Identification of a new outflow channel on Mars in Syrtis Major Planum using HRSC/MEx data

Transcription

1 Planetary and Space Science 56 (2008) Identification of a new outflow channel on Mars in Syrtis Major Planum using HRSC/MEx data N. Mangold a,, V. Ansan a, D. Baratoux b, F. Costard a, L. Dupeyrat a, H. Hiesinger c, Ph. Masson a, G. Neukum d, P. Pinet b a IDES, UMR8148, CNRS, and Université Paris-Sud, Batiment 509, Orsay, France b DTP-OMP, UMR5562, CNRS, Toulouse, France c Institut für Planetologie, Münster, Germany d Freie Universität, Berlin, Germany Received 20 August 2007; received in revised form 24 January 2008; accepted 28 January 2008 Available online 16 February 2008 Abstract Syrtis Major Planum is a volcanic plain dominated by lava flows. High resolution stereo camera (HRSC) images of the northern Syrtis Major region display erosional features such as grooves, teardrop-shaped islands and valleys. These landforms are characteristics of outflow channels seen on Mars, therefore implying that a flood event took place in this region. The flow of 100 km long and a few kilometer wide followed the local slopes in most locations. Maximum flood discharges estimated from images and topography vary from about to m 3 /s, and therefore are in the range of terrestrial mega-floods in the Scablands or Lake Bonneville. In North Syrtis Major, the relationships with surrounding lava flows and the timing of the flood coeval to Syrtis Major volcanic activity suggest that it could be related to the subsurface water discharge mobilized by the volcanic activity. The proximity of Noachian age basement rocks 20 km away from the flood and below lava flows might have played a role in its formation and water presence. r 2008 Elsevier Ltd. All rights reserved. Keywords: Mars; Outflow; Water; Volcanism 1. Introduction Very energetic floods suggesting catastrophic episodes of concentrated water flows formed outflow channels on Mars in a similar way as the release of water by glacial surges on Earth. However, the exact origin and conditions under which the outflow channels formed on Mars are still controversial (e.g., Carr, 2006). Outflow channels of the Xanthe Margaritifer region, east of Valles Marineris, date back to the Hesperian period (43 Gy ago) and chaotic terrains of unknown origin characterize their source regions. Most models suggest a formation related to groundwater release either from aquifers, or permafrost melting resulting from increased geothermal flux, or magmatic activity (Carr, 1979; Baker et al., 1992; Rodriguez et al., 2005; Coleman, 2005). Outflows such as Corresponding author. address: mangold@geol.u-psud.fr (N. Mangold). Mangala Valles (Southwest of the Tharsis region) or Athabasca Valles (southeast of Elysium region) have channel characteristics similar to those of Xanthe Terra but their source areas consist of a single fracture, probably related to volcano-tectonic activity (Burr et al., 2002; Head et al., 2003) or tectonic processes (Hanna and Phillips, 2006). The identification of new areas of similar floods is important to complete the classification and geographic distribution of Martian outflows and to understand their relationships with magmatic activity, groundwater distribution and climatic events. The high resolution and spatial coverage of the high resolution stereo camera (HRSC) allows us to study outflow channels in more detail or discover new locations of similar flood activity. The studied region is located at the transition from Syrtis Major Planum and the Nili Fossae regions (161N, E), in the eastern hemisphere, west of the Isidis plain (Fig. 1). No landform indicative of volcanoice interactions (pseudocraters, table mountains, etc.) or /$ - see front matter r 2008 Elsevier Ltd. All rights reserved. doi: /j.pss

2 N. Mangold et al. / Planetary and Space Science 56 (2008) Fig. 2. (a) Overall image of the studied area (HRSC mosaic ). At this low resolution the deep valley (DV) in the bottom center of the image is the only landform visible suggesting a lava tube associated to the lava flows. Details shown in Figs. 3 5 show other erosional features not consistent with volcanic processes. Fig. 1. (a) MOC wide-angle mosaic of the Syrtis Major region. (b) Closeup over the studied region (centered 171N, 761E). fluvial/glacial erosion (channels, valleys) has been previously observed on these Syrtis Major volcanic plains. Only a few studies suggest a possible interaction of volcanism with volatile-rich sediments in the Isidis Syrtis area (Ivanov and Head, 2003), as well as possible subsurface water storage hypothesized by the presence of lobate ejecta craters (e.g. Baratoux et al., 2005). In our study, we focus on erosional features such as grooves, teardropshaped islands, and a deep valley located close to the boundary between the Syrtis Major smooth plains and the northern Noachian highland. On the basis of our investigations, we interpret these features as being the result of an aqueous flood event because these landforms are typical of energetic outflows. We estimate discharge rates and discuss the age and origin of these landforms. 2. Data and geologic context The studied HRSC images are orbits 988, 1347 and 1593 with nadir resolution from 15 to 25 m. A mosaic of the three images has been completed at 25 m/pixel in the region of interest. The Digital Elevation Model (DEM) of each orbit has been computed using tools developed for the HRSC instrument (Gwinner et al., 2005). Nevertheless, the accuracy is very low over smooth plains which lack roughness, limiting the number of correlation points (e.g. Ansan et al., 2008). We therefore use MOLA data for topographic measurements rather than HRSC DEMs on these smooth plains. The Nili Fossae region belongs to Noachian highlands that can be easily identified by their rough and hilly landscape in contrast to the dark smooth lava plains of Syrtis Major (Fig. 2). A north south lineation, visible in the top center of Fig. 2, is a fracture parallel to the main Nili Fossae system. This fracture is probably due to the same tectonic stress as Nili Fossae and it is buried by lava flows to the south, showing that it pre-existed lavas. The highlands locally display valleys with alluvial fans and possible associated lakes (Fassett and Head, 2005; Mangold et al., 2006, 2007). OMEGA/MEx spectral data show that the highlands outcrops contain hydrated minerals such as clays, suggesting an alteration of the ancient crust by liquid water (Poulet et al., 2005). In contrast, Hesperian aged lava flows of Syrtis Major do not show any sign of hydrous minerals. This implies a history devoid of long-term liquid water activity. The presence of minerals such as pyroxenes and olivine confirm the volcanic origin of the material (Bandfield, 2002; Mustard et al., 2005; Poulet et al., 2007; Baratoux et al., 2007; Pinet et al., 2007). 3. Observation of erosional landforms on volcanic plains At regional scale (Fig. 2), the most interesting erosional landform is a valley that crosses the northern part of the

3 1032 N. Mangold et al. / Planetary and Space Science 56 (2008) lava plain close to the contact with the Noachian highlands. This narrow and deep valley is 60 km long with an EW orientation; we will refer to it as deep valley (DV). The DV, as observed on wide-angle images (Fig. 2), could be interpreted as a volcanic feature such as a large collapsed lava tube. However, in the following section, we will present observational facts that render a volcanic interpretation less likely (Figs. 3 8) Erosional landforms on HRSC images Observations at the deep valley source area The western part of DV shows some branching patterns a few kilometers long and a low sinuosity suggesting a more complex formation than expected at regional scale (Fig. 3). The valley head (Fig. 4b) is characterized by a theater-head shape, but a young crater modified the source area complicating the interpretation. We see a sinuous valley continuing on the northwestern side of the crater that branches at the valley head, showing that the valley head has not just a single source. Some striations, or grooves, cover a large part of the area in the west and northwest of the valley head (Figs. 3 and 4a, b). These grooves are relatively narrow (few tens of meters) compared to their length of several kilometers. These grooves are present all around the valley head and follow locally the same direction (Fig. 4b). South of the deep valley, other grooves can be observed, with apparently smaller size, and a NW SE direction (Fig. 4d). Such Fig. 3. (a) Portion of HRSC image 1347 over the lava plain showing a deep valley and many grooves (G) from top left edge through bottom right edge. landforms are typical of erosion into a resistant bedrock. Grooves and valley might have been generated by the same process, as they show same directions. Another type of erosional landform is the teardropshaped islands A and B (Fig. 4). The most obvious example (A in Fig. 4a) is a small topographic feature with a teardrop shape that asymmetrically extends away from a 600 m diameter impact crater. The second small island (B in Fig. 4d) is less developed and is observed to the south, in a region where grooves are also less dense. Typical lava flow features include lobes, dikes, and rilles. Flow lobes of Syrtis Major lavas are identified by typical lobate shape in the southwest corner of the image only (Figs. 3, 4c). In contrast, no lobate shapes are observed in association with the DV, the grooves or the teardropshaped islands. Wind streaks are the only clear eolian landforms visible (Fig. 4a). They form by differential dust deposition and erosion behind the relief created by impact craters. The observed erosional landforms are elongated and their orientation is variable. Grooves have a N S direction in the northern part of Fig. 3 (Fig. 4a) and change to NW SE in the southern part (Fig. 4d). The direction of the two teardrop-shaped islands are exactly similar to that of the grooves direction. In addition, these islands show a sense of erosion from the north to the south. These directions are consistent with the valley system in the western part, but grooves extending south of the valley show a different orientation than the valley. The wind streaks indicate the orientation of the current wind direction from east to west and the lava lobes indicate an orientation of lava flows from the southwest to the northeast. Thus, these two last types of landforms, volcanic lobes and current eolian landforms, are, nonetheless distinct in shape, but also distinct in orientation of the elongated grooves and teardrop-shaped islands Observations of the deep valley eastern part In the eastern part (Fig. 5), the DV is sinuous with locally straight parts and is interrupted by a 10 km diameter impact crater that formed after the valley. Overall, DV has a nearly constant width of m and few tributaries with apparent theater-heads (Fig. 5b). These tributaries branch together and connect to the main valley. These tributaries do not exist everywhere along the main valley, and do not show a dendritic geometry. The eastern termination of DV is enigmatic (Fig. 5c) because it stops abruptly after the connections with tributaries. Due to image resolution and quality, it is difficult to know what exactly occurred at this point. It seems that the valley disappears on the plain but a few lineations are still visible east of that point in Fig. 5c and more to the east. Indeed, a few other grooves and valleylike features can be observed further east. These lineations may represent the continuation of the flow to the eastern part of the studied area but they remain poorly visible features that should be interpreted with caution. In

4 N. Mangold et al. / Planetary and Space Science 56 (2008) Fig. 4. Detailed views of landforms on HRSC image 1347: (a) A: teardrop-shaped island indicating an N to S erosion together with grooves (G) elongated in the same direction over a 10 km wide area (only grooves only is pointed by arrow). (b) Grooves (G) at the head of the deep valley. A sinuous valley (SV) is observed with same main direction as grooves. The wind streak indicates an E to W dominant wind. (c) Lobate features typical of lava flows lobes (LL). (d) Grooves south of the deep valley with a second smaller teardrop shaped island (B). addition to this deep valley a few sinuous landforms seen as negative relief are also identified in diverse locations of the smooth plain, especially to the north, northwest, and east of the central area of Fig. 3; they are mapped in the geomorphic map as valley-like lineations (dotted lines in Fig. 6) Comparison between erosional landforms directions and slopes from topographic data The previous observations show the presence of a 60 km long DV with sinuous shape and few branching tributaries. Close to the source area, DV is surrounded by elongated grooves and tear-dropped islands both typical of erosion. The directions of flow indicated by these features show a NS direction in the western part of the study area changing into a WE direction in the eastern part (Fig. 6). These morphologically derived flow directions can be further investigated with topographic data. The landforms observed are located at MOLA elevations of 1000 m at the west to 2000 m at the eastern end of DV. Regional slope azimuths have been derived from MOLA gridded data (Fig. 7). In Fig. 7, arrows (in blue in the online version) indicate directions of erosional landforms (valleys, grooves and teardrop-shaped landforms) that can be compared to lines with measured slope azimuths (in yellow in the online version). The lava flows lobes are present in the southwest part of the grooved plain (in red in the online version). The orientations of the erosional landforms seem consistent with the overall current topography. A more detailed comparison is done in Fig. 7c. In the region of the teardrop-shaped island A, the regional slope (in yellow) is of 0.11 to the south and increases to 0.51 where we observe the most developed grooves. To the east, other grooves also developed on a 0.51 slope to the south. At the head of the DV, there is a gentle step in the topography with a southern azimuth and a slope reaching here South to the DV, grooves are still present on a 0.51 slope. East of the DV end, the slope is nearly constant at Thus, from A to the eastern end, directions given by landforms are consistent with the direction of the main regional slopes. Nevertheless, notice a small saddle (mapped with S) that indicates a small relief here (100 m of elevation difference

5 1034 N. Mangold et al. / Planetary and Space Science 56 (2008) Fig. 5. (a) Eastern part of the deep valley (DV). (b) Streams connecting (CS) to the deep valley with small theater-shaped head. (c) The deep valley ends suddenly with poorly visible lineations to the right of the image. Fig. 6. Interpretative geomorphic map of the studied area. Grooves (gray lines) and valleys (black) indicate a NS direction at left changing to a WE direction at the bottom central part of the image. Putative erosional landforms (dotted lines) indicate a continuation of the flow to the east and possible indication for the flow source area in the northwest.

6 N. Mangold et al. / Planetary and Space Science 56 (2008) Fig. 7. (a) MOLA color map of the area with elevation contours every 200 m. (b) Regional slopes azimuths (yellow arrows), lava flows direction (red arrows) and fluvial landforms direction (blue arrows). Main slopes are consistent with directions observed for aqueous landforms all over the area. S is the saddle north of the dense grooves of Fig. 4a. The lava front (in red) is determined from the difference in shape of the lava plain (rougher inside red lines). (c) Close-up of MOLA topography contours and HRSC image on the location of Fig. 3. at maximum). North of A, the plain is very flat with the slope lower than 0.11; this area is devoid of any obvious erosional landform. 4. Interpretation of erosional landforms 4.1. Teardrop-shaped islands origin The two teardrop-shaped landforms A and B are similar in shape to the teardrop-shaped islands observed in classical outflow channels (e.g. Baker et al., 1992). The crater creates a zone of non-erosion, or at least of limited erosion, in the shadow behind the crater (Fig. 8a, b). Compared to Ares Vallis, a major flood event, the observed teardrop-shaped landforms are more than ten times smaller (Fig. 8c). The same shadow effect can occur at different scales depending on the intensity of the flood, thus suggesting a smaller flood at Syrtis Major than in Ares Vallis. By comparison, the teardrop-shaped island observed on Fig. 8d in Athabasca Valles, a more limited but recent flood (Burr et al., 2002), is 1 km wide, only two times larger than the similar landforms of the Syrtis Major region. The length/width ratio (L/W) has been widely used to compare these landforms to other processes (e.g. Baker and Kochel, 1979). Values of L/W ratio are known to be of 3.15 for Scablands (Washington State) and 3.25 for the Martian islands in average (Baker and Kochel, 1979; Komar, 1984). They are generally slightly lower than fluvial islands in classical river streams with 4.3 (Komar, 1984). The L/W ratio of the first one (Fig. 8a) is approximately 3.0 fitting to the range of L/W ratio on earth and for the Martian

7 1036 N. Mangold et al. / Planetary and Space Science 56 (2008) Fig. 8. Comparison of landforms on Syrtis Major lava plain and other ouflow channels on Mars. (a,b): Close-ups of the two teardrop-shaped islands of North Syrtis area (see Fig. 4 for localization). (c,d): Teardrop-shaped islands in Ares Vallis floor (HRSC 1607) and Athabasca Vallis (HRSC 2121). (e) Close-up on grooved area in Syrtis area. (f) Grooves in Ares Valles (HRSC 1000). channels. The second one (Fig. 8b) has slightly lower ratio, with 1.9, and is less similar as noticed in the description. Could other processes explain these landforms? The plains material is of volcanic origin and volcanic landforms can locally be erosional. However, the presence of streamlined islands is not possible from lava flows themselves. These two craters obviously formed after the plains formed, or they would not have shown erosional patterns associated. Thus, teardrop-shaped landforms are not volcanic in origin. Current eolian activity follows E W direction different from the grooves and islands, but a past eolian erosion might have followed other pathways. Teardrop-shaped islands and wind streaks are both due to the shadow created by the crater topography but they do not show similar morphologies, suggesting the teardrop-shaped islands are not wind related features. Small tear-dropped shapes behind craters can locally be due to eolian processes in dusty regions (see for example MOC E ). However, these landforms are very small compared to the observed case, as they affect craters of typically 50 m wide or less. In addition, the outcrops of Noachian highlands do not show any sign of any NS direction of erosion, suggesting the observed erosion was limited to the plains material. This characteristic is more difficult to explain if this was created by wind. Eventually, a difference due to the difference in material property between highlands and plains might explain difference of erosion but this should indicate a contrary trend, because highlands are likely more erodable than lava flows. From the difference of scale and shape, it is therefore unlikely that wind erosion can explain the observed teardrop-shaped landforms in Fig Grooves origin Some eroding forces such as wind, glacier abrasion or any aqueous flows often form grooves. Energetic aqueous floods can create grooves, similar to those frequently observed in other outflow channels (Fig. 8e, f). Here, the example of Ares Vallis is not larger than the grooves of Syrtis, in contrast to the teardrop-shaped islands. These grooves are located on the external edge of the valley, at a location where the flood was probably not as much developed as in the main course channel, therefore suggesting a discharge rate of lower amplitude than the Ares Vallis main channel system. Grooves follow the topographic slope as would aqueous flows too. Wind action forms erosional features such as yardangs, i.e. elongated hills formed in weak material parallel to the wind direction. However, wind streaks indicate E W prevailing winds, which are not consistent with the N S orientation of the grooves. One can argue that ancient wind directions could have been different, but yardangs are typically formed by erosion into weak material and are

8 N. Mangold et al. / Planetary and Space Science 56 (2008) much more difficult to create on a lava plain. As explained for tear-dropped islands, no wind erosion of a past N S direction is seen in the relics if highlands visible inside the plain, an argument not favorable to explain grooves by this process too. The grooves follow the main slopes and change of direction sharply with it, despite the slope is very low, a characteristic also difficult to explain with a wind erosion origin. Volcanic grooving is not common, but might occur when lavas are sequestered along a given topography, leading to more erosion than deposition. In Fig. 4c, in the southwestern part of the studied area, lava flows are clearly visible with a lobate shape and individual flows can be followed. However, where plains are grooved and cut by valleys, no lobes are visible, and no individual flows can be followed. Moreover, the presence of streamlined islands associated to the grooves, showing the same direction, does not favor a volcanic origin due to the gap of time necessary to form these two craters after the plains. Thus, the overall lack of any volcanic features due to the lava flows visible at the surface, and associated to the erosional landforms supports the idea that the grooves are not volcanic in origin, and formed after the emplacement of the lava bedrock. Glacial abrasion forms grooves at different scales with associated features such as drumlins and moraines (e.g. Allen, 1997). This hypothesis is possible from the geometry of grooves, but no residual moraines are observed in this region whereas moraines should be present when ice sublimates or melts. Given the slope of o0.51, any glacier would be very thick: Given the basal strength of glacier of 50 kpa, such a slope would invoke a 1.6 km thick glacier which should have left other landforms such as moraines. Despite glacial grooves should not be fully excluded, erosional grooves from an aqueous erosion is easier to explain these landforms Geometry of the valley system The valley by itself resembles lava tubes at low resolution. However, the presence of a few branching tributaries indicates a geometry different from usual lava tubes, generally isolated and not connected. DV has a geometry and a width similar from beginning to end, with only a few tributaries, indicating an origin different from usual overland flows which are generally characterized by an increasing width and more dendritic patterns. The apparent theater-heads of several tributaries and the nearly constant width could suggest a formation by subsurface sapping, i.e. the seepage of water and regressive erosion forming large and poorly hierarchized valleys (Laity and Malin, 1985). A similar landform is visible in the Kasei Valles outflow channels and is interpreted mainly as a late episode of subsurface sapping (Williams et al., 2000). However, the observed grooves point towards an erosion by surface flows, hence a mechanism different from subsurface sapping. Fig. 4b indicates that grooves and DV have approximately same directions at the valley head. In order to reconcile these apparently contradicting observations, an interpretation is proposed: DV might be a local incision of a catastrophic flood that predominantly created grooves. Comparison can be made with the Holocene sub-glacial volcanic flood of the Jo kulsa a Fjo llum in Iceland (Waitt, 2002a, b). Therefore, the surface erosion was followed by the incision of a nearly 1 km large and 100 m deep canyon about 100 km away from the flood source region. Such change of flow mode might occur, for example, when the bedrock strength or the slope changes, or when a preexisting landform such as a lava tube exist and is enlarged and re-incised. The increase of regional slopes to 0.71 at the valley head is consistent with a flow mode change (Fig. 7). These interpretations might also explain the direction difference between DV and grooves south of the valley (Fig. 4d). The flood was not fully sequestered in the valley but part of the flood continued as a sheet flow south to the DV. In summary, the presence of widespread grooves and associated tear-dropped islands that are consistent with the regional slopes is evidence for a high-energy aqueous flow having carved these landforms. We interpret the deep valley as formed by progressive surface erosion, and/or infiltration of volatiles inside the subsurface and backward sapping. In this interpretation, (1) aqueous flows formed first as a sheet flow forming grooves, and tear-dropped islands, over the smooth plains, and (2) aqueous flows carved the DV and its tributaries in specific location controlled by slope or local geology. This interpretation might also explain the abrupt end of the DV. Indeed, catastrophic floods often disappear progressively because the flow discharge decreases progressively in intensity from the source area and fluids infiltrate in the ground or evaporate (e.g. Baker et al., 1992). 5. Putative source areas The teardrop-shaped island A and erosional grooves indicate that the flood originated north of them. Following the erosional features to the north, we observe a gentle topographic saddle less of 80 m high over 20 km (S in Fig. 7). Strong floods might climb low slopes (here only over 20 km) when over pressurized at the source area. An alternative, and probably more realistic explanation, is that this topographic saddle formed after the flood, thus explaining the apparent local discrepancy between the flow directions and the slopes. In fact, the saddle is located at backward slope of a wrinkle ridge (compressive structure) (Fig. 3). Wrinkle ridges have typically elevation differences on the order of several hundred meters that could explain the existence of this relief (e.g. Watters, 1993; Mangold et al., 1998). They have also longer wavelength relief such as broad arches that create small bulge inside volcanic plains. Moreover, these lava plains are formed by lava flows that follows main slopes. As there is no apparent vent or dikes

9 1038 N. Mangold et al. / Planetary and Space Science 56 (2008) feeding lavas at the saddle top, there is no reason that the lavas would have shown this topography during their formation. Wrinkle ridges in Syrtis Major Planum might have formed at the Late Hesperian epoch (Mangold et al., 2000), so after the lava flows, and perhaps after the erosional landforms formed. In addition, the plains north of A are very flat (typically 0.11) while we observe grooves occurring on slopes of It is possible that the very flat relief could have impeded the formation of erosional landforms because of the relative lower discharge involved by this flat topography, therefore rendering difficult the identification of the flow origin. An interesting landform is the chaotic terrain (Fig. 9a) observed 50 km north of A. Here, the smooth plains are no more flat and continuous, but are broken in many pieces that appear to be uplifted from a previously flat surface. A 100 m deep depression formed in these lava blocks. Chaotic terrains are found in many source areas of other outflow channel on Mars. Chaotic terrains at the source of outflows in Xanthe Terra are much larger and deeper than the one of Fig. 9b, with usually more than 2 km in depth and more than 50 km in width, but the resulting outflow channels are much wider too. By analogy with Xanthe Terra, we suggest that this terrain might be a possible source area. Nevertheless, we do not observe any flow features right at the opening of this chaotic region. Several outflows such as Mangala or Athabasca Valles are born on large fractures; i.e. Cerberus Fossae in the case of Athabasca Valles. Similar fractures cross the highlands (Fig. 2). However, none is seen crossing the lava plains that could be a potential source area. Northwest of A, there are only few signs of lineations possibly corresponding to fluvial flows in three areas, especially close to the contact with Noachian highlands. Here, the grooves indicate a flow, perhaps coming from beneath the lava layers (Fig. 9b), thus suggesting the presence of aquifers beneath the uppermost lava layers, but this interpretation remains speculative because of the limitation in the image resolution. In summary, the source region of the aqueous flow is very difficult to identify. Signs of flows exist, but they are not continuous to the region of abundant grooves. 6. Estimation of discharge rates Discharge rates are important for our understanding of the magnitude of outflow events. A main question for this calculation is: do we observe the effect of several episodes of flows or just a single one? DV might have formed as a second stage in the evolution of the outflow, but this stage can be part of the same episode of flow. Grooves visible at the scale of HRSC images required an energetic flow, but a suite of medium-size floods is not excluded. We estimate discharge rates as produced by a single episode of flow, but do not exclude the possibility of multiple episodes. As for other Martian outflow channels, we assume aqueous flows to be liquid water mixed with rocks, but this does not exclude other more exotic volatiles (e.g. Baker et al., 1992; Coleman, 2005) Maximum discharge rate in the deep valley Fig. 9. Close-up on possible flow-related features north of the main landforms. (a) Chaotic zones as a possible origin for the flow. (b) Flow lineations at the highlands lavas boundary. We can calculate the maximum channel discharge with the well-known Manning equation assuming a bankfull discharge in the channel. This must be taken as a rough maximum that might overestimate the true discharge rate. We use the equation modified for Mars by Komar (1979) which corrects for Martian gravity: Q ¼ Aðg m sr 4=3 =g e n 2 Þ 1=2 where A is the flow cross-sectional area, g m and g e are gravity on Mars and Earth, respectively, s is the local slope, n is the Manning roughness coefficient and R is the hydraulic radius, defined as the ratio of flow cross-sectional area to wetted perimeter. Use of this equation to determine flow discharges has been used extensively for Martian outflow channels (e.g. Carr, 1996). A and R are calculated assuming a rectangular shape of the valley, using the depth of 50 m measured with MOLA data, and an average width of 500 m. At this location, the slope s is of The main approximation in this calculation is the coefficient n.

10 N. Mangold et al. / Planetary and Space Science 56 (2008) Taking n ¼ 0.04 as commonly used for a rough non-vegetated bedrock, the result gives a value of Q ¼ m 3 / s. Taking extreme values of n ¼ 0.01 and 0.07 would result in values from 0.3 to m 3 /s. The value of QE m 3 /s should thus be taken as a rough estimate of the maximum discharge rate possible through the valley Discharge rate on the flood plain On the grooved plains, the flood appears as a sheet flow and might not have been as deep as in the valley. The largest teardrop island (A) and grooves are erosion markers that can be used to estimate the flow depth. However, they are topographically subtle structures not visible on HRSC or MOLA DEM. The teardrop-shaped island is also invisible on MOLA profiles because it is located in a gap in between individual profiles. Grooves can be compared to the surrounding landforms in Fig. 7c. Here, the 100 and 200 m high highland residual hills are not affected by the flow limiting the flow thickness to smaller depth. On HRSC close-ups, grooves are apparently of similar height as the rim of the 500 m diameter crater in their surroundings (Fig. 4a). Estimating the rim height of this crater size would give an order of magnitude for the grooves height, so for the flow depth too. The height of crater rim (Hr) is related to the crater diameter (D) according to Hr ¼ 0.036D 1.01 on the moon (Melosh, 1989) and Hr ¼ 0.03D 0.96 for small Martian craters (Garvin et al., 1999). On the basis of the diameter of the impact crater of 500 m, the height would be about 18 m using the first relation or 12 m using the second. Thus, we use a value of 1575 m for the flow thickness. We assume for the calculation a flow width of 10 km, which is the lateral extent of the grooves, and a slope of 0.51 measured locally. The calculation gives a value of about m 3 /s for n ¼ 0.04 and a depth of 15 m, with extreme values of a minimum of m 3 /s using n ¼ 0.07 and 10 m of depth to a maximum of m 3 /s using n ¼ 0.01 and 20 m of depth. This estimation of about QE m 3 /s is consistent with the value obtained for the valley of QE m 3 /s: discharge rates are consistent with an initial sheet flow that was later partially sequestered inside the valley to the east while it was still flooding the plains to the southeast Comparison with other floods The calculated peak discharge rates in the range of magnitude of m 3 /s for the flood system are times lower than those calculated for large Martian outflows such as Kasei or Ares Vallis with values of up 10 9 m 3 /s (Komar, 1979), but they are similar to the values found for late episodes of floods in Kasei Vallis (Williams et al., 2000). These discharge rates are also in the range of calculated peak discharges of terrestrial megafloods, smaller than the glacial surge of Lake Missoula in the Scabland region with m 3 /s (Baker, 1973), and similar to the overflow of Lake Bonneville with m 3 /s (O Connor, 1993) and to the largest reported sub-glacial eruption in Iceland with m 3 /s (Waitt, 2002a, b). These two latter comparisons show that the North Syrtis outflow, despite being not as large as most of the Martian outflows, is an important flood. 7. Discussion: Stratigraphic relationships, age and origin the flood 7.1. Age relationships The Early Hesperian epoch is defined by a crater density of N(5)= , i.e craters larger than 5 km in diameter per million square kilometers, or N(2)= , i.e craters larger than 2 km per million square kilometers (Tanaka, 1986). Crater densities found for Syrtis Major Planum are of N(5) ¼ (Hiesinger and Head, 2004), or N(5) ¼ (Mangold et al., 2000), or N(5) ¼ 100 for the main activity episode (Greeley and Guest, 1987), which all put Syrtis Major volcanic plains in the Early Hesperian. Using the new HRSC images, the area of 4300 km 2 is too small to derive a N(5) age, whereas our crater count gives an N(2) ¼ , which also indicates an Hesperian age, but seems to be slightly older than previous estimates. The age of the volcanic plains does not give the age of the flood itself but a maximum age. Most craters are fresh and not affected by the aqueous flows, but the presence of the two teardrop-shaped islands shows that two impact craters, at least, were present before the flood, thus implying a hiatus of time between the lava plain formation and the catastrophic flood. It is therefore unlikely that the floods formed as a direct consequence of the underlying lava emplacement. Nevertheless, the presence of fresh lava lobes in the SW and North of the studied region (Fig. 6) suggests that the volcanic activity continued in the region after the flood occurred. In that case, the flood might have formed in between two distinct episodes of activity of Syrtis Major lavas, in the Hesperian epoch Origin of the flood The origin of this flood over Syrtis lava flows is difficult to understand as it is for many other outflow channels, but the interpretation is complicated by the poor knowledge of the source area. Here, different hypotheses can be proposed such as (1) a subsurface origin due to ground ice melting or groundwater release by either increasing geothermal activity or heating by lava flows; (2) a subsurface origin due to tectonic pressurization by seismic activity; (3) a regional rise of the water table; or (4) a melting of glaciers present at the surface by incoming lava flows. Hypotheses involving a lake overflow are not discussed hereafter because of the lack of evidence in favor of such a lake in the surroundings.

11 1040 N. Mangold et al. / Planetary and Space Science 56 (2008) The magmatic activity and presence of dikes driving lavas to the surface might be the best candidate to trigger outflows on Mars (Wilson and Head, 2004). The presence of chaotic terrain (Fig. 9b), or putative aqueous flows coming from underneath lavas (Fig. 9a) may favor a subsurface origin. These putative source areas suggest that water was stored in the ground of Syrtis either as liquid water, or as water ice that subsequently melted due to the magmatic activity. This questions if such water was stored inside Syrtis lavas or in the underlying bedrock. An important parameter is that the lava plains on which the outflow formed are located in the vicinity, only 20 km, of the Noachian highlands lavas boundary. Clay minerals are known to exist in the crust of the Nili Fossae region (Poulet et al., 2005; Mangold et al., 2007) suggesting that water might have existed in this region before the Syrtis Major lavas were emplaced. Clay minerals present below the lava flows could have contributed to increase the liquid water storage ability of the subsurface. Thus, at this location, lava could have buried a Noachian crust in which abundant water, as liquid or ground ice, was stored. A recent work (Hanna and Phillips, 2006) shows that outflows might form when the tectonic pressurization due to fault movements would progressively create the subsurface drainage under pressure (Hanna and Phillips, 2006). This hypothesis requires water to be liquid to occur. A large fracture (Fig. 2) appears to cross the highlands north of the study region and might have been a source of seismic activity. However, this fracture does not cross the surface of lava plain. Whether this fracture played as a blind fault after its burial is impossible to demonstrate, and the source of fluid is difficult to explain from a blind fault. Thus, the tectonic pressurization hypothesis is here not as convincing as it is for outflows clearly controlled by major fractures such as for Mangala or Athabasca Valles. A third hypothesis is that the remobilization at surface of subsurface liquid water might have occurred due to a regional rise of the groundwater table. The exact reason of a water table rise is uncertain, and the result of rovers study on the Terra Meridiani area suggests that modifications of the water table on Mars might have occurred (McLennan and 31 colleagues, 2005). Sapping valleys exist in the Nili Fossae region showing the possibility of groundwater activity (e.g. Mangold et al., 2006, 2007). Nevertheless, we do not observe other flood signatures connected to these sapping valleys, so this hypothesis does not well explain the occurrence of the single outflow channels on the volcanic terrain. An alternative explanation to the three subsurface processes proposed is that surface ice stored as glaciers over lava flows might have triggered floods. Glacial surges are the most frequent processes to trigger catastrophic floods on Earth (e.g. O Connor, 1993). This hypothesis could explain the sudden presence of flood features north of the DV: if a glacier was present at this location, the first grooves would be located at the glacial front. However, we do not observe any glacial landforms such as moraines or eskers that could support this hypothesis either locally at the north of DV source area, or at any locations of the Nili Fossae-Syrtis Major region. This hypothesis is therefore much more speculative than the subsurface triggered floods. From the present set of observations, we conclude that the exact formation mode of the flood incising Syrtis Major lava flows remains uncertain. We favor the first hypothesis, the role of the volcanic heating, involving water ice melting and/or liquid water pressurization below lavas, as suggested by the presence of the chaotic terrain, the restriction of the aqueous flows to lava plains and the presence of lava flows unaffected by fluvial features that might form later. The presence of this outflow close to the region of Nili Fossae at a location where the lava is likely thin is certainly not a coincidence: the source of water might be found in the basement below the lava flows. This hypothesis could also explain why the inferred flood features are only observed in the Syrtis region. Indeed, the overall volcanic region of Syrtis Major is devoid of any other fluvial landforms, as observed with new datasets. The flood might sign an activity at the periphery of the bulge due to interactions with the basement. At this point, it should be noted that most outflow channels on Mars are observed in connection with volcanic regions. Late Amazonian outflows have been identified in the Cerberus plain (Athabasca Valles, Marte Valles) in close relation with recent volcanic activity (e.g., Burr et al., 2002). Other examples were found on HRSC images at the southeastern margin of Olympus Mons (Basilevsky et al., 2006). Many outflow channels are connected to Hesperian volcanism, such as Dao Vallis and Harmakhis Vallis around Hadriarca Patera, or Hebrus Vallis around Elysium Mons. Recently, a paleoflood activity has been identified using HRSC images over the Hesperia Planum. Hesperian plain (Ivanov et al., 2005) and volcanic cones have been identified in the chaotic region of Margaritifer Terra (Meresse et al., 2008). All these outflows seem to be genetically related to the formation of volcanic plains or their thermal consequences. Nevertheless, Syrtis Major Planum, as well as Hesperia Planum, are some of the oldest locations on Mars where water activity seems to have been negligible. According to the identification of new outflows in Hesperia Planum (Ivanov et al., 2005) and now in Syrtis Major Planum (this study), it appears that some locations previously identified as among the driest on Mars once had local hydrological activity characterized by quick fluid discharge and flood at the surface. This activity might outline the interactions between the volcanic activity and the water rich basement presently buried beneath the lavas. 8. Conclusion HRSC images of the northern Syrtis Major region at the contact with the Nili Fossae highlands display erosional features such as grooves, teardrop-shaped islands and valleys. These landforms display similar directions

12 N. Mangold et al. / Planetary and Space Science 56 (2008) indicating a likely coeval activity different from lobate lava flows where observed. We interpret these landforms as due to a flood event that likely took place in the Hesperian period. Flow landforms of 410 km long and few km wide are consistent with main topographic slopes in most locations. Peak discharge estimates of about m 3 /s are in the range of terrestrial mega-floods. To date, the most likely origin invokes subsurface water mobilization and discharge due to volcanic activity and subsequent pressurization. The observation of an aqueous flood is unusual in the Syrtis Major region, which is a generally dry region dominated by volcanic landforms and mafic minerals. This identification is comparable to the recent identification of outflow channels by HRSC/MEx data in different volcanic plains in the Hesperia Planum, Elysium Mons or Tharsis regions. It shows how frequent is this process even in regions previously supposed to be very dry and devoid of aqueous flows. This outflow occurred in the periphery of the region, possibly as an expression of water storage buried in the underlying basement rocks. Acknowledgments We acknowledge the helpful reviews by D. Burr and an anonymous reviewer. We acknowledge the effort of the HRSC Co-Investigator Team members and their associates who have contributed to this investigation in the preparatory phase and in scientific discussions within the teams. The French authors are granted by the Programme National de Plane tologie (PNP) of the Institut National des Sciences de l Univers (INSU) and the Centre National d Etudes Spatiales (CNES). References Allen, P.A., Earth Surface Processes. Blackwell Science, Malden, MA. 404pp. Ansan, V., Mangold, N., Masson, Ph., Neukum, G., and the HRSC Coinvestigator team, Topography of valley networks on Mars: comparison between MOLA and HRSC DTM. LPSC38th, Houston, USA, #1660. Baker, V.R., Paleohydrology and sedimentology of lake Missoula flooding in Eastern Washington. Geological Society of America Special Paper 144, 79pp. Baker, V.R., Kochel, R.C., Martian channel morphology Maja and Kasei Valles. J. Geophys. Res. 84, Baker, V.R., Carr, M.H., Gulick, V.C., Williams, C.R., Marley, M.S., Channels and valley networks. In: Kieffer, H.H., et al. (Eds.), Misc. Univ. of Ariz. Press, Tucson, pp Bandfield, J.L., Global mineral distributions on Mars. J. Geophys. Res. 107 (E6), Baratoux, D., Mangold, N., Pinet, P., Costard, F., Thermal properties of lobate ejecta in Syrtis Major, Mars: implications for the mechanisms of formation. J. Geophys. Res. 110, Baratoux, D., Pinet, P., Gendrin, A., Kanner, L., Mustard, J.F., Daydou, Y., Vaucher, J., Bibring, J.-P., The mineralogical structure of the subsurface of Syrtis Major from OMEGA observations of lobate ejecta blankets. J. Geophys. Res., in press, doi: /2007je Basilevsky, A.T., Werner, S., Neukum, G., Van Gasselt, S., Head, J.W., Ivanov, B.A., Potential life habitat at the eastern flank of the Olympus Mons as seen in Mex Hrsc and Mgs Moc images of Mars. Lunar Planet. Sci. Conf. 37th, Houston Clear Lake, abstract #1179. Burr, D.M., McEwen, A.S., Sakimoto, S.E.H., Recent aqueous floods from the Cerberus Fossae, Mars. Geophys. Res. Lett. 29 (1), Carr, M.H., Formation of Martian flood features by release of water from confined aquifers. J. Geophys. Res. 84, Carr, M.H., Water on Mars. Oxford University Press, New York, 229pp. Carr, M.H., The Surface of Mars. Cambridge University Press, Cambridge, 308pp. Coleman, N.M., Martian megaflood-triggered chaos formation, revealing groundwater depth, cryosphere thickness, and crustal heat flux. J. Geophys. Res. 110, E12S20. Fassett, C. I., Head III, J. W., Fluvial sedimentary deposits on Mars: ancient deltas in a crater lake in the Nili Fossae region. Geophys. Res. Lett. 32, L14201, doi: /2005gl Garvin, J.B., Sakimoto, S.E.H., Schnetzler, C., Frawley, J.J., Global geometric properties of Martian impact craters: a preliminary assessment using Mars orbiter laser altimeter (Mola). Lunar Planet. Sci. Conf. 30th, #6163. Greeley, R., Guest, J.E., Geologic map of the eastern equatorial region of Mars, USGS map I-1802B, Gwinner, K., Scholten, F., Giese, B., Oberst, J., Jaumann, R., Spiegel, M., Schmidt, R., Neukum, G., Hochauflösende Digitale Geländemodelle der Marsoberfläche auf der Grundlage von Mars express HRSC Daten. Photogrammetrie Fernerkundung Geoinformation (PFG), 5/2005, S Hanna, J.C., Phillips, R.J., Tectonic pressurization of aquifers in the formation of Mangala and Athabasca Valles, Mars, JGR-Planets, in press, doi: /2005je Head, J.W., Wilson, L., Mitchell, K.L., Generation of recent massive water floods at Cerberus Fossae, Mars, by dike emplacement, cryosphere cracking, and confined aquifer groundwater release. Geophys. Res. Lett. 30 (11), Hiesinger, H., Head, J.W., The Syrtis major volcanic province, Mars: synthesis from Mars global surveyor data. J. Geophys. Res. 109, E Ivanov, M.A., Head, J.W., Syrtis Major and Isidis basin contact: morphological and topographic characteristics of Syrtis Major lava flows and material of the Vastitas Borealis Formation. J. Geophys. Res. 108 (E6) (Art. No. 5063). Ivanov, M.A., Korteniemi, J., Kostama, V.-P., Aittola, M., Raitala, J., Glamoclija, M., Marinangeli, L., Neukum, G., Major episodes of the hydrologic history in the region of Hesperia Planum. Mars J. Geophys. Res. 110 (E12s21). Komar, P.D., Comparisons of the hydraulics of water flows in Martian outflow channels with flows of similar scale on Earth. Icarus 37, Komar, P.D., The lemniscate loop-comparisons with the shapes of streamlined landforms. Journal of Geology 92, Laity, J.E., Malin, M.C., Sapping processes and the development of theater-headed valley networks on the Colorado plateau. Geol. Soc. Am. Bull, Mangold, N., Allemand, P., Thomas, P.G., Wrinkle ridges of Mars: structural analysis and evidence for shallow deformation controlled by ice-rich décollements. Planet. Space Sci. 46 (4), Mangold, N., Allemand, P., Thomas, P., Vidal, G., Chronology of compressional deformation on Mars: evidence for a single and global origin. Planet. Space Sci. 48 (12 14), Mangold, N., Poulet, F., Mustard, J.F., Bibring, J.P., Langevin, Y., Gondet, B., Ansan, V., Masson, P., Hoffman, H., Neukum, G., and the OMEGA and HRSC co-i teams, Geological relationships between phyllosilicates and olivine outcrops in Nili Fossae region. Mars. European. Geophysics. Union 3rd Assembly, EGU06-A-06832, Vienna. Mangold, N., Poulet, F., Mustard, J.F., Bibring, J.P., Langevin, Y., Gondet, B., Ansan, V., Masson, P., Hoffman, H., Neukum, G., 2007.