By continuing to browse this site you agree to us using cookies as described in About Cookies
Notice: Wiley Online Library will be unavailable on Saturday 7th Oct from 03.00 EDT / 08:00 BST / 12:30 IST / 15.00 SGT to 08.00 EDT / 13.00 BST / 17:30 IST / 20.00 SGT and Sunday 8th Oct from 03.00 EDT / 08:00 BST / 12:30 IST / 15.00 SGT to 06.00 EDT / 11.00 BST / 15:30 IST / 18.00 SGT for essential maintenance. Apologies for the inconvenience.
 Recent Mars Orbiter Laser Altimeter (MOLA) data have provided a new picture of the Martian northern lowland basin topography and surface roughness. In order to assess detailed topographic structure important in understanding the formation and evolution of the northern lowlands, we have removed regional slopes from the topography to produce a series of maps highlighting the local topographic and geologic structure. We find that (1) the northern lowlands are underlain by a regional unit containing a basin-wide system of subparallel wrinkle ridges; (2) this unit contains highly modified craters, the number of which suggests an Early Hesperian age; (3) this unit is laterally contiguous with Early Hesperian-aged ridged plains in the southern uplands; (4) the orientation and location of the wrinkle ridges in the North Polar Basin complete a global circum-Tharsis ridge system forming a band ∼7000 km wide and extending over the whole circum-Tharsis region; and (5) several subareas of the northern lowlands show individual wrinkle-ridge patterns (e.g., Isidis and Utopia are basin-like). The recognition of this unit and the superposed very degraded craters permits us to assess the stratigraphy and geometry of subsequent units and structure. We find that (1) the present spacing and height of wrinkle ridges and geometry of buried craters suggest that the Vastitas Borealis Formation is a thin sedimentary unit superposed on regional ridged plains (Hr) and that its minimum average thickness is ∼100 m; (2) the polar terrain completely obscures the wrinkle-ridge system and some circumpolar deposits partially obscure it, supporting the interpretation that some circumpolar deposit thicknesses exceed several hundred meters and that wrinkle-ridge formation was not active in the Amazonian; (3) Late Hesperian-aged outflow channels entering Chryse Planitia are controlled by the orientation and topography of wrinkle ridges deep into the basin, indicating that wrinkle ridges had largely formed by this time; (4) Amazonian-aged smooth plains units, particularly in Amazonis Planitia, further bury and obscure the underlying wrinkle ridges, and (5) fretted terrain, particularly in the Deuteronilus Mensae region, formed subsequent to the Early Hesperian-aged ridged plains, and remnants can be seen to extend beneath the Vastitas Borealis Formation. The recognition of these units and their stratigraphic relationships permits us to outline a new perspective on the history of the northern lowlands: (1) in the Early Hesperian the majority of the northern lowlands was filled with volcanic plains similar to those presently exposed in the southern uplands; outcrop patterns of buried Noachian-aged terrain and volcanic flooding models suggest an average thickness of 800–1000 m; (2) these plains were deformed soon thereafter by Tharsis-circumferential and basin-related wrinkle ridges; (3) circum-Chryse outflow channels were emplaced in the Late Hesperian, forming subdued channels whose course was largely controlled by wrinkle-ridge orientation and height, and these channels deposited water and sedimentary material in the basin; (4) loss of the outflow channel water resulted in formation of the Vastitas Borealis Formation as a residual sedimentary deposit on top of the ridged plains; little evidence is seen for the presence of any massive near-surface residual ice deposits remaining from the outflow channel effluent; (5) Amazonian volcanic plains were emplaced, primarily in Amazonis Planitia and Utopia Planitia, further obscuring the Hesperian ridged plains; and (6) Late Amazonian polar and circumpolar deposits formed, further obscuring the structure of the underlying Hesperian ridged plains. The widespread emplacement of the Hesperian-aged ridged plains of apparent volcanic origin is interpreted to mean that the volcanic phase represented by this unit was global in nature and resurfaced the northern lowlands, in addition to the ∼10% of the planet previously known, for a total resurfacing of ∼30% of Mars. This remarkable event increases dramatically the amount of water and other volatiles that might have been degassed into the atmosphere during the Early Hesperian and further underlines the global significance of the contractional deformation typical of this period.
 The northern lowlands of Mars cover about one third of the surface of the planet and are the central part of a larger drainage basin that has composed about two thirds of the planet's surface for most of its history [Smith et al., 1998; Zuber et al., 1998]; the northern lowlands may have contained a large standing body of water in earlier Mars history [Parker et al., 1989, 1993; Baker et al., 1991] and have been proposed to be the location of plate tectonic activity in earliest Mars history [Sleep, 1994].
 In ascending stratigraphic order the exposed surface units of the northern lowlands [Scott and Tanaka, 1986; Tanaka and Scott, 1987; Greeley and Guest, 1987] are dominated by (1) Noachian-aged remnants, including a small unit dissected by channels in Acidalia Planitia, old degraded crater rims, and Scandia Colles, a collection of domes interpreted to be old degraded Noachian crust, (2) the Hesperian-aged Vastitas Borealis Formation, an unusual unit interpreted to be degraded lava flows and sediments, (3) Hesperian-aged outflow channel deposits at the margins of the lowlands in Chryse Planitia, (4) various local Amazonian-aged plains units, (5) Elysium channel and flow deposits, (6) the north polar cap, consisting of Late Amazonian ice and layered terrain deposits, and (7) Amazonian-aged circumpolar mantling material. The predominantly sedimentary nature of many of these northern plains units combined with modification processes since their formation has obscured the character and age of underlying terrain and thus its origin and evolution. Two fundamental issues remain unresolved: (1) when and how did the distinctive northern lowlands form, and (2) what were the processes involved in their evolution?
1.2. Mars Orbiter Laser Altimeter Data
 The Mars Orbiter Laser Altimeter (MOLA) experiment on board the Mars Global Surveyor mission has provided a new global characterization of the topography and slopes of Mars [Smith et al., 1998, 1999; Aharonson et al., 1998; Zuber et al., 1998; Garvin et al., 1999; Kreslavsky and Head, 1999, 2000]. The combination of gravity and topography has provided important new insights into crustal structure and thermal evolution [Smith et al., 2001; Zuber et al., 2000]. These data (Figures 1a and 1b) have shown that (1) the northern lowlands are extremely flat [Smith et al., 1998]; (2) they are irregular in shape and consist of two basins, the ancient, circular, heavily modified Utopia Basin of impact origin [McGill, 1989] and the irregularly shaped North Polar Basin [Head et al., 1999]; (3) the dichotomy boundary (the distinctive morphologic and local topographic slope boundary between the southern heavily cratered uplands and the northern lowlands) does not coincide exactly with the crustal thickness trends [Zuber et al., 2000]; and (4) the surface of the northern lowlands is exceptionally smooth at scale lengths from hundreds of meters to tens of kilometers [Aharonson et al., 1998; Kreslavsky and Head, 1999, 2000]. As a further step in the analysis of the origin and evolution of the northern lowlands, we have used recent MOLA data to analyze basin topography and surface roughness characteristics, removing regional slopes to assess subtle topographic variations.
1.3. Representation of Topography
 MOLA vertical accuracy is ∼0.3 m [Smith et al., 2001]. This means that 1-m-high features can be detected and mapped in smooth regions. Topographic variations across the northern lowlands, excluding deep impact craters, are ∼3000 times greater, of the order of 3 km. This high dynamic range resulting from the excellent MOLA data quality actually produces some problems for the global visual presentation of the data and their interpretation for geomorphological analysis.
 Contour presentation of topography (as actual contour lines or color shaded renditions) is the manner in which topographic information is commonly portrayed for geomorphological and geological analyses (Figure 1b). This technique is not suitable, however, for obtaining a simultaneous overview of a wide range of spatial scales. A gray-scale image can present only major topographic variations, because the human eye can distinguish only 15–20 shades of gray. Use of color-coded images increases the perceptible dynamic range a few times. The color-coded images are less suitable, however, for several aspects of visual geomorphologic analysis. This is because the same small features displayed as variations of dark gray and light gray are automatically accepted by the human eye as the same, while the same features displayed as shades of different colors are not. Gray scales or color scales can be repeated a few times. This increases the perceptible dynamic range but decreases the suitability for geomorphological analysis.
 The use of derivatives, that is, imaging of slopes rather than, or in addition to, mapping of elevations, provides a set of powerful methods for visual presentation of topographic data of a wide dynamic range. Such methods are based on a common property of planetary topography: smaller features usually have steeper slopes. Among these methods the most suitable for geomorphological analysis is the presentation of simulated shaded relief (“gradient maps”) showing steep (small-scale) features overlain with color-coded topography showing regional-scale features. Figure 2a shows an example of artificial topography with a set of features of different scale and steepness. In the images shown in this example and in this paper, the brighter shades denote higher relative elevations. Simulated shaded relief for this example (Figure 2b) demonstrates an important characteristic of this type of data presentation: it can emphasize or deemphasize some features depending on the mutual orientation of the slopes and the simulated light source.
 There are numerous topographic features on the northern plains of Mars whose lateral scale is of the order of tens of kilometers but whose characteristic slopes are not much steeper than those of the regional relief. Slope-imaging methods do not give the best visual presentation of such features (Figure 2b). Our aim is to present the topographic data in a form uniformly showing topographic features with gentle slopes, lateral scales of tens of kilometers, and vertical scales of tens of meters in the northern plains of Mars. To do this, we applied a set of filters that remove long-wavelength (regional) topography from 16 pixels per degree and 32 pixels per degree gridded topography data.
 Application of any filter modifies topography and can produce artifacts in the filtered images. Linear high-pass filters produce distortions related to sharp breaks in regional slopes; for example, deep troughs appear in the filtered topography along the polar cap base or the dichotomy boundary. Figure 2c shows an example of the application of a simple linear high-pass filter: for each point the mean elevation in a circular vicinity of the point (filter window or core) was subtracted from the elevation at this point. It is seen that small-scale details along the base of a steep large-scale slope are obscured. Other kinds of linear high-pass filters give similar results. To reduce such unwanted effects, instead of linear filters we applied a “median high-pass filter”: for each point, the median elevation in a circular window was subtracted from the elevation at this point. Our artificial example (Figure 2d) shows that in this case it is possible to trace subtle features (two low ridges) through the break in the large-scale slope.
 In this paper we study the northern lowlands using topography to which was applied “median high-pass filters” with circular windows of 50, 100, and 200 km in diameter. We refer to these data as “detrended” topography. We present it as gray-scale images (Figure 1c). We stretched the shade scale to show features tens of meters high. Steep features at the tens of kilometer scale, e.g., impact craters, are saturated: their higher parts are “whiter than white,” and their lower parts are “blacker than black.” Knobs and grooves, as well as the previously detected 3-km-scale background topography of the Vastitas Borealis Formation [Kreslavsky and Head, 1999, 2000], are not completely resolved in the data used and sometimes are not seen in the detrended images. For example, in regions of polygonal terrain a number of linear depressions are seen, but they do not form a continuous network, while higher-resolution images clearly show a polygonal groove pattern.
 All smooth topographic features much larger than the window size are filtered out and are not seen in the images at all. Examples of such completely removed features are gentle global slopes and regional slopes along the dichotomy boundary and in the Utopia basin. The filtering procedure leaves unchanged all topographic features where both dimensions are sufficiently smaller than the window. Understanding of the filtering procedure and comparison of the detrended topography produced with different window sizes allow one to distinguish between real topographic features and the types of candidate artifacts described above (see also Figure 2). We are confident that all of the features discussed in this contribution are real.
1.4. Scope of Analysis
 Removal of the regional slopes has allowed us to detect and assess subtle topographic variations and geologic structure in the northern lowlands [Head and Kreslavsky, 2000], and we report on these analyses here. We have used these techniques to explore the regionally very flat and smooth topography of the northern lowlands [e.g., Smith et al., 1998, 1999] in order to detect important topographic elements that might be obscured by regional topographic trends in normal contour map portrayals. Specifically, we use these data to analyze (1) the presence and distribution of linear ridges and wrinkle-ridge-like structures, (2) the presence of fretted terrain and its extension into the northern lowlands, (3) the traces of outflow channels down into the northern lowlands, (4) the nature of structural patterns on the floors of the Isidis, Utopia, and North Polar Basins, (5) the presence of subtle and buried impact craters, (6) the role of volcanic plains in the history of the northern lowlands, (7) the nature of the Vastitas Borealis Formation, and (8) the relationship of these units, structures, and features to the stratigraphy and history of the northern lowlands.
2.1. Northern Lowlands: Geologic Characteristics and Structure
 The dominant trend in the North Polar Basin (Figures 1c, 1d, 1e, and 4) is formed by a very distinctive set of linear parallel ridges striking across the basin in a broadly arcuate pattern between Tharsis and Utopia (Figures 1a–1c). The trend is first seen in the lowlands in Chryse Planitia and extends from there for a distance of 1500 km through Acadalia Planitia, where it is interrupted by the island composed of a Noachian-aged dissected unit. It then stretches for another 2000 km deep into the North Polar Basin until it reaches the vicinity of the North Pole, where it is obscured by overlying polar and circumpolar deposits [Tanaka and Scott, 1987; Fishbaugh and Head, 2000] for a distance of ∼1250 km. It emerges from beneath these deposits north of Alba Patera and then extends through Arcadia Planitia for ∼1750 km to the edge of Amazonis Planitia, where it is clearly covered by smooth units of the Amazonian-aged Arcadia Formation for a distance of ∼1500 km. The trend reemerges from beneath the Arcadia Formation in Amazonis [see Plescia, 1993] and extends for a distance of ∼1000 km into the highlands region in the vicinity of Medusae Fossae. The total distance of this trend in the North Polar Basin is ∼9000 km, and the pattern is clearly arcuate around the northern part of Tharsis (Figures 1a, 1c, 3, and 4).
 To a first order these data show that the almost complete lack of structure in the northern lowlands [e.g., Chicarro et al., 1985; Watters, 1993; Scott and Tanaka, 1986; Tanaka and Scott, 1987; Greeley and Guest, 1987] (Figure 3) was apparently due to the inability of Viking image data to reveal structure. This may have been due to the commonly hazy atmosphere above the lowlands, superposition of younger units, such as the Vastitas Borealis Formation, or the eolian modification and smoothing of much of the surface. The distinctive and pervasive structure revealed by the MOLA detrended data appears to complete much of the regional circumferential trend around Tharsis previously mapped outside the northern lowlands [e.g., Chicarro et al., 1985; Watters, 1993].
 In general, the map patterns of the surface structures in the northern lowlands can best be described as linear to anastomosing (Figures 4 and 5). Individual features are broadly parallel, but they wax and wane in their distinctiveness and width along strike and have lengths ranging up to several hundreds of kilometers. The vast majority of the occurrences would be described as parallel, in which linear to somewhat anastomosing parallel ridges dominate. In some places (e.g., the eastern part of Vastitas Borealis (0°) (Figure 5d); the eastern part of Acidalia (Figure 5e); and Chryse Planitia (Figure 5f)) the pattern is somewhat more polygonal. The similarity of shapes in map view and patterns of distribution between the linear features in the northern lowlands and wrinkle ridges elsewhere on Mars [Watters, 1993] is striking and lends support to the interpretation of these ridges as wrinkle ridges.
 Some of the characteristics of these individual features can be determined from their geometry and relationships in these detrended maps. In order to investigate the spacing and heights of these features, we compiled six profiles of the detrended data spaced across their zone of occurrence (Figures 1f and 5a–5f). Each profile was oriented normal to the trend of the ridges and extended for ∼4000 km in length. These profiles were placed underneath the detrended image and compared; the peak corresponding to each linear feature was identified where it crossed the profile, and these data provided the basis for compiling a histogram of linear feature heights and spacing, taking into account and correcting for the projection and orientation of features relative to the profile. On the basis of 162 measurements across a total linear distance of ∼14,000 km, the mean spacing of these features is ∼83 km (Figure 6).
 Using the detrended MOLA data, we made similar measurements of wrinkle-ridge spacing in Lunae Planum and Syria Planum (Figures 5g and 6b). On the basis of six profiles crossing 69 ridge structures, we found that the mean spacing was ∼50 km, consistent with the results of Allemand and Thomas  and Watters . Comparison of the mean spacing between ridges in the circum-Tharsis rise region and the northern lowlands (Figure 6) shows that the mean northern lowlands spacing (∼83 km) is almost 1.7 times that of ridges in the Eastern Solis Planum and Lunae Planum (∼50 km).
 The shapes and patterns of distribution of the linear features are also seen and can be compared to wrinkle ridge patterns in Lunae Planum and Syria Planum (Figure 5). Wrinkle-ridge patterns can be described in three main categories: (1) parallel, in which linear parallel ridges dominate (e.g., Syria Planum, Lunae Planum, and Coprates), (2) orthogonal, in which there is a major linear parallel trend and an orthogonal minor trend, and (3) polygonal, in which the linear parallel trends are interrupted by variable angular orientations, crater rim structures, and other features which degrade parallel or orthogonal trends (e.g., Hesperia Planum). Polygonal trends often occur in regions where more heavily cratered terrain or degraded highland terrain lies at shallow depths below the surface (e.g., Hesperia Planum [Watters, 1993, Figure 6] and Arcadia Planitia [Plescia, 1993]).
 The detailed topographic structure of these features can be determined using individual MOLA altimetric profiles and gridded altimeter data (in this area the grid data point density is 1/32 degree). We examined the height of all 162 structural elements along the individual profile traverses where the spacing measurements were made. These data show that the linear elements are subdued ridges that have a mean height of ∼48 m (Figure 6c). As shown in selected profiles (Figure 7), their shape is very similar to wrinkle ridges elsewhere on Mars [e.g., Watters, 1993]. Comparison of this mean height (∼48 m) to measurements of ridges in the detrended topography in the Eastern Solis Planum and Lunae Planum region (Figure 6d) (mean ∼65 m) shows that the ridges are typically ∼1.4 times higher outside the North Polar Basin.
 These comparisons show that there is a tendency for the ridges we measured in the northern lowlands to be more widely spaced and somewhat lower than wrinkle ridges in Hesperian-aged ridged plains in Eastern Solis Planum and Lunae Planum. The differences are not dramatic (mean separation distances 83 km versus 50 km, or 1.7 times; mean heights 48 m versus 65 m, or 1.4 times), as shown in Figures 6a–6d.
 The characteristics of these linear features that we have mapped in the northern lowlands in terms of their spacing, shapes, and patterns in map view, heights, and cross-sectional profiles, together with their general continuity with the circumferential trends around Tharsis (compare Figures 1e, 3, and 4), support the interpretation that these features are wrinkle ridges. This interpretation is further supported by previous mapping that shows that wrinkle ridges in the circum-Tharsis area outside the northern lowlands lie predominantly in the Hesperian-aged ridged plains (Hr) and that this unit underlies younger stratigraphic units and deposits of the northern lowlands (primarily the Vastitas Borealis Formation) [e.g., Scott and Tanaka, 1986; Tanaka and Scott, 1987; Greeley and Guest, 1987; Plescia, 1993; Rotto and Tanaka, 1995].
 If these features are so similar to wrinkle ridges, why were they not detected before? Ridges on Hesperian ridged plains often consist of a wide arch and a narrow sharp ridge [e.g., Watters, 1991]. Ridges observed in the detrended data in the northern lowlands are usually just arches and do not typically have sharp narrow ridges on top. Elsewhere, the sharp narrow ridges are commonly much more readily observed in images owing to their steeper slopes. In the northern lowlands the atmosphere is usually less transparent owing to hazes and the polar hood. Because of this, when the Sun is low, scattered light obscures surface topography, and when the Sun is high, the gently sloping topography of arches is not seen and the steeper ridges are subdued.
 The trend toward broader spacing and more subdued heights of the ridges in the northern lowlands (Figure 6), together with the sedimentary and volcanic nature of many of the subsequent units, suggests that blanketing, burial, and subdual are also among the explanations. Examination of individual areas and profiles (Figure 5) further documents some of the candidate explanations. In Amazonis Planitia (Figure 5a), ridge spacing typical of the west central region increases, and heights decrease to the east, where the ridges are partly, and then wholly, buried by Amazonian-aged plains of the Arcadia Formation. To the west in this profile, a similar trend is seen as the ridges are buried by Amazonian-aged flows from Elysium Mons [see also Plescia, 1993]. In Arcadia Planitia (Figure 5b; lower left part of image) and Vastitas Borealis (180°) (Figure 5c; left part of image), Late Hesperian flows from Alba Patera embay and subdue the ridges. In the Vastitas Borealis (180°) and (0°) areas (Figures 5c and 5d), polar and circumpolar deposits subdue and cover the ridges. In the Chryse Planitia region (Figures 5d and 5f), superposed Late Hesperian channels and channel deposits have eroded and subdued the ridges.
 Amazonian modification of the surface is also seen in abundance in Mars Orbiter Camera (MOC) images. We examined numerous MOC images covering key wrinkle ridges in Lunae Planum and seen in the detrended altimetric data. The narrow width of the MOC images means that most high-resolution strips cover only a part of the ridge, and virtually no evidence of the wrinkle ridges can commonly be seen in images of the northern lowlands. Instead, one observes a variety of very lightly cratered deposits and textures that suggest the recent reworking of a significant part of the surface [Kreslavsky and Head, 2000]. Clearly, this serves as an important factor in obscuring the wrinkle ridges in the images.
 The orientation of these features can also be determined from angular measurements made along these profiles (Figure 8). This plot also emphasizes the parallelism of the linear trends seen in the northern lowlands and shows that they curve in an arcuate pattern around Tharsis, essentially completing the arc to the north, with a center of curvature located at about the highest part of the Tharsis rise (8°S, 95°W). This trends stands out distinctly (Figure 1c) from others in the Utopia Basin, in Isidis Planitia, and in the area north of Deuteronilus Mensae (near Lyot crater), which we discuss further below.
 On the basis of these data and comparisons, we conclude that the North Polar Basin contains extensive linear features that are most plausibly interpreted as wrinkle ridges and that the arcuate trend documented here provides a major missing section of the previously mapped circumferential trend around Tharsis [e.g., Chicarro et al., 1985; Watters, 1993] (Figures 3 and 8). The exposed circum-Tharsis wrinkle ridges previously mapped [e.g., Chicarro et al., 1985; Watters, 1993] are predominantly on the ridged plains unit (Hr) of Early Hesperian age. The generally obscured and more subdued nature of these features in the northern lowlands appears to be attributable to modification by later Hesperian (Vastitas Borealis Formation; Alba Patera Formation; channel deposits) and Amazonian-aged sedimentary and volcanic units. Comparison of the ridges located in these two areas illustrates this and suggests that sedimentary and volcanic modification of a surface that was originally similar to the Hesperian-aged ridged plains (Hr) could produce the observed differences in ridge spacing and height (Figure 9). For example, in the central part of the Figure 5a image, prominent wrinkle ridges of Hesperian ridged plains (Hr) are exposed and strike normal to the profile and are seen as distinctive peaks and troughs in the profile. To the left of this area, the wrinkle ridges are progressively buried by younger volcanic and fluvial deposits in Amazonis Planitia (in this area, note the exposed wrinkle ridges at the top of the image and the buried remnants along the profile and below). This type of configuration is similar to the modification scenario illustrated in Figure 9 (middle) in which volcanic flooding dominates. In comparison, examination of Figures 5c and 5d shows that the ridges deeper in the North Polar Basin are more subdued than those seen in the Hesperian ridged plains in the middle of Figure 5a but are not surrounded or completely buried as suggested by the relationships in Amazonis Planitia seen in the left part of Figure 5a. The type of configuration seen deeper in the North Polar Basin (Figure 5c and 5) is similar to the modification scenario illustrated in Figure 9 (bottom), in which sedimentary deposition dominates and ridges become subdued and partly obscured. On the basis of these examples we interpret much of the difference in ridge height and spacing characteristics (Figures 5 and 6) to be due to postformation volcanic and sedimentary modification processes, as illustrated in Figure 9.
Montési and Zuber  have suggested that differences in ridge spacing between the ridged plains and the northern lowlands reported in this study are primary and reflect a localization instability due to different crustal and lithospheric structure between the two areas. We believe that the relatively small differences in mean heights and spacings between the two populations (Figure 6) are reasonably interpreted in terms of the subdual of a northern lowland population of wrinkle ridges by subsequent sedimentary and volcanic deposition (e.g., Figure 9). A more comprehensive analysis of the morphometry of northern lowlands ridges is necessary to determine if there is a statistically significant difference in primary ridge spacing, as required by the interpretation of Montési and Zuber .
2.1.2. Isidis Basin trends
 Isidis is the site of a large impact basin [Schultz and Frey, 1990]. The Isidis Basin interior is characterized by Hesperian-aged Vastitas Borealis Formation (Hvr, ridged member), Amazonian smooth plains (Aps), and Hesperian ridged plains (Hr) (along its northeastern margin) [Greeley and Guest, 1987]. Wrinkle ridges were not generally mapped in the interior of the Isidis Basin in previous regional and global studies [e.g., Chicarro et al., 1985; Watters, 1993]. The detrended topographic map of the Isidis Basin, however, reveals a remarkable configuration of ridges (Figure 10). In the outer portions of the interior of the basin, a series of irregular radial spoke-like features dominates the trend. In the central interior part of the basin (the inner 300–400 km), the ridges display a more random orientation. Superposed on these trends is a general WNW-ESE ridge trend extending through the middle of the basin. Finally, it is clear that circular ridge configurations are more common near the margins of the basin interior, with diameter distributions similar to impact craters in the adjacent cratered terrain. This last observation is interpreted to mean that wrinkle ridges are influenced by the rim crest topography of buried craters, a phenomenon that is more common in shallowly buried regions such as basin margins [e.g., Allemand and Thomas, 1995; DeHon and Waskom, 1976].
 We thus interpret the Isidis Basin ridges detected in the detrended topography to be a system of wrinkle ridges, showing many of the same characteristics of wrinkle ridges as seen in the North Polar Basin. In Isidis, however, the pattern is largely controlled by the circular configuration of the basin, in contrast to the more linear nature of the North Polar Basin. The pattern in Isidis is similar to those seen in wrinkle ridges in the lunar maria (see maps of Solomon and Head ). The lunar wrinkle ridges are attributed to lithospheric loading by mare basalt accumulation within the basins, with superposition of a global thermal stress that shifted to compressional as the Moon changed from net expansion to net contraction in its early history [Solomon and Head, 1980; Watters, 1993, Figure 6].
 Why were these wrinkle ridges not detected in earlier analyses [e.g., Chicarro et al., 1985; Watters, 1993]? Apparently, the veneer of deposits represented by Hvr and Aps was sufficient to obscure both the morphologic and topographic expression of these features in Viking Orbiter images. Specifically, Grizzaffi and Schultz  mapped hillocky and small-scale ridged terrain throughout the interior of Isidis which they interpreted as a transient air fall deposit of ice and dust which obscured the underlying units and subsequently underwent sublimation and partial removal [see also Tanaka et al., 2000]. Such an overlying residual blanket could readily modify the morphology of wrinkle ridges in much the same way as it appears to have done in the North Polar Basin. Contiguous outcrops of the Hesperian-aged ridged plains in the northeastern part of the basin [Greeley and Guest, 1987] strongly suggest that this unit underlies the subsequent units (e.g., Hvr and Aps) and is the unit on which the wrinkle ridges detected in the basin interior are formed.
2.1.3. Utopia Basin circumferential and radial trends
 The Utopia Basin was previously predicted to be the site of a large and degraded impact basin on the basis of geological structure [McGill, 1989], a hypothesis confirmed by MOLA data [Frey et al., 1999; Thomson and Head, 2001]. The interior of the Utopia Basin is characterized by the Hesperian Vastitas Borealis Formation (all four members) overlain by Amazonian deposits associated with Elysium (lava flows and other lobate and smooth deposits) [Greeley and Guest, 1987]. Along the southern margin of the Utopia Basin a band of Hesperian ridged plains (Hr) about 2000 km long and 500 km wide is exposed [Greeley and Guest, 1987]. Although a few structural elements were mapped within Utopia, the general distribution of structures and wrinkle ridges was minor to nonexistent [e.g., Chicarro et al., 1985; Watters, 1993].
 The detrended topographic map of the Utopia Basin (Figure 11), however, reveals abundant ridges and a distinctive configuration apparently related to processes similar to those seen in Isidis and the North Polar Basin. To the north of the basin is seen the general linear trend of the North Polar Basin wrinkle ridges (compare top of Figure 11 and Figures 1c and 1e). As one approaches the margins of the basin, the ridges become more circumferential, and this broad pattern gives way to a series of generally radially oriented ridges with some zones of more randomly oriented ridges in the southwestern part of the basin adjacent to the heavily cratered terrain. In one such area of the basin (west central) a buried crater forms a circular structure with ridges extending away like spokes, suggesting that the ridge-forming activity was partly controlled by the presence of impact craters. The Hesperian-aged deposits of the Vastitas Borealis Formation contain the most distinctive and abundant ridges. The Amazonian-aged deposits of the Elysium Formation contain few ridges, and the unit appears to embay those units (primarily Hv) with ridges [Tanaka et al., 1992b; Thomson and Head, 2001]. The morphology, exposure patterns, and distribution of the majority of the ridge structures are consistent with those of wrinkle ridges, as previously described in other areas of the northern lowlands. The ridges forming the radial system appear wider, less sinuous, and more asymmetric in profile in comparison to many wrinkle ridges. Some concentric ridges may be related to aqueous processes or to modification of tectonic features by water and ice processes [e.g., Head et al., 1999; Thomson and Head, 2001]. The lack of detection of the vast majority of these structures in previous studies can also be attributed to the modifying and subduing effect of the Vastitas Borealis Formation and the virtually complete burial by the Amazonian-aged deposits of the Elysium Formation [e.g., Thomson and Head, 2001]. The exposure of the pre-Vastitas Borealis Hesperian-aged ridged plains (Hr) at the southern edge of the basin (and the wrinkle ridges seen there) [Greeley and Guest, 1987] strongly suggests that the unit on which the wrinkle ridges are formed is the Hesperian ridged plains (Hr), covered and obscured by later units (particularly the Vastitas Borealis Formation).
2.1.4. Polar and circumpolar deposits
 Polar deposits of Amazonian age (permanent ice, Api, and layered terrain, Apl), as well as Amazonian circumpolar mantling deposits, are superposed on the Vastitas Borealis Formation in the region of the North Pole [Dial, 1984; Tanaka and Scott, 1987; Fishbaugh and Head, 2000]. Structural elements associated with the Vastitas Borealis Formation (other than the features defining the various facies) were not observed prominently or in abundance in regional maps [Dial, 1984; Tanaka and Scott, 1987]. The MOLA detrended topography (Figure 12; see also Figures 1, 4, and 5c) shows, however, that abundant linear ridges can be seen in the Vastitas Borealis Formation in the vicinity of the North Pole. Ridges interpreted to be wrinkle ridges in earlier sections cross the North Polar Basin in the vicinity of these deposits and are further covered and obscured by the circumpolar deposits and completely buried where they underlie the polar deposits. This suggests that by the time of the emplacement of these Late Amazonian-aged deposits, wrinkle ridging had ceased, supporting the interpretation that the majority of wrinkle ridging took place relatively soon after the emplacement of the Hesperian plains (Hr).
2.2. Northern Arabia Terra-Deuteronilus Mensae
 The highland-lowland boundary in northern Arabia Terra (Figure 1a) is one of the most complex along the dichotomy boundary. Noachian-aged plains of the plateau sequence have been resurfaced by Noachian (Nplr) and Hesperian ridged plains (Hr), and these in turn have been cut by several stages of Hesperian channel development to produce the distinctive texture of the fretted terrain [Greeley and Guest, 1987; McGill, 2000]. Fretted terrain erosion occurred most likely in the Early Hesperian and involved liquid water [McGill, 2000]. These deposits are embayed and overlain by the Late Hesperian Vastitas Borealis Formation (Hvr and Hvk) in Acidalia Planitia. A significant part of the region has been modified by Amazonian-aged mass wasting, smooth plains emplacement, and formation of debris flows on the channel floors [McGill, 2000], as well as by ejecta from the crater Lyot. The orientation of wrinkle ridges in the upland Noachian and Hesperian ridged plains is NNW to NW; no wrinkle ridges are mapped in the Vastitas Borealis Formation in the adjacent lowlands [Chicarro et al., 1985; Greeley and Guest, 1987; Watters, 1993].
 Examination of the detrended topography in this area (Figure 13) reveals a remarkable transition across the dichotomy boundary. To a first order, the crater Lyot and its ejecta dominate part of the region, but the distinctive texture of the fretted terrain can be traced several hundred kilometers into the lowland areas mapped as the Vastitas Borealis Formation. The individual bumps of Hvk, the knobby member of the Vastitas Borealis Formation, dominate the upper left-hand part of Figure 13, but near this can be seen broad linear ridges extending in a NW direction, ridges that extend toward the North Pole and converge with the circum-Tharsis trend in the North Polar Basin (see Figure 1). These trends are parallel to the linear trends represented by the wrinkle ridges of Nplr and Hr in northern Arabia Terra [Greeley and Guest, 1987] and were apparently continuous across the complex transition zone now represented by the fretted terrain.
 Thus, on the basis of these new data (Figure 13), we conclude that (1) the complex transition zone of the fretted terrain extends several hundred kilometers out into the lowlands beneath the Vastitas Borealis Formation, (2) wrinkle ridges predating the formation of the fretted terrain extend from the uplands across the transition zone and into the basement of the northern lowlands, and (3) on the basis of the stratigraphy of this region [Greeley and Guest, 1987; McGill, 2000] and the relationships revealed by these new data, it is likely that Hesperian ridged plains (Hr) exposed in the adjacent uplands extend well into the northern lowlands in this region below the Vastitas Borealis Formation (Figures 1 and 13).
2.3. Hesperian Outflow Channels of Chryse Planitia
 A series of major Hesperian-aged outflow channels extends from the cratered uplands and across Hesperian-aged ridged plains (Hr) into the northern lowlands in Chryse and Acidalia Planitia [Scott and Tanaka, 1986; Rotto and Tanaka, 1995] (Figure 1a). The northern lowlands in this area is characterized predominantly by the Vastitas Borealis Formation (all four members), a kipuka of Noachian dissected plains, and several Amazonian units. Wrinkle ridges of the Hesperian ridged plains (Hr) trend in the broad circum-Tharsis mode (e.g., Lunae Planum, parts of Xanthe Terra) or more NNW (northern Terra Meridiani). Only a very few wrinkle ridges have been mapped into the northern lowlands [Scott and Tanaka, 1986; Chicarro et al., 1985; Watters, 1993].
 The detrended topography in this area (Figure 14) reveals an abundance of linear ridges that are continuous with the wrinkle ridge trends of the adjacent uplands and shows how the traces of the channels have continued into the northern lowlands, beyond the major change in channel morphology [e.g., Ivanov and Head, 2001]. The detrended topography data also show that virtually all of the channels were influenced in their flow patterns by the presence of wrinkle ridges, confirming previous interpretations [Scott and Tanaka, 1986; Rotto and Tanaka, 1995] that the channel-forming events postdated the emplacement of Hr and the formation of the wrinkle ridges on them [e.g., Head and Kreslavsky, 2000]. On the basis of these new data, we conclude that Chryse Planitia and Acidalia Planitia were floored by extensive deposits of Hesperian-aged ridged plains continuous with those observed in the adjacent uplands and overlain by outflow channel deposits and the Vastitas Borealis Formation.
2.4. Amazonis Planitia
 This region, located west of Olympus Mons and east of Elysium (Figure 1), is one of the smoothest areas in the northern lowlands, smoother even than the Vastitas Borealis Formation [Aharonson et al., 1998; Kreslavsky and Head, 1999, 2000]. The region is characterized by members of the Arcadia Formation, an Amazonian-aged unit with visible flow fronts in Viking images, and interpreted to be volcanic in origin [Scott and Tanaka, 1986]. Recent MOC images of the Elysium Planitia to the west have shown evidence of extensive smooth plains of volcanic origin [e.g., Keszthelyi et al., 2000] of recent Amazonian age (within the last few tens of millions of years) [Hartmann and Berman, 2000]. Additional detrended topography data from MOLA provide evidence that the deposits of Amazonis Planitia are laterally continuous with those of Elysium Planitia [Head and Kreslavsky, 2001; Head et al., 2001a]. Underlying these Late Amazonian-aged deposits in the northern lowlands are various members of the Vastitas Borealis Formation and Hesperian-aged ridged plains (Hr) east of Phlegra Montes and the Elysium Rise [Scott and Tanaka, 1986]. The wrinkle ridges of Hr are associated with peaks or crater rims, suggesting relatively shallow burial of a cratered surface by volcanic plains, and subsequent deformation [e.g., Plescia, 1993]. The trends of these features form part of the circum-Tharsis system mapped in the northern lowlands (Figures 1e, 5, and 8). The detrended topographic data reveal the presence of ridges in Amazonis Planitia (Figure 5a, center left; compare to Figures 5c and 5d), but these ridges are more irregular, more widely spaced, and lower than others in the North Polar Basin. We interpret these data to indicate that Hesperian-aged ridged volcanic plains (Hr) underlie both the Vastitas Borealis Formation and Amazonian-aged volcanic plains units (Aa) in this area and that the superposition of the Amazonian-aged volcanic plains on the Vastitas Borealis Formation has further increased ridge spacing and reduced ridge height (e.g., Figure 9) [Head et al., 2001a].
2.5. Vastitas Borealis Formation
 About 45% of the northern lowlands are presently occupied by the Late Hesperian-aged Vastitas Borealis Formation [Tanaka and Scott, 1987], which in turn is covered by the Early Amazonian-aged materials of the Elysium Formation in the Utopia Basin, by the Arcadia Formation in Arcadia and Amazonis, and by Late Amazonian-aged polar and circum-polar deposits [Scott and Tanaka, 1986]. Superposition relationships described above clearly show that the northern lowlands wrinkle-ridge system predates the Elysium Formation (Figure 11), the Arcadia Formation (Figure 5), and the polar and circumpolar deposits (Figure 12). The topography of the wrinkle ridges is observed in the Vastitas Borealis Formation, although in this unit ridge spacing tends to be greater and topographic expression more subdued than wrinkle ridges exposed elsewhere (Figures 6 and 17). A variety of data show that the surface of the northern lowlands is unusually smooth at all scale lengths compared to terrains exposed elsewhere on Mars [Smith et al., 1998; Kreslavsky and Head, 1999, 2000; Aharonson et al., 1998; Head et al., 1999].
 Study of kilometer-scale roughness [Kreslavsky and Head, 2000] has shown that the Vastitas Borealis Formation is characterized by a distinctive dependence of roughness on scale. The dependence of the median differential slope used as a measure of roughness by Kreslavsky and Head  on baseline length has a pronounced maximum at ∼3 km baseline (Figure 15). Virtually no other terrains on Mars possess such a kilometer-scale roughness signature. This characteristic 3-km-scale surface roughness is represented mostly by gently sloping knobs. This knobby pattern is mostly undersampled in the gridded topography maps. A specific texture of the Vastitas Borealis Formation, however, is already seen in 32 pixel per degree detrended topography images (see upper left part of Figure 13). Owing to this specific texture, the boundaries of the Vastitas Borealis Formation are usually clearly seen in the detrended topography maps. The roughness map [Kreslavsky and Head, 2000] allows one to outline the boundaries even more reliably but at lower resolution. At many locations the topography and roughness maps enable one to outline the Vastitas Borealis Formation more reliably than the images. In some areas there are discrepancies between the boundaries seen in the roughness and topography maps and the boundaries mapped by Scott and Tanaka  and Greeley and Guest . It is interesting that the ∼1400-km-long boundary between the Vastitas Borealis Formation and the Alba Patera Formation (Figure 4) is expressed as a distinctive topographic step 10–30 m high. This step clearly shows that the Vastitas Borealis Formation material is superposed on top of the Alba Patera Formation and that its thickness, at least near the edge, is several tens of meters [Kreslavsky and Head, 2001a].
 Geological mapping discussed above shows that Hesperian ridged plains (Hr) are abundant in the uplands surrounding the northern lowlands [Greeley and Guest, 1987; Tanaka and Scott, 1987; Scott and Tanaka, 1986]. In addition, the stratigraphic columns in these maps place the Hesperian ridged plains (Hr) as Early Hesperian in age and the Vastitas Borealis Formation as Late Hesperian in age, essentially contemporaneous with the formation of outflow channels and their deposits. Stratigraphic evidence derived from the new detrended data and cited above strongly suggests that Hesperian ridged plains (Hr) underlie units presently exposed in the northern lowlands in Chryse Planitia, Amazonis Planitia, Acidalia Planitia, and Utopia Planitia. This raises the question of whether the unusual roughness characteristics of the Vastitas Borealis Formation [Kreslavsky and Head, 1999, 2000] could be related to a combination of features associated with this unit and an underlying unit (Hr).
 To address this question, we examined the median differential slope of Hr exposed outside the northern lowlands (Figure 15) and compared it to that of the Vastitas Borealis Formation. Hesperian ridged plains are generally rougher than the Vastitas Borealis Formation, but the Vastitas Borealis Formation is very similar to Hr at intermediate scale lengths (2–5 km) and has lower differential slopes at shorter (<1 km) and longer (>8 km) scale lengths. How thick would a layer of overlaying material have to be to provide enough smoothing to make Hr appear like Hv? To address this question, we used Hesperia Planum as a typical example of Hr.
 Smoothing at a larger spatial scale requires addition of a thicker layer. In addition, the difference between roughness of ridged plains and the Vastitas Borealis Formation increases with increase of baseline length (Figure 15). Hence the result of simulations crucially depends on the largest spatial scale that can be considered as a scale of roughness. We found that for an 80-km baseline the median differential slope is almost the same for any chosen MOLA profile across Hesperia Planum (as well as for shorter baselines), while for a baseline twice as long the median differential slope differs from profile to profile by more than a factor of 2. We conclude that for large patches of ridged plains, slopes at 80-km baseline still reflect the intrinsic surface roughness, while slopes at larger baselines are defined by regional topography.
 Finding the minimal amount of material necessary to reach a given smoothness is a difficult mathematical problem, and so we addressed this question using simulations. We chose several MOLA profiles across Hesperia Planum to characterize the ridged plains, Hr. The total number of points in the segments chosen exceeded 10,000, which provided a robust roughness estimation. Then we modified the profiles by “adding some material” in a way that provided an “economical” smoothing of the profiles. We tried several heuristic algorithms for smoothing by “adding some material.” The most effective algorithm among those tried successively replaces randomly chosen concave segments of profiles with straight segments. We feel that the amount of added material in this algorithm is rather close to the rigorous minimum. Our simulations showed that application of a layer with a mean thickness of ∼90 m can reduce the roughness at 80-km baseline from the level of Hr in Hesperia Planum to the level of the Vastitas Borealis Formation. The algorithm applied produces a very smooth surface at short scales, and an additional 10-m-thick layer is necessary to provide the characteristic 3-km-scale topography (Figure 15). The resulting average thickness of 100 m provides a lower boundary for the mean thickness of the Vastitas Borealis Formation material. Obviously, the thickness can be somewhat higher, because the emplacement process probably does not work like an effective smoothing algorithm.
 Thus we conclude that on the basis of stratigraphic relations and surface roughness analyses, the Vastitas Borealis Formation could plausibly represent a thin veneer of ∼100-m minimum thickness superposed on Hesperian ridged plains (Hr) in the northern lowlands.
 What might be the nature of such a veneer? Among the contemporaneous units in the northern lowland region are the Late Hesperian channel deposits (Hc), representing the formation of the outflow channels and their debouching into the northern lowlands. Although the detailed nature of these events is controversial (e.g., see review by Carr ), all agree that their formation involved emplacement of sediment into the northern lowlands. Carr et al.  estimated that around the Chryse basin (the area of the most significant channel input into the northern lowlands), the volume of sediment eroded and emplaced into the lowlands was about 4 × 106 km3. If the Vastitas Borealis Formation represents a sedimentary unit emplaced over the northern lowlands with a minimum average thickness of ∼100 m, then its minimum total volume would be about 3 × 106 km3, an amount less than, but close to, this value. On the basis of these several considerations, we thus adopt as a working hypothesis the concept that the Vastitas Borealis Formation may represent at least part of a sedimentary layer emplaced during outflow channel formation [see also Tanaka and MacKinnon, 1999]. In the following sections we assess this working hypothesis.
2.6. Impact Craters
 Abundant impact craters with a large variety of morphologic characteristics are observed on Mars (see review by Strom et al. ). The density and size-frequency distribution can be used to date the surfaces of geologic units and place them in the context of the global stratigraphy of Mars [e.g., Tanaka et al., 1992a]. The morphologic characteristics of craters can be used to assess their mode of formation, influence of the substrate, and aspects of the cratering process (e.g., see review by Strom et al. ). Analysis of the detrended topographic data clearly shows the distribution and morphology of the relatively fresh craters mapped in the northern lowlands [e.g., Greeley and Guest, 1987; Scott and Tanaka, 1986; Tanaka and Scott, 1987; Barlow, 1988; Garvin et al., 2000a, 2000b; J. B. Garvin et al., Geometric characteristics of Martian impact craters from the Mars Orbiter Laser Altimeter (MOLA), submitted to Journal of Geophysical Research, 2001] (Figures 1d, 5, 10–14, and 16). Details of the structure of large craters, especially their fluidized ejecta patterns, are clearly seen in the maps. Also revealed in the detrended data, however, is a large number of circular features with a range of diameters and a very subdued morphology (Figure 16d). We interpret these as hidden, or “stealth,” circular depressions, most likely to be subdued and buried impact craters (Figure 16). In this section we analyze the detrended topographic data to assess the morphology and number of impact craters and stealth features and their relationship to the different units observed in the area.
2.6.1. Stealth craters
 There are many smooth, shallow, flat-floored circular depressions in the northern lowlands (Figure 16). Their perfect circular form and apparently random scattering around the surface indicate their impact origin. We called such objects stealth craters. Some of them can be recognized in Viking low-resolution mosaics, while many of them are too smooth to be visible in the images, as is the case with most of the wrinkle ridges. In the detrended topography maps the stealth craters clearly differ from fresh craters: all fresh craters have distinctive rims and ejecta patterns. Stealth craters appear only within the boundaries of the Vastitas Borealis Formation (Figure 16e). We found no stealth craters in the Hesperian ridged plains of Lunae Planum, suggesting that such craters did not form during the emplacement of Hr. Rather, these craters were probably formed before or during the Vastitas Borealis Formation emplacement and then heavily modified and obscured by the Vastitas Borealis Formation material.
 We systematically studied the stealth crater population and compared it to the population of real craters within the Vastitas Borealis Formation. For the purposes of this study we took the boundaries of the Vastitas Borealis Formation from the geological maps [Scott and Tanaka, 1986; Tanaka and Scott, 1987; Greeley and Guest, 1987]. For several segments of the outer boundary, where the edge of the characteristic roughness signature in the roughness map [Kreslavsky and Head, 2000] clearly deflected from the mapped boundary, we adjusted the boundary. We included the circumpolar mantling (unit Am of Tanaka and Scott ) and large craters mapped as separate units into the Vastitas Borealis Formation outline. The total area within this outline is 16.9 × 106 km2.
 We identified and mapped all stealth craters in the northern lowlands using the detrended topography maps (Figure 16e) [Kreslavsky and Head, 2001a]. We calculated the diameter of each stealth crater as the diameter of a circle having the same area as the mapped depression. The largest stealth crater is 169 km in diameter (this object is located on the western slope of Utopia basin and was mentioned in section 2.1.3); the second largest stealth crater is 75 km. The smallest identified stealth craters are ∼6 km. Identification of stealth craters smaller than 10–15 km is not completely reliable owing to the limited resolution of the topographic data. There are several clusters of irregularly shaped smooth depressions in the region under study. Large stealth craters can be easily distinguished from such depressions owing to their perfect circular shape. However, the shape of small depressions cannot be reliably assessed because of the limited actual resolution of gridded topography data. The same reason causes the diameter estimation of the small craters to be inaccurate. For quantitative characterization of the stealth craters population we used only craters larger than 16 km. There are 140 such stealth craters, and their size-frequency distribution is shown in Figures 16a and 16b. The mean density of the stealth craters within the Vastitas Borealis Formation boundaries is 8.3 per 106 km2.
 Straightforward use of these data for age estimation is not possible, because our diameter measure is different from that used to establish the age scale on Mars [Tanaka, 1986]. Rim crest-to-rim crest diameter is traditionally used as a size measure in all crater population studies, but this cannot be simply applied to stealth craters because they do not have distinctive rims. Strictly speaking, we do not know the details of the crater modification process; hence we cannot unambiguously tie our measured diameters to any morphometric characteristics of pristine craters. We proceed by making the reasonable assumption that the size of a stealth crater is equal to the size of inner depression of a pristine crater. We identified all real craters within the same outline using the detrended topography maps and measured their sizes in the same manner as we did for the stealth craters. Of course, we included central peaks and peak rings in the inner depression. The total number of real craters larger than 16 km is 80. Their density is 4.7 per 106 km2.
 We compared the lists of craters obtained in this study with the crater catalog kindly provided to us by N. Barlow and described by Barlow . All objects identified as real craters in our study except one were present in the catalog. Crater sizes measured through the area of the inner depression turned out to be 5–30% smaller than the rim-to-rim diameters listed in the catalog. Among 140 stealth craters, only 14 were present in the catalog; 13 of them are situated near the outer boundaries of the Vastitas Borealis Formation, mostly in western Elysium Planitia and southeastern Utopia Planitia.
 The size-frequency distribution of the stealth craters turned out to be remarkably similar to that of real craters (Figures 16a and 16b). It is probable that we see almost all craters formed on the substrate before or during the emplacement of the Vastitas Borealis Formation. If some craters were totally erased by the Vastitas Borealis Formation emplacement, we would expect that smaller craters would be more likely to be erased than large craters; hence we would find a noticable relative paucity of small stealth craters. Thus we can use the total number of stealth and real craters (220, which gives a density of 13.0 per 106 km2) to estimate the age of the substrate, as discussed below.
 We measured the depth of the craters as the difference between the elevation of the surroundings and the elevation of the crater floor. The elevation of the crater floor was calculated as the median elevation within a circle around the crater center with the diameter twice smaller than the crater diameter. We checked that for the distinctive, nonstealth craters, the elevation calculated in this way usually reflects the elevation of the floor rather then the elevation of the central peak or peak ring. The elevation of the surroundings was calculated as the median elevation within a ring between 4 and 4.25 crater diameters from the center. For the real craters, such a ring is usually outside the ejecta. The measured depth is plotted against crater size in Figure 16c. Several pedestal craters have negative depths: their floor is higher than their surroundings. The depth of real craters is widely scattered, though large craters are generally deeper. The depth of the stealth craters is scattered around 100 m, and no dependence of the depth on crater size is observed. Only two stealth craters are deeper than 200 m, and one of them is the largest stealth crater. The very shallow nature of the stealth craters is consistent with their extensive erosion and infill after their formation on an underlying unit (most likely Hr). The depth distribution of the fresher craters suggests that they have not undergone the same level of erosion and infilling. The fresher craters are superposed on the Late Hesperian Vastitas Borealis Formation. The contrast in crater depths between these two populations suggests that the stealth craters underwent modification and shallowing during Late Hesperian, most likely coincident with the emplacement of the Vastitas Borealis Formation. Among the candidate processes of modification of these unusual craters are (Figure 9) (1) erosion and deposition associated with the emplacement of the outflow channels (which might degrade the rims and infill the crater interiors), (2) impact into a standing body of water [e.g., Lindström et al., 1996] (which might involve catastrophic backwash, tending to cause erosion of the rims and infilling of the interior), (3) glacial planing [e.g., Kargel et al., 1995] (which might remove the rim and infill the crater), (4) enhanced sedimentation in standing bodies of water (extreme sedimentation rates of 50 cm/yr are known in terrestrial submarine craters [e.g., Thatje et al., 1999]), and (6) eolian redistribution of sediment (which has not affected the subsequent craters in the same manner).
2.6.2. Estimation of ages of surfaces
 Crater densities for the chronostratigraphic series on Mars were defined by Tanaka  for craters larger than 1, 2, 5, and 16 km in diameter. Densities for craters larger than 16 km were established for the Noachian only, because for younger terrains the density is too small to be useful in most cases. For the whole region of the Vastitas Borealis Formation, however, the total number of craters >16 km in diameter is large enough for statistical interpretation.
 For smaller craters the Hesperian spans a factor of 3 in crater density [Tanaka, 1986]. We assume that the density of craters >16 km changes in approximately the same way from the lower to the upper boundary of the Hesperian. The density of real craters within the Vastitas Borealis Formation is 2.75 times lower than the density of real + stealth craters. Thus, if we postulate the Vastitas Borealis Formation to be Upper Hesperian and close to the Amazonian boundary, the substrate will be Lower Hesperian and close to the Noachian boundary.
 To use the value of crater density itself rather than the density ratio, we need to compensate for the bias due to the difference in the crater size definition. N. Barlow's catalog contains 167 craters larger than 16 km for the rim-to-rim diameter within the unit outline. Among them, 26 craters do not have noticeable rims or ejecta patterns in the detrended topography, including 14 stealth craters larger than 16 km, according to our survey described above. We interpret that the rest (167 − 26 = 141) of the craters in the catalog postdate the Vastitas Borealis Formation. This number is 1.76 times greater than our number of real craters. This difference is due to the difference in crater size definition, and the factor 1.76 can be used to compare our crater densities with the commonly used system. Applying this factor to our density of real + stealth craters, we obtain 23 per 106 km2, just a little less than the Hesperian/Noachian boundary defined by Tanaka  at 25 per 106 km2. This result is in agreement with that which we obtained above using the density ratio. A population of larger subdued quasi-circular depressions described by Frey et al.  and interpreted to represent an underlying Noachain-aged surface is discussed in more detail later.
2.6.3. Interaction of ridges and craters
Allemand and Thomas  studied crater-ridge relationships for the Lower Hesperian ridged plains in the Coprates region. They found a number of examples of ridge-crater intersections, where (1) ridges postdate craters, (2) ridges predate craters, and (3) ambiguous crosscutting relationships occur. An analogous study for the northern lowlands is made difficult by the subdued character of all features and the relatively low map resolution; most of crater-ridge intersections would be classified as ambiguous. However, one of the intersection types where ridges postdate craters, namely, “ridges stopped or modified by crater intersection,” is clearly identifiable in the detrended topography maps. There are several examples where the ridge pattern definitely “knows” about the existence of a stealth crater. The best example is the previously mentioned spoke-like ridges near the largest stealth crater in Utopia (section 2.1.3; Figure 11). Several other stealth craters serve as terminations of the ridges. There are a few sites where ridges split or cross each other at stealth craters. These observations show that ridge formation occurred (or continued) some time after the deposit of the substrate material. The observations do not provide good constraints on this time; however, it is clear that it could be a measurable part of the Hesperian. There are no convincing examples of ridges postdating real craters. We found two sites (61°N, 272°W and 60°N, 224°W) where ridges might be thought to be influenced by real craters, but these examples are ambiguous. There are numerous examples where ridges are covered and obscured by crater ejecta. This suggests that wrinkle ridge formation had ceased by some time in the Late Hesperian, although the exact duration and time of cessation is unknown.
3. Discussion and Synthesis
3.1. Northern Lowlands Wrinkle-Ridge System
 Detrended topographic data provide substantial evidence for the presence of a comprehensive system of structural features throughout the northern lowlands (Figure 1c) that was essentially undetected in previous data sets (Figure 3) [Chicarro et al., 1985; Watters, 1993]. Evidence supporting the interpretation of these features as wrinkle ridges includes their patterns in map view (Figures 1c and 5), cross-sectional topography (Figure 7), spacing (Figure 6), and continuity with wrinkle ridges in the Tharsis region (Figure 8) and elsewhere (Figures 10–14).
3.2. Subdivisions of the Northern Lowlands Wrinkle-Ridge System
 Trends of wrinkle ridges in the northern lowlands (Figures 1c and 1e) can be subdivided into (1) Isidis Basin, a generally equidimensional system (Figure 10), (2) Utopia Basin, a system exhibiting both concentric and radial ridges (Figure 11), (3) circum-Tharsis, a system that stretches for 7000 km in an arcuate pattern around the Tharsis rise (Figures 4 and 8), and (4) Northern Arabia Terra, a system that extends NW from Arabia, crosses the fretted terrain into the northern lowlands, and merges with that of the North Polar Basin (Figure 13). The Isidis and Utopia systems are similar to patterns observed in lunar impact basins due to loading of the lithosphere by mare basalt impact basin filling and flexural deformation [e.g., Solomon and Head, 1979, 1980]. The circum-Tharsis system (Figure 8) is more comparable to the extensive linear and arcuate patterns observed in Oceanus Procellarum on the Moon [Whitford-Stark and Head, 1980] and completes the broad pattern of wrinkle ridges mapped in the circum-Tharsis region [Chicarro et al., 1985; Watters, 1993] forming a pattern with a radius of 7000 km. This pattern is also broadly similar to the circum-Aphrodite wrinkle-ridge pattern on Venus [e.g., Bilotti and Suppe, 1999].
3.3. Nature of the Substrate Deformed by Wrinkle Ridges
 Several lines of evidence suggest that the Vastitas Borealis Formation represents a veneer on a preexisting unit deformed by wrinkle ridges. We interpret this unit to be Hesperian ridged plains (Hr) on the basis of the following points.
 Examination of the distribution of circum-Tharsis units shows that unit Hr (wrinkle-ridged plains of Early Hesperian age) forms part of the annulus around Tharsis. Where this unit abuts the northern lowlands (e.g., Chryse Planitia), the wrinkle ridge patterns and orientations in Hr appear continuous with those observed in the immediately adjacent overlying Vastitas Borealis Formation, suggesting that Hr underlies the Vastitas Borealis Formation (Figures 8, 11, 13, and 14).
3.3.2. Surface roughness
 The characterization of surface roughness and slopes by Kreslavsky and Head [1999, 2000] shows that the northern lowlands and the Vastitas Borealis Formation are unusually smooth at all scale lengths. Characterization of Hr (Hesperian ridged plains) indicates that these units are also smooth, but rougher at several scale lengths than the Vastitas Borealis Formation. We have shown above that addition of ∼100 m of material to units with typical Hr roughness would be enough to produce a roughness scale length plot comparable to that of the Vastitas Borealis Formation (e.g., Figure 15).
3.3.3. Models for wrinkle-ridge formation
 Wrinkle ridges are thought by most workers to be contractional tectonic features. There is, however, no commonly accepted mechanical model for their formation. A comprehensive review of such models has been recently presented by [Schultz 2000, section 4]. Most models for the formation of wrinkle ridges [e.g., Zuber and Aist, 1990; Watters, 1988, 1991, 1993; Zuber, 1995; Mouginis-Mark et al., 1992; Mangold et al., 1998; Montési et al., 2000] show that wrinkle-ridge formation is favored in a deforming medium in which a strong layer overlies a weaker layer (e.g., lava/megaregolith) and that they are inconsistent with a deforming medium where a weak layer overlies another weak layer (e.g., sediment/megaregolith). Thus the occurrence of wrinkle ridges in the northern lowlands is interpreted to be more consistent with their formation in an Hr-like volcanic plains unit than in a wholly weak layer (e.g., sediment, megaregolith).
3.3.4. Wrinkle ridge height and spacing
 Circum-Tharsis wrinkle ridges in the northern lowlands tend to be more widely spaced and more vertically subdued than in previously mapped [Chicarro et al., 1985; Watters, 1993] circum-Tharsis exposures of Hr (e.g., Lunae Planum, Syria Planum) (Figures 6 and 7). One explanation for this is that wrinkle ridges in the northern lowlands have been subdued and partly buried by subsequent emplacement of the Vastitas Borealis Formation, thus tending to increase ridge spacing and decrease ridge height (e.g., Figure 9). Wider spacing could also be due to a localization instability related to different crustal and lithospheric structure between the two areas [Montési and Zuber, 2001], but a more comprehensive analysis of the ridge morphometry is necessary to determine if there is a statistically significant difference in primary ridge spacing, as required by this interpretation.
3.4. Age of Northern Lowlands Substrate Deformed by Wrinkle Ridges
 Superposed impact craters and stratigraphic relationships indicate a Late Hesperian age for the Vastitas Borealis Formation [Scott and Tanaka, 1986]. Our analysis of the detrended topography data, however, shows that there are a number of previously undetected stealth craters in the northern lowlands. We found 140 such subdued craters >16 km in diameter, and these craters appear to form part of the wrinkle-ridged surface that we interpret to predate the Vastitas Borealis Formation. Using the detrended altimetry data, we also counted craters whose morphology indicated that they were relatively fresher than the stealth craters. The combined population of stealth and relatively fresher craters provides numbers that are consistent with an Early Hesperian age for the substrate, an age essentially identical to the Early Hesperian ridged plains (Hr) (Figures 16a, 16b, and 17). We thus conclude that these data support the interpretation that the vast areas of the northern lowlands deformed by wrinkle ridges represent regions underlain by, and in several cases laterally equivalent to, Hr, the wrinkle-ridged plains of Early Hesperian age.
3.5. Stratigraphy and History of the Northern Lowlands
 On the basis of these observations and interpretations, we review the stratigraphy and history of the northern lowlands as previously mapped and as reconstructed from these new results (Figure 17). Previous mapping [Scott and Tanaka, 1986; Tanaka and Scott, 1987; Greeley and Guest, 1987] has shown that the northern lowlands are surrounded to the south (south of the dichotomy boundary) by three major units in order of decreasing abundance (measured in length of boundary), Noachian cratered terrain (Npl1), Hesperian ridged plains (Hr), and Noachian ridged plains (Nplr).
 According to this mapping, the oldest unit presently exposed within the northern lowlands is Npld, a dissected unit similar to the heavily cratered unit in the southern uplands but more highly modified by small channels and troughs. The only occurrence of this is in Acidalia Planitia, where a 300-km-diameter outcrop occurs which is annular in planform and stands ∼300–800 m above the surrounding plains. The second oldest unit is HNu, which is an undivided unit that spans the Noachian and Early Hesperian and forms highstanding hills and irregular mesas that have diameters of several kilometers to more than 10 km. Exposures of this unit form Scandia Colles and the degraded rims of large craters in the vicinity [Tanaka and Scott, 1987]. These are interpreted by Tanaka and Scott  as remnants of ancient material heavily eroded by mass wasting that are projecting through younger units. In these previous maps, there are no surface exposures of the units so abundant in the adjacent uplands in the vicinity of the boundary with the northern lowlands (e.g., Noachian cratered terrain (Npl1), Noachian ridged plains (Nplr), or Hesperian ridged plains (Hr) [Scott and Tanaka, 1986; Tanaka and Scott, 1987; Greeley and Guest, 1987]).
 The next younger unit is Hal, the lower member of the Alba Patera Formation, which spans much of the Hesperian. This unit occurs on the northern flanks of Alba Patera and consists of degraded lobate scarps, fractures, and graben; these are interpreted to be structures and flows from Alba Patera [Tanaka and Scott, 1987]. The upper part of Hal is laterally equivalent to Hchp (channel floodplain materials) and Hv (the Vastitas Borealis Formation). Our data show that in part, Hv postdates Hal [Kreslavsky and Head, 2001a]. A unit interpreted to be channel floodplain materials from fluvial channels south of the area (Hchp) occurs along the western edge of Acidalia Planitia and Chryse Planitia and is continuous with Hch (channel materials) in Chryse Planitia. These channel deposits are shown by Tanaka and Scott  and Scott and Tanaka  to be laterally equivalent to the Vastitas Borealis Formation (Hv), which is Late Hesperian in age. Hv forms vast subpolar plains deposits and consists of several members (Hvm, mottled; Hvg, grooved; Hvr, ridged; and Hvk, knobby [Scott and Tanaka, 1986]). This unit and its subunits are thought [Scott and Tanaka, 1986; Tanaka and Scott, 1987; Greeley and Guest, 1987] to be formed by several (multiple?) processes, including eolian, fluvial, periglacial, alluvial, volcanic, tectonic, and desiccation.
 Overlying these units (Figure 17) are several units of Amazonian age, including the Arcadia Formation (Aa, smooth plains of lava flow and sedimentary origin), the Elysium Formation (Ael, a variety of units of Early Amazonian age which are of volcanic origin and associated with the Elysium rise and related deposits in the Utopia Basin), smooth plains (Aps), which may be of eolian origin, and Late Amazonian polar and circumpolar deposits.
 Among the most notable characteristics of this previously mapped stratigraphic sequence [Scott and Tanaka, 1986; Tanaka and Scott, 1987; Greeley and Guest, 1987] (Figure 17) are (1) the extremely low abundance of Noachian-aged units in the northern lowlands (These units make up ∼48% of the surface of Mars outside the northern lowlands but <3% of the northern lowlands.), (2) the lack of occurrence of previously mapped Hr (ridged plains) in the northern lowlands, while outside the northern lowlands Hr is a major unit making up ∼11% of the surface of Mars, (3) the common abundance of these units in the southern uplands adjacent to the northern lowlands, and (4) the unique occurrence of the Vastitas Borealis Formation, occurring only in the northern lowlands and making up ∼45% of the area of this region.
 These observations raise several questions that we have partly addressed in previous sections: Could the northern lowlands have been the site of extensive flooding by Early Hesperian volcanic plains which were subsequently deformed by wrinkle ridges? Could such flooding explain the paucity of Noachian-aged units in the northern lowlands? Could the Vastitas Borealis Formation be a veneer superposed on such Hesperian-aged ridged plains?
 In order to analyze further these questions, we performed experiments in which we used MOLA topography data to simulate the emplacement of regional plains (Hr) in several typical areas of the southern uplands. We took a series of randomly chosen profiles across Noachis Terra. This region is dominated by Npl1, the widespread cratered unit of Early to Middle Noachian age [Scott and Tanaka, 1986; Greeley and Guest, 1987], which makes up ∼16% of the area outside the northern lowlands. We then simulated application of a layer of material in the same manner as outlined in section 2.5, where we assessed the application of a deposit on the ridged plains and compared it to the characteristics of the Vastitas Borealis Formation. We found that the minimal mean thickness of the layer necessary to reduce the 80-km-scale roughness from the level of Noachis Terra to the level typical for the ridged plains is 350 m. This lower limit of Hr material thickness is certainly far below the real thickness, because volcanic resurfacing is not likely to operate in a manner similar to the optimal smoothing algorithm, and also because the 100-km-scale topographic pattern resulting from the simulation does not resemble the observed topography of northern plains.
 We then simulated the regional volcanic flooding of these regions to determine how much lava would be required to bury typical southern upland terrain to a depth that would completely bury it or that would form outcrop patterns similar to those presently seen in the northern lowlands. First, we examined an area of 1 × 106 km2 in Noachis Terra, west of the Hellas basin (Figure 18). This region is dominated by Npl1, the widespread cratered unit of Early to Middle Noachian age [Scott and Tanaka, 1986; Greeley and Guest, 1987], which makes up ∼16% of the area outside of the northern lowlands. We used the digital MOLA topography as a basis (Figure 18) and flooded it by creating equipotential surfaces on the existing topography in three 500-m increments. This technique is known as a “porous” flooding model [see Head, 1982] and is equivalent to having vents virtually everywhere in the area. A separate approach is to have a smaller number of vents in specific areas and to assess how much lava might build up locally before it overtops local topography and begins to flood adjacent lowlying areas. In the approach we used, the 500-m increments are added from the top of the preexisting low point (or equipotential surface). The sequence of flooding steps shows that ∼41% of the area is resurfaced after 500 m of flooding and ∼65% after 1000 m and that after 1500 m of flooding, over 90% of the area has been resurfaced. The topography of the Npl1 unit remaining exposed forms patterns dominated by arcuate crater rim segments and a single plateau about 100 × 250 km in dimension. These patterns are strikingly similar to those seen in the northern lowlands (Figure 1) in association with outcrops of Npld and HNu (see description above and Scott and Tanaka , Tanaka and Scott , and Greeley and Guest ). Note that although the maximum thickness of volcanic deposits would be ∼1500 m, the range is from 0 to 1500 m, and the average thickness of volcanic plains over the whole area under consideration (1 × 106 km2) would be ∼920 m.
 Next, because unit Npld (the Early to Middle Noachian-aged dissected unit which is similar to Npl1 but more dissected by small valleys) makes up a significant percentage of the area south of the northern lowlands (∼11%), we chose an area of this terrain in northern Terra Tyrrhena, just west of Hesperia Planum (the type area of Hr, the Hesperian-aged ridged plains), and performed the same type of flooding experiment. Beginning with the initial step in Figure 19, we flooded the terrain in 500-m increments. This initial flooding reduces the underlying topography exposed in this area to 64%, while an additional 500 m reduces the area exposed to 34%, and the last 500-m increment reduces the area exposed to 6% (Figure 19). Again, note that although the maximum thickness of volcanic deposits is ∼1500 m, the range is from 0 to 1500 m, and the average thickness of volcanic plains in this case over the whole area considered would be ∼820 m. The terrain remaining exposed after this last step shows outcrop patterns of portions of crater rim crests and a few linear to rectangular intercrater areas. The addition of another 500 m of lava (not shown in Figure 19) reduces the outcrops to <2% of the area and removes most of the small plateau areas, leaving a few isolated arcuate crater rim crests as kipukas. These patterns are again similar to those few outcrops of HNu and Hpld observed in the northern lowlands (Figure 1).
 In summary, we flooded typical Noachian-aged cratered and dissected terrain (units which together make up ∼27% of the area outside of the northern lowlands) by a maximum of ∼1500 m of volcanic material and an average of ∼820–1000 m. This flooding produced an area with a small percentage of Noachian-aged terrain surrounded by volcanic plains, with patterns of Noachian unit outcrops similar to those seen in the northern lowlands. These results suggest that if the characteristics of the Noachian-aged local to regional topography of the northern lowlands were similar to those of typical areas of the southern uplands, then Early Hesperian volcanic flooding with thickness of the order of 800–1000 m of lavas could explain the present distribution of units. In addition, a population of large subdued quasi-circular depressions described by Frey et al.  was interpreted by them to represent an underlying Noachain-aged surface. The emplacement of the Hesperian-aged ridged plains could be a major factor in the burial and subdual of the original craters to produce the now-subdued quasi-circular depressions described by Frey et al. , as discussed in more detail later. We previously raised the following questions: Could the northern lowlands have been the site of extensive flooding by Early Hesperian volcanic plains which were subsequently deformed by wrinkle ridges? Could such flooding explain the paucity of Noachian-aged units in the northern lowlands? On the basis of these analyses, we conclude that these questions can be plausibly answered in the affirmative.
 We now return to the question of whether the Vastitas Borealis Formation could be a veneer superposed on such Hesperian-aged ridged plains. Previous workers have shown that the Vastitas Borealis Formation is time-correlative with the Hesperian-aged channel deposits (Figure 17) and suggested that it may have several origins and represent the operation of several processes (including eolian, fluvial, periglacial, alluvial, volcanic, tectonic, and desiccation) [Scott and Tanaka, 1986; Tanaka and Scott, 1987; Greeley and Guest, 1987]. Our data are consistent with the emplacement of extensive deposits of Hesperian-aged ridged plains (Hr) in the northern lowlands and its subsequent degradation by smoothing processes to produce the presently observed smoothness characteristics [Kreslavsky and Head, 1999, 2000] and wrinkle ridge characteristics (tendency toward greater spacing and lower heights in the northern lowlands; Figures 6, 9, and 15). The close association in space and time of the Hesperian-aged outflow channel deposits and the Vastitas Borealis Formation in the northern lowlands (Figure 17) supports the conclusion that the channels were largely emplaced on top of the Hesperian-aged ridged plains (a relationship seen along the margins of the uplands in places like Lunae Planum [e.g., Scott and Tanaka, 1986]). Most workers agree that the emplacement of the outflow channels involved a significant sediment load (see review by Carr ), although the exact percentage is a matter of controversy.
 We conclude that the emplacement of the outflow channels provided sufficient water and sediment to the northern lowlands to veneer significant parts of the basin (Figure 9, bottom) [Kreslavsky and Head, 2001b]. Processes in the period subsequent to the Late Hesperian (a period possibly as much as 3.3–3.5 billion years long [Tanaka et al., 1992a; Hartmann and Neukum, 2001]) could have redistributed some of this material by eolian and other processes. In addition, other processes have clearly operated on deposits in the northern lowlands to produce additional Amazonian-aged units, including (1) mass wasting and lateral transport of material from the margins of the dichotomy boundary into the northern lowlands, (2) volcanic activity, (3) polar deposit emplacement and modification to produce the circumpolar deposits, and (4) periglacial processes [e.g., Rossbacher and Judson, 1981; Squyres and Carr, 1986; Kreslavsky and Head, 2000].
3.6. Implications for the Evolution of Mars
 On the basis of the above discussions, we interpret the wrinkle-ridge distribution documented in this paper to represent the emplacement of a significant volume of Early Hesperian-aged volcanic plains in the northern lowlands that subsequently were deformed by wrinkle ridges. If this interpretation is correct, then there are some important implications for the history of Mars and the northern lowlands.
 Subsequent to the completion of our analysis, Frey et al.  described a population of circular arcs and structures in the northern lowlands revealed in MOLA data which they interpreted to be buried impact basins. They found 644 quasi-circular depressions in the northern lowlands whose diameters are >50 km and concluded that “the smooth and sparsely cratered Hesperian and younger lowland plains are likely only a thin veneer overlying a much older crust.” They further concluded that the buried lowlands surface is at least as old as the exposed Noachian highland surface [Frey et al., 2001]. The vast majority of stealth craters mapped in our analysis are <50 km in diameter, and thus the population mapped by Frey et al.  is complementary in size distribution to the populatiom of stealth craters that we mapped here (Figure 16). On the basis of this comparison, we interpret the population of quasi-circular depressions mapped by Frey et al.  to predominantly represent features superposed on the underlying Noachian-aged surface that was subsequently flooded by Hesperian-aged ridged plains, as outlined in this study. The population of craters represented by the stealth structures (Figure 16) was then superposed on the ridged plains surface during the Early Hesperian and modified before or during the emplacement of the Vastitas Borealis Formation. The population of relatively fresh craters superposed on the Vastitas Borealis Formation (Figure 16) represents those that accumulated on it subsequent to its emplacement in the Late Hesperian.
 Outside the northern lowlands there is an observed early planetwide volcanic emplacement phase of Late Noachian-Early Hesperian age. For example, Late Noachian ridged plains (Nplr) comprise ∼3% of the presently exposed surface of Mars and Hesperian-aged ridged plains (Hr) make up another 10%. The thickness of the Hesperian ridged plains (Hr) has been estimated to be generally ∼300–600 m in Lunae Planum [Frey et al., 1991] and 0.5–1.5 km in eastern Tharsis [DeHon, 1982], and recent observations of layers interpreted to be volcanic flows in the walls of Valles Marineris have led to the suggestion that thicknesses could be even greater [e.g. McEwen et al., 1999].
 Thus, if Hesperian-aged ridged plains underlie the northern lowlands, then this would mean that the total percentage of the planet resurfaced by Hesperian-aged volcanic plains would be the 10% presently exposed in the southern uplands and the 20% making up the northern lowlands, for a total of 43.5 × 106 km2, or 30% of the surface of Mars. Tanaka et al. [1988, 1992a] estimated that Early Hesperian volcanic plains covered 19.3 × 106 km2 of Mars (∼13% of the surface area; includes both exposed and buried deposits). These new estimates would more than double the area covered by Hesperian ridged plains.
 If the southern upland plains are typically ∼500 m thick and those in the northern lowlands are ∼900 m thick (on the basis of the average thickness of the flooding models), then the total volume of Hr would be of the order of 3.3 × 107 km3. These estimates represent an increase in the importance of Hesperian-aged volcanic plains from previous estimates [e.g., Tanaka et al., 1992a; Greeley and Schneid, 1991] of more than twice the area and almost twice the volume. Greeley  estimated gas exsolution from volcanic effusion and input of volatiles into the atmosphere as a function of time during Mars history and showed that the peak input was during the Early Hesperian. These new data suggest that this peak volatile input into the atmosphere was significantly more important than previously recognized.
 These results further emphasize the importance of events occurring during the Early Hesperian. On the basis of a variety of data, it appears that significant volcanic activity, perhaps the peak global flux [Greeley and Schneid, 1991; Tanaka et al., 1992a], occurred during this period. This was accompanied and followed closely by regional and global contraction to produce the circum-Tharsis and other wrinkle-ridge systems [e.g., Chicarro et al., 1985; Watters, 1993]. On the basis of our interpretations here, the northern lowlands played a very important role in the Early Hesperian volcanic activity and in the subsequent period of regional and global-scale contraction. In addition, flooding of the northern lowlands by extensive Hesperian-aged lavas could easily obscure previous structures, rendering ancient impact basins or tectonic structures more difficult to detect.
3.7. Implications for the Oceans Hypotheses
 Several researchers have proposed that the northern lowlands were once the site of an extensive standing body of water or ocean [e.g., Parker et al., 1989, 1993; Baker et al., 1991; Scott et al., 1992, 1995; Clifford and Parker, 2001]. Estimates of the ages of the proposed standing bodies of water range from Noachian to Amazonian. Tests of the hypotheses of Parker et al. [1989, 1993] have found that several of the predictions of the hypothesis are consistent with the new data but that others are not [Head et al., 1999, 2001b]. The new data and interpretations presented here do not directly address the presence or absence of large standing bodies of water in the northern lowlands. They do, however, provide additional possible insight into the nature and evolution of the region and the origin of some of its unusual properties.
 For example, one of the most unusual characteristics of the northern lowlands is its extreme smoothness at several scale lengths, comparable to abyssal plains on Earth [e.g., Smith et al., 1998; Aharonson et al., 1998]. The presence of Hesperian-aged ridged volcanic plains mantled by the Vastitas Borealis Formation (Figure 15) provides an alternate interpretation for the timing and mode of formation of this smoothing. Instead of long-term sedimentation in a standing body of water, the general regional smoothing could have taken place as part of the Early Hesperian volcanic resurfacing, and the later fine-scale smoothing could have occurred as a result of the emplacement of outflow channel sediments and their subsequent modification.
 Although providing an alternative explanation for some of the smoothing, the emplacement of these volcanic plains does not preclude the presence of oceans. They could have existed prior to this time (Noachian [e.g., Clifford and Parker, 2001]), during this time (in which case the volcanic plains might have been emplaced subaqueously), or following the emplacement of the ridged plains as a result of outflow channel basin filling [e.g., Baker et al., 1991]. In addition, some of the channel-like mass deficits observed in Mars Global Surveyor (MGS) gravity data in the northern lowlands [e.g. Zuber et al., 2000, Figure 5] suggest that the history of the northern lowlands prior to the emplacement of the Hesperian-aged volcanic plains may have been quite different, with substantial amounts of water and sediment being transported to the northern lowlands before the cessation of resurfacing [Zuber et al., 2000]. The presence of water in the northern lowlands could be responsible for the removal of stealth crater rims before or during the emplacement of the Vastitas Borealis Formation (Hv).
Carr  pointed out that substanital amounts of the floodwater that was emplaced in the northern lowlands during the Late Hesperain outflow channel events could have frozen and then reached thermal and diffusive stability owing to a thin cover of sediment or lavas. Clifford and Parker [1999, 2001] concluded that the presence of such massive ice deposits beneath the northern plains was consistent with a variety of observations, including their smoothness and geomorphology of landforms. The results and interpretations reported here do not favor the hypotheses of massive ice deposits beneath the northern lowlands. First, if thick frozen residues of the outflow channel floodwaters existed in the northern lowlands, they should largely bury underlying topography. However, the detected presence of the underlying Early Hesperian plains, stealth craters, and wrinkle ridges suggests that any layer overlying these plains must be thin, less than a few hundred meters in thickness. A more likely explanation is that the overlying material (the Vastitas Borealis Formation) is composed of the sedimentary residue of the sediment/water effluent of the outflow channels [e.g., Kreslavsky and Head, 2001b]. In this case, a significant part of the smoothness of the northern lowlands is due to an Early Hesperian volcanic infilling of a Noachian-aged surface (Figures 16, 17, and 18). Buried massive ice deposits below the Hesperian ridged plains cannot be ruled out, but the large number of quasi-circular depressions dating from the Noachian [Frey et al., 2001] would seem to argue against the presence of such a deposit.
 Detrended MOLA altimetry data have provided a new picture of the Martian northern lowland basin topography, morphology, evolution, and relation to the history of Mars. We interpret our results to indicate that the northern lowlands are underlain by a regional unit containing a basin-wide system of subparallel wrinkle ridges and arches. This unit is laterally contiguous with Hesperian-aged ridged plains in the southern uplands and contains highly modified craters, the number of which suggests an Early Hesperian age. The orientation and location of the wrinkle ridges in the North Polar Basin completes a global circum-Tharsis ridge system forming a band approximately 7000 km wide and extending over the whole circum-Tharsis region. Several subareas of the northern lowlands show individual patterns (e.g., basin-like areas of Isidis and Utopia). The present spacing and height of wrinkle ridges and geometry of buried craters in the northern lowlands suggest that the Late Hesperian Vastitas Borealis Formation is a sedimentary unit superposed on Hr (regional plains). Hesperian-aged channels entering Chryse Planitia are controlled by the orientation and topography of wrinkle ridges deep into the basin, indicating that wrinkle ridges had largely formed by Late Hesperian. These channels are among the strongest candidates for providing the material of the Vastitas Borealis Formation. Amazonian-aged smooth plains units of volcanic origin, particularly in Amazonis Planitia, further bury and obscure the underlying wrinkle ridges and the Vastitas Borealis Formation.
 Recognition of these units and their stratigraphic relationships provides a new perspective on the history of the northern lowlands (Figure 17). In this scenario, in the Early Hesperian the majority of the northern lowlands was filled with volcanic plains similar to those presently exposed in the southern uplands (Hr). As with those deposits in the southern uplands, evidence for volcanic vents was scant or subdued in topography. The Hesperian-aged plains in the northern lowlands were pervasively deformed soon thereafter by Tharsis-circumferential and basin-related wrinkle ridges. Circum-Chryse outflow channels formed in the Late Hesperian following courses largely controlled by wrinkle ridge orientation and height and deposited material in the basin to form a major contribution to the Vastitas Borealis Formation.
 Widespread emplacement of the Hesperian-aged ridged plains of apparent volcanic origin is interpreted to mean that the volcanic phase represented by this unit was global in nature and resurfaced the northern lowlands, in addition to the ∼10% of the planet previously known, for a total resurfacing of about 30% of Mars. This remarkable event increases by a factor of 2 the amount of volatiles that might have been degassed into the atmosphere during this time period of peak volcanic flux.
 We gratefully acknowledge the MOLA instrument team and the MGS spacecraft and operations teams at the Jet Propulsion Laboratory and Lockheed-Martin Astronautics for providing the engineering foundation that enabled this analysis. We particularly thank Greg Neumann for superior professional performance in data reduction and preparation and presentation. We also thank the MOLA science team and Co-PI's David Smith and Maria Zuber for productive scientific discussions. Detailed reviews by Michael Carr and an anonymous reviewer improved the clarity of the manuscript and are greatly appreciated. Nadine Barlow kindly provided a copy of her impact crater catalog. Anne Côté rendered the final sketch maps, and Peter Neivert assisted in figure preparation. This work was partially funded by a grant from the National Aeronautics and Space Administration.