In order to investigate sources of lava and water to the Cerberus plains of Mars, geomorphological mapping on High Resolution Imaging Science Experiment (HiRISE) images was carried out to reveal the history of activity of fissures and pits which lie upstream of channels and deposits associated with a wrinkle ridge near Cerberus Fossae. The fissures and pits are superbly exposed and imaged, and flows and channels emanate directly from them, interpreted as clear evidence that these are vents. The mapping establishes stratigraphic relationships between the plains and the channels and deposits originating from the vents, establishing the vent history. For example, to the south of the wrinkle ridge, both incised channels and leveed flows extend onto the southern plain and are clearly the final phase of plains-forming activity. Conversely, to the north, vent-sourced channels only incise the plain close to the ridge—beyond that, they are overlain by large-scale regional flows that appear to have originated from the direction of Athabasca Valles. In the southeast, there is evidence of contemporaneity between vent-sourced activity and large-scale plains-forming flow that was not sourced from the vents, indicating that activity here was part of a broader process of Cerberus plains formation from multiple sources. Crater counts show all the activity to be Late Amazonian, with the latest activity tentatively dating to circa 11 Ma. Thus, this study implies that very recent outflows from these vents contributed to the formation of the Cerberus plains and constrains the timing and local flow direction of plains-forming deposits from other sources.
 The Cerberus plains, also known as Central Elysium Planitia, have long been recognized as one of the youngest surfaces on Mars on the basis of their sparse impact cratering [Tanaka and Scott, 1986; Plescia, 1990]. Early Viking imagery was used to suggest that they formed by fluvial processes [Tanaka and Scott, 1986] or by lava flows overlying earlier fluvial channels [Plescia, 1990]. As higher resolution imagery and Mars Orbiter Laser Altimeter (MOLA) topographic data have become available, theories regarding their formation method have become more complex, with the surface of channel areas such as Athabasca Valles being attributed to either mixed fluvial and volcanic processes [Burr et al., 2002a] or volcanic processes alone [Leverington, 2004, 2011], and other areas to fissure and shield volcano-sourced plains volcanism [Plescia, 2003; Vaucher et al., 2009a]. The relative youth of the surface has also been confirmed by counting craters at higher resolution, in some cases indicating model ages of as little as a few million years old [Berman and Hartmann, 2002; Vaucher et al., 2009b].
 However, there is considerable debate regarding the source of the lava and/or water which formed the plains. MOLA topography and Mars Orbiter Camera and High Resolution Imaging Science Experiment (HiRISE) imagery indicate that flows forming Athabasca Valles originated either at the Cerberus Fossae [Plescia, 1990; Burr et al., 2002a] or to their north [Berman and Hartmann, 2002; Edgett and Malin, 2003], but these flows do not extend over the entire Cerberus plains (Figure 1). They appear to have flowed from northeast to southwest (Figure 1) and do not extend further to the southeast than a ridge centered on 157°E, 7°N [Vaucher et al., 2009b]. On the far side of the ridge, low shield volcanoes may have been the source of some of the surface material, but flows of a viscosity high enough for form such positive features are likely to have had limited spatial extent [Plescia, 2003; Vaucher et al., 2009b]. Martian lava surfaces similar to those of the Cerberus plains have been likened to terrestrial flood lavas, such as that erupted at Laki, Iceland, in 1783–1784 [Keszthelyi et al., 2000]. These originate from fissures with a subdued topographic expression so their sources are vulnerable to burial by younger lava flows [Plescia, 1990]. It is advisable to look for still-visible sources at topographic highs where the potential for burial is lower.
 Recent high resolution imagery has revealed that the topographic high which divides the Athabasca Valles region from the plains to its southeast is indeed cut by a series of pits and fissures. Each pit/fissure has associated channels and deposits indicating outflow onto the plains surrounding this ridge. They are therefore identified as vents which may be the source of the Cerberus plains material surrounding the ridge. Detailed examination of their stratigraphic relationships with the plains is necessary to test this theory but had not been conducted prior to the present study: the geologic map by Tanaka et al.  gives general unit names for the ridge and plains deposits but is at a large scale and is based on earlier, lower resolution imagery.
 The present study produces large-scale geomorphological maps to establish the stratigraphic relationships between the outflows from the ridge and the Cerberus plains surrounding it. It is shown that deposits from the vents do overlie and/or modify the plains to the south and immediate north of the ridge and thus are the source of a part of the Cerberus plains. Crater counting is also conducted to supplement the stratigraphic observations and show that the ages of plains to the south of the ridge and deposits from vent-sourced outflows are similar. The most recent deposits and landforms formed by the flows from the vent coincide with a large regional flow from the east, probably lava, which may be as young as 11 Ma.
2 Geological Setting
 The study area is in the western Cerberus plains of Mars (Figure 1), also known as Central Elysium Planitia. The Elysium Rise lies to the northwest, with Elysium Mons at a distance of circa 1150 km. Activity at the Elysium Rise central volcanoes is generally much earlier than deposition of the Cerberus plains [Werner, 2009], and lava flows at the edifices are of restricted extent [Platz and Michael, 2011], so the volcanoes themselves are not the source of the plains material. However, the Cerberus plains, if volcanic, may represent the most recent phase in the long history of eruptive activity in the Elysium Volcanic Province after a shift in the locus of activity to the south [Platz and Michael, 2011].
 The Cerberus Fossae lie northeast of the study area (Figure 1). The fossae are a series of sublinear graben trending approximately 110°, stretching over 1000 km from northwest to southeast. The Athabasca Valles outflow channel originates from at or to the north of the southwest part of the graben system [Plescia, 1990; Berman and Hartmann, 2002], and flow features and channel morphology indicate flow was towards the southwest [Burr et al., 2002b].
 The study area for this paper lies between 156.8°E, 6.72°N and 158°E, 8.0°N and centers on a ridge with a maximum elevation of 200 m above the surrounding plains according to MOLA data (Figure 2). The ridge is a broad sublinear asymmetric rise topped by a narrower ridge. This is consistent with it being an arch topped by a wrinkle ridge, a compressional landform which may be the surface expression of a relatively old blind thrust fault [e.g., Schultz, 2000], although the results of this paper do not rely on this interpretation. In order to distinguish this from other topographic highs, this association will be referred to as the “wrinkle ridge” in this paper. It is orientated approximately concentric to the Elysium rise and is shallowly dipping to the southeast. The large-scale geological map of Tanaka et al.  assigns it to the Late Hesperian Utopia Planitia 2 unit (HBu2) which is considered to be colluvium in other areas but has an unknown origin and source in this region.
 The wrinkle ridge is surrounded by smooth, subhorizontal, lightly cratered plains with some upstanding heavily cratered inliers. The plains have been assigned to the Late Amazonian Cerberus Fossae 3 unit (AEc3) [Tanaka et al., 2005] which is the youngest unit in the region and is suggested to be composed of platy ridged lava flows with fluvial deposits in the outflow channels. Flow features indicate regional inflow from the north of the study area (Figure 1) [Vaucher et al., 2009b], terminating just short of the ridge. To the south of the ridge, the source of the plains material is not clear, though the orientation of lobate margins in flows to the southeast may indicate inflow of some material from east of the study area. These flow directions are consistent with the present gentle slope of the region to the southwest.
 The ridge is cut by 13 fissures, pits, and broader depressions of varying depth (Figure 2). Channels run from these downslope onto the plains, either incising into the ridge or deposited between levees on its surface (examples of each type can be seen in Figure 3). The fissures, pits, and broad depressions are collectively referred to as “vents” in this paper due to their clear genetic association with the channels and their deposits. The vents and channels, in contrast to the ridge, have crisp margins and are relatively uncratered.
 In order to constrain the processes, age and history of activity of the vents, photogeological mapping was carried out to establish a relative stratigraphy by units' superposition relations and an absolute stratigraphy through detailed crater counts. Vents within the wrinkle ridge were mapped utilizing high resolution imagery where available, and individual maps were combined into a regional map.
3.1 Large-Scale Mapping
 Base images were obtained from the High Resolution Imaging Science Experiment (HiRISE) onboard Mars Reconnaissance Orbiter (Table S1 in the supporting information). “Map projected” JP2 products were used, an equirectangular projection of the image onto a two-dimensional surface. Mapping was conducted at 12.5% of the full scale of these images. The full-sized HiRISE images and THEMIS (Thermal Emission Imaging System, Mars Odyssey) daytime IR imagery covering the area were consulted when necessary to resolve stratigraphic questions. The final maps were then scaled down to a practical scale for publication, 1:80,000 or 1:40,000.
 The mapping performed was geomorphological, with the aim of constraining the stratigraphic relationships between channels, deposits, and the vents. Units were defined by following contacts and noting changes in surface texture and elevation. They were named on morphological type: Ff for fissure, pit, or depression fill; Cf for channel fill; E for eroded areas; and D for deposits unrelated to channels. The heavily cratered wrinkle ridge basement was named B. Units were numbered primarily on the basis of their position in the stratigraphic column within a map and secondarily from west to east across the maps. Number does not indicate stratigraphic relationships across maps or between unrelated units. Deposits lateral to channels, vents, and eroded areas and interpreted to have been deposited during their formation were prefixed L, LF, and LE, respectively, and given the same number as their associated unit. Where the vent or channel landforms themselves are referred to in unit descriptions, they are referred to as Fn or Cn where n is the number ascribed to their fill units.
 A stratigraphic tree was produced for each map indicating superposition and erosive contacts. Though these trees young upwards, the horizontal position of a unit relative to one with which it does not have a contact does not indicate relative age.
3.2 Regional Mapping
 A regional map was produced based on CTX images (Mars Reconnaissance Orbiter Context Camera) as listed in Table S2. These were projected in an equirectangular system and overlain to form a mosaic base image.
 Geomorphological mapping was conducted as on the large-scale maps. Numbering of new units continued on from that reached in the large-scale maps.
3.3 Depth Calculations From Shadow Lengths
 In order to study depth of channel incision, shadows were measured at channel edges. Relief was calculated from these measurements using equation (1).
where h is the relief, d is the shadow length measured in the direction of the solar azimuth of the image, and θ is the solar incidence angle. d and θ where obtained from http://viewer.mars.asu.edu/planetview/inst/ctx/[image_name]#start and adjusted to the scale and orientation of the map in question.
 Shadow length was measured on the large-scale maps. Locations where chosen where shadow margins were sharp and where there were no dark deposits on the surface which could be mistaken for shadows.
3.4 Crater Counting
 Craters >8 m diameter were mapped and counted for units of specific interest. Crater diameters and sample area sizes were determined using CraterTools software [Kneissl et al., 2011]. Sample areas are indicated on the maps (supporting information) and were chosen on the basis of the following:
 Relevance of the unit to questions raised by the mapping.
 Homogeneity of the surface texture within the sample area.
 Size—minimum area 2 km2.
 Absence of abundant superficial unconsolidated material. In particular, some areas of the B ridge surface had a more subdued texture and lower number of small craters than others. This was taken to indicate a mantling deposit, and these areas were avoided.
 Absence of surface textures such as circular patterns which may obscure impact craters.
 The study area is 525 km from the recent Zunil crater, and many small craters (<100 m diameter) in this region have been shown to be secondary craters from that impact [McEwen et al., 2005]. It was therefore necessary to avoid counting secondary craters where they could be identified. The following were omitted from counts:
 Clusters of similar-sized craters of a similar degradation state, especially where they shared an ejecta blanket.
 Clusters of a similar degradation state and size forming linear chains.
 Craters with strongly directional ejecta blankets: these form through oblique impact [Gault and Wedekind, 1978] which is characteristic of secondary impacts. This method of discrimination is only applicable to younger craters which have visible ejecta blankets, but as Zunil crater formed recently (within the past 5 Ma [McEwen et al., 2005]), this is a relevant attribute for identifying its secondaries.
 To avoid counting nonimpact-related features, circular depressions were only counted if they had raised rims. All other craters which intersected or lay within a sample area were mapped and counted, irrespective of degradation state.
 Crater size-frequency data were plotted using the Craterstats2 program [Michael and Neukum, 2010]. In order to determine tentative absolute crater model ages, data plots were compared to isochrons based on the most recent version of the Martian crater production function [Hartmann, 2005] and the Hartmann and Neukum's  chronology function. Error bars were calculated using a 1-sigma error on N(1), which was taken as 1/sqrt(n) where n = number of craters in the range used for the statistical fit.
 Counts of large craters are very low in this area, so much of the data relates to small (<100 m) craters. Though steps were taken to avoid counting secondary craters, these are abundant at small diameters, so the model ages indicated by the production function are likely to be older than the surface's real age.
 Figure 4 summarizes the sources of material to the study area. This is based on the detailed large-scale maps and regional map which can be found in the supporting information. The relationships between plains units and units sourced from fissures, pits, and depressions on the wrinkle ridge differ at different locations around the wrinkle ridge as shown below.
4.1 Southwest End of the Wrinkle Ridge
 Channels and deposits from the vents at the west and center of the ridge incise and overlie the plains to the south. Figure 5 shows the relationship at the southwest of the ridge in detail. The stratigraphic relationships are given in Figure 6, and unit descriptions and interpretations in Table 1. All the channel units here, Cf1-8, originate from vents F1 and F2 on the wrinkle ridge and overlie the D1 plain to the south. The most recent are Cf5 and Cf8, the upstream regions of which are seen in Figure 3. C8 incises deeply into the wrinkle ridge in its upstream reaches: shadow measurements at i–iii on Figure 3 give depths of 32 m, 11 m, and 14 m, respectively. Downstream, where it passes from the B ridge basement onto the D1 southern plains, it incises more shallowly, widens, and deposits material over its banks and on its bed. C5 is a constructional leveed channel which overlies the basement (B) upstream and the D1 plains in the south. Several earlier erosive and constructed channels, Cf1–4 and Cf7, also overlie or incise this plain, though their upstream and downstream reaches are disconnected by later erosion at the ridge-plains interface. Clearly, a complex series of flows was produced by F1 and F2 and deposited material on the southern plains.
Table 1. Descriptions and Interpretations of Units in Figure 5
Channel Fills and Associated Deposits
Coarse-grained deposits within the shallower downstream reaches of C8, breaking up into angular fragments choking the channel downstream.
Material deposited by flow in an erosive channel. Transport was less efficient downstream.
Deposits lateral to C8d with finely fingered margins and local high-standing flat-topped areas with steep linear margins.
Overbank deposits from C8d. Morphology suggests the material at the margins was derived from the high-standing areas, either as breakouts during lava inflation or, if the deposits are ice-rich, through postdepositional freeze-thaw processes.
Distally coarsening dimpled deposits lateral to Cf8d.
Overbank deposits from C8d.
Surface of C5. Rough, angular material with margin-parallel lineations along the channel.
Lineations formed by deposition of material in channelized flow.
Multiple superimposed lobes of rough material lateral to Cf5b.
Levee deposits along C5.
Incised linear regions with smooth, longitudinally lineated beds.
Channels formed by erosion.
Superimposed layers of material with lobate margins, smoother than L5b.
Earliest overbank deposits from C5.
Surface of an incised channel with deposits in stepped ridges and local rough, flow-parallel deposits. Margins are scalloped.
Deposits from flow in C8. Steps suggest deposition ebbed over time, layering newer deposits gradually further towards the center of the channel.
Distally fining deposits lateral to Cf8c; patchy and sinuous at the channel margin.
Overbank deposits from flow in C8, partially eroded by fluid from the channel after deposition.
Shallowly incised region with smooth, polygonal texture.
Early phase of C8 channel deposits. Shallow incision suggests brief flow or low flow rate.
Lobes of medium-rough material lateral to Cf8b.
Overbank deposits from C8b.
Distally coarsening semiconsolidated/unconsolidated material within C7. Isolated southern segment has a smooth, polygonal texture.
Channel which is erosive upstream and depositional downstream. A section of its deposits has been removed during formation of D4.
Rough material with lobate margins. Lateral to and forming the uppermost part of the banks of C7.
Overbank deposits from C7.
Smooth, slightly dimpled deposit lateral to C7 with indistinct margins.
Earliest phase of overbank deposits from C7.
Surface with margin-parallel deposits of rough material on a smooth, higher albedo surface. Stepped morphology at its east margin.
Channel deposit. Overlies Cf2 in the west and may have formed by a late stage of flow in C2: relationship is obscured by erosion in the region of D2. Steps at east margin suggest erosion into a layered substrate.
Elevated deposit with a rough, angular surface, a honeycomb texture and bank-parallel ridges.
Deposit of material during channelized flow within channels previously eroded by C1.
Multiple superimposed lobes of rough material along the east of Cf2.
Levee deposits of Cf2. Lacks the texture of L2i, so may have formed under different conditions.
Multiple superimposed lobes of rough, pitted material with a honeycomb texture at the highest elevations.
High viscosity levee deposits of Cf2.
Partially eroded surface with patchy irregular material along C4 and on obstacles within it.
Erosional and depositional surface from earliest flow in C4.
Locally discontinuous material with smooth but uneven surface.
Overbank deposits from earliest flow in C4. Formed of lower yield-strength material than L4b.
Smooth surface with darker flow-parallel lineations.
Fine material deposited during erosive flow in C1.
Semirough flat-lying material with lobate margins.
Overbank deposits from C1.
Smoothly dimpled subhorizontal deposit with convex-sloping lobate margins and closed hollows. Truncates channel C7, which overlies D3.
Erosional/depositional unit which removed Cf7 and D3 deposits. Hollows suggest formation through lava inflation or secondary material loss.
Deposits at the margins of D3, forming lobes with finely fingered margins.
These may have formed as terminal deposits during the emplacement of D3, or through postdepositional erosion of D3.
Subhorizontal surface with coarse polygonal texture, polygons approximately 70 m across, and some rough, dark areas. Margins are steep and convex.
Lava or ice-rich deposit. LD3 at its margins, convex scarps, and polygonal texture may have formed through pahoehoe inflation or postdepositional processes in an ice-rich body.
Subhorizontal surface with a smoothly polygonal texture, closed hollows, and sinuous grooves.
Scarps down to this in surrounding earlier deposits suggest it was partially formed by erosion. Surface features suggest either freeze-thaw of icy deposits or inflated lava.
Surface with both smooth, high albedo areas and rough deposits.
Primarily erosional surface formed by fluid flow from the east.
Subhorizontal surface with dark, rough higher elevation areas and light, smooth lower elevation areas. Both surface types are crossed by irregular ridges.
Surface formed by a widespread flow with syndepositional or postdepositional breakup causing the variation in surface type and ridges.
Heavily cratered smoothly irregular surface which forms an asymmetric ridge through the center of the region.
Older basement upon which units are deposited and into which channels are cut.
 Regional mapping (supporting information) shows that deposits from vent F7 in the center of the ridge also overlie regional plains deposit D1. A shallow channel (Cf24) extends south from the erosive area around Ff7 (E6). Extending from its margins is a smooth surface with a polygonal texture which gradually blends into the D1 surface in the south (D18). This vent-sourced deposit thus forms the surface of the plain in this area.
 A contrast-enhanced Context Camera (CTX) mosaic (Figure 7) reveals that the surface of the plains up to 20 km from the wrinkle ridge features irregular ridges and areas of high and low albedo fanning to the south of the units seen in Figure 5 and the regional map.
4.2 Southeast of the Wrinkle Ridge
 In the southeast region, large-scale flows from outside the study area overlie the majority of vent-sourced channels and form the plains in this region. However, the plains deposits are locally incised by the very youngest vent-sourced units.
 Figure 8 shows the relationship between a large flow unit and channels from F9. D15 is a sheet-like deposit with a varying surface: dark, coarse-grained and ridged in some areas, smooth with a polygonal texture in others and with some smooth-floored channels. Higher albedo deposits at its margins are sinuously grooved and have bluntly fingered lobate margins. This deposit overlies the channels here (Cf20), truncating them. D15 does not show surface modification where it meets the channels, so it postdates flow through them.
 The relationship of vent-sourced units and this large regional flow is more interactive at the western margin of D15, where it meets channels and deposits from the broad pit F6 (Figure 9). Though D15 overlies the earlier channel C19 (regional map—supporting information), D15 is slightly incised by E4, an erosive area adjacent to F6. The higher albedo marginal deposits of D15 fringe E4, suggesting that formation of these marginal deposits was controlled by the position of E4.
4.3 Northwest of the Wrinkle Ridge
 Vent-sourced channels extend north and northwest from the wrinkle ridge onto the plains, incising into them shallowly (Figure 4). For example, the channel filled by Cf9b (regional map—supporting information) begins at vent F4, incises through the ridge, and curves round to run subparallel to the northern margin of the ridge. It is bounded in the north by two smooth-surfaced units, L9Ei and L9Eii, which have unconstrained lobate margins which point towards the north (Figure 10) and are interpreted to be overbank deposits from C9.
 These channel-lateral deposits overlie D10, which is a coarse-grained unit with grooves within it diverging towards the south. These structures may be distributary and suggest this earlier surface was formed by flow of material from the north. In the north of Figure 10, D10 is overlain by D12. This if formed of thick lobes of a material which appear similar to D10 but with a much crisper morphology. The orientation of its margins suggests that it was emplaced from the north, and this is borne out by the regional mapping (Figure 4; regional map—supporting information): it is the southernmost extent of a large lobate unit extending from the north of the study area. To the west of this unit (D11–12), there is another widespread unit with a low albedo, rough surface and structures indicating flow from the north, D20. D20 breaks up into fingers where it meets D11–12, indicating that D20 formed after emplacement of D11–12. D20 overlies the youngest channel in this region, Cf25 (Figure 11), showing that large-scale flows from the north postdate vent-sourced flows to the north of the wrinkle ridge.
4.4 Crater Count Age Constraints
 Crater size-frequency plots for units in the study area are shown in Figure 12. All plots show a lower proportion of small craters than predicted by the production function, suggesting an anomalously high level of superficial resurfacing.
 The surface of the ridge (B) has a higher crater density that the other units, plotting distinctly from vent-sourced units and regional plains. This shows that the ridge surfaces are older and this is supported by its crater morphology: the rims are in places smoothed, and ejecta blankets are indistinct or incomplete. As its plot crosses isochrons, B was not considered a good candidate for obtaining an absolute age.
 Plots for vent-sourced units and regional plains units are similar, suggesting approximately contemporaneous formation. The southern plains D1 and D3 and the later erosive area D4 plot in very similar positions to vent-sourced activity forming Cf8d and Cf9b. The large southeast regional flow, D15, plots towards the lower end of the range shown by plains and channel deposits and is also the largest sample of the nonbasement units (26 km2). Cumulative (Figure 12a) and differential (Figure 12b) fits were therefore made to this data. Fits were made on a diameter range of 16 m–31 m to minimize fit error caused by superficial mantling of small craters and the low density of larger craters. These gave a model age of 11.1 ± 1.9 Ma (differential) and 10.9 ± 1.4 Ma(cumulative). The sample is small, so these ages should be treated as tentative.
4.5 Overall Stratigraphy
 Figure 13 shows overall unit age relationships on the basis of both crater count and stratigraphic data.
 The oldest surface in the region is the wrinkle ridge itself. The vent-sourced channels show flow directions and variations in depth of incision in accordance with the present topography, indicating that ridge formation occurred prior to the period of vent-related activity.
 The oldest plains in the region are D1 and D3, southwest of the ridge. These are overlain by deposits from the western vents, F1 and F2. There was some erosion of these deposits at the ridge-plains interface near the end of the period of vent activity but they are largely intact.
 The stratigraphically earliest activity in the east originates from vents F3, F4, and F9. The periods of activity from these vents overlap, with deposits ceasing towards the north from F9 first but towards the south from F3 and F4 first. On the basis of crater counts, outflow from these vents was approximately contemporaneous with the latest activity from the western vents.
 To the north, the widespread regional flow deposit, D20, postdates all vent-sourced activity.
 To the south of the ridge, deposits and channels from F3, F4, and F9 are truncated by shallower pits F6 and F7. Early deposits associated with these pits and with the eastern vent F9 are overlain by the regional plain D15, but the youngest vent-sourced unit, E4, incises D15, so it slightly postdates emplacement of the southeast regional plain.
 All surfaces were modified by later deposition of fine-grained material. Some of material may have derived from the vents via airfall (if vent activity was volcanic) or flooding (if it was hydrous). The action of aeolian erosion and deposition then further modified the surfaces by reworking this material and introducing material from greater distances. Formation of dark streaks on slopes has also occurred, with new steaks appearing on mounds in the 4 years between CTX image T01_000867_1873_XN_07N202W and HiRISE image ESP_016427_1875.
 Overall, the result implies that a complex series of flows from the fissures, pits, and depressions along this wrinkle ridge have contributed material to form part of the Cerberus plains through episodic flow. Identification of these vents and channels reveals the source of the material forming the Cerberus plains beyond the southeast extent of Athabasca Valles and indicates that outflow was by at least two mechanisms: erosive flow and constructive, leveed flow. Their observed relationships with large-scale units emplaced from the north and east of the study area constrain the relative ages of other plains-forming flows in the region.
5.1 Implications for the History of Formation of the Cerberus Plains
 Outflows from the vents in the study area are the source of the Cerberus plains in the south and immediate north of the wrinkle ridge.
 In the south, constructive flows in C5 overlie the plain and Cf8 (Figure 5) and Cf24 (regional map—supporting information) incise it and deposits material onto it. Beyond these units, though individual channels are not visible, the surface of the plain shows features suggesting earlier deposition of material from the same direction (Figure 7). This suggests that the southern plain was modified by channels and/or deposits from the wrinkle ridge over a wide area. This is supported by the appearance of D1 around Cf8 (Figure 5): it is similar to D15 in that it is made up of darker, coarse-grained regions with irregular ridges and lower smoother areas with a polygonal texture, but it is smoother and has a higher albedo and fewer dark, ridged areas than D15. This suggests it was originally similar to D15 but has been draped by a layer of fine-grained material. This clastic material could be either airborne or deposited by hydrous flow, but the presence of features linking it to the vents favors the latter explanation.
 In the north, deposits from the ridge have a purely local effect on the plains. Within a maximum of 8 km from the edge of the wrinkle ridge, they are overlain by large-scale flows from the north. Thus, though material from the vents here does not form the plains to the north, it does constrain the timing of these large-scale northern flows: they occurred after the last vent-sourced flows to the north of the wrinkle ridge.
 In the southeast, channels C20 and C19 are overlain by a large-scale flow from the east, D15, and as in the north outflows from the vents are not plains-forming. However, E4, an erosive area associated with F6, incises this deposit and appears to dictate the location of its marginal deposits. This indicates that formation of E4 occurred during and continued after deposition of D15, and so shows that the latest vent-sourced activity was contemporaneous with large-scale flow units from the east.
 The crater counts give some indication of the absolute age of this activity. Sample sizes are small, leading to large error margins for larger craters and a reliance on counts of small craters. For these reasons, model ages must be viewed as tentative and should be reviewed as isochrons are refined in future. However, current isochrons indicate a model age of circa 11 Ma for D15, which constitutes the largest sample area of all of the nonbasement units (26 km2) (Figure 12). As some vent-sourced activity has been established to postdate this deposit, the youngest activity from the vents can also be tentatively ascribed an age of circa 11 Ma. This is supported by the restriction of the crater size-frequency plots of other vent-sourced channel units to between the 100 Ma and 10 Ma isochrons in Figure 12. This is also an upper age constraint for the large-scale northern flows.
 This is within the range of ages obtained for this region (ACy) by Vaucher et al. [2009b] and is in particular consistent with the model ages for surfaces to the south (15.2 ± 3 Ma) and east (13.5 ± 3 My) of the study area. These findings indicate that there was a recent period when flows occurred from a variety of sources to form the Cerberus plains. The vents on this wrinkle ridge emitted material to form the plains to its south. Large-scale flow deposits composed of ridged plates and smooth regions were emitted by an as yet unidentified source to the east. After this activity, another unit of large-scale deposits were emplaced by flow from the north. The location, flow direction, and thick, lobate morphology of these northern units is consistent with lava deposits originating near Athabasca Valles mapped by Vaucher et al. [2009b].
 The orientation of the vents on the ridge is similar to that of the Cerberus Fossae, which are often cited as the source of the Athabasca Valles flows [Plescia, 1990; Burr et al., 2002a]. This suggests they formed under the same regional stress regime and that this regime allowed the release of the large quantities of lava and/or water which formed the Cerberus plains during this period. This has implications for ongoing thermal and tectonic evolution of Mars.
5.2 Nature of the Outflows From the Vents
 There has been much discussion of whether the surface of the Cerberus plains was entirely formed by lava flows [Plescia, 1990, 2003; Jaeger et al., 2007, 2010] or whether parts were formed by hydrous flow and were subsequently modified by periglacial processes [Page, 2010; Balme et al., 2010]. Many of the attributes of the channels and deposits in the study area could be formed by either: polygonal textures by lava cooling [Peck and Minakami, 1968] or postdepositional freeze-thaw [Levy et al., 2010], erosive channels by water or low viscosity lava [Leverington, 2004], small mounds within D18 by periglacial processes [Page and Murray, 2006] or water-lava interaction [Bruno et al., 2004] and smooth and sinuously grooved deposits at flow margins by lava inflation [Keszthelyi et al., 2000] or postdepositional freeze-thaw.
 Evidence from surface morphology and crater counts that there is a mantle of fine-grained material on surfaces around the vents and channels may also be attributed to either volcanic or hydrous processes: pyroclastic fall or flood deposition. The variation in this mantle by location does not clearly favor either process. In addition, if this mantle could be shown to be airfall, then this would not necessarily be sourced from the vents: there are several small mounds on the wrinkle ridge which lack vent features but are possible volcanic edifices and may have erupted pyroclastic deposits (D7 in Map S3, D8 in Map S4—supporting information). Their surfaces show a similar density of impact cratering to vent-sourced units, indicating a similar surface age and thus period of activity.
 It is clear, however, that channels were formed by at least two processes: erosion and construction of leveed channels on the basement. Twenty-two of 25 mapped vent-sourced channels were incised into the basement. These all originate from linear fissures. Shadow lengths indicate a depth of incision of at least 32 m in C8 (Figure 3). The presence of less-eroded terrace-like areas above the deeper channels (e.g., those from Fissure F2 in Figure 3) suggests either localization of the channel after an initial broader incision or periods of bankfull flow punctuating longer periods of shallower flow within the channel. Where there are deposits lateral to these channels, they are flat-lying and lobate. All this suggests these channels were formed by turbulent flow of a low viscosity liquid. Though this is consistent with fluvial emplacement, it has been suggested that lava can form erosive channels on Mars [Leverington, 2004] and low viscosity lava flows have been identified in this region [Vaucher et al., 2009a]. Therefore, lava cannot be ruled out as the erosive agent.
 Three channels, Cf4, 18, and 19, were constructed on the basement. These were sourced from shallow fissure F1 in the west and the broader depressions F6 and F7 in the southeast. The channel banks are formed of superposed lobes of rough material which piled up to form levees, and where channel termination is seen in Cf5 (Figure 4, regional map—supporting information), it is abrupt and occurs without widening of the channel. This morphology suggests these channels were formed by a higher yield strength fluid than the erosive channels, consistent with lava or a hyperconcentrated sediment flow. It is not clear, however, where the material to form C5 by sediment flow would have come from: fissure F1 is small in relation to the C5 deposits, and the eroded area around it incises C5 and L5 deposits, not the basement. For this reason, it is more probable that these flow deposits are lava. Extraction of topographic data from HiRISE anaglyphs may help quantify the volume of these units to resolve this issue.
 The morphology of the source vents for the two channel types is distinct: the eroded channels issue from deep linear fissures and the constructive channels from the shallower fissure F1, the broader linear fissure F7 and the broad depression F6. The morphology of F6 and F7 may be due to their location: they are at the interface between the ridge and the plains, while the other vents are nearer the crest of the ridge. The physical properties of the substrate here may have been different than at the locations of the deeper fissures. However, it is also possible that these differences in morphology relate to differences in the style of release or source depth of material issuing from these vents. This supports the conclusion that two distinct outflow processes occurred along this wrinkle ridge during the same time period.
 The orientation of the linear vents is similar to that of Cerberus Fossae. This suggests that the regional strain regime controlled their orientation. However, the location of the majority of the vents near the crest of the ridge arch suggests that the ridge also played a controlling role: fault planes associated with the ridge may have provided conduits for fluid transfer from depth.
 The Cerberus plains south of a wrinkle ridge which borders the southeast boundary of Athabasca Valles were sourced from vents in the ridge. The material to form the plains was supplied from these fissures, pits, and broad depressions through episodic flow in a complex series of incised and constructed channels.
 This study confirms that large-scale sheet flow units to the north of this wrinkle ridge have a source to the north of the study area, consistent with an association with Athabasca Valles and Cerberus Fossae. This deposit overlies channels sourced from the ridge vents and so postdates their period of activity.
 Another widespread sheet flow meets the ridge through flow from the east, indicating a further source of the Cerberus plains in that direction. This overlies some channels sourced from the wrinkle ridge but is incised by and has its margins dictated by the youngest activity from these vents. This indicates that this large-scale plains-forming deposit and ridge-sourced activity were contemporaneous in this region.
 Outflow activity from the vents was recent and may have occurred up to 11 Ma.
 The majority of the ridge-sourced channels are incised, with a minority forming constructive flows. This indicates that channel-forming fluid from the vents had two distinct viscosities and may be lava and water or high- and low-viscosity forms of lava or hydrous flow.
 Higher resolution topographic and compositional data for the region is required to resolve the question of the nature of these plains deposits. Whether lava or water was erupted by these vents, however, these landforms and deposits are evidence that Mars was geologically active very recently, opening the possibility that it is even now.
 This study was conducted as an undergraduate project at Birkbeck, University of London. The author would like to thank Gerald Roberts of Birkbeck, University of London for his invaluable help in formulating the scope and format of this paper and his comments on the work itself.