The origin of the rough-textured aureoles that surround the immense Olympus Mons volcano on Mars is controversial. We present data from the Mars Global Surveyor and Mars Odyssey missions to demonstrate that at least two of the aureole lobes are derived from the volcano's flanks in large and probably catastrophic mass movement events, leaving behind headwalls that constitute the basal scarp. This evidence stems from the morphology and internal structure of aureole blocks, which exhibit remnants of volcanic flow units on their surfaces. Our claim is supported by plausible reconstructions of the prefailure flanks. Structural analogs to known flank failure events at Hawaiian volcanoes suggest that repeated cycles of flank growth and collapse at Olympus Mons allow generation of the observed aureoles from a protoedifice similar in size and shape to the present one.
 Olympus Mons is an immense volcano, reaching up to 23 km height above base and 600 km in diameter, located to the northwest of the Tharsis Rise on Mars (Figure 1a). The volcanic edifice is partially bounded by an escarpment of height up to 10 km, known as the Olympus Mons basal scarp. Lobate deposits with rugged morphology, known as the Olympus Mons aureole deposits, extend outward from the base of the scarp for hundreds of kilometers, with greatest extents and widths to the northwest of the edifice. While the proximity of scarp segments to aureole lobes is suggestive of a causal relationship between the two, the limited resolution of available Viking Orbiter (VO) data sets has made it difficult to conclusively demonstrate such a link. However, recent topography data from the Mars Orbiter Laser Altimeter (MOLA) allow greatly improved characterizations of the structure and volume of aureole lobes. Furthermore, image data from the Mars Orbiter Camera (MOC) show features that strongly link aureole materials of two proximal aureole lobes to structures now evident on the edifice and at the basal scarp.
 Plausible emplacement mechanisms for the Olympus Mons aureole deposits fall broadly into two categories. Under the first category, flank failure, the aureoles are interpreted as material derived from the slopes of the Olympus Mons edifice either as landslides, possibly catastrophic [Lopes et al., 1980, 1982], or as thrusting or gravity spreading of local sediments or flank material at low strain rates [Francis and Wadge, 1983; Tanaka, 1985]; the latter model invokes the presence of ice as a basal lubricant for gravity sliding. In mass movement scenarios the origin of the basal scarp (as the head scarp of the slides) is intimately related to the origin of the aureoles. Several mass movement scenarios for aureole generation require a substantial elongation of the protoedifice toward the northwest to generate sufficient driving forces for gravity sliding [Francis and Wadge, 1983] or to match volumes of aureole lobes with corresponding failed flank sectors [Lopes et al., 1982]. Lopes et al. [1980, 1982] noted that giant submarine landslides off the Hawaiian ridge detected by Moore  were the closest terrestrial analogs in size and morphology to the Olympus Mons aureoles. More detailed mapping of the landslides and slumps on the flanks of Hawaiian volcanoes [Moore et al., 1989] confirmed the analogy; furthermore, recent geodetic, seismicity, and reflection seismic data indicate the presence of slipping decollements at the bases of the edifices, which have enabled outward spreading of Hawaiian volcano flanks [e.g., Swanson et al., 1976; Owen et al., 1995, 2000; Denlinger and Okubo, 1995; Morgan et al., 2000]. Structural modeling of stresses in edifices with detached basal boundaries led McGovern and Solomon  to suggest analogies between the aureoles and the Hawaiian landslides and between the high-angle Pali faults on Kilauea and the Olympus Mons basal scarp. Under the second category of emplacement mechanisms the aureoles are volcanic products, e.g., pyroclastic flows or eroded effusive volcanic flows [Morris, 1982; Carr, 1973; Morris and Tanaka, 1994] emplaced locally, i.e., from a source beneath the deposits themselves, possibly predating the construction of the Olympus Mons edifice. For such scenarios the basal scarp has no direct relation to the aureoles. Flow directions inferred from the distributions of interlobe ridges and boundary features allowed identification of potential source vents within lobes [Morris, 1982]. The apparent absence of bedding in VO images of aureole ridges led to their characterization as “thick, easily eroded deposits typical of unwelded ash flow tuffs” [Morris, 1982]. Recently, Wilson and Mouginis-Mark  asserted that ridges near the edge of the north aureole lobe originated via explosive eruptions initiated by intrusion.
2. View From Mars Global Surveyor
 The high-resolution topography and image data sets from Mars Global Surveyor [Smith et al., 2001] allow for new insights into the structure and evolution of the Olympus Mons aureoles. Investigation of MOLA data has revealed that the aureoles of Olympus Mons are among the roughest terrains on the planet [Smith et al., 2001]. Here we focus on the aureole lobe directly to the north of Olympus Mons (Figure 1b), which on the basis of superposition relations is likely one of the youngest (see section 3) and which appears to be less modified by erosion and volcanic embayment than other lobes [Morris and Tanaka, 1994]. High relief is often observed at aureole boundaries. For example, at the eastern boundary of the northern aureole lobe near the Olympus Mons edifice (label 1 in Figure 1b), there is a sharp step downward (with almost a kilometer of relief) from the regional plains to the level of material between the basal scarp and the northern lobe. Such a large downward offset toward the center of a lobe is inconsistent with an eruptive (e.g., topography building) aureole formation mechanism. Instead, the structure likely accommodates right-lateral shear resulting from the greater extent of flow in the central part of the lobe relative to the periphery [Lopes et al., 1982; Francis and Wadge, 1983]. Several other lineaments appear to accommodate shear from differential motions of lobe segments, with modest or negligible (labels 2 and 3 in Figure 1b, respectively) cross-strike topographic offsets. A prominent topographic step extends across the full width of the lobe (label 4 in Figure 1b). Prominent ridges parallel the step to the south, whereas the lower terrain to the north exhibits a more subdued, less linearly ridged pattern. The lateral and distal lobe boundaries consist of another sharp topographic step (label 5 in Figure 1b).
 High-resolution (several meters/pixel) narrow-angle (NA) MOC images [Malin and Edgett, 2001] reveal the small-scale structure of the aureoles. Much of the north and northeast aureole lobes consist of narrow linear ridges (e.g., between labels 1 and 3 in Figure 2a) or broader rectilinear blocks with subhorizontal surfaces (e.g., label 4 in Figure 2a), the former perhaps the result of block tilting. The blocks are cohesive and fragmented, with no evidence for a constructional origin. The topographic lows between blocks and/or ridges are commonly covered with smooth-textured materials, as are the lower flanks of many ridges and blocks. The smooth materials can be mobilized, as evidenced by slope streaks (e.g., Figure 2a): Such streaks are abundant and currently forming in the aureoles [e.g., Schorghofer et al., 2002]. However, coarse-textured and erosion-resistant materials are often seen at the edges of blocks and ridges (Figure 2a and close-up in Figure 2b). Some such materials (usually dark) appear as laterally coherent layered outcrops (Figure 2a, labels 1 and 2, and Figure 2b, label 1), refuting earlier assertions based on VO images that aureole materials are predominantly unbedded and poorly consolidated [e.g., Morris, 1982]. Layering is found on only one side of the ridge shown in Figure 2b (although artifacts of exposure and illumination may obscure layering elsewhere on the ridge); such a finding is consistent with edge-on exposure of a tilted block with internal layering (and inconsistent with erosion of horizontally bedded materials). The layering shown here is not consistent with ridge formation by phreatomagmatic dikes [Wilson and Mouginis-Mark, 2003], and the creation of blocks is inexplicable via such a scenario.
 The presence of lava flows with sinuous raised margins, or “levees,” on the exposed surfaces of aureole blocks strongly links the aureole lobes to Olympus Mons flank materials. Lava flows with leveed margins are ubiquitous on the Olympus Mons edifice [Wadge and Lopes, 1991; Rawling et al., 2003] and are particularly prominent in MOC images of the edifice and scarp (see Figures 3a and 3b) but are largely absent elsewhere in the region [Rawling et al., 2003]. In the low or very gently sloping lava plains between the edifice and the aureole, NA MOC images show little evidence for intact leveed flows; instead, leveed flows appear to be preferentially emplaced on moderate slopes such as those of the Olympus Mons edifice [Chittenden and McGovern, 2004]. However, pairs of raised parallel lineations that resemble the margins of leveed lava flows are visible on aureole blocks (e.g., Figures 4a and 4b). The lineations appear to terminate at block edges rather than continuing down the sides of the blocks, indicating emplacement before the blocks detached. The overall surface morphology of such blocks, with multiple leveed flow remnants and impact craters (Figure 4b), is similar to surfaces on the main edifice (Figure 3a) and to flow units draped over sections of the basal scarp (Figure 3b). Possible leveed flow remnants also occur on less topographically prominent rock exposures, as at label 2 of Figure 2b. The presence of effusive leveed flows that predate block formation is inconsistent with an explosive volcanic origin [Morris, 1982] for the aureoles. Thus the aureole blocks exhibiting leveed flow remnants are most likely to have been derived from the flanks of the Olympus Mons edifice.
 Aureole block surfaces are also covered by linear fractures and faults (Figure 2a, near label 4, and Figure 4b). The largest such lineations do not appear to continue beyond block edges; however, a few small linear lineations appear to cut the edges of blocks (although they do not continue into the surrounding low terrain). This evidence suggests that fault and fracture formation preceded (or perhaps was contemporaneous with) block creation. Small grabens observed on aureole blocks should not be confused with larger-scale linear depressions, sometimes also called grabens in the literature [e.g., Lopes et al., 1982], that cut aureole lobes and in many cases appear to accommodate lateral movement of one aureole sector past another (e.g., the feature labeled “3” in Figure 1b).
 Further evidence for detachment of edifice materials to form aureole lobes may be seen in MOC narrow-angle images of the Olympus Mons basal scarp. Grabens oriented perpendicular to the edifice rim and small faults on the uppermost scarp (Figure 3c, labels 1 and 2, respectively) indicate an environment of extensional faulting. Such faults may be analogs to faults seen on the surfaces of blocks near the basal scarp [e.g., Morris and Tanaka, 1994, Figure 8] and in the aureole (Figure 2a). Resistant dark layers, similar to those on the margins of aureole blocks and ridges, are seen lower on the basal scarp (Figure 3c, label 3). The upper scarp has a characteristic spur and gully morphology [e.g., Tanaka, 1985]. Farther down the scarp, several spurs broaden into triangular facets (Figure 3c, labels 4), the bases of which consist of the dark resistant layers. The facets may represent the exposed faces of scarp-forming faults, as for similar facets at the bases of Valles Marineris troughs [Blasius et al., 1977].
3. Volume Balance Between Edifice and Aureole
 The connection between aureole material and the distal flanks of Olympus Mons has been called into question on volumetric and geometric grounds. Under early mass movement scenarios, substantial elongation of the proto-Olympus Mons flank to the northwest was required to achieve flank/aureole volume balance [Lopes et al., 1982] or to allow aureole lobes of the observed extents to be emplaced at low strain rate [Francis and Wadge, 1983]. Summarizing such arguments, Morris and Tanaka  asserted that mass movement aureole formation scenarios required a two-stage evolution for Olympus Mons: an initial highly elongated, asymmetric shield, followed by flank failure generating the aureoles and edifice seen today. However, our current understanding of flank failure in terrestrial volcanoes indicates that repeated cycles of flank growth, oversteepening, lateral spreading, and collapse contribute to the formation of landslides and slumps throughout the lifetime of a volcano [Denlinger and Okubo, 1995; Bulmer and Wilson, 1999; Morgan et al., 2003; Morgan and Clague, 2003]. The resulting edifices can grow in a self-similar manner; changes in dimensions of the order of 50% [e.g., Francis and Wadge, 1983] are not required, although these sometimes occur, as at Molokai in the Hawaiian chain [e.g., Holcomb and Searle, 1991; Satake et al., 2002].
3.1. Reconstruction of the Prefailure Flanks of Olympus Mons
 This principle can be demonstrated via reconstruction of the failed flank of Olympus Mons, using the method of Satake et al. . We reconstruct the north flank sector and the adjoining north aureole lobe under the assumptions that the whole lobe was derived as a landslide from the concave, north facing embayment in the basal scarp and that gentle upper flank slopes continued farther northward. The original topography of this region is shown in Figure 5a, along with the masks used to define the slide area (i.e., the aureole lobe, orange line) and reconstructed flank area (red line) [Satake et al., 2002]. A continuous curvature gridding algorithm [Wessel and Smith, 1991] was applied to the topography outside the masks but also including a line of control points (yellow line in Figure 5a) representing the approximate upper boundary (e.g., top of the basal scarp) of the reconstructed edifice. We set the control point elevation at 6 km, roughly that of the scarp in neighboring sectors. The gridding routine generated smooth surfaces for the preslide surface and reconstructed flanks (Figure 5b). The dimensions of the reconstructed flank area and the positions of edifice control points were adjusted to fulfill two conditions: first, that reconstructed lower flank and scarp slopes are similar to those observed (see Figure 6), and, second, that calculated volumes for the aureole lobe (8.8 × 104 km3) and reconstructed flank (8.4 × 104 km3) agree closely. These volumes are similar to the Viking-era estimate of 7.7 × 104 km3 [Lopes et al., 1982] for this lobe (despite differences in lobe area definition). Thus it is plausible to derive aureole lobes from an edifice of dimensions and flank slopes similar to those observed at present.
3.2. Failure Geometry and Stratigraphy
 The boundary we have chosen for the lobe due north of the edifice differs from that in the most recent geologic map of the area [Morris and Tanaka, 1994]. Delineation of the boundary between units Aoamb-c (containing most of the “north” lobe, “N” in Figure 1) and Aoau (the proximal “northwest” lobe, “NW” in Figure 1) in this map (white line in Figure 5a) may have been influenced by a particularly strong shadow in the VO base image mosaic for the region (this shadow is much less prominent in MOC wide-angle images taken with different illumination conditions). Topography data (Figures 1 and 5) are not subject to such biases and thus offer an alternative, objective means for defining units, well suited for rugged terrains like the Olympus Mons aureoles. Aureole units on either side of the Morris and Tanaka  boundary (roughly along lineament labeled “2” in Figure 1b) have similar height and characteristic roughness, whereas the proposed new boundary (orange lines in Figure 5a) separates regions of differing height and roughness. This redefinition inverts the previously defined stratigraphic relationship between the two units [Morris and Tanaka, 1994] such that the north lobe (originally Aoamb-c) is younger than the northwest one (originally Aoau). Under the assumption of a mass movement origin for the aureoles the unit definition of Morris and Tanaka  requires that the Aoau lobe emanate from two mutually perpendicular scarps facing roughly west and north, (labels 1 and 2, respectively, in Figure 5a) and that the Aoamb-c lobe emanate from the remainder of the north facing scarp. Redefining the boundary (Figure 1b) not only establishes a strong one-to-one correspondence between the north lobe and the north facing scarp but also allows the reconstructed flank to form extended headscarps for the older NE and NW lobes (labels 3 and 4, respectively, in Figure 5b).
4. Hawaiian Analogs
 Hawaiian volcanoes offer direct analogs to many features observed at Olympus Mons. The Nuuanu slide, an enormous debris avalanche derived from the north flank of Oahu [e.g., Satake et al., 2002], contains flat-topped blocks (e.g., Tuscaloosa Seamount, the large block labeled 1 in Figure 7a and in the foreground of Figure 8a) and narrow ridges that resemble those in Olympus Mons aureole lobes, as seen in the MOC NA images in Figures 2a, 2b, and 4a and in an image from the Mars Odyssey Mission's Thermal Emission Imaging System (THEMIS) visible spectrum, or VIS, imager (Figures 7b and 7c). The landslide block structures at Oahu and Olympus Mons are also comparable in size: The large aureole block in Figure 7c has horizontal dimensions about half those of Tuscaloosa seamount, and several of the elongated aureole blocks/ridges in Figure 7b are nearly as long as the blocks/ridges northeast of Tuscaloosa Seamount (label 3 in Figure 7a). The block in Figure 7c is about 300–500 m high, similar to the heights of blocks near label 3 in Figure 7a, although less than the 2 km height of Tuscaloosa Seamount. The characteristic scales of blocks/ridges in the Nuuanu slide are therefore comparable to those in the north aureole lobe (see also Figure 8b), although the overall volume of the latter (8.8 × 104 km3) is more than an order of magnitude larger than the volume of the former (3 × 103 km3 [Satake et al., 2002]).
 The landslide block in Figure 7c exhibits subtle lineations aligned roughly west-northwest-east-southeast (roughly from top to bottom in Figure 7c). These may be remnants of leveed lava flows as seen in Figures 2–4. If so, their orientations suggest that the block has rotated during aureole lobe emplacement, since radial flows on the failed edifice sector would have a predominantly north-south orientation. Such a rotation is consistent with the sense of shear induced by the differential motion of the central part of the lobe past the margins (e.g., Figure 1) [Lopes et al., 1982; Francis and Wadge, 1983].
 Hawaiian edifices exhibit seaward facing fault systems that may be analogs to the detachment faults that exposed the Olympus Mons basal scarp. The best examples are the Hilina Pali faults on Kilauea [e.g., Swanson et al., 1976] (see Figure 8c). Many such faults are buried and smoothed by subsequent volcanic flows in a manner similar to that observed at certain flank sectors (e.g., the northeast flank between the labels “NE” and “E” in Figure 1a) and radial faults (e.g., the fault scarp in Figure 3a) on Olympus Mons. Additionally, volcanic strata within the Olympus Mons basal scarp may have been uplifted and back tilted [Tanaka, 1985; Morris and Tanaka, 1994] above distal thrust ramps, similar to the frontal bench structures seen on the lower flanks of Kilauea (Figure 8c) and Mauna Loa [Smith et al., 1999; Denlinger and Okubo, 1995; Morgan et al., 2000; J. K. Morgan and P. J. McGovern, Discrete element simulations of gravitational volcanic deformation: 1. Deformation structures and geometries, submitted to Journal of Geophysical Research, 2004, hereinafter referred to as Morgan and McGovern, submitted manuscript, 2004]. Many of these frontal benches have experienced subsequent detachment and breakup resulting in the emplacement of modest debris fields [Morgan and Clague, 2003]. Such materials (e.g., in the foreground of Figure 8c) are potential aureole analogs, although the volumes of the former are much smaller than those of the latter. We note that Borgia et al.  proposed a thrust-based scarp formation mechanism for Olympus Mons, specifically as the front of a fault bend fold or fault propagation fold. However, these workers explicitly discounted a connection between the scarp and the aureole, because of difficulties in producing the observed volumes of aureole lobes from the fronts or tops of the folds. Given the links between aureole and edifice material presented above, aureole emplacement scenarios that lack a connection between the edifice (and bounding scarp) and the aureole are untenable.
5. Rapid or Slow Flank Failure?
 Given the evidence that aureole materials are derived from the flanks of Olympus Mons, the question of deformation rate remains. Both slow (e.g., creep and fault slip [Francis and Wadge, 1983; Tanaka, 1985]) and rapid (e.g., catastrophic [Lopes et al., 1980, 1982; McGovern and Solomon, 1993]) flank-derived aureole formation mechanisms have been proposed. The presence of tilted and possibly overturned blocks exhibiting noncoherent orientations with respect to originally subhorizontal surfaces (e.g., Figures 2a, 2b, 4a, and 4b), plus the distance of the distal aureole margins from the basal scarp, favors a rapid, high-energy emplacement mechanism, as does the resemblance of the aureole block and ridge morphology to the Nuuanu slide (Figures 7a and 8a), an event commonly held to be catastrophic [Moore et al., 1989; Moore and Clague, 2002]. In contrast, a thrust sheet mechanism [Francis and Wadge, 1983; Tanaka, 1985] would be expected to leave at least somewhat coherent and laterally extensive folds, whereas the purported “folds” in the aureoles are now seen to be individual blocks of former edifice materials. Further, thrust sheet scenarios require a substantial elongation of the paleoedifice to the northwest such that material can be delivered throughout the currently observed extent of the aureole lobes [e.g., Francis and Wadge, 1983]. The difficulties in explaining the growth of such an asymmetric edifice have been cited in critical assessments of mass movement aureole scenarios [e.g., Morris and Tanaka, 1994], but as demonstrated above, repeated cycles of edifice growth and catastrophic collapse allow generation of broad aureole lobes from an edifice not greatly different in size and shape from the current one. Thus a catastrophic mass movement origin for the Olympus Mons aureole deposits is consistent with the evidence from Mars Global Surveyor presented above and with our understanding of volcanic flank failure processes.
Harrison and Grimm  modeled Martian landslides subject to combinations of frictional, Bingham, and power law rheologies. They found no model that matched the topographic profile of the north Olympus Mons aureole lobe (Figure 6) and concluded that the aureoles were emplaced at low strain rates, as suggested by Francis and Wadge  and Tanaka . However, as argued above, the topographic and image data are consistent with a rapid, high-energy mass movement origin for the aureoles. A flow consisting of large blocks of varying size and geometry, such as those observed in the Olympus Mons north and east aureole lobes (Figures 2, 4, and 7), might not be readily simulated by the continuum rheologies used in the computations of Harrison and Grimm . Thus the inability of such models to match observed profiles does not necessarily rule out catastrophic scenarios for aureole emplacement. We also acknowledge that the Olympus Mons edifice is likely subject to low-strain rate processes, such as basal slip and flank creep (e.g., Morgan and McGovern, submitted manuscript, 2004). In particular, inward tilting strata at the edifice rim [Morris and Tanaka, 1994; Morgan and McGovern, submitted manuscript, 2004] and large blocks of edifice material that have slumped partway or completely down the scarp [Tanaka, 1985] may have resulted from edifice spreading or flank failure at low strain rates.
6. Flank Weakening Mechanisms
 The relatively low slopes of Olympus Mons (average values around 5–6°) and Hawaiian volcanoes do not favor gravitationally driven flank failure. A flank weakening mechanism is therefore suggested. Rheological weakening due to rock properties [Francis and Wadge, 1983] or the presence of water in solid [Tanaka, 1985] or liquid form [Lopes et al., 1982] has been invoked. The existence of a weak basal decollement underlying Hawaiian volcanic edifices [e.g., Dieterich, 1988; Denlinger and Okubo, 1995] has been attributed to a mechanically weak pelagic sediment layer [Nakamura, 1980] or the presence of excess pore fluid pressure Pf along a low-hydraulic-diffusivity basal detachment underlying the edifices [Iverson, 1995]. High Pf reduces effective friction, enabling slip at relatively low shear stresses. If excess pore fluids are responsible for weakening a basal decollement under the distal flanks of Olympus Mons, the presence of liquid water is implied. In the Hawaiian case the pore water was entrained within marine clay sediments at the time of deposition and trapped by the growing volcano [Iverson, 1995]. At Olympus Mons, potential sources of water include outflow channels from the adjacent Tharsis rise [Mouginis-Mark, 1990] and preedifice basement sediments, such as those currently covering large portions of the northern lowlands, the latter perhaps deposited by standing bodies of water [e.g., Fuller and Head, 2002]. Moreover, recently discovered fluvial features distributed preferentially around the margins of the youngest aureole lobes [Chittenden and McGovern, 2004] suggest expulsion of significant volumes of water from the bases of those lobes, consistent with a catastrophic flank movement rooted along an overpressured decollement.
7. Application to Other Aureole Lobes
 The evidence presented above favors a mass movement origin for the north and northeast aureole lobes. By analogy, we argue that the remaining aureole lobes of Olympus Mons, although not examined in detail here, are, nonetheless, likely to have formed by the same mass movement mechanism. For example, the northwest lobe (labeled NW in Figure 1a) has ridge and trough morphology and dimensions, and associated basal scarp relief, similar to those of the north (Figure 1b) and northeast lobes, implying a similar formation scenario. While the morphology of the “lower aureole member” [Morris and Tanaka, 1994], which includes the oldest aureole units, has broad affinities to that of the younger lobes, we acknowledge that significant differences exist. For example, ridges in the broadest lobe (labeled “W,” or west, in Figure 1a) tend to be shorter, narrower, and more rounded than those in the north, northwest, and northeast lobes, and there are more secondary sets of radially oriented ridges in the west lobe in contrast to the primarily margin-parallel ridges in the upper aureole members [e.g., Morris and Tanaka, 1994]. The lower aureole member thins toward its distal edge, whereas the middle and upper members are bounded by sharp topographic offsets (see Figures 1b and 6) [Morris and Tanaka, 1994]. Basal scarp relief associated with the lower aureole member varies by sector; in some sectors the expression of the scarp is subdued or absent entirely (Figure 1a).
 We have argued above (supported by the geologic mapping of Morris and Tanaka ) that the north aureole lobe is probably the one least affected by erosion. If older lobes were also formed by a landslide mechanism, modification subsequent to emplacement may account for observed morphological differences. The geologic record indicates that erosive processes, including aeolian and fluvial, were active in the lower aureole member [Morris, 1982; Morris and Tanaka, 1994; Chittenden and McGovern, 2004]; such activity may have reduced the size of ridges, smoothed ridge and lobe boundaries, and modified the orientations of ridges and troughs. Thus the current morphology of the lower aureole member could have been produced by erosion of block and ridge structures similar to those currently seen in the north and northeast lobes (Figures 2 and 4). Repeated emplacement of lava flows (e.g., Figures 1a and 3b) has reduced and smoothed basal scarp relief in many sectors corresponding to the lower aureole member, accounting for the contrast with the steep scarp bounding the north lobe (Figures 1b and 6). Alternatively, the emplacement mechanics of the lower aureole member may have differed from those of the younger lobes, favoring the production of smaller blocks and thinning margins in distal regions. The proximal sectors of the lower aureole lobes may contain structures similar in size and morphology to those of the north lobe (Figures 1b and 8b); such structures would now be obscured by younger aureole lobes, volcanic plains flows, and aeolian materials [e.g., Morris and Tanaka, 1994]. Both of these scenarios allow for a flank failure mass movement origin for the west lobe and other elements of the lower aureole member.
 We have identified evidence linking blocky structures in the north and northeast aureole lobes to the surface of the Olympus Mons edifice, thereby favoring a mass movement flank failure origin for these structures. Specifically, parallel lineations on the upper surfaces of some aureole blocks resemble the leveed or channelized lava flows that are ubiquitous on the surface of the Olympus Mons edifice. Other lineations seen on aureole block surfaces may be the remnants of faults such as those seen at the margin of the Olympus Mons edifice near the top of the basal scarp. We determined a plausible prefailure geometry for the edifice by matching the volume of a reconstructed edifice section to that of the north aureole lobe. A modest (∼60 km) northward expansion of the current edifice is sufficient to account for the volume of the landslide that formed the north lobe, demonstrating that repeated failure events of an edifice not greatly different in size or shape from the present one are capable of generating the observed aureole deposits. The other aureole lobes probably formed by the same mechanism proposed here for the north and northeast lobes; postemplacement modification and burial of proximal aureoles and the adjoining basal scarp sectors can account for observed differences. The Hawaiian Islands offer structural analogs to the Olympus Mons aureole deposits (Nuuanu slide), basal scarp faults (the Pali faults), and lower edifice benches (the Hilina slump). We suggest that a basal detachment enabled by high-pore fluid pressure, as inferred for Hawaiian volcanoes, allowed flank failure and aureole emplacement at Olympus Mons. Evidence for analogous flank failure processes on Mars and Earth demonstrates that repeated failures and outward spreading are universal phenomena on volcanic constructs.
 We thank Donielle Chittenden for comments and image manipulation help and Brian Fessler, Greg Neumann, and Anton Ivanov for help and advice on coregistering MOC and MOLA data sets. We also thank Ken Tanaka, Rosaly Lopes, and Francis Nimmo for very helpful reviews. This work was sponsored by NASA grants NAG5-12226 (to PI P. J. M. and all other authors) and NASW-4574 (to the Lunar and Planetary Institute). LPI contribution 1204. SOEST contribution 6402.