The Flood Lavas of Kasei Valles, Mars

Abstract

Both the northern and southern arms of Kasei Valles are occupied by platy-ridged flood lavas. We have mapped these flows and examined their morphology to better understand their emplacement. The lavas were emplaced as high-flux, turbulent flows (exceeding 10⁶ m³ s⁻¹). Lava in southern Kasei Valles can be traced back up onto the Tharsis rise, which is also the likely source of lavas in the northern arm. These eruptions were similar to, but somewhat smaller than, the Athabasca Valles flood lava in Elysium Planitia, with estimated volumes of >1200 km³ here and 5000 km³ in Athabasca Valles. The flood lavas in both Kasei and Athabasca Valles have evidence for distal inflation as well as widespread drainage or volume loss in medial areas; this may be an important characteristic of many large, recent Martian eruptions. Despite their great size and flux, the Kasei Valles flood lavas are only a late modification to the valley system capable of only modest local erosion. The more vigorous Athabasca Valles lava may have been capable of somewhat more erosion in its smaller valley system.

1. Introduction

Flood lavas are a fundamental part of the geologic history of Mars. They cover roughly half of the surface and make up much of the volume of the crust (Greeley and Schneid, 1991; McEwen et al., 1999; Keszthelyi and McEwen, 2007; Caudill et al., 2012). In places, layering in the central uplifts of craters has been used to infer stacks of hundreds or thousands of lava cooling units (Caudill et al., 2012), and the walls of Valles Marineris may expose an 8-km-thick stack of flood lava flows (McEwen et al., 1999). The volatiles released from flood basalt eruptions are likely to have had major climate effects (Plescia, 1993; Halevy and Head, 2014).

The best-preserved and best-documented Martian flood lava flow is the Athabasca Valles lava described by Jaeger et al. (2007; 2010), as summarized below. This flow is an example of “platy-ridged” lava (Keszthelyi et al., 2000; 2004; 2006; 2008), with a surface crust broken into detached plates. This lava type is thought to be equivalent to terrestrial rubbly pahoehoe (Keszthelyi et al., 2006). The flow coats the surface of the Athabasca Valles outflow channel, having mostly drained as the flow waned (Jaeger et al., 2007). Although deflation or drainage is apparent for several hundred km from the vent, distal edges of the flow and late breakouts show classic morphologies of inflated lava. The flow volume was ~5000 km³ and the flow traveled ~1400 km from the vent, making it comparable to some of the largest individual flows in terrestrial flood basalt provinces. Analytical modeling of the flow through the lava-coated channel requires that it was emplaced turbulently with a peak lava flux of ~10⁷ m³ s⁻¹, implying an eruption lasting weeks to months. However, this one example is insufficient to determine if this mode of emplacement is typical of Martian flood lavas.

Kasei Valles is another channel that has been occupied by young flood lavas, providing a good example to compare to the Athabasca Valles and broaden our perspective on flood lava emplacement on Mars and elsewhere. The floor of Kasei Valles has extensive well-preserved platy-ridged lava surfaces, mapped as Late Amazonian (c. 100–200 Ma) by Chapman et al. (2010b). Chapman et al. reported platy flows found over a 2100 km region from Echus Chasma to Sharonov crater. Dundas and Keszthelyi (2014) used several Digital Terrain Models (DTMs) to estimate lava fluxes of order 10⁶ m³ s⁻¹ in the southern arm of Kasei Valles. Some flux measurements in southern Kasei Valles approached 10⁷ m³ s⁻¹ but could have been due to catastrophic breaching of lava dams since fluxes calculated upstream of those locations were lower. This is similar to the “fill-and-spill” emplacement described by Hamilton et al. (2015) for various lavas on Mars and Earth. The lava in south Kasei Valles locally showed evidence for drainage, as well as erosion of the substrate. The behavior of this flow was broadly similar to that in Athabasca Valles: a high-flux lava flow that was at least locally turbulent.

In this paper, we examine lava in the Kasei Valles region (Fig. 1) in more detail, mapping the full extent of the exposed lava flow in the southern arm of Kasei Valles (south Kasei Valles; SKV) and comparing it with lava in north Kasei Valles (NKV) and Sacra Sulci. We also compare the flow behavior and morphology to the Athabasca Valles lava.

Figure 1.

Overview of the Kasei Valles region, with major features labeled. The basemap is a shaded-relief image of MOLA topography colorized by elevation, with low-lying areas in blue. Echus Chasma and Echus Palus are commonly treated as part of the Kasei Valles system, but in the formal International Astronomical Union definition (and as used herein) the name Kasei Valles refers only to the well-defined troughs. Numbered points indicate the location of figures.

2. Data

We used several data sets for this project. The most important was Mars Reconnaissance Orbiter (MRO) Context Camera (CTX; Malin et al., 2007) images (6 m/pix) covering the region of Kasei Valles and extending west to near Tharsis Tholus, which formed the primary data set for outlining the lava. There are some minor seams and offsets between the map-projected but uncontrolled images, but these positional errors are mostly less than a few hundred meters, insignificant on the scale of the lava flows in question. A THEMIS (Thermal Emission Imaging System; Christensen et al., 2004) daytime-IR controlled mosaic (100 m/pix) of the Lunae Planum region (Fergason et al., 2013) was used as a supplementary basemap. Images from the MRO High Resolution Imaging Science Experiment (HiRISE; McEwen et al., 2007) were used to examine detailed morphologies at 25 cm/pix in the scattered locations where they are available.

Topographic data were also used to understand the flow behavior. We used both gridded maps and individual shot points from the Mars Orbiter Laser Altimeter (Zuber et al., 1992), as well as Digital Terrain Models (DTMs) produced from CTX and HiRISE stereo image pairs. HiRISE stereo (typically 1 m/post) follows the approach of Kirk et al. (2008), and CTX methods are similar. Some CTX DTMs suffer from jitter due to un-modeled errors in the spacecraft pointing, which introduces long-wavelength undulations in the topography.

The Athabasca Valles lava flow was already mapped in detail by Jaeger et al. (2010), with flux calculations made near the vent. We supplement the previous work with additional, more distal flux calculations and observations from several HiRISE and CTX DTMs and images to better compare the lavas.

3. Methods

3.1. Mapping

We mapped the Kasei Valles lavas in ESRI ArcMap®. The primary mapping data set was CTX images, covering most of the region between ~270°–305° E and ~11°–28° N. (All coordinates herein are planetocentric latitude and east longitude.) We worked at a scale of 1:20,000 to enable us to trace flow margins with precision of <50 m, suitable for this flow, which ranges between 5 and 40 km across in most places. The shapefile is included as supplementary material.

From the SKV lava’s well-defined terminus near 26.2°N, 304°E (in northeastern Kasei Valles), we traced the flow’s margins west and southwest for roughly 2,300 km upstream to near 12°N, 272°E (in northeastern Tharsis). At this location, subsequently emplaced lavas have covered the flow’s uppermost reaches, thus preventing us from identifying a source vent. In most places, the flow margin is clearly discernable as an abrupt change in texture and/or elevation from the adjacent surfaces; this is particularly evident in the flow’s easternmost ~1500 km, where the pre-existing SKV channel floor was covered edge-to-edge by a continuous sheet of lava with distinct edges. Where not confined to a channel, the flow margin is often expressed as lobate extensions responding to local decameter-scale topography. We successfully traced the flow’s outer margin through its entire length, however its precise location was locally ambiguous in circumstances where CTX image quality is poor, at channel constrictions where lava depths were most variable and mechanical erosion may have occurred (Dundas and Keszthelyi, 2014), and where the flow has encountered similar-looking lavas that erupted from different sources at geologically near-concurrent times (judging from indistinguishable textures and preservation states). To aid with defining flow margins at these ambiguous locations, we examined a THEMIS Day-IR mosaic as well as HiRISE images, which helps delineate the contact between the flow’s edge and pre-existing terrain in most cases. After tracing the flow’s outer perimeter, we attended to the interior and traced-out the margins around all un-covered outcrops (kipukas) larger than ~100 m.

We repeated this approach with lava in NKV, beginning with a well-defined flow terminus near 26.1°N, 297.9°E. We traced this lava up-flow into the Sacra Sulci region. In this region (green in Fig. 2) it became difficult to demarcate the flow because of overlapping flow lobes. Lava with similar textures is present throughout Sacra Sulci, with multiple feeder channels from the west. It is likely that several of these are from the same eruption that fed the lavas in SKV, but it is possible that other similar age lavas have also entered the region. The current image coverage and quality make it difficult to reliably address these complexities. Consequently, we did not attempt to map this lava further towards its source. However, we discuss additional observations of this flow, as it provides an additional example of a large lava flow on Mars fed by a high-flux eruption.

Figure 2.

Mapped Kasei Valles lava flows. Lavas originated from the Tharsis province at left and flowed to the north-east. The South Kasei Valles (SKV) lava is outlined in yellow from the terminus near Rongxar crater to the emergence point near Tharsis Tholus. The southern margin of the lava in Echus Palus is partially buried but the buried flow area is considered unlikely to be extensive. Light blue extensions indicate underlying lava beyond well-defined flow fronts, likely due to a previous flow. The North Kasei Valles (NKV) lava is outlined in dark blue from the terminus to the Sacra Sulci region. Lava in Sacra Sulci (likely several flows, undifferentiated) is in green, with inflow paths shown schematically in dark red. The sources of the lava are not known, as they are buried (yellow) or not mapped all the way to the source. Flows further to the south, and higher on Tharsis, were not mapped. The basemap is a THEMIS daytime infrared mosaic.

3.2. Flux Estimates

We calculated lava fluxes at several locations using the same fluid flow equations used in several previous publications (Keszthelyi and Self, 1998; Keszthelyi et al., 2006; Jaeger et al., 2010; Dundas and Keszthelyi, 2014). The key measurements are determination of the lava depth and slope, and estimation of the lava properties. The methods for determining slope and cross-section follow Dundas and Keszthelyi (2014) with some minor updates. Briefly, we use a topographic profile across the flow (orthogonal to the direction of flow) to determine the hydraulic radius for input into the flow equations, and estimate the lava surface slope by using the elevation difference and along-flow distance of points upstream and downstream from the measured cross-sectional profile. Lava velocity and thus flux are determined with commonly used equations for turbulent flow through a channel (Shaw and Swanson, 1970; Keszthelyi and Self, 1998; Keszthelyi et al., 2006; Jaeger et al., 2010):

$$\langle v\rangle ={\left(gd\mathit{\text{sin}}(\theta )/C_f\right)}^{1/2}$$ (1)

$$C_f=(1/32){\left({\mathit{\text{log}}}_{10}{(6.15(2Re+800)/41)}^{0.92}\right)}^{-2}$$ (2)

$$Re=\rho d\langle v\rangle /\eta$$ (3)

where <v> is the mean velocity, C_f is a friction factor (Goncharov, 1964) which is determined recursively, and Re is the Reynolds number. In a few cases the estimated high-lava margins on each side of the flow are at slightly different elevation. This could be due to DTM errors (e.g., unrecognized jitter in CTX DTMs) or real physical effects (e.g., superelevation of lava). We remediate this by applying a slight tilt to the profile to give a level upper surface, which is approximately accurate (and does not artificially increase or reduce flow depth) regardless of whether the DTM is in error or there was actually a slight cross-flow slope to the lava surface. We used the geodesic profiling tool available at https://astrogeology.usgs.gov/facilities/mrctr/gis-tools (beta version downloaded July 6, 2017) for drawing profiles. (When applied to the profile locations from Dundas and Keszthelyi (2014), this tool results in some minor differences, but they are small compared with the other uncertainties and have no meaningful effect on the results of the computations that use these input.) Calculated flux values are given with two significant figures for relative comparisons, but realistically they should be regarded as order-of-magnitude estimates.

As is standard practice in hydrology we calculate fluxes at locations where the entire flow is concentrated into a well-defined single channel (i.e., at a constriction). These are also locations where the lava has largely drained, allowing the best estimates of the sub-lava topography. These locations also typically have steeper slopes, which reduces the error in flux estimates arising from small uncertainties in the lava surface slope. The velocities determined at such locations are expected to be higher than in other reaches of the flow. However, the fluxes at such locations should be typical, except under conditions where lava was pooling or abruptly draining.

The lava properties are not well known (composition is discussed in section 4.2), but variations across the full parameter range appropriate for mafic to ultramafic lavas did not affect conclusions about the behavior of lava in either Athabasca or Kasei Valles (Jaeger et al., 2010; Dundas and Keszthelyi, 2014). Reasonable variations lead to flux differences of a factor of a few, which are not sufficient to affect the interpretations here. For simplicity all calculations in this paper are made assuming a density of 2100 kg m⁻³ and a viscosity of 100 Pa s. Spatial and temporal variations in these parameters within a given lava flow should be small compared to the overall uncertainties in the parameter values. The density is similar to basaltic Martian meteorite parent magmas (McSween, 1994) with ~25 vol% vesicles. Basaltic lava in Gusev crater investigated by the Spirit rover had an estimated liquid viscosity of 2.3 Pa s, a factor of ~40 less than a representative terrestrial tholeiitic flood basalt (Greeley et al., 2005). McSween (1994) reviewed several studies of Martian basaltic meteorites that reported similar or slightly greater liquid viscosities. However, the effective viscosities would have been higher due to the effects of vesicles and suspended crystals. Greeley et al. (2005) calculated that the latter effect alone could have raised the effective viscosity of the Gusev crater lavas to 50 Pa s with 25% crystals. A value of 100 Pa s thus appears to be a reasonable estimate for effective viscosity given available information about Martian lavas, incorporating moderate stiffening from crystals and vesicles relative to liquid alone. It is at the fluid end of estimates used in terrestrial modeling, as values of 100–1000 Pa s have been used in previous studies of long terrestrial basaltic lava flows (e.g., Keszthelyi and Self, 1998; Keszthelyi et al., 2006). Morphometrically estimated viscosities for nearby Ascraeus Mons lavas are orders of magnitude higher (e.g., Zimbelman, 1985; Hiesinger et al., 2007), but those models rely on simplistic assumptions about lava flow behavior that are not valid for most real lava flows (Keszthelyi, 2012) and hence the numbers that they produce are not appropriate inputs for these fluid flow equations.

4. The Kasei Valles Flood Lavas

4.1. Overview

The SKV lava flow can be traced from the terminus near Rongxar crater, through Kasei Valles and Echus Palus, and up onto the Tharsis rise (Fig. 2). The mapped flow has a length of >2200 km and an area of ~59,000 km², making it among the longest well-mapped lava flows in the Solar System. (Canali on Venus reach up to 6,800 km and are probably lava channels but are not well understood (e.g., Baker et al., 1992)). It descended nearly five km in elevation, but much of this occurred at a series of cataracts and constrictions (Fig. 3). The flow vanishes under younger lavas near Tharsis Tholus.

Figure 3.

Longitudinal profile of lava in South Kasei Valles based on MOLA DTM. The profile approximates the main lava flow path. Small spikes or apparent uphill sections are due to either post-flow impacts or imperfections in the DTM, such as poorly resolved constrictions.

The SKV flow thickness is variable. MOLA shots along the flow edge in the Tharsis region suggest that relief of 20 m is typical. In Echus Palus the flow edge commonly rises 30–40 m above the adjacent lavas. These are typical values and both are variable along the flow margin and may understate the flow thickness if there have been subsequent lavas emplaced around the flow we are mapping. Within the SKV channel, the lava has commonly receded from the lava high stands, but continuation of the valley wall slopes from above the lava suggests flow thicknesses of many tens of meters near channel center. It is possible, however, that some of this depth was filled by previous lavas or sediments that are now buried, and the valley bottom may not ever have been as deep as simple extrapolations suggest. For a volume estimate we use a thickness estimate of 20 m, which yields a flow volume of ~1200 km³. This is a conservative thickness estimate and does not include any contribution from the unknown area of buried lava up-slope from the mapped flow, so this should be considered a lower bound. Given the uncertain vent location and variable thickness, we cannot accurately determine an upper bound, but it should not be more than a factor of a few higher.

We traced the NKV lava back from the terminus along the northern arm of Kasei Valles (Fig. 2) but had to stop our mapping in Sacra Sulci. The flow must have passed through that region, but it divides into multiple lobes in a region where several of the available CTX images have poor signal-to-noise ratios and other data are inadequate. The mapped flow length is thus over 700 km, but the true flow length is likely much greater, as there are no candidate vent structures nearby. Within the well-defined NKV channel, we observe no distinct flow fronts, indicating that it was occupied by a single large lava flow with behavior similar to that in SKV. The two are also morphologically similar, with rough surface textures suggestive of a rubbly cover.

4.2. Composition

Much of the lava in Kasei Valles, and particularly to the west on the Tharsis rise, is covered in dust that prevents spectral determination of the composition. Dust cover is reduced near the eastern ends of the flows. In this region, the Thermal Emission Spectrometer (TES) shows both “basalt” and “andesite” signatures in parameter maps (Bandfield et al., 2000), which we inspected in JMARS (Java Mission-planning and Analysis for Remote Sensing; Gorelick et al., 2003). The latter may represent weathered basalt (Wyatt and McSween, 2002; Kraft et al., 2003). The “andesite” signature has no obvious correlation with terrain or geomorphology; the “basalt” signature has some correlation with the valley bottom but does not uniquely correspond to the mapped lava. Compact Reconnaissance Imaging Spectrometer for Mars (CRISM) near-infrared spectral analyses are also frustrated by dust cover but have revealed the presence of mafic minerals in patches exposed by fresh craters in the Tharsis region from which the flow descends (Viviano-Beck et al., 2017). Similarly-textured lava in Echus Chasma also has a basaltic spectral signature (Mangold et al., 2010). Gamma Ray Spectrometer data from Tharsis are dominated by the mantling dust (Gasnault et al., 2010), and within the Kasei Valles region the flows are too narrow to affect the composition sensed at the very coarse resolution of the instrument. Thus, direct evidence for the composition of the Kasei Valles lavas is limited but consistent with a mafic composition. Given the generally basaltic nature of Mars (e.g., McSween et al., 2009), it is likely that the lavas in Kasei are some variety of basalt.

4.3. Sites of Interest

Here we discuss the behavior of lava at several key locations in the SKV flow and explain our interpretations at several of the most complex locations.

Echus Palus:

The lava forms a broad plain in the northern part of Echus Palus. Part of the southern margin of the SKV flow in Echus Palus is buried by a later flow. Since the buried and unburied parts of the margin there are roughly aligned, it is unlikely that a significant portion of the SKV flow is concealed. Furthermore, the lava would have to flow up against the regional slope to extend any significant distance to the south. We conclude that the volume of any buried lavas is trivial compared to the total volume of the SKV flow. At several locations, lava occurs beyond the well-defined flow fronts that we have mapped as the edge of the SKV flow (Figs. 2 (light blue) and 4). The additional lava could be due to a breakout or earlier stage of the SKV flow, or to a previous flow with similar extent, now mostly buried by the mapped flow. The sharp demarcation of the flow fronts suggests the latter, but if not, then the SKV flow was modestly more voluminous than the lower bound given here.

Figure 4.

Margin of the mapped SKV lava on the northern edge of Echus Palus (16.5°N, 283.8°E). Lava beyond the flow margin could be from an older flow, or a breakout or earlier stage of the SKV lava. (Mosaic of CTX images P06_003408_1977_XI_17N076W and B18_016752_1986_XN_18N076W. North is up and illumination from the left.)

South–north linkages:

An arm of SKV lava extending north of Echus Palus fills in a plain lying ~80 m above a lower-lying lava surface that connects with the northern lavas in Sacra Sulci. The two are linked by a small connection (17.2°N, 282.7°E) that descends a scarp (Fig. 5a). We have not been able to definitively identify the flow margin at this location but treat it as the end of the mappable SKV flow. This is because the lower-lying plain rises tens of meters in traversing to the north, which would require the lava to flow uphill. There is no evidence for a much-thicker flow that could have later drained, and this local low would have been a sink. Moreover, the narrow connection between the two surfaces cannot have carried a large lava flux. The most likely interpretations of this location are that a small amount of lava descended from the east (the SKV flow) and was then subsequently buried by lava from the north (Sacra Sulci), with an ambiguous contact because of mantling and/or the topographic complexity of the linkage, or that one of the candidate termini in Fig. 5 is the true flow edge.

Figure 5.

A) Contact between the SKV flow and lava from Sacra Sulci (17.2°N, 282.7°E). Black arrows show the flow of SKV lava from Echus Palus down the steep, grooved scarp. White arrows show possible locations for the contact between the two. Box shows location of Fig. 5b. B) Enlargement of the location of the southern white arrow from A. (Both panels are subsections of HiRISE image ESP_044404_1975. North is up and illumination from the left.)

A second connection to the north also has obscure contacts at the northern end (near 18.6°N, 283.1°E). This contact is also narrow and could not have carried a high lava flux, but threads its way through grooved terrain for roughly 70 km. This connection lies beyond one of the Echus Palus flow margins noted above and thus is likely from a previous flow rather than the mapped SKV lava.

South Kasei Valles cataract and constrictions:

The SKV lava passes over a cataract near 17.9°N, 285.8°E and subsequently passes through a series of four constrictions where wall material blocked the channel floor. Dundas and Keszthelyi (2014) showed evidence for lava erosion at the cataract and possibly the third constriction. Between the cataract and first constriction, Dundas and Keszthelyi (2014) noted evidence for substantial drainage (~200 m) and suggested that the lava could have eroded the cataract, leading to breaching of a dammed pond. A HiRISE DTM (DTEED_032193_1990_031837_1990_A01) supports this interpretation, demonstrating that the high-lava mark remains near-level all the way to a spur that might represent the pre-flow location of a cataract cut into the constriction. The volume of lava temporarily ponded here would have been several tens of km³, a small fraction of the total downstream lava volume.

The NKV lava also has several locations of interest:

• Crater and constriction: At 23.8°N, 287.1°E the NKV channel was largely blocked by a 12.5 km-diameter impact crater. Lava eventually flowed into and out of the crater through breaches in the rim, but it is not clear whether these were eroded by the lava or were pre-existing. Kilometer-scale slabs of material within the crater do not appear to have been covered by the lava, and may have been removed from the wall or floor of the crater. The valley passes through a narrow constriction about 25 km further downstream (Fig. S1), and the lava filled in the region between crater and constriction to significant depth before subsequently draining. The high-lava mark lies at least 60 m above the present lava surface. The (largely drained) constriction was used for a flux measurement.

• Cataract: The lava passes through a second cataract (fed by an inner channel) near 26°N, 289.4°E (Fig. S2). This provides a location for a second flux calculation since the lower reach of the inner channel feeding into the cataract has largely drained from the high-lava marks.

4.4. Morphologies

The typical surface of the Kasei Valles lavas is rough and ridged, with various indications of deformation such as internal shearing and separation into plates. The crust clearly translated in places, but we observed few wakes or other indications of crumpling against obstacles. One striking location is shown in Fig. 6a. At this location, the lava in NKV flowed past a pre-existing impact crater, as shown by the distinct lava margin embaying the rim. Unlike some similar locations in Cerberus Palus (filled by the Athabasca Valles lava), there is no upstream pile-up of broken crust or downstream wake of missing material, only some modest disruption of the plates moving within the main channel. Fig. 6b shows a location near the terminus of SKV where a crater obstructed flow and produced a long (~20 km) crumple zone of moderately roughened crust upstream, rather than a concentrated pile-up. This suggests that the crust crumpled and built ridges over a protracted period when it was only slightly mobile, allowing the affected zone to build outward. The crust texture is also different downstream of the crater in a region that was likely sheltered and less mobile than the main flow channel. It should be noted that the preserved crust is from the last stage of crust evolution and does not necessarily give a good direct indication of the behavior of the crust during the bulk of the lava’s emplacement.

Figure 6.

A) Lava in northern Kasei Valles flowing past a crater on the channel floor (27.1°N, 292.5°E). The rim of the crater is well-preserved, indicating that there was little erosion of the channel floor between formation of the crater and emplacement of the lava. B) Interactions of lava and a crater in southern Kasei Valles (25.3°N, 301.5°E). The presence of the crater influenced the formation of crust textures both upstream and down. (A: CTX image F04_037454_2072_XN_27N067W. B: CTX image D18_034078_2048_XN_24N058W. North is up and illumination from the left.)

To understand the spatial variation of processes within the lava, we inspected HiRISE images of the SKV downstream of the region of constrictions. In this reach the dynamics were quite simple, as the lava flowed along a relatively straight, uniform valley floor without major disruptions other than the cataract near 25°N, 298.5°E. This simplicity allowed us to assess the variations with distance. In the more proximal section from 288–296°E, the flow surface was generally a roughly uniform (at large scale) mix of ridges and fractures of various sizes, often separated into discrete plates that moved independently of each other (Fig. 7a). The plates grade from barely detached slabs that can clearly be linked to adjacent lava to highly crumpled rafts that appear totally independent of their surroundings. In the western (proximal) part of this reach, large-scale flow topography is imperceptible. In such cases the lava crust might have raised or lowered uniformly, but there are no clear morphological indicators of inflation or lava drainage. Further east the lava clearly lies somewhat below high-lava marks on the bank, suggesting deflation or drainage. Between 296–299°E, the flow has spatial patterns suggestive of localized inflation (Fig. 7b), except where it passed through the cataract noted above. The transition between morphologies appears gradational and we attribute the textural difference to local changes within the same flow, rather than a flow margin. It may have been caused by topographic effects on the flow, since the profile first flattens, and then steepens as it approaches the cataract. The amoeboid circular patterns are suggestive of inflation features but fine-scale diagnostic morphologies are lacking. From 299–303.5°E, the lava again forms a sheet of ridges and fractures (Fig. 7c) with some plates, and mesoscale topography that is generally imperceptible in image data. Discrete plates are less common than in the more proximal sections of the flow. Finally, in the last stretch of the flow from 303.5–304°E there are locally uplifted sections with spatial patterns indicating inflated lava, including one section where plates of the crust rafted apart, intervening material froze, and then the entire mass was uplifted (Fig. 7d). The inflated sections of the flow often have sharp edges, but lack visible dilation fractures like those normally found in tumuli or at the edge of lava-rise plateaus. This may be due to the presence of a rubble crust suppressing such fine-scale morphologies, or simply to degradation (e.g., infilling by dust). These variations along-flow are likely caused by two factors: changes in the local slope, and variations in the extent to which liquid lava drained from under a crust. For instance, inflated morphologies occur near the terminus where there was nowhere for liquid lava to drain away. The transition to lateral confinement within the main Kasei Valles channel also contributes to these factors.

Figure 7:

South Kasei Valles lava morphologies. See text for discussion. (All panels are subsections of HiRISE images with north up and illumination from the left. A: ESP_046052_2030, 22.6°N, 293.6°E. B: ESP_036056_2045, 24.4°N, 297.4°E. C: ESP_019850_2060, 25.5°N, 302.1°E. D: ESP_035568_2065, 26.2°N, 303.7°E.)

4.5. Lava Fluxes

4.5.1. South Kasei Valles

As a reference, we recalculated the flux at cross-section A-A´ of Dundas and Keszthelyi (2014) using the methods and lava fluid properties described above. The results are consistent with previous work, with a flux estimate of 6.7×10⁵ m³ s⁻¹ and Reynolds number of 2900, indicating turbulent flow. This demonstrates that the parameters used here do not lead to any relevant changes in the results of that work.

4.5.2. North Kasei Valles

The constriction and cataract in NKV are both largely drained, with distinct high-lava marks, and provide ideal locations for flux calculations. Using the parameters discussed in section 3.2, the flux in the constriction was 3.1×10⁶ m³ s⁻¹, and at the cataract 1.5×10⁶ m³ s⁻¹. This could indicate a lower peak flux downstream within the flow, but given the uncertainties, we consider these values effectively identical. In both locations the flux was clearly in the turbulent regime, with velocities exceeding 40 m s⁻¹ and Reynolds numbers above 30,000.

5. The Athabasca Valles Flood Lava

The Athabasca Valles lava has already been mapped and characterized by Jaeger et al. (2010). Here we discuss a few new observations for comparison with the Kasei Valles lavas.

We examined morphologies within the western arm of lava just north of Aeolis Planum. Here, as in SKV, the lava was confined to a long, straight trough for over four hundred kilometers. The present slope of this reach is essentially zero: local variations in the lava surface topography (due to terrain draping, etc.) are larger than the change over the entire baseline. In the more proximal region from 140–146°E, the lava appears to drape the surface and the margin is thin and often indistinct in relief. A DTM in this region (around 1000 km from the vent) shows that the high-lava mark is up to 20 m above the nearby lava surface (Fig. 8). This zone terminates at a breached low ridge near 4.5°N, 139.8°E (Fig. 9a). Past this point the lava spreads onto a broader terminal plain and the margin regularly shows classic inflation features (primarily an inflated sheet flow) and is fringed by small breakouts, although the lava surface did eventually lower enough to show the position of buried crater rims. High lava marks just upstream from the ridge are discontinuous, suggesting failure of wall material, which very likely occurred while the flow was liquid because there are no corresponding deposits on the flow surface (Fig. 9b). This suggests that the lava caused failure of slopes adjacent to its margins, strengthening the possibility that the ridge was breached or the gap enlarged by erosive effects of the lava. Additionally, the previously noted DTM shows a curious ridge paralleling the lava margin in places (Fig. 8). This resembles the edge of an inflation plateau but with a depressed interior; this may be an inflation plateau that subsequently collapsed when a major breakout downstream (such as due to ridge erosion) led to drainage of the liquid core. This suggests that initial emplacement of lava in this reach was relatively quiescent.

Figure 8:

Distal part of the Athabasca Valles lava (4.1°N, 141.2°E), colorized by elevation. The lava covered the surface but subsequently drained or deflated, producing ~20 m of surface relief in the lava-coated terrain. (Orthorectified HiRISE image PSP_008015_1840; north is up and illumination from the left.)

Figure 9:

Location where the Athabasca Valles lava passed through a breached ridge and formed its most distal western arm (4.5°N, 139.8°E). Flow was from right to left. A) Broad view of the entire breach region. B) Detail of the high-lava mark along the northern edge. (Both panels are subsections of HiRISE image ESP_013830_1845. North is up and illumination from the left.)

We made flux calculations for distal parts of the Athabasca Valles lava, at a constriction near the beginning of Lethe Vallis (Fig. S3; 3.8°N, 153.8°E) and a cataract in the eastern part of the flow (Fig. S4; 2.1°N, 157.3°E). These locations are both far from the vent, and unlike Kasei Valles the lava was not strongly channelized. Thus, the lava spread out over a broad region and these locations saw only small parts of the flow go past. (For reference, the fluid properties used here, combined with slope, depth, and cross-section estimates from Jaeger et al. (2010), would give a near-vent lava flux of 8.7×10⁶ m³ s⁻¹ and Reynolds number of 5100, consistent with their results.) Lethe Vallis nonetheless saw a high flux of 2.3×10⁵ m³ s⁻¹ and a Reynolds number of 1100 as lava spilled eastward out of Cerberus Palus. This indicates that this channel experienced vigorous and likely marginally turbulent flow. The eastern cataract experienced a flow that was still more diminished (9.3×10³ m³ s⁻¹), but even here the Reynolds number of 680 places the flow in the initial stages of turbulence. Incomplete lava drainage may be more important at this location than others because of the relatively small scale of the channel; this could moderately increase the flux and Reynolds number.

6. Discussion

6.1. Lava or Mud?

Williams and Malin (2004) suggested that the flow material that we have mapped within Kasei Valles could have been a mudflow deriving from the waning stages of flood waters passing through the valley. We consider this interpretation unlikely for several reasons. The platy-ridged material has a strong morphological resemblance to terrestrial rubbly pahoehoe lava flows (e.g., Keszthelyi et al., 2000; 2004). The SKV flow descended from the Tharsis rise, which is generally considered to be volcanic (e.g., Tanaka et al., 2014), and there are not fluviatile features in the vicinity of the flow before it enters Echus Palus. Additionally, there is evidence for a significant time separation between the emplacement of the rough flow material and the last major valley incision: the flow embays or buries a number of craters on the valley floor. If this viscous flow was the waning stage of a significantly larger aqueous flood, the peak stage should have been erosive enough to overtop and modify or destroy those craters. In particular, the rims of craters like that in Fig. 6 should have been heavily modified by any flood much larger and deeper than the mapped flow. In addition, the flow margins are typically quite sharp and show no evidence of grading into modifications of a larger flooded area. Similar issues regarding the identification of lava, mudflows, and aqueous floods have also recently been addressed for Hrad Vallis on Mars, where a lahar (mudflow) interpretation is also not favored (Hamilton et al., 2018).

6.2. The Kasei Valles Flood Lavas

The overall history inferred for the large flood lavas in Kasei Valles is as follows: lava in both NVK and SKV erupted on the Tharsis rise and flowed west or northwest to reach the valley system. The well-defined SKV flow (yellow in Fig. 2) covered northern Echus Palus and then spilled into the SKV channel. Both this lava and a previous, underlying flow have several narrow connections to the Sacra Sulci region, but none of these carried a large flux; instead, flows in Sacra Sulci (green in Fig. 2) and NKV (blue) descended from the west. Where they can be estimated, fluxes in both NVK and SKV were high.

The SKV lava is buried by younger flows near Tharsis Tholus. This has two implications. First, we were unable to trace the SKV lava to its vent, so the length and volume given here are lower limits. The trends of adjacent flows ultimately lead to a region south of Ascraeus Mons (source region for the flow mapped by Garry et al. (2007)), but it is not possible to say whether the vent was there or nearer to Tharsis Tholus. Second, despite the rather fresh appearance of the lava in Kasei Valles, some number of Tharsis lava flows are younger. The NKV flow has not been mapped as thoroughly as that in SKV, but they appear similar in age and emplacement style.

The mapped lava flows form a minor, late veneer in the overall history of Kasei Valles. There cannot be a thick series of lava flows within the Kasei Valles channels, because the lavas have a small effect on the overall topography. This implies that only a few flows from Tharsis (perhaps only one each in the northern and southern arms) have been able to reach so far, suggesting that these were somewhat atypically large eruptions (although the Early Hesperian Lunae Planum surface around Kasei Valles may also be volcanic (Chapman et al., 2010a; Tanaka et al., 2014)). An alternative possibility is that a water flood eroded previous lava infill within Kasei Valles; if this is the case then there could have been more lava flows extending through the valley. This is less favored because at several locations the lavas in Kasei Valles cover or were deflected by crater rims, implying that there was little erosion or infill during some significant interval between the main valley incision and the emplacement of recent lava flows. A combination of these effects is possible; what can be stated definitively is that there have been no more than a handful of lava flows reaching so far through the valley system since the last major incision of the Kasei Valles channels. (Chapman et al. (2010b) estimate the age of older channel floor material (their unit Ach) to be ~1 Ga.) It is notable that the large lava flows that we mapped are minor modifications to Kasei Valles, and cover only small fractions of the grooved, incised surface around Sacra Sulci and Echus Palus. (For comparison, Robinson and Tanaka (1990) estimated a water flux of 10⁹ m³ s⁻¹ in NKV.) This suggests that the flows that eroded the valley system (whether water, lava, or mud) were orders of magnitude larger than the lavas we discuss here.

Chapman et al. (2010b) gave crater-count ages for lavas in Kasei Valles but mapped based on texture and using somewhat lower-resolution imagery, resulting in some map units that differ from the flow mapping herein. However, sections of their platy flow unit Apf correspond to the distal part of the Kasei Valles flows. Their age estimates for the sections of Apf unit in SKV were 114–155 Ma. The mapping herein indicates a single flow in SKV, implying that the apparent age differences are simply due to target property effects (Dundas et al., 2010), secondary craters (McEwen and Bierhaus, 2006), or other sources of error. Regardless of uncertainties in crater ages, these are clearly young flows relative to the geologic history of Mars. Chapman et al. (2010b) noted that their age estimates for Apf were roughly consistent with radiometric age estimates for the youngest shergottite meteorites (cf. Nyquist et al., 2001), and there are several craters >1 km diameter on the mapped SKV flow that could have ejected material. The best candidates are located at 19.8°N, 286.8°E, 14.4°N, 282.2°E, and 14.6°N, 284.2°E. However, definitive connections between Martian meteorites and particular craters or regions remain challenging (e.g., Hamilton et al., 2003; Tornabene et al., 2006; Ody et al., 2015), and while a connection is plausible there is no evidence that the SKV lava is any more likely than other young flows to be a shergottite source.

Chapman et al. (2010b) argued that platy-ridged flow material within Kasei Valles was erupted from Echus Chasma. Mangold et al. (2010) and Leverington (2018) also considered Echus Chasma a possible volcanic source region, which would indicate a major volcanic source in Valles Marineris and support igneous origins for other features there. Our mapping supports earlier interpretations (summarized by Coleman and Baker, 2009) that the lava in Kasei Valles descended from the Tharsis rise. All workers agree that much of the latest lava fill in Echus Palus arose on Tharsis. All of this lava infill (possibly >200 m thick in Echus Palus since it lies above the floor of Echus Chasma, although some subsidence process could have occurred there) likely postdates the last major erosional event in the Kasei/Echus system. The present floor of Echus Chasma is at lower elevation than either the SKV lava in northern Echus Palus or the erosional grooves in Sacra Sulci. These relationships are most consistent with the current surface of Echus Chasma being the distal terminus of one or more lava flows from Tharsis, not a vent. There is some evidence, such as fissures and fractured, domed structures in the lava, that the Echus Chasma/Echus Palus region has experienced unusual volcanic processes (e.g., Chapman et al., 2010b; Leverington, 2018); however, these do not necessarily require that it was a vent. An alternative possibility is that these are massive lava inflation structures or other non-vent deformation. Similar structures have been described in the Elysium region (Keszthelyi et al., 2008).

The Kasei Valles lava flows are a useful reference location for studies of Martian stratigraphy and the geologic timeline. They represent an accessible surface partway through the Tharsis lava pile (although in the uppermost section) with relatively little dust cover and postdate outflow channel formation. Safe landing and in situ exploration on such a rough surface would be challenging, but the combination of Late Amazonian volcanism (including, via stratigraphic analysis, information about the eruption frequency and magmatic flux from subsequent flows), Hesperian outflow channel formation, and ancient channel wall and floor materials that were once several km deep in the crust means that the region could be used to understand several important processes.

6.3. Comparison with the Athabasca Valles Flood Lava

The gross dimensions of the SKV lava are on the same order as the Athabasca Valles flow. The volume is lower by a factor of ~4 or less (since the volume estimate above is conservative), but the SKV lava is considerably longer since it was tightly confined over much of its length. Unlike the Athabasca Valles lava, the Kasei Valles lavas are postdated by shorter flows. This suggests that the Athabasca Valles lava is not an extreme outlier, but it is at the large end of the size distribution of recent Martian eruptions. It is intriguing that it has not been superposed by any subsequent smaller flows, but this lends further support to the interpretation that the Athabasca Valles lava is extremely young. Further study of more typical sized flows may reveal regional differences between the Tharsis and Elysium volcanic provinces but they produced very similar very long lava flows.

Inflation appears to have been a characteristic of the terminal regions of both the Kasei and Athabasca flood lavas. However, identifiable local inflation features such as tumuli and lava rise plateaus are characteristic of only the distal reaches (<10% of the flow length) and internal breakouts. (The inflation midway along the SKV flow is likely the latter because the contact is gradational in places, but in other cases the inflated material appears to superpose the platy-ridged surface.) For most of the length of the flows, the lava crust behaved as a uniform sheet detached from the ground, and in many cases that sheet was ultimately draped upon underlying topography as the lava flowed away and/or lost volume due to gas escape and phase change to solid rock. This suggests that deflation or drainage are important and widespread processes for these high-flux turbulent flows. These characteristics may help to understand the behavior of less-well-preserved flows: the deflation and drainage are characteristic of large parts of the flow, and so when indications of this are observed, it is suggestive of this eruption style. Small-scale inflation features are concentrated in distal areas and local breakouts.

Athabasca Valles is a small outflow channel (~160 km³ eroded volume; Hanna and Phillips, 2006), and is entirely within the drained proximal region of the infilling lava flow, which had a volume 1.5 orders of magnitude larger than the eroded material and extended 4× the length of the valley. This leaves open the possibility that lava was the major erosive fluid in that system, perhaps via several flows; further investigation is needed (cf. Leverington, 2011; Hurwitz and Head, 2012; Keszthelyi et al., 2017). It is intriguing that lava flow Reynolds numbers near distal erosional features were near the transition to turbulence. Both are close enough to the transition that a higher viscosity might place them in the laminar regime, but the possibility of turbulent flow (more effective at erosion) is consistent with erosion not only of the main Athabasca channel but also the lesser distributaries and distal features. In contrast, the eroded volume of the Kasei Valles system is approximately 7×10⁵ km³ (Carr and Head, 2015), and it is more than 2000 km long. The mapped Kasei Valles lava flows are orders of magnitude smaller than their channel and had a net constructive (infilling) effect, although they also demonstrate local erosion at cataracts (Dundas and Keszthelyi, 2014) and it is possible that bed erosion occurred beneath the lava infill. Formation of Kasei Valles or other large outflow channels by lava erosion (Leverington, 2011; 2018) would require lava flows much larger in volume and flux than any well-mapped flows on Mars, as the well-documented lava erosion in Kasei Valles (and the possible erosion in Athabasca Valles) is on a much smaller scale.

Leverington (2018) suggested a simple reference model of erosion of Kasei Valles by 200 flows, each lasting five days with a volume of 25,000 km³, sourced from Echus Chasma. This is only a factor of a few larger than the Athabasca Valles lava, but does imply a mean flux of 5.8×10⁷ m³ s⁻¹, nearly an order of magnitude higher than the peak flow in Athabasca; this is a substantial extrapolation but not impossible. Very high lava fluxes have been suggested elsewhere in the Solar System (e.g., Baker et al., 1997; Byrne et al., 2013), although the data does not yet exist to study these systems at the level of individual flows, and the highest fluxes are part of a wide parameter space. However, the estimated erosion rate of 1.5 m/day used by Leverington (2018) is based on slopes significantly steeper than those along most of the length of Kasei Valles, since most of the descent occurs at cataracts (Fig. 3). As noted above, the present lava-covered floor of Echus Chasma lies below most of Echus Palus and the first expressions of young-lava-free grooved terrain in Sacra Sulci, so the slope over this reach would have been essentially zero in the later stages of incision, but even in other sections the channel floor slope is <0.05°. Moreover, this reference model does not account for infill by smaller flows or the waning stages of the hypothesized large flows, but the lavas we have mapped resulted in tens of meters of net construction in typical sections of Kasei Valles. Such events could more than cancel out the 7.5 meters eroded per large flow envisioned in this reference scenario. These factors suggest that lava erosion would have required substantially larger flows than this reference scenario, and/or less viscous flows that drained through the system more effectively. However, as discussed above, the mapped flows in Kasei Valles were likely the largest and longest in the valley system in the last ~1 Ga. It is possible that lava flows early in Mars’ history were more erosive, being erupted in a more active era when lavas may have been less viscous; however, it is also possible that maximum lava fluxes have increased over time as Mars’ lithosphere thickened (Keszthelyi et al., 2014). Fluid komatiite lavas on Earth are thought to have been erosive in some conditions (e.g., Williams et al., 2011).

These difficulties do not definitively rule out the possibility that Kasei Valles was formed by erosive lava flows (particularly as lava erosion processes are incompletely understood; see Dundas and Keszthelyi, 2014), but they point to challenges in reconciling such a hypothesis with known lava flows on Mars. Moreover, aqueous interpretations have been proposed for near-surface sediments (Salvatore and Christensen, 2014) and unusual landforms (e.g., Farrand et al., 2005; Moscardelli et al., 2012; Komatsu et al., 2016) in the Chryse-Acidalia region, in addition to many features of the Kasei Valles system (e.g., Baker and Kochel, 1979). However, there is evidence for ~1–2 km of mafic material in the northern plains (Pan et al., 2017), and a secular change in magmatism is possible. The sources of this material and whether it was emplaced as erosive flows are not certain.

7. Conclusions

Kasei Valles is occupied by some of the longest lava flows ever mapped in the Solar System and some of the youngest and best-preserved lavas on Mars. In addition to their own volcanological interest, these flows provide an important comparison with large lava flows elsewhere on Mars. In both Athabasca and Kasei Valles, there is evidence for high-flux, turbulent lava flows, with signs of inflation confined to the most distal portions. Deflation or drainage was important in both flows and is likely an important trait of high-flux, high-volume eruptions. There is also evidence for local erosion by the lava, and flux estimates are consistent with local turbulence in constrictions far from the vent, which may have contributed to local erosion of the substrate. The Kasei Valles lavas emanated from Tharsis rather than Echus Chasma and cannot be the waning stages of the flow that eroded the main channels of Kasei Valles. The well-preserved lava is merely a late veneer on the floor of Kasei Valles, incapable of erosion on the scale of the channel system.