. The four stages are as follows:
 Stage 1 involves the earliest observed degradation of Noachian to perhaps Early Hesperian highland materials (e.g., Figure 4). This degradation probably began no earlier than the end of the Noachian, as it consistently and deeply erodes materials of Late Noachian age. Highland material (unit HNu) became deeply broken up along structural weakness. This activity led to development of the knobby unit (unit HNk) comprised of highland remnants and intervening slope and plains materials. At the base of the knobs, boundary plains unit 1 (unit Hb1) formed as colluvium shed from the highland rocks, at maximum elevations of <−2.0 to −2.9 km. This unit extends as much as a few hundred kilometers down slope into the northern plains below which the unit is buried or eroded by later activity. Ice-lubricated creep in most places may have transported the clastic debris forming the unit, as fluvial channels cutting into highland rocks along the HLB are rare. Thus it seems unlikely that the unit entirely filled the northern plains and would have terminated before or at reaching the lowest part of the basin floor accessible to down-slope transport. Unit Hb1 is Early Hesperian according to crater counts (Table 2). Ghost craters in the unit may represent deeply eroded and buried craters that may have been degraded in part by pre-Stage 1 activity. Possibly, outflow channel discharges occurred through Chryse and Amazonis Planitiae during this stage [Rotto and Tanaka, 1995; Dohm et al., 2001a]; however, that erosional record may have been largely erased by later channel erosion. Unit Hb1 occurs along most of the highland boundary, with the following exceptions: (1) Along the margins of Alba Patera and the Elysium rise, the emplacement of older lobate material (unit AHl) obscures the Early Hesperian geologic record. (2) In southern Isidis Planitia, the unit either is missing, poorly developed, or at anomalously low elevation; the highland material (unit HNu) that makes up that part of the HLB consists of Early Noachian basin massifs that may have been resistant to degradation. (3) Along the northwestern margin of Arabia Terra (Figure 5), the unit either occurred at much lower elevation and was later buried or destroyed by the formation of boundary plains unit 2 (unit Hb2) or the low surface gradient prevented wholesale degradation of highland material. Evidence for possible incipient Stage 1 (or later) degradation of northwestern Arabia includes irregular depressions and fractured crater floors (Figure 5).
 Stage 2 appears to result mostly from degradation of boundary plains unit 1 (unit Hb1) and in places of highland rocks (units HNu and HNk) during the Late Hesperian. The lower margin of unit Hb1 appears rugged and locally collapsed and eroded in southern and western Utopia Planitia and southern Amazonis Planitia. Largely at the base of these degraded zones, beginning at −2.7 to −3.9 km elevation, the Late Hesperian boundary plains unit 2 (unit Hb2) apparently forms the depositional plains resulting from this degradation. Like unit Hb1, it might have covered only areas down slope from source materials. At Acidalia Mensa, lightly cratered, smooth material forms the western part of the mensa and embays unit HNu. This relation suggests that Stage 2 degradation (and also that of Stage 1 if it had occurred here) developed more deeply here, below −4.4 to −4.8 km elevation. At Deuteronilus Mensae, fretted trough floors and plains appear topographically continuous with the unit Hb2 surface along northwestern Arabia Terra (Figure 5). Here, the unit may have formed by mass flow of volatile-rich highland material [e.g., Sharp, 1973; Lucchitta, 1984]. In eastern Amazonis Planitia and surrounding the western and southern flanks of the Elysium rise, unit Hb2 is missing and may be buried by younger volcanic materials. Erosion and perhaps deposition by outflow channels in Chryse Planitia, sourced from chaotic depressions (mapped as older chaos material, unit Hct), appears to have erased unit Hb2 and represents later Stage 2 activity.
 Following the formation of boundary plains unit 2 and the end of outflow channel activity at the end of the Hesperian and the beginning of the Amazonian, Stage 3 ensued as the only demonstrably spatially continuous resurfacing in the northern plains below −3.1 to −4.1 km elevation as defined by the Vastitas Borealis Formation. Unlike the apparent mass-wasting origin of older, higher rocks and down-slope transport typical of Stages 1 and 2, we think that Stage 3 activity mainly consisted of in situ reworking of older materials (including channel material, boundary plains units, and other miscellaneous volcanic, sedimentary, polar, and impact materials) to form the troughed and hummocky topographies of the marginal and hummocky members (units AHvm and AHvh) of the VBF. The VBF locally rises to −3.1 km northeast of Alba Patera in the vicinity of Tantalus Fossae. Although it is tempting to propose that this rise was due to local tectonic uplift [e.g., Head et al., 1998, 1999; Tanaka et al., 2001], the nearby higher, older AHl/Hb2 contact does not show evidence for a similar elevation rise. Surrounding Acidalia Mensa, the VBF contact is significantly lower (<−4.6 to −4.8 km elevation). VBF resurfacing may have largely destroyed smaller, pre-VBF landforms such as ridges, scarps, troughs, and channels by deformation, karst development, and explosive discharges all stemming from activities involving shallow volatiles. Ghost craters in the unit may be substantially relaxed, pre-Stage 3 craters; their largely consistent floor depth in spite of diameter indicates that the modification process was highly efficient and perhaps controlled by near-surface rheology. Stage 3 activity seems to have occurred continuously below the VBF contact elevation throughout the northern plains, with the exception of the relatively highstanding and broad Acidalia Mensa and Scandia Colles that rise several tens to hundreds of meters above the regional plains.
 Finally, Stage 4 resurfacing was local, resulting in younger chaos and tholi materials (units Act and At, respectively) postdating the VBF in the Isidis, Cydonia, Scandia, and Acidalia Mensa regions during the Early Amazonian. Maximum elevation of the chaos margins ranges from −3.8 to −5.0 km. The chaos includes zones of apparent collapse and uplift that may be due to movement of ground volatiles, and the tholi may represent discharges of subsurface slurries.
3.2. Plains-Filling Oceans?
 Various workers have contended that a standing body of water, ice, or debris, or some complex mixture or sequence thereof, had once contiguously filled the basins of Borealis, Utopia, and, in some scenarios, Isidis as a result of Chryse outflow-channel dissection. In these scenarios, the VBF is usually assumed to be the submarine or debris-flow deposits resulting from outflow-channel dissection (hereafter, we loosely apply “ocean” to include all of these scenarios unless otherwise stipulated; we do not consider possible oceans formed earlier [e.g., Clifford and Parker, 2001] although some of the arguments may apply to them). Suggested evidence for oceans includes proposed paleoshorelines and marine deposits by Parker et al. [1989, 1993], Edgett and Parker , and Clifford and Parker ; the nearness of Parker et al.'s [1989, 1993] Contact 2 to an equipotential surface [Head et al., 1998, 1999], including the potential convergence of outflow-channel base levels at this elevation [Head et al., 1999; Ivanov and Head, 2001]; coastal spits and bars and wave-cut terraces along proposed shoreline levels and ocean stillstands [Parker et al., 1993; Head et al., 1998]; the topographic smoothness of the northern plains below Contact 2 [Head et al., 1999], and rebound of Utopia basin following removal of the ocean to form polygonal-trough terrain [Hiesinger and Head, 2000; Thomson and Head, 2001]. One notable argument made against Contact 2 being a paleoshoreline is the general lack of expected shoreline features seen in MOC images; instead, the contact in places appears to suggest a lobate deposit emplaced in the upslope direction [Tanaka, 1997; Malin and Edgett, 1999, 2001]. Also, features previously interpreted as terraces resulting from stillstands of a retreating ocean may instead be contractional tectonic features [Thomson and Head, 2001; Withers and Neumann, 2001] or gently tilted layers [Malin and Edgett, 2001].
 In our study, Contact 2 (also designated the Deuteronilus Shoreline by Clifford and Parker ) generally is the same as the upper margin of the VBF; we are able to map this “contact” more completely with the MOLA DEM. In the following, we note a number of observations that indicate that (a) the Vastitas Borealis Formation cannot be strictly defined as a deposit resulting from outflow-channel dissection but consists of an undifferentiated mixture of lowland materials that have been largely reworked by later processes, and (b) the processes responsible for in situ VBF resurfacing have largely obscured any subtle evidence for prior oceans that may have existed below the VBF margin.
 1. The VBF postdates the outflow channels according to crater-density data (Table 2) and to geologic relations in which the VBF in southern Acidalia Planitia appears to onlap, degrade, and deform the channel material (unit Hch) as seen in Viking and MOC images and MOLA data (Figures 6 and 12) [Tanaka, 1997; Malin and Edgett, 1999, 2001]. Based on crater densities and their standard deviations and the 2.0 to 3.4 Ga age assigned to the conclusion of the Hesperian [Hartmann and Neukum, 2001], the difference in the mean ages of the channel material (unit Hch) and the hummocky member of the VBF (unit AHvh) ranges from 450 ± 240 to 760 ± 410 m.y. Thus we conclude that virtually all of the putative shoreline landforms occurring within the VBF likely postdate outflow-channel activity; in turn, the development of VBF structures probably has obscured much of the possible evidence for pre-VBF bodies of water that may have occurred in the northern plains up to the level of the VBF margin. The ridges suggested to be coastal spits [Parker et al., 1993] that occur within the marginal member of VBF (unit AHvm) therefore cannot be contemporaneous with channel activity; MGS data indicate that the features may have a deformational origin instead (Figure 12).
 2. The lowest parts of the saddles between Utopia and Borealis and Isidis basins show a lack of intense dissection that would be expected due to catastrophic overspill of water between the basins (Figure 7); however; basin infilling by debris flows having no erosive capacity would not produce dissection [e.g., Jöns, 1991; Tanaka et al., 2001]. The lowest elevations of the saddles are −4340 m between Utopia and Borealis and −3520 m between Utopia and Isidis basins. An ocean at the VBF margin elevation would have breached the Utopia/Borealis saddle and one filling up to the proposed Arabia Shoreline [Clifford and Parker, 2001] would have breached both saddles, forming broad, deeply incised channel scars and destroying and streamlining topographic features such as craters and possible preexisting ridges. Later VBF development may have obscured subtle channel structure in the Utopia/Borealis saddle, but in Acidalia Planitia within the VBF, Chryse outflow-channel streamlined landforms are still detectable (Figure 6). Given the lack of ghost craters down slope of the Chryse channel system in Borealis basin (Figure 3) but their general presence elsewhere on VBF, it appears that outflow-channel sediments thick enough to bury ghost craters were restricted to that area. The ghost-crater and basin-saddle observations suggest to us that effluent from Chryse outflow-channel dissection—both the water and sediment—was likely restricted to Borealis basin.
 3. We suggest that the Chryse channel system does not have a definable erosional base level at −3600 to −3900 m elevation that coincides with the VBF margin as proposed by Ivanov and Head . At those elevations, the individual channel systems debouch from the highlands into Chryse Planitia, where they coalesce into one broad, shallow channel system that can be traced continuously through Chryse Planitia and into Acidalia Planitia at least as far as 40°N down to an elevation of −4230 m. Thus the channel system can be traced well into the VBF (Figures 6 and 12). Since subaqueous density flows may also produce scouring, the existence of low-elevation scour features neither lends support to nor refutes the proposal that floods or mass flows entered into an ocean.
 4. The smoothness of the VBF in MOLA point-to-point elevation measurements has been used to support the contention that the VBF resulted from submarine sedimentation [Head et al., 1999; Kreslavsky and Head, 1999, 2000]. However, we note that the same data show that lava plains in Elysium and Amazonis Planitiae have the same range of 0.6-km slope roughness as the VBF. Also, these data show that the surface of VBF appears rougher at 0.6-km slopes than the adjacent boundary plains surfaces above the unit along the margins of Isidis and southern Utopia Planitiae (see also Figure 4). Elsewhere, the roughness mapping between VBF and older plains units is less distinctive, which may be due in part to terrain softening and mantling, particularly above 30–50°N [Squyres and Carr, 1986; Zimbelman et al., 1989; Kreslavsky and Head, 2001; Malin and Edgett, 2001; Mustard et al., 2001]. MOC images show that the VBF has a hummocky surface that includes thumbprint terrain, pitted domes, polygonal troughs, and complex pitting (see Figures 9–16) that we and other workers have attributed to periglacial-like deformation, local volatile and sediment discharge, and thermokarst. Thus the VBF smoothness at MOLA resolutions relates to modification of the plains surface rather than to the possible depositional environment of the VBF.
 5. The MOLA DEM has revealed many scarps having relief of tens of meters or more in the northern plains, but no consistent stillstand-like wave-cut terraces [Parker et al., 1989, 1993] or ice-margin strandflats [Thomson and Head, 2001] can be mapped out consistently along and below the VBF margin, whether on plains surfaces, crater rims, or the large highland inlier, Acidalia Mensa (Figure 9). In the case of benches surrounding knobs in Cydonia proposed to be wave-cut terraces, the MOLA DEM and MOC images show that they occur within a broad depression and appear to result from collapse of surrounding material, resulting in the bench scars and polygonal troughs (Figure 15) [Tanaka, 1997]. Elsewhere, MOC NA images show that the apparent benches seen in Viking images by Parker et al.  actually form discontinuous undulations [Malin and Edgett, 2001]. Terraces proposed to represent a pre-Deuteronilus (Contact 2) Arabia Shoreline in the Cydonia area [Clifford and Parker, 2001] occur on a surface that is much less cratered and thus younger than the outflow channel floor entering the northern plains, as shown in MOC images (Figure 20); we therefore think that such terraces may have a mass-wasting origin instead. The scarps that do occur prominently in the northern plains appear to be either asymmetric ridges belonging to contractional tectonic systems or irregular scarps outlining depressions produced perhaps by local erosion or collapse (see Figures 2a, 4, 7, 9, and 15).
 6. The confinement of the pronounced polygonal troughs of Cydonia Labyrinthus within a 100- to 300-m-deep depression about 100 × 450 km across (Figure 15) indicates that these features may form by collapse of near-surface material and do not necessarily require tectonic rebound to form.
 Thus we conclude that no strong, direct geomorphic evidence exists for a Late Hesperian, outflow-channel fed ocean on Mars, because either an ocean did not form in the first place or VBF resurfacing has destroyed the evidence for it.
4. Waxing and Waning of North Polar Deposits
4.1. Hesperian (?) to Middle Amazonian
 MOC images and MOLA shaded-relief maps indicate a sequence of basal, polar layered deposits 1 (unit Apl1) in Planum Boreum that underlie the uncratered (down to a few hundred meters diameters) polar layered deposits 2 (unit Apl2) [Edgett and Malin, 2000; Herkenhoff and Plaut, 2000; Kolb and Tanaka, 2001; Byrne and Murray, 2002]. Unit Apl1 is characterized in MOC images by what appears to be cross bedding and a relatively low albedo [Byrne and Murray, 2002]. It is clearly superposed by two large impact craters (24 and 18 km diameters respectively at 81.4°N, 295.4°E and 81.3°N, 254.7°E) whose ejecta blankets are visible, and by a 6-km crater (82.3°N, 290.5°E) that may be partly buried and/or degraded.
 The polar layered deposits 1 approaches 1000 m in thickness west of the mouth of Chasma Boreale, where it forms an abrupt scarp. This scarp apparently formed by extensive back wasting, and the unit may have extended southward across perhaps hundreds of kilometers of the adjacent plains. Beyond Planum Boreum, mapped outcrops of material form either outliers of unit Apl1 or perhaps polar deposits of different age and/or origin, as follows. (1) About 200 km south of Chasma Boreale, a 23-km-diameter crater (77.0°N, 304.8°E) rests on an isolated plateau about 600–700 m high that may be an outlier of unit Apl1 (Figure 13). Other large knobs forming Abalos Colles west of this plateau have summit craters and may be remnants of unit Apl1 preserved as eroded pedestal craters; alternatively, they may be volcanoes [Dial and Dohm, 1994; Hodges and Moore, 1994; Sakimoto et al., 2001]; (2) Another possible broad outlier of unit Apl1 occurs at 76–82 °N., 105–145 °E (Figure 7, right). A few small craters within this outcrop may be resting on unit Apl1 or perhaps on the underlying VBF. (3) The dunes of Olympia Planitia form a generally thin layer resting above layered deposits [Zuber et al., 1998; Fishbaugh and Head, 2000]. Exposed outliers of these deposits show a smooth, layered surface characteristic of unit Apl1. However, a well-preserved, 20-km crater (81.7°N., 190.0°E) in the middle of Olympia Planitia is probably superposed on layered deposits; since craters are rare on unit Apl2, unit Apl1 probably underlies this crater indicating that unit Apl2 is very thin to non-existent in this region. (4) The Scandia unit (unit Hs) forms what appears to be a broad band of degraded, potential Hesperian paleopolar deposits >1000 km across and 20 to 200 m thick roughly between Alba Patera and Olympia Planitia (<60–77°N, 200–275°E.). (5) Numerous small pedestal craters (Figure 3) supposedly overlie blankets of material that have come and gone across the plains; perhaps these blankets include unit Apl1. Most of the pedestals occur near the topographic divide between north polar and Utopia basins (60–75°N, 60–110°E.). These various outcrops suggest a potentially rich history of waxing and waning of polar deposits, but it remains unclear whether all of the outcrops are layered and were formed by polar-type dust and ice deposition.
4.2. Late Amazonian
 Polar layered deposits 2 (unit Apl2) form the primary record of youthful, finely preserved layers as viewed in MOC images. The deposits may easily exceed 2000 m in thickness, if the underlying unit Apl1 does not thicken within Planum Boreum. The surface of unit Apl2 is largely mantled by thin polar residual ice (unit Api) and appears extremely smooth and undulating in MOLA DEMs. Outliers apparently made up of unit Apl2 occur >300 km south of Planum Boreum and Olympia Planitia as crater fill (e.g., within 84-km diameter Korolev crater at 73°N, 164°E) and as an isolated, irregular mound nearly 90 km long north of Utopia Planitia (at 74.5°N, 98°E; Figure 7, right). Also, hummocky material (unit Ah) south of Chasma Boreale (Figure 13) forms a few broad patches of smooth material mantling the underlying VBF and may be remnant outliers of unit Apl2. Thus unit Apl2 probably was at one time considerably more extensive than at present. Surrounding the polar region down to 30–50°N, surfaces appear mantled in MOC NA images by a meters-thick mantle that probably is made up of dust and ice [Kreslavsky and Head, 2000; Malin and Edgett, 2001; Mustard et al., 2001]; these deposits may be an expression of a much more transient and dispersed style of polar deposition.
 Dune material (unit Ad) surrounding Planum Boreum has a low albedo similar to that of polar layered deposits 1 (unit Apl1) and in MOC images appears to originate from erosion of unit Apl1 in depressions within Planum Boreum near the margin of Olympia Planitia and at the head of Chasma Boreale [Byrne and Murray, 2002].
 Our new geologic mapping of the northern plains of Mars based on the rich data set from Mars Global Surveyor has provided us with many new and unexpected insights into the region's resurfacing history. These findings appear to require major revisions in the interpretation of the origin and age for a host of features, although we acknowledge that many of our new interpretations must be viewed as tentative and subject to further analysis. First and foremost in importance is the sequence of four stages of resurfacing during the Early Hesperian to Early Amazonian revealed by the geologic and topographic mapping of boundary plains units, the VBF, and chaos and tholi materials. These episodes appear to result from widespread degradation and erosion of progressively topographically lower rocks and seem to relate to a variety of volatile-induced erosional, deformational, and discharge processes. The first two stages involve successive degradation of the highland-lowland boundary, first in highland rocks and then also including deposits shed along the boundary. The Chryse outflow dissection occurred during Stage 2 and may have deposited thick sediments only down slope from the channels in Borealis basin as indicated by the paucity of ghost craters therein. During Stage 3 the VBF formed, apparently as a result of near-surface water and perhaps CO2 beneath a thin cryosphere that permitted thin-skinned deformation and ubiquitous sediment and water discharges to form pitted domes, polygonal troughs, thermokarst pits, and other landforms. This activity likely obliterated any possible subtle evidence for a Stage 2 ocean below the VBF margin; thus we cannot determine conclusively if an ocean had formed or not. Stage 4 resurfacing activity was more localized but formed some particularly spectacular chaos zones and tholi features attesting to uplift, collapse, and discharge. Volcanic flows emanating from the long-lived Tharsis and Elysium volcanic rises resurfaced parts of Utopia and Amazonis Planitiae during the Amazonian. Pedestal and infilled craters indicate that debris mantles tens to hundreds of meters thick covered parts of the plains at times. Discontinuous deposits forming Planum Boreum and surrounding outliers suggest that waxing and waning of two major sequences of polar layered deposits has occurred. The newly observed sets of scarps and ridges demonstrating widespread tectonic deformation in the northern plains appear to be mostly Late Hesperian to Early Amazonian in age, except for possible Early Hesperian ones along the HLB, even though most wrinkle ridges elsewhere on Mars are thought to be Late Noachian to Early Hesperian. One final surprise is the possible discovery of strike-slip faults in southern Amazonis Planitia. We fully expect that these interpretations will be further refined and revised as we complete more detailed geologic mapping based on the MGS data set and begin to examine the new data from Mars Odyssey.
 We would like to thank Jeff Kargel, Nick Hoffman, Eric Kolb, Ken Herkenhoff, Bruce Murray, Jim Head, Kate Fishbaugh, Shane Byrne, Elizabeth Fuller, and Mikhail Kreslavsky for helpful discussions regarding the various aspects of the geology of the northern plains and the north polar deposits. Discussions with Bob Anderson, Richard Schultz, and Chris Okubo have been helpful to us in addressing the origin and timing of structures in the northern plains. James Dohm, Vic Baker, Mary Chapman, and an anonymous reviewer provided helpful many comments that improved the paper. Nadine Barlow generously provided us with her crater database of Mars from which we gathered our crater statistics. We are greatly indebted to the MOLA and MOC instrument teams for providing us with a rich data set to investigate as well as a research grant from the Mars Data Analysis Program that made this study possible.