Journal of Geophysical Research: Planets

Constraints on the history of open-basin lakes on Mars from the composition and timing of volcanic resurfacing

Authors


Abstract

[1] Abundant evidence exists for valley network-related fluvial activity near the Noachian-Hesperian transition on Mars, and areally significant quantities of volcanic ridged plains were emplaced during this period as well. Thus, it is worthwhile to explore the hypothesis that lava-water interaction occurred on the surface of Mars at this time. We analyzed the morphology, physical properties, composition, and surface ages of thirty open-basin lakes (topographic lows with both an inlet valley and an outlet valley) that were also resurfaced by volcanic flows. Hyperspectral imaging data from the Compact Reconnaissance Imaging Spectrometer for Mars (CRISM) and Observatoire pour la Minéralogie, l'Eau, les Glaces et l'Activité (OMEGA) instruments indicate that of the 30 basins, 12 exhibit the spectral properties of basaltic lithologies with diagnostic absorptions of olivine, high-calcium pyroxene and low-calcium pyroxene. An olivine/high-calcium pyroxene mixture is the most commonly identified mineral assemblage, consistent with other Hesperian-aged volcanic units. Therefore, our mineralogical results for over a third of the open-basin lake floors analyzed support prior interpretations that the basins were resurfaced by volcanic flows. Supporting evidence for resurfacing by volcanic flows is also given by the observed morphology and physical properties (e.g., surface roughness, thermal inertia) of the resurfaced open-basin lake floors. In these 30 examples, however, no evidence was found for lava-water interaction. The ages of emplacement, derived through counts of superposed craters, of all 30 of the open-basin lake volcanic resurfacing units show that resurfacing began in the Late Noachian, near the Noachian-Hesperian boundary, was concentrated in the Hesperian, and continued into the Early Amazonian. The lack of geologic features indicative of lava-water interaction suggests that the basins were likely to have been mostly devoid of water at the time of the latest phase of volcanic resurfacing. We conclude that there is no geologic evidence that suggests the fluvial activity associated with the studied paleolakes was coeval with the emplacement of the observed volcanic resurfacing units, several of which date to the Late Noachian-Early Hesperian.

1. Introduction

[2] Extensive volcanic plains units are one of the oldest known and best-documented features of the Martian surface [e.g.,Greeley and Spudis, 1981; Scott and Tanaka, 1986; Greeley and Guest, 1987]. These volcanic units are typically Late Noachian to Hesperian in age, and are thought to have resurfaced at least ∼30% of the Martian surface during this time period [Scott and Tanaka, 1986; Greeley and Guest, 1987; Head et al., 2002]. The emplacement of volcanic plains in the Hesperian directly follows a period earlier in Martian history when fluvial activity was common, recorded by distinct morphologies such as large valley networks and paleolake basins [e.g., Pieri, 1980; Goldspiel and Squyres, 1991; Cabrol and Grin, 1999, 2001; Howard et al., 2005; Irwin et al., 2005; Fassett and Head, 2008a, 2008b]. It is thought that much of the fluvial activity that created such features ended near the Noachian-Hesperian boundary [Irwin et al., 2005; Fassett and Head, 2008b; Hoke and Hynek, 2009; Mangold et al., 2012], and so preserved evidence of interaction between fluvial and volcanic activity at this critical junction in Martian history is an intriguing possibility.

[3] Paleolake basins provide a topographic depression that lava may have ponded in, and many paleolake basins are contained within ancient impact craters [e.g., Goldspiel and Squyres, 1991; Cabrol and Grin, 1999, 2001; Irwin et al., 2005; Fassett and Head, 2008a; Hauber et al., 2009]. Impact craters can enable surface volcanic eruptions through reductions in the crustal thickness, which may be further aided by deeply fractured zones under impact basins [e.g., Pike, 1971; Schultz, 1976, 1978; Head and Wilson, 1992]. Open-basin lakes, defined as hydrological basins with both an inlet valley and an outlet valley, require that water must have ponded in the basin to at least the topographic level of the outlet valley head before overflowing [e.g.,Cabrol and Grin, 1999; Fassett and Head, 2008a].

[4] In situ and orbital observations show that volcanic resurfacing has affected many open-basin lakes [Goldspiel and Squyres, 1991; Squyres et al., 2004; Fassett and Head, 2008a; Goudge et al., 2012]. A recent study of the morphology of a catalog of 226 open-basin lakes has shown that a variety of geologic resurfacing and modifying processes have played an important role in the post-lacustrine activity history of open-basin lakes on Mars [Goudge et al., 2012]. From this study, it was shown that the emplacement of volcanic plains units in particular was one of the dominant geologic processes responsible for post-lacustrine activity resurfacing, with 96 of the 226 (∼42%) open-basin lakes examined classified as volcanically resurfaced based on a distinctive morphology of the basin interiors [Goudge et al., 2012]. In this investigation, we present a detailed study of 30 open-basin lakes classified as volcanically resurfaced byGoudge et al. [2012], to further assess their morphology, physical surface properties (e.g., thermal inertia), composition and age of emplacement. The 30 basins chosen for this analysis are broadly geographically distributed across the Martian surface (Figure 1 and Table 1), and have relatively large surface areas, allowing for improved crater counting statistics compared to open-basin lakes with smaller floor areas.

Figure 1.

Distribution of all mapped open-basin lakes fromFassett and Head [2008a] and Goudge et al. [2012](yellow symbols), and the subset of 30 resurfaced open-basin lakes analyzed in this work (black symbols). Background is MOLA topography overlain on MOLA hillshade [Smith et al., 2001].

Table 1. Locations of Analyzed Resurfaced Open-Basin Lakesa
Basin #Longitude (E)Latitude (N)References
  • a

    Table lists basin number (assigned in this study), basin location (north positive latitude and east positive longitude) and appropriate references.

1−174.86−14.63Forsythe and Zimbelman [1995] and Goudge et al. [2012]
2152.75−11.53Irwin et al. [2007] and Goudge et al. [2012]
360.9421.10Fassett and Head [2008a] and Goudge et al. [2012]
4−12.32−21.67Goldspiel and Squyres [1991] and Goudge et al. [2012]
5−8.5925.57Fassett and Head [2008a] and Goudge et al. [2012]
6−7.21−8.84Fassett and Head [2008a] and Goudge et al. [2012]
72.77−10.63Fassett and Head [2008a] and Goudge et al. [2012]
884.96−2.42Fassett and Head [2008a] and Goudge et al. [2012]
989.71−0.09Cabrol and Grin [1999, 2001] and Goudge et al. [2012]
10110.86−2.71Cabrol and Grin [1999, 2001] and Goudge et al. [2012]
11−23.53−23.12Fassett and Head [2008a] and Goudge et al. [2012]
12102.25−3.42Fassett and Head [2008a] and Goudge et al. [2012]
133.88−27.90Fassett and Head [2008a] and Goudge et al. [2012]
1431.7224.39Fassett and Head [2008a] and Goudge et al. [2012]
1533.5716.72Fassett and Head [2008a] and Goudge et al. [2012]
1677.7018.38Fassett and Head [2005] and Goudge et al. [2012]
17175.39−14.40Grin and Cabrol [1997], Cabrol and Grin [1999, 2001], Squyres et al. [2004], and Goudge et al. [2012]
18135.38−6.91Fassett and Head [2008a] and Goudge et al. [2012]
19134.92−9.45Cabrol and Grin [1999, 2001] and Goudge et al. [2012]
20−165.61−10.22Cabrol and Grin [1999, 2001] and Goudge et al. [2012]
2162.17−2.56Fassett and Head [2008a] and Goudge et al. [2012]
22−18.29−26.87Fassett and Head [2008a] and Goudge et al. [2012]
23171.33−17.44Cabrol and Grin [1999, 2001] and Goudge et al. [2012]
24−20.51−22.45Goldspiel and Squyres [1991] and Goudge et al. [2012]
25−100.69−38.57Mangold and Ansan [2006] and Goudge et al. [2012]
2628.76−0.03Cabrol and Grin [1999, 2001] and Goudge et al. [2012]
27−11.15−26.81Fassett and Head [2008a] and Goudge et al. [2012]
28−5.25−21.26Irwin et al. [2007] and Goudge et al. [2012]
29158.54−22.89Fassett and Head [2008a] and Goudge et al. [2012]
3039.745.16Fassett and Head [2008a] and Goudge et al. [2012]

[5] The major goals in this study are: (1) to document evidence of volcanic resurfacing of these open-basin lakes using multiple remotely sensed data sets, (2) to help further understand the timing and emplacement of the resurfacing units within paleolake basins, and (3) to assess the possibility of lava-water interaction at these sites. These analyses are designed to provide further insight into the history of open-basin lakes on Mars, and help to illuminate aspects of the hydrological cycle during the Noachian and Hesperian periods.

2. Data Used and Methods

[6] The basin and resurfacing unit morphology, topography, surface roughness, thermal inertia, bolometric albedo, composition, and age of emplacement were assessed for 30 open-basin lakes identified as volcanically resurfaced byGoudge et al. [2012]based on their morphology. The suite of data sets were all geographically referenced and analyzed in ESRI's ArcMap Geographic Information System (GIS) software, which allows for on-the-fly co-registration and manipulation of diverse data sets.

2.1. Assessment of Basin Resurfacing Unit Morphology

[7] The morphologies of the open-basin lakes and their associated resurfacing units were assessed using a combination of ∼6 m/pixel images from the Context Camera (CTX) instrument aboard the Mars Reconnaissance Orbiter (MRO) spacecraft [Malin et al., 2007], <50 m/pixel images from the High Resolution Stereo Camera (HRSC) instrument aboard the Mars Express (MEx) spacecraft [Neukum et al., 2004], and the 100 m/pixel global mosaic of daytime infrared (IR) images from the Thermal Emission Imaging System (THEMIS) instrument aboard the Mars Odyssey spacecraft [Christensen et al., 2004]. Detailed morphology was also investigated using ∼0.25 m/pixel images from the High Resolution Imaging Science Experiment (HiRISE) instrument aboard the MRO spacecraft [McEwen et al., 2007].

[8] The assessment of morphology focused on the recognition of distinctive characteristics indicative of volcanic resurfacing (Figure 2), such as lobate margins and embayment relationships, as well as wrinkle ridges, which are commonly found in Martian volcanic units [e.g., Greeley and Spudis, 1981; Scott and Tanaka, 1986; Greeley and Guest, 1987; Watters, 1991; Head et al., 2002], but are not necessarily diagnostic of volcanic resurfacing alone.

Figure 2.

Open-basin lake at −11.53°N, 152.75°E (Basin 2 inTable 1) [Irwin et al., 2007; Goudge et al., 2012] that displays morphologic characteristics suggesting it has been volcanically resurfaced. North is up in both images. (a) Overview of volcanically resurfaced open-basin lake. Basin outline, as defined by a MOLA topographic contour [Fassett and Head, 2008a], is shown in blue. White arrows indicate prominent wrinkle ridges. Note the smooth surface texture and high crater retention. Location of Figure 2b is indicated by white box. Image is a mosaic of HRSC nadir image h8425_0000 and CTX images B20_017403_1658_XN_14S206W and P07_003703_1682_XN_11S206W overlain on the THEMIS daytime IR global mosaic [Christensen et al., 2004]. (b) Basin perimeter embayment in the southwestern portion of the basin shown in Figure 2a. The smooth plains unit contained in the basin is clearly embaying the rougher, older terrain exterior to the basin. Image is HRSC nadir image h8425_0000.

2.1.1. Morphologies Indicative of Lava-Water Interaction

[9] The image data described above were also examined closely for features that might indicate lava-water interaction [e.g.,Head and Wilson, 2002, 2007; Wilson and Head, 2007; Wilson et al., 2012], with focus on: (1) lava deltas, which are formed by lava flowing into standing water [Moore et al., 1973; Mattox and Mangan, 1997; Skilling, 2002], (2) littoral cones, which form due to explosive eruptions of steam as lava flows into a standing body of water [Moore and Ault, 1965; Fisher, 1968; Jurado-Chichay et al., 1996], (3) rootless cones, which form as lava flows onto water-saturated sediment [Thorarinsson, 1953; Greeley and Fagents, 2001], (4) tuyas, which form as lava erupts under a glacial ice sheet [Mathews, 1947; Jones, 1968; Smellie, 2007; Jakobsson and Gudmundsson, 2008], and (5) maar craters, which form due to phreatomagmatic eruptions from the interaction of rising magma and groundwater or surface water [Lorenz, 1973, 1986; Gutmann, 1976; White, 1989]. All of these feature classes should be large enough to be resolved in the images examined for this study.

[10] Lava deltas are formed as erupting lava flows into standing water, typically the ocean, which causes the rapid cooling of the lava and the creation of a depositional form that is roughly similar in morphology to a lacustrine delta or alluvial fan [Moore et al., 1973; Mattox and Mangan, 1997]; however, the morphology of many lava deltas is influenced by the fact that they are commonly fed by subsurface lava tubes as opposed to incised channels on the topset of the delta, allowing them to be distinguished from deltas deposited through fluvial activity [Mattox and Mangan, 1997; Skilling, 2002; Leeder, 2011]. Additionally, lava deltas are prone to the failure or collapse of the delta toe, making them even more distinguishable from fluvial deltas [Mattox and Mangan, 1997; Skilling, 2002].

[11] Another possible geologic consequence of lava flowing into standing water in the open-basin lakes is the formation of littoral cones. In addition to the creation of a lava delta [Moore et al., 1973; Mattox and Mangan, 1997], such a scenario can result in the rapid heating of the surrounding water. If the water surrounding the lava flow is vaporized quickly enough, the explosive expansion of the water vapor will fragment the lava flow, forming a cone-shaped deposit composed of pyroclastic material [Moore and Ault, 1965; Fisher, 1968; Jurado-Chichay et al., 1996]. Littoral cones can be formed either by lava directly entering a standing body of water [e.g., Moore and Ault, 1965; Fisher, 1968], or through the growth of the cone structure fed by lava-tubes [e.g.,Jurado-Chichay et al., 1996]. Littoral cones are large constructional features that are found on Earth in the nearshore regions of islands with active volcanism, particularly Hawai'i, and can also occur in association with lava deltas [Moore and Ault, 1965; Fisher, 1968; Jurado-Chichay et al., 1996]. Littoral cones on Earth are typically hundreds of meters in diameter, and tens to up to ∼100 m in height [Moore and Ault, 1965; Fisher, 1968; Jurado-Chichay et al., 1996].

[12] If the basin instead contained only small amounts of standing water or water in the pore space of saturated or partially saturated sediment at the time of volcanic resurfacing, rootless cones or pseudocraters might have formed in the volcanic unit. As lava flows onto water-saturated sediment, the excess heat from the lava will cause the pore water to vaporize, and upon reaching a pressure equal to the tensile strength of the cooling lava, the water vapor will explode through the cooling lava unit, creating a rootless cone or pseudocrater [Thorarinsson, 1953; Greeley and Fagents, 2001]. Rootless cones are well studied on Earth, especially in Iceland [e.g., Thorarinsson, 1953; Greeley and Fagents, 2001; Hamilton et al., 2010a, 2010b], and features similar to these have been inferred on the surface of Mars, ranging in size from ∼30–1000 m in diameter and >25–60 m in height [Frey et al., 1979; Greeley and Fagents, 2001; Lanagan et al., 2001; Fagents et al., 2002; Fagents and Thordarson, 2007]. Rootless cones also typically occur in very dense clusters [Thorarinsson, 1953; Frey et al., 1979; Hamilton et al., 2010b] and are constructional in nature [Thorarinsson, 1953; Greeley and Fagents, 2001], making them easy to distinguish from impact craters. One potential limit on the production of pseudocraters is the emplacement of a very thick lava flow unit, which might be able to resist the vapor pressure of the superheated water, preventing a phreatic eruption; however, it is reasonable to expect that at the margins of the lava flow (i.e., near the basin rims), the thickness would not be so great so as to prevent pseudocrater formation [Thorarinsson, 1953; Greeley and Fagents, 2001].

[13] If instead lava reaches the surface from below the paleolake, a tuya or a maar crater may result. Tuyas will be formed if the rising magma body is able to erupt on the surface below a large ice sheet. This will create a subglacial volcanic construct that has steep sides and a flat to slightly domed top, which is termed either a tuya or a table mountain. Tuyas form as a subglacial eruption melts the overlying ice sheet, creating a pool of standing water, which the lava will erupt into. As the volcanic edifice is built up, it may breach the top of the ice sheet, allowing for the emplacement of sub-aerial lava flows, thus resulting in the flat top. The tuya will also build out laterally by the creation of lava deltas on its flank with very steep foresets that are a consequence of the geometry of the circumferential pool of standing water, resulting in the steep-sided tuya [Mathews, 1947; Jones, 1968; Wilson and Head, 2002; Smellie, 2007; Jakobsson and Gudmundsson, 2008; Wilson et al., 2012]. Tuyas are very large features, with lengths of several to tens of kilometers, widths of several kilometers and heights of several hundred meters [Smellie, 2007]. It has also been hypothesized that tuyas exist on Mars, having typically larger dimensions than terrestrial tuyas [e.g., Chapman and Tanaka, 2001; Ghatan and Head, 2002; Smellie, 2007].

[14] Alternatively, if the rising magma encounters groundwater or surface water via draining by fissures, a phreatomagmatic, or sometimes phreatic, eruption may occur, resulting in the formation of a maar crater. On Earth, these eruptions form low, broad craters, with diameters between ∼100 and 200 m, and depths from ∼10–200 m [Lorenz, 1973, 1986]. Maar craters are also associated with specific lithofacies, such as ejecta deposits, which form a raised ‘rim’ tens to ∼100 m in height above the ground surface [Lorenz, 1973, 1986; Gutmann, 1976; White, 1989]. While maar craters are not readily observable on Mars [e.g., Wilson and Head, 1994], they should theoretically form if rising magma interacts with groundwater or surface water. However, one complication in identifying maar craters is their potential morphologic similarity to impact craters. Although this is the case, the steep maar crater walls typically contain distinct morphologic features such as pyroclastic deposits and lahar flows due to the nature of their formation [Lorenz, 1973, 1986; Gutmann, 1976; White, 1989] that cause them to differ in morphology from impact craters. Additionally, maar craters often form in multiple episodes, which can result in crater morphologies that are not perfectly circular or have additional cuspate features around their rims [Lorenz, 1973, 1986; Gutmann, 1976; White, 1989], making them further distinct from impact craters.

2.2. Assessment of Basin Resurfacing Unit Surface Properties

[15] Topography of the basin floors was analyzed using Mars Orbiter Laser Altimeter (MOLA) gridded topography and MOLA-derived slope maps [Smith et al., 2001], as well as HRSC stereo-derived topography [Neukum et al., 2004; Gwinner et al., 2010]. Additionally, the surface roughness of the basin floors and resurfacing units were evaluated using the MOLA-derived roughness map ofKreslavsky and Head [2000], which uses variations in topography at three length scales (0.6, 2.4 and 19.2 km) to estimate the roughness of the Martian surface.

[16] The main physical surface properties of the basin resurfacing units investigated were the thermal inertia and the bolometric albedo. The thermal inertia values of the basin resurfacing units were investigated using the global thermal inertia map produced by Putzig and Mellon [2007]. This map was derived from data from the Thermal Emission Spectrometer (TES) instrument aboard the Mars Global Surveyor spacecraft [Christensen et al., 2001; Putzig and Mellon, 2007], and offers a quantitative measure of the competence of the surface unit, with high thermal inertia values indicating competent surface units, such as bedrock, and low values indicating units of low competence, such as loose granular material or dust [e.g., Putzig et al., 2005; Putzig and Mellon, 2007]. Bolometric albedo values derived from the TES instrument [Christensen et al., 2001] were also analyzed.

2.3. Assessment of Basin Resurfacing Unit Composition

[17] The composition of the basin resurfacing units were studied using hyperspectral data from both the Compact Reconnaissance Imaging Spectrometer for Mars (CRISM) instrument aboard the MRO spacecraft [Murchie et al., 2007] and the Observatoire pour la Minéralogie, l'Eau, les Glaces et l'Activité (OMEGA) instrument aboard the MEx spacecraft [Bibring et al., 2004]. These imaging spectrometers provide spectral reflectance data in the visible to near-infrared (VNIR) region of the spectrum (∼0.36–3.9μm for CRISM, and ∼0.38–5.1 μm for OMEGA) at spatial resolutions of ∼18–36 m/pixel (targeted CRISM observations) [Murchie et al., 2007] and ∼0.3–5 km/pixel (OMEGA) [Bibring et al., 2004], and so offer two unique spatial perspectives on the composition of the basin resurfacing units.

[18] OMEGA data were corrected for photometry, known instrument artifacts [Bibring et al., 2005; Bellucci et al., 2006] and for atmospheric contributions. The atmospheric gas correction assumes that the contribution of the atmosphere and the surface are multiplicative, and that atmospheric gas absorptions vary with atmospheric path length following a power law [Bibring et al., 1989]. Using this assumption, spectra taken from the summit of Olympus Mons and from a lower portion of the volcano are ratioed to compute an estimate of the atmospheric transmission spectrum, which is then removed from each observation to estimate the surface reflectance [Bibring et al., 2005; Mustard et al., 2005].

[19] The CRISM data were first corrected to I/F by dividing the radiance data returned from the spacecraft with a solar spectrum [Murchie et al., 2007, 2009]. The I/F data were then corrected for photometry, and the effects of atmospheric absorptions were removed through a method similar to that used for the OMEGA data as described above (i.e., through the use of a volcano scan correction) [Mustard et al., 2008; Ehlmann et al., 2009; McGuire et al., 2009].

[20] The compositional analyses of the basin floor resurfacing units were performed using both spectral parameter maps [Pelkey et al., 2007; Salvatore et al., 2010], and detailed spectral analysis. During the spectral analysis, spectra from basin floors were ratioed to exterior terrain with spectrally bland surface materials in an attempt to remove instrument noise and to bring out the spectral diversity of the area of interest [e.g., Mustard et al., 2008; Ehlmann et al., 2009], which in this case is the resurfacing unit on the basin floor. During the spectral analysis, an emphasis was put on identifying crystal field absorptions in the 1 to 2 μm range, which are caused by electronic crystal field transitions of octahedrally coordinated Fe2+ in the mineral structure of silicate minerals [Burns, 1993].

2.4. Assessment of Basin Resurfacing Unit Emplacement Ages

[21] Emplacement ages of the basin resurfacing units were evaluated using counts of superposed craters and models of crater production for Mars that allow estimates of both relative ages and absolute model ages [e.g., Tanaka, 1986; Ivanov, 2001; Hartmann and Neukum, 2001; Hartmann, 2005; Werner and Tanaka, 2011]. Crater counts were performed on CTX, HRSC and/or THEMIS global IR mosaic images.

[22] Superposed craters were mapped using the CraterTools extension for ArcMap, where a best fit circle for each crater is mapped and the crater diameter is calculated in a local sinusoidal projection, thus offering an accurate assessment of the crater's diameter [Kneissl et al., 2011]. Craters were only included for model age determination if they exhibit a raised rim, a depressed center, and superpose the basin resurfacing unit. All craters that were readily identifiable as secondaries, such as highly clustered craters or craters occurring in chains [Oberbeck, 1971; Oberbeck et al., 1975], were excluded from age determinations. The final crater counts were analyzed using the software CraterStats, which determines a best fit model age based on a nonlinear least squares fit to a cumulative crater size-frequency distribution over a given range of crater diameters [Michael and Neukum, 2010].

[23] In order to determine the most accurate model age of emplacement, small crater diameters were not used in the final analysis due to the commonly observed downturn in small craters on Mars due to crater obliteration [e.g., Hartmann, 1971; Carr, 1992; Hartmann and Neukum, 2001; Fassett and Head, 2008b; Smith et al., 2008]. For the majority of the basins analyzed in this work, only craters larger than 1 km in diameter were used in the model age determinations; however, a portion of the basins (8; ∼27%) have very few craters larger than 1 km, and so in an attempt to improve the crater counting statistics, a smaller diameter range was used for these basins to ensure that at least 5 counted craters were used in the model age determination. Using this methodology, the smallest crater diameter included in our model age determinations was 600 m. Model age determinations were made using the Neukum production function reported by Ivanov [2001]. The calculated model ages for the resurfaced open-basin lake floors were also compared to the model ages of the end of valley network activity fromFassett and Head [2008b] derived from the Neukum production function [Ivanov, 2001].

[24] In addition to analyzing model ages, period determinations were calculated from the stratigraphic age boundaries of both Hartmann and Neukum [2001] and Werner and Tanaka [2011]. The period determinations were made using the derived model ages from the Neukum production function [Ivanov, 2001]. Additionally, period determinations were made using the cumulative number of mapped craters greater than or equal to 1 km in diameter normalized to an area of 106 km2 (i.e., N (1)) and the period boundaries defined by both Tanaka [1986] and Werner and Tanaka [2011].

3. Results

3.1. Resurfacing Unit and Basin Floor Morphology

[25] The results of the morphologic survey of the 30 resurfaced open-basin lakes (Figure 1 and Table 1) continue to support the interpretation that all 30 have a distinct morphology that is diagnostic of volcanic resurfacing (Figures 2 and 3a), consistent with previous studies [Fassett and Head, 2008a; Goudge et al., 2012]. This morphology includes: (1) lobate margins that embay both the basin perimeter, often defined by crater walls, and older material, such as sedimentary units associated with lacustrine activity or central peaks/peak rings in basins defined by ancient impact craters (Figure 2b) [Goudge et al., 2012]; (2) wrinkle ridges (Figure 2a, white arrows), which are commonly observed on Martian volcanic surfaces [e.g., Watters, 1991]; and (3) high crater retention, especially at small crater diameters, suggesting a competent surface unit [e.g., Fassett and Head, 2008a].

Figure 3.

Open-basin lake at −22.89°N, 158.54°E (Basin 29 inTable 1) [Fassett and Head, 2008a; Goudge et al., 2012] displaying physical characteristics indicative of volcanic resurfacing. North is up in all images. (a) THEMIS daytime IR global mosaic [Christensen et al., 2004] showing the smooth floor unit that embays the basin perimeter. (b) MOLA gridded topography [Smith et al., 2001] overlain on the THEMIS IR global mosaic [Christensen et al., 2004] showing the smooth floor of the basin interior with minimal topographic variation. (c) MOLA-derived roughness map ofKreslavsky and Head [2000]showing a dark basin interior, indicative of a smooth surface similar to other volcanic plains units across the surface of Mars. Image is a false color RGB where the red channel is roughness at a baseline of 19.2 km, green is roughness at a baseline of 2.4 km and blue is roughness at a baseline of 0.6 km, and darker colors indicate lower surface roughness. (d) MOLA-derived slope map [Smith et al., 2001] overlain on the THEMIS IR global mosaic [Christensen et al., 2004] showing a relatively flat, horizontal interior with slopes typically <0.5° across much of the basin interior. (e) THEMIS nighttime IR global mosaic [Christensen et al., 2004] showing a relatively warm signature in the basin interior, indicative of more competent material than the relatively cooler exterior terrain. (f) TES-derived thermal inertia [Putzig and Mellon, 2007] showing higher thermal inertia in the basin interior, indicative of more competent material than the exterior terrain. (g) OLINDEX2 parameter map [Salvatore et al., 2010] derived from OMEGA observation ORB1445_4 overlain on the THEMIS IR global mosaic [Christensen et al., 2004] showing enrichment of olivine in the basin interior compared to the exterior terrain.

Figure 3.

(continued)

[26] Based on the survey results, we found no evidence for the presence of lava deltas [Moore et al., 1973; Mattox and Mangan, 1997; Skilling, 2002], littoral cones [Moore and Ault, 1965; Fisher, 1968; Jurado-Chichay et al., 1996], rootless cones [Thorarinsson, 1953; Greeley and Fagents, 2001], tuyas [Mathews, 1947; Jones, 1968; Smellie, 2007; Jakobsson and Gudmundsson, 2008], or maar craters [Lorenz, 1973, 1986; Gutmann, 1976; White, 1989]. All five of these features should have been resolvable with the resolution of images used, suggesting that they never formed in these locations, or if so, were buried by later lava flooding. Alternative hypotheses for the lack of these features indicative of lava-water interaction are discussed insection 4.4.

3.2. Basin Resurfacing Unit Surface Properties

[27] The analysis of several data sets that give estimates of basin floor surface properties (e.g., surface roughness, thermal inertia) indicates that the open-basin lake resurfacing units have physical properties that are distinct from their surrounding terrain (Figure 3).

[28] Based on the analysis of MOLA gridded topography, MOLA-derived slope maps [Smith et al., 2001], HRSC stereo-derived topography [Neukum et al., 2004; Gwinner et al., 2010] and MOLA-derived roughness [Kreslavsky and Head, 2000], all 30 analyzed basins have smooth, flat-lying interiors with minimal topographic variation (Figures 3b–3d). This smooth topographic signature (Figures 3b and 3c) is consistent with topographic and roughness signatures identified for larger volcanic units on the surface of Mars [e.g., Kreslavsky and Head, 2000; Head et al., 2002; Hiesinger and Head, 2004], and the relatively low slopes of the basin interiors (Figure 3d), typically <0.5°, suggest that the interior resurfacing units were deposited by a mechanism that results in a gravitational equipotential surface, such as a fluid lava flow. Additionally, all of the open-basin lakes analyzed represent distinct topographic lows in comparison to the exterior terrain (Figure 3b). This is in accordance with how the basins were originally defined by Fassett and Head [2008a], where the authors required that “the observed feature must remain a basin on the basis of its present topography.” This distinction is important when considering alternate hypotheses for the formation of the valley networks associated with the basins and the smooth, volcanic floor units, as have been discussed by previous workers [e.g., Leverington and Maxwell, 2004; Leverington, 2006].

[29] The thermal properties of the basin floor units suggest that the basin resurfacing materials have high values of thermal inertia compared to their surrounding terrain (Figures 3e and 3f). This is indicated in both THEMIS nighttime IR data (Figure 3e) [Christensen et al., 2004] and TES-derived thermal inertia (Figure 3f) [Putzig and Mellon, 2007]. In the THEMIS nighttime IR data, the basin resurfacing materials all have a relatively ‘warm’ signature, indicating that they retain more heat at night than the surrounding terrain [Christensen et al., 2004], which is expected of competent material.

[30] This observation can be quantified using thermal inertia, which shows that the basin interiors have higher values of thermal inertia than the surrounding terrain (Figure 3f), again suggestive of a competent resurfacing unit [Putzig et al., 2005; Putzig and Mellon, 2007]. Of the 30 basins, 26 have a nighttime IR and thermal inertia signature that is warmer/higher than the exterior terrain. The four outliers are all located in Arabia Terra, a region of Mars that is known to be very dusty [e.g., Tanaka, 2000; Ruff and Christensen, 2002; Arvidson et al., 2003]. These four basins do show many of the same morphologic indications of volcanic resurfacing observed in the other studied basins (Figure 2), so it is possible that there is a thin surficial dust layer that is overlying the volcanic resurfacing unit, resulting in lower thermal inertia values. Looking at the distribution of thermal inertia values for only the 26 basins with high thermal inertia interiors, the frequency distribution is skewed toward the higher end of the global data set (Figure 4a), with a mean value of 264.77 (standard deviation = 72.58) J m−2 K−1 s−1/2. This shows the relatively high thermal inertia values for these units, consistent with a mixture of sand, rocks, boulders, and bedrock, and also consonant with the presence of a competent volcanic unit in the basin interiors below some surficial layer of granular material [Putzig et al., 2005; Putzig and Mellon, 2007].

Figure 4.

(a) Frequency distribution of TES-derived thermal inertia values [Putzig and Mellon, 2007] for the 26 volcanically resurfaced open-basin lakes with high thermal inertia interiors (blue curve) compared to the global data set (gray curve). Note that the volcanically resurfaced open-basin lake curve is skewed toward the higher end of the thermal inertia range. (b) Frequency distribution of TES-derived bolometric albedo values [Christensen et al., 2001] for the 26 volcanically resurfaced open-basin lakes with high thermal inertia interiors (blue curve) compared to the global data set (gray curve). Note that the volcanically resurfaced open-basin lake curve is skewed toward the lower end of the albedo range.

[31] For the 26 basins with high thermal inertia, the frequency distribution of bolometric albedo values is skewed toward the lower end of the global data set (Figure 4b), with an average value of 0.17 (standard deviation = 0.04) [Christensen et al., 2001]. This average bolometric albedo was coupled with the average thermal inertia value to obtain a unit competence classification based on the work of Putzig et al. [2005]. The basin resurfacing units fall into the category of Unit B from Putzig et al. [2005], which they interpret as “sand, rocks and bedrock; some duricrust.” The majority of the Syrtis Major and Hesperia Planum regions on Mars, both of which are large expanses of Hesperian-aged volcanic plains [Greeley and Spudis, 1981; Scott and Tanaka, 1986; Greeley and Guest, 1987], are also classified as Unit B by Putzig et al. [2005]. The physical properties of the resurfacing units within the studied basins match those of other volcanic plains units, and are consistent with volcanic resurfacing.

3.3. Basin Resurfacing Unit Composition

[32] Based on the analysis of CRISM and OMEGA VNIR spectral reflectance data, 12 of the studied resurfaced open-basin lakes (40%) have clear spectral signatures associated with the resurfacing units in their interior, all of which indicate an enhancement of mafic minerals in their interior compared to the surrounding terrain (Figure 3g and Table 2). These 12 basins have spectral signatures that exhibit diagnostic crystal field absorptions caused by electronic crystal field transitions of octahedrally coordinated Fe2+ in the structure of silicate minerals [Burns, 1993] (Figure 5 and Table 3).

Table 2. CRISM and OMEGA Observations for Basins with Identified Mafic Mineralsa
BasinCRISM Observation IDOMEGA Observation ID
  • a

    Table lists the basin number (from Table 1), and the ID numbers of the OMEGA and CRISM (where such data exist) observations used for the spectral identification of mafic minerals within the basin interiors.

3FRT0000B7B7ORB0488_4
6No CoverageORB1194_4
7No CoverageORB0485_2
10HRL00018A62ORB1310_2
11No CoverageORB1348_3
16FRT0001FB74ORB2272_4
21HRL0000491EORB1435_0
22FRT0001EC17ORB0353_1
24FRT0000503EORB2262_2
27No CoverageORB1238_5
28FRT0000A4C4ORB0551_2
29No CoverageORB1445_4
Table 3. CRISM and OMEGA Observation Numbers, Pixel Locations, and Window Sizes for the Numerator and Denominator Spectra Used to Calculate Ratio Spectra Shown in Figure 5a
Basin #CRISMOMEGA
 NumeratorDenominator NumeratorDenominator
Observation IDLatitude (N)Longitude (E)Pixel WindowLatitude (N)Longitude (E)Pixel WindowObservation IDLatitude (N)Longitude (E)Pixel WindowLatitude (N)Longitude (E)Pixel Window
  • a

    Table lists the basin number (from Table 1; also shown in Figure 5), the observation ID number for the analyzed observations over the basin (CRISM and OMEGA), and the latitude (north positive), longitude (east positive) and pixel window size of the numerator and denominator spectra used for producing the ratioed spectra shown in Figure 5.

3FRT0000B7B719.56061.84525 × 2519.42261.85225 × 25ORB048 8_419.56061.8585 × 519.49567.97115 × 15
16FRT0001FB7418.33277.47215 × 1518.03277.51315 × 15ORB227 2_418.33277.4837 × 711.98676.85015 × 15
21HRL0000491E−2.65761.80315 × 15−2.71361.80815 × 15ORB143 5_0−2.65261.80211 × 111.38761.27035 × 35
22FRT0001EC17−27.294−18.07015 × 15−27.255−18.06715 × 15ORB0353_1−27.293−18.06815 × 15−22.706−18.12535 × 35
24FRT0000503E−22.490−20.35215 × 15−22.444−20.34815 × 15ORB226 2_2−22.488−20.3497 × 15−25.278−20.40915 × 15
Figure 5.

Representative ratioed CRISM (Figure 5a) and OMEGA (Figure 5b) spectra from the interiors of five different volcanically resurfaced open-basin lakes (middle plots) compared with the spectra of Syrtis Major volcanic smooth plains material (bottom plots) [Skok et al., 2010]. The CRISM and OMEGA spectra are from the same five basins, and are taken from areas that are as geographically close to one another as possible. See Table 3 for full image information, as well as spectra (numerator and denominator) locations. Listed basin numbers are from Table 1. (a) Top plot shows example numerator and denominator spectra used for computing a ratioed CRISM spectrum. Numerator is characteristic of the mafic rich, volcanic resurfacing unit in Basin 21, while denominator is spectrally bland material. Lower two plots show CRISM ratioed spectra from five different volcanically resurfaced open-basin lakes (middle plot) compared with a spectrum of Syrtis Major volcanic smooth plains material (bottom plot) [Skok et al., 2010]. (b) Top plot shows example numerator and denominator spectra used for computing a ratioed OMEGA spectrum. Numerator is characteristic of the mafic rich, volcanic resurfacing unit in Basin 21, while denominator is spectrally bland material. Lower two plots show OMEGA ratioed spectra from five different volcanically resurfaced open-basin lakes (middle plot) compared with a spectrum of Syrtis Major volcanic smooth plains material (bottom plot) [Skok et al., 2010].

[33] Specifically, we have identified the presence of the mafic minerals olivine, low-calcium pyroxene (LCP) and high-calcium pyroxene (HCP) (Figures 5 and 6). Olivine is identified by the presence of a broad, complex absorption centered near 1 μm, with the exact band center and shape varying with changes in Fe content in the mineral structure [King and Ridley, 1987]. Pyroxene is identified by two broad absorptions centered at ∼1 and 2 μm respectively [Adams, 1974]. In the case of pyroxene minerals, the band center of the ∼1 and 2 μm absorptions is most closely related to Ca content in the mineral structure, with HCP minerals, such as diopside (MgCaSi2O6) and hedenbergite (FeCaSi2O6), having band centers at longer wavelengths (near 1.05 and 2.3 μm), and LCP minerals, such as enstatite (Mg2Si2O6) and ferrosilite (Fe2Si2O6), having band centers at shorter wavelengths (near 0.9 and 1.8 μm) [Adams, 1974; Klima et al., 2007, 2011].

Figure 6.

Laboratory spectra of primary mafic minerals that have signatures similar to those derived from CRISM and OMEGA ratioed reflectance spectra from the open-basin lake volcanic resurfacing units (Figure 5). Spectra are from the CRISM spectral library [CRISM Science Team, 2006]. Diopside (HCP) is shown in red and is sample LAPP97. Enstatite (LCP) is shown in orange and is sample C2PE30. Forsterite and fayalite (olivine) are shown in green, and are samples C3PO61 and C3PO59 respectively. Spectra are offset for clarity.

[34] The spectral signatures identified in the basin resurfacing units show a range of different band shapes and centers, which we interpret to represent a range in relative proportions of olivine, LCP and HCP, with the most common spectral signature observed being indicative of a combination of olivine and HCP (Figures 5 and 6). This mineral assemblage is consistent with the composition of Hesperian-aged volcanic plains identified across the surface of Mars [e.g.,Mustard et al., 2005; Baratoux et al., 2007; Rogers and Christensen, 2007; Poulet et al., 2009b; Salvatore et al., 2010; Skok et al., 2010] (Figure 5, bottom plots). While other workers have determined relative proportions of the mafic minerals present based on identified spectral signatures [e.g., Sunshine et al., 1990; Sunshine and Pieters, 1993; Mustard et al., 1997; Baratoux et al., 2007; Poulet et al., 2009a; Skok et al., 2010], such a detailed analysis is beyond the scope of this work. An additional observation is that the spectral signatures of the basin floor units are relatively consistent across CRISM and OMEGA data within each basin (Figure 5), which has implications for compositional homogeneity.

3.4. Basin Resurfacing Unit Emplacement Ages

[35] From the analysis of crater counts and cumulative size-frequency distributions for the 30 basins, model ages and stratigraphic periods were determined that represent the emplacement age of the basin resurfacing units (Figures 7 and 8 and Tables 46). The model ages and period determinations show that the resurfacing of the open-basin lakes occurred over a wide range in Martian history, beginning at approximately the Noachian-Hesperian boundary and continuing into the Early Amazonian (Figure 8 and Tables 5 and 6). The majority of basins appear to be resurfaced during the Hesperian (Figure 8 and Tables 5 and 6), consistent with large emplacements of volcanic plains during this time, which resurfaced ∼30% of the planet [Scott and Tanaka, 1986; Greeley and Guest, 1987; Head et al., 2002]. Additionally, it is interesting to note that some of the basin resurfacing ages are similar to the time period when valley network activity on a regional-to-global scale ceased [Fassett and Head, 2008b].

Figure 7.

Example count areas and counted craters, and cumulative size-frequency distributions for two different volcanically resurfaced open-basin lakes analyzed in this work. North is up in both images. (a) Volcanically resurfaced open-basin lake at −8.84°N and −7.21°E (Basin 6 inTable 1) [Fassett and Head, 2008a; Goudge et al., 2012]. Basin perimeter, defined by a MOLA topographic contour [Fassett and Head, 2008a], is indicated by the green line, count area is indicated by the orange line and counted craters used in the model age determination (D > 1 km) are shown as red circles. Image is THEMIS daytime IR global mosaic [Christensen et al., 2004]. (b) Cumulative size-frequency distribution for craters superposed on the volcanically resurfaced open-basin lake shown in Figure 7a. Only counted craters used in the model age determination are shown. Model age determination is from the Neukum production function [Ivanov, 2001]. The calculated model age isochron is shown in red, and the Hesperian-Amazonian boundary (3.46 Gyr) and the Noachian-Hesperian boundary (3.74 Gyr) are shown in black [Werner and Tanaka, 2011]. (c) Volcanically resurfaced open-basin lake at −21.26°N and −5.25°E (Basin 28 inTable 1) [Fassett and Head, 2008a; Goudge et al., 2012]. Basin perimeter, defined by a MOLA topographic contour [Fassett and Head, 2008a], is indicated by the green line, count area is indicated by the orange line and counted craters used in model age determination (D > 1 km) are shown as red circles. Image is THEMIS daytime IR global mosaic [Christensen et al., 2004]. (d) Cumulative size-frequency distribution for craters superposed on the volcanically resurfaced open-basin lake shown in Figure 7c. Only counted craters used in the model age determination are shown. Model age determination is from the Neukum production function [Ivanov, 2001]. The calculated model age isochron is shown in red, and the Hesperian-Amazonian boundary (3.46 Gyr) and the Noachian-Hesperian boundary (3.74 Gyr) are shown in black [Werner and Tanaka, 2011].

Figure 8.

Plot of model ages for emplacement of the volcanic resurfacing units within the studied open-basin lakes. Ages shown here were calculated using the Neukum production function [Ivanov, 2001]. See Table 4 for a complete list of basin ages. Period boundaries (green lines) are as defined by Werner and Tanaka [2011], where LN = Late Noachian, EH = Early Hesperian, LH = Late Hesperian, EA = Early Amazonian and MA = Middle Amazonian. Top plot shows a summary of all calculated model ages for emplacement of the volcanic resurfacing units for the 30 basins analyzed (blue points). Shaded red area indicates the range of model ages for the cessation of valley network activity [Fassett and Head, 2008b], derived from the Neukum production function [Ivanov, 2001], excluding the four youngest valley networks, which are thought to be unrelated to open-basin lake activity [Fassett and Head, 2008a, 2008b]. Error bars are indicated in black. Bottom plot shows a direct comparison between the volcanically resurfaced open-basin lakes with the oldest (i.e. >3.4 Ga) model ages of emplacement (blue points) and the model ages for valley network cessation, excluding the four youngest valley networks, fromFassett and Head [2008b] (red points). Note that while there is some overlap in the two populations of ages, the population of volcanic resurfacing model ages is noticeably younger than the population of valley network cessation ages. Error bars are indicated in black.

Table 4. Model Ages of Emplacement for the Analyzed Volcanic Resurfacing Unitsa
Basin #Dmin (km)N(D ≥ Dmin)Neukum Production FunctionN (1)b
AgeLower Error BarUpper Error BarValueError
  • a

    Table lists the basin number (from Table 1), the minimum crater size diameter used in model age calculations (Dmin), the number of craters used for the model age determinations (i.e., N(D ≥ Dmin)), the model age derived from the Neukum production function [Ivanov, 2001] with errors, and the N(1) value with error, calculated as √n, both normalized to 106 km2. Model age errors are calculated in CraterStatsby converting the +/−1-sigma error on the model fit forN(1) (which itself is calculated as 1/√n) into uncertainties in the model age based on the utilized production function [Michael and Neukum, 2010]. Note also that the large errors in many of the N(1) values are due to the small count areas of the resurfaced open-basin lake floors.

  • b

    N(1) values presented in the table are the measured N(1) values for all basins with Dmin = 1 km. For all basins with Dmin < 1 km, the presented N(1) value is the model N(1) value and calculated 1-sigma error obtained from the best fit model age fromCraterStats [Michael and Neukum, 2010].

10.773.111.100.291059594
21133.580.100.063150874
31493.480.060.05946135
41123.460.200.092033587
51103.660.090.0639411246
61283.640.050.042228421
71143.510.130.071981529
8183.650.110.0636241281
90.762.591.000.661260510
100.953.430.730.1323701050
110.953.201.300.251700753
121153.740.060.0455111423
131343.650.040.033278562
141333.560.060.042143373
151173.560.090.062364573
160.753.450.670.1224401080
171313.450.100.062101377
181103.690.080.0562791986
190.661.880.760.76918371
201133.700.070.053209890
21193.620.110.062730910
22183.600.130.072450866
23173.800.090.0556482135
240.863.520.280.1029401190
251163.560.090.062315579
261113.740.070.0566632009
271103.710.080.0537761194
281393.590.050.041310210
290.853.201.300.251700751
301153.690.060.043408880
Table 5. Period Determinations for Emplacement of the Analyzed Volcanic Resurfacing Unitsa
Basin #Neukum Production Function Model Age DeterminationN(1) Determination
Hartmann and Neukum [2001] PeriodWerner and Tanaka [2011] PeriodTanaka [1986] PeriodWerner and Tanaka [2011] Period
1EAEAEAEA
2LHLHEHEH
3LHLHEAEA
4LHLHLHEA
5EHEHEHEH
6EHLHLHLH
7LHLHLHEA
8EHEHEHEH
9EAEAEAEA
10LHEALHLH
11EAEALHEA
12LNLNLNLN
13EHEHEHEH
14LHLHLHLH
15LHLHLHLH
16LHEALHLH
17LHEALHLH
18EHEHLNLN
19MAEAEAEA
20LNEHEHEH
21EHLHLHLH
22EHLHLHLH
23LNLNLNLN
24LHLHLHLH
25LHLHLHLH
26LNLNLNLN
27LNEHEHEH
28LHLHEAEA
29EAEALHEA
30EHEHEHEH
Table 6. Summary of Period Determinations for Emplacement of the Analyzed Volcanic Resurfacing Unitsa
PeriodNeukum Production Function Model Age DeterminationN(1) Determination
Hartmann and Neukum [2001] PeriodWerner and Tanaka [2011] PeriodTanaka [1986] PeriodWerner and Tanaka [2011] Period
  • a

    Table lists number of model ages of emplacement for the analyzed volcanic resurfacing units within each stratigraphic period for the four different methods presented in Table 5.

Late Noachian5344
Early Hesperian8777
Late Hesperian12121410
Early Amazonian4859
Middle Amazonian1000

[36] The derived ages are consistent with previous model ages for the basin floors of several open-basin lakes determined through crater counting [e.g.,Cabrol et al., 1998; Cabrol and Grin, 2001; Irwin et al., 2002; Fassett and Head, 2008a], although some previous workers have suggested that these ages are representative of the termination of lacustrine activity [e.g., Cabrol et al., 1998], as opposed to ages of resurfacing, as we conclude here.

4. Discussion

4.1. Definitive Evidence for Post-lacustrine Volcanic Resurfacing

[37] It is clear from these observations of the morphology (Figures 2 and 3a), physical properties (Figure 3), and composition (Figures 3g and 5) of the resurfacing units in the interiors of the studied open-basin lakes that they are indeed volcanic in origin, consistent with previous morphologic studies of resurfaced open-basin lakes [e.g.,Fassett and Head, 2008a; Goudge et al., 2012]. These findings further confirm the importance that resurfacing had in the history of open-basin lakes on Mars [Goldspiel and Squyres, 1991; Fassett and Head, 2008a; Goudge et al., 2012], which have been devoid of fluvial activity for ∼3.6–3.7 Gyr [Fassett and Head, 2008a, 2008b]. These results also confirm the hypothesis that volcanic resurfacing of paleolake basins was a widespread process on the surface of Mars, proposed based on the analysis of both orbital data [e.g., Goldspiel and Squyres, 1991; Fassett and Head, 2008a; Goudge et al., 2012] and in situ exploration at the Gusev crater paleolake [e.g., Squyres et al., 2004].

[38] While the work we have discussed thus far falls into the paradigm of sustained fluvial activity on the surface of Mars forming the inlet and outlet valley networks that feed these open-basin lakes [e.g.,Goldspiel and Squyres, 1991; Cabrol and Grin, 1999; Craddock and Howard, 2002; Irwin et al., 2005; Fassett and Head, 2008a, 2008b], it has also been suggested by previous workers that some of the observed valleys and associated open-basins were instead formed by the flow of low viscosity, effusive lava [e.g.,Leverington and Maxwell, 2004; Leverington, 2006].

[39] The strongest argument against the formation of the valley networks and open-basins through volcanic processes is the observation that these basins remain as topographic lows on the Martian surface [Fassett and Head, 2008a] (Figure 3b), a point discussed by Irwin et al. [2005]. In order for the outlet channels of these basins to have formed, the liquid that filled the basin interior must have ponded to at least the topographic level of the outlet valley head before breaching the basin perimeter and flowing out on to the exterior terrain, requiring a sustained input of the liquid responsible for filling the basin [Fassett and Head, 2008a]. If flowing lava was the liquid responsible for forming these inlet and outlet valley networks, and such large volumes of ponding lava did occur within these basins, it would be expected that at least the upper portions of the lava would cool by conduction, creating a solid thermal boundary layer, similar to terrestrial lava lakes [Peck and Minakami, 1968; Wright et al., 1968; Peck et al., 1977]. When this boundary layer solidified, it would have been at approximately the topographic level of the outlet valley head, which is typically on the order of hundreds of meters above the current basin floor (Figure 3b).

[40] In order for the basin to remain a topographic depression, as is observed [Fassett and Head, 2008a], there would have to have been substantial subsidence of the solid, thermal boundary layer, which would have resulted in distinctive lava deflation morphologies [Wright et al., 1968; Thordarson and Self, 1998] as well as ‘high-lava marks’, neither of which are observed within these basins, although examples can be found elsewhere on Mars [see, e.g.,Jaeger et al., 2010]. Furthermore, it is likely that more than just an upper thermal boundary layer of the flowing lava would have solidified within the basins due to the requirement of sustained flow to form the outlet valley [Fassett and Head, 2008a], thus making it even more difficult to explain the topographic lows within these basins if the inlet valleys and outlet valley were formed by flowing lava.

[41] Stratigraphic relationships within these basins also show clear embayment of the basin perimeter (Figure 2), as well as older lacustrine deposits, where they occur (Figure 9), indicating that the basins must have experienced some period of fluvial activity prior to volcanic resurfacing. Therefore, while it is known that low viscosity, effusive lava flows can cause erosion of the underlying terrain on Mars to form sinuous channels [e.g., Williams et al., 2005; Hurwitz et al., 2010], we do not believe that such lava erosion could have created the observed morphologic features (i.e., basins with inlet channels and an outlet channel) studied in this work. This conclusion is also in agreement with previous arguments against a volcanic origin for such open-basins [Irwin et al., 2005; Fassett and Head, 2008a].

Figure 9.

Volcanically resurfaced open-basin lake in Jezero Crater, 18.38°N, 77.70°E (Basin 16 inTable 1) [Fassett and Head, 2005], showing embayment of the older deltaic deposit by the younger volcanic resurfacing unit. North is up in both images. (a) Overview of the delta deposit in Jezero Crater showing the smooth, volcanic plains unit on the crater floor. Location of Figure 9b is indicated by white box. CTX image P03_002387_1987_XI_18N_282W. (b) A portion of the Jezero Crater delta showing embayment by the volcanic resurfacing unit. White arrows indicate locations of lava embayment of the delta, where the lava flow has subsequently deflated. HiRISE image PSP_003798_1985.

4.2. Implications of Volcanic Resurfacing Unit Composition

[42] The analysis of the composition of the volcanic resurfacing units from CRISM and OMEGA hyperspectral data indicate that where clear spectral signatures are present, the resurfacing units always appear enriched in mafic minerals compared to their surrounding terrain (Figure 3g). The most commonly observed spectral signature for these units is a mixture of olivine and HCP, consistent with previously identified examples of Hesperian-aged volcanic smooth plains, and distinct from LCP rich Noachian-aged units [Mustard et al., 2005; Baratoux et al., 2007; Rogers and Christensen, 2007; Poulet et al., 2009b; Salvatore et al., 2010; Skok et al., 2010] (Figure 5, bottom plots). This composition also supports the model ages derived from crater counting and cumulative size-frequency distributions, which indicate a primarily Hesperian age of emplacement for these units (Figure 8 and Tables 5 and 6).

[43] The spectral analysis of the open-basin lake resurfacing units shows that the CRISM and OMEGA spectral signatures are very similar when looking at individual basins with both CRISM and OMEGA coverage (Figure 5). We ensured that the spectra extracted from both CRISM and OMEGA are from the same latitude and longitude, within ∼0.01° (Table 3), which equates to ∼600 m on the Martian surface at the equator; however, the difference in spatial scale between the two instruments, a factor of ≥∼15 [Bibring et al., 2004; Murchie et al., 2007], precludes the extraction of spectra from precisely the same area. Considering such differences in spatial scales, the most probable explanation for the correspondence in spectral signatures from the CRISM and OMEGA instruments (Figure 5) is that the resurfacing units within each basin are relatively compositionally homogenous.

[44] Previous workers have shown that variations in Fe/Mg content in olivine [Skok et al., 2012] and HCP to LCP ratios [e.g., Baratoux et al., 2007; Kanner et al., 2007; Poulet et al., 2008, 2009a, 2009b; Skok et al., 2010] are detectable through CRISM and OMEGA orbital spectroscopy, and so it would be expected that if such variations were present in the resurfacing units, they would be resolved in the analyzed spectra. The fact that such spectral variations are not observed between CRISM and OMEGA data suggests that when the basins were resurfaced, the volcanic unit within each basin was emplaced as a compositionally homogeneous unit across the entire basin floor.

4.3. Timing of Volcanic Resurfacing

[45] The model ages for the emplacement of the volcanic resurfacing units show that the process of volcanic resurfacing began in the Late Noachian, was concentrated in the Hesperian, and then continued into the Early Amazonian in some locations (Figure 8 and Tables 5 and 6). These ages correspond well with the global Martian stratigraphy, with most of the large volcanic provinces emplaced in the Noachian to Hesperian, followed by localized volcanic activity in the Amazonian [Greeley and Spudis, 1981; Scott and Tanaka, 1986; Tanaka, 1986; Greeley and Guest, 1987; Head et al., 2002]. While the model ages for emplacement of the volcanic resurfacing units appear concentrated in the Hesperian (Figure 8 and Tables 5 and 6), this does not necessarily imply a peak in volcanic resurfacing at this time compared with earlier in Martian history, as the dated volcanic resurfacing units are the youngest of a potential series of volcanic resurfacing units; however, the model ages for emplacement do imply that volcanic resurfacing in the Hesperian was more prevalent than in the Amazonian.

[46] There does not appear to be a correlation between geographic location of the basins and the emplacement age of the volcanic resurfacing units (Figure 10). This suggests that the temporal signature of emplacement ages (i.e., with volcanic resurfacing beginning in the Late Noachian, being focused in the Hesperian and continuing into the Early Amazonian) is not influenced by particular anomalous regions of volcanically resurfaced open-basin lakes, and is rather more reflective of the true temporal signature associated with the volcanic resurfacing of open-basin lakes.

Figure 10.

Geographic distribution of period determinations for the emplacement ages of the 30 volcanically resurfaced open-basin lakes analyzed in this work. Note the lack of correlation between geographic location and emplacement age. Period determinations are those derived from absolute model ages from the Neukum production function [Ivanov, 2001] and the Werner and Tanaka [2011] absolute age boundaries (Table 5). Background is MOLA topography overlain on MOLA hillshade [Smith et al., 2001].

[47] The emplacement ages give further evidence that most of the valley networks that feed these open-basin lakes were formed by flowing water prior to the volcanic resurfacing of the paleolake basin floors, as the valleys themselves were active until approximately the Noachian-Hesperian boundary [Irwin et al., 2005; Fassett and Head, 2008b; Hoke and Hynek, 2009; Mangold et al., 2012], while the majority of basin resurfacing units have Hesperian emplacement ages (Figure 8 and Tables 5 and 6). This can be readily seen when directly plotting the population of valley network cessation model ages and volcanic resurfacing unit emplacement model ages against one another (Figure 8, bottom plot). This plot shows that the population of model ages for emplacement of the volcanic resurfacing units is largely younger than the population of model ages for valley network activity cessation [Fassett and Head, 2008b]. Although this is the case, there are a small number of basins that have model ages of emplacement that are similar to, or only slightly younger than, the ages of valley network cessation [Fassett and Head, 2008b] (Figure 8).

4.3.1. Relationship Between Ages of Valley Networks and Lacustrine Activity

[48] In order to make any inference about the relationship between the volcanic resurfacing ages for the open-basin lake interiors and the timing of lacustrine activity within these paleolakes, we have made the assumption that the valley network ages presented byFassett and Head [2008b]are representative of this period of lacustrine activity. We feel that this is a valid assumption for two primary reasons. The first reason is that 15 of the 30 studied open-basin lakes are in fact fed and drained by valley networks dated byFassett and Head [2008b]. Furthermore, all 30 of the studied open-basin lakes are fed and drained by valley networks that are included in the catalog ofHynek et al. [2010]. Hynek et al. [2010]compiled a global map of valley networks across the surface of Mars, and concluded that the large majority of these valley networks formed during the Late Noachian-Early Hesperian, consistent with the work ofFassett and Head [2008b], as well as other works dating the ages of Martian valley networks [e.g., Irwin et al., 2005; Hoke and Hynek, 2009; Mangold et al., 2012]. As the lacustrine activity within the studied open-basin lakes was the result of the fluvial activity within their inlet and outlet valley networks [Fassett and Head, 2008a], it follows that the period of lacustrine activity within these paleolakes must have been coeval with the period of fluvial activity within the associated valley networks.

[49] Second, work by many previous authors directly dating the timing of fluvial activity associated with Martian valley networks have all concluded that the primary episode of fluvial erosion in the cratered highlands of Mars occurred early in the planet's history, ceasing by approximately the Noachian-Hesperian boundary [e.g.,Irwin et al., 2005; Fassett and Head, 2008b; Hoke and Hynek, 2009; Mangold et al., 2012]. While these works do not date every individual valley network across the surface of Mars, they all conclude based on similarities of morphology, degradation state and stratigraphic relationships, that a Noachian-Hesperian boundary cessation age is likely to be representative for the majority of Martian highland valley networks [Irwin et al., 2005; Fassett and Head, 2008b; Hoke and Hynek, 2009; Mangold et al., 2012]. Additionally, the multitude of different workers that have reached the same conclusion that Martian valley network formation largely ceased following the Noachian-Hesperian boundary [Irwin et al., 2005; Fassett and Head, 2008b; Hoke and Hynek, 2009; Mangold et al., 2012] adds confidence to the notion that this age is representative of all Martian highland valley networks. It is important to note that the term ‘valley network’ as used in the context above is distinct from younger and morphologically different fluvial features observed across the surface of Mars, such as midlatitude valleys probably associated with the melting of ice [e.g., Fassett et al., 2010; Howard and Moore, 2011] or channels on impact ejecta blankets of fresh craters [e.g., Morgan and Head, 2009; Mangold, 2012].

4.4. Implications of the Lack of Lava-Water Interaction

[50] While there is some overlap between model ages for emplacement of the volcanic resurfacing units and valley network activity cessation (Figure 8), we have observed no geologic features that would indicate lava-water interaction, with the primary focus of our attention being on detecting evidence such as remnant lava deltas, littoral cones, pseudocraters, tuyas, and/or maar craters.

[51] Lava deltas form as lava flows enter a body of standing water, creating a deposit that appears roughly similar in morphology to lacustrine deltas from the rapid cooling of the lava [Moore et al., 1973; Mattox and Mangan, 1997; Skilling, 2002]. Littoral cones are cone-shaped deposits composed of pyroclastic material that form due to the explosive expansion of vaporized water in the nearshore regions of areas where lava flows enter standing water [Moore and Ault, 1965; Fisher, 1968; Jurado-Chichay et al., 1996]. Pseudocraters are constructional features that form in dense clusters due to phreatic eruptions as lava flows onto water-saturated sediment [Thorarinsson, 1953; Greeley and Fagents, 2001; Hamilton et al., 2010b]. Tuyas are essentially subglacial volcanoes or volcanic constructs that result in large mesas with steep sides and a flat to slightly domed top [Mathews, 1947; Jones, 1968; Smellie, 2007; Jakobsson and Gudmundsson, 2008]. Maar craters form as rising magma encounters groundwater or surface water that has drained to depths via fissures, causing explosive phreatomagmatic eruptions, resulting in low, broad craters [Lorenz, 1973, 1986; Gutmann, 1976; White, 1989]. We examined each of the resurfaced open-basin lakes for all five of these geologic landforms, and found no evidence for intact or remnant features indicative of lava-water interaction.

[52] The lack of morphologic evidence for lava-water interaction of the types detailed above suggests that the basins were largely devoid of water, either standing or contained in the pore space of partially saturated sediment, at the time of resurfacing. If this is the case, it has potentially important implications for the timing of fluvial activity on Mars at the Noachian-Hesperian boundary, and so we have investigated several alternate hypotheses for this observation. The three alternate hypotheses for explaining the lack of features indicative of lava-water interaction investigated in this section are:

[53] 1. The studied basins are not a representative sample of the volcanically resurfaced open-basin lakes initially identified byGoudge et al. [2012], and the majority of the unstudied basins do contain features indicative of lava-water interaction.

[54] 2. Features indicative of lava-water interaction that were initially formed have since been eroded.

[55] 3. Features indicative of lava-water interaction that were initially formed have since been buried by younger lava flows.

4.4.1. Unrepresentative Sample

[56] One possible explanation for the lack of observed features that would indicate lava-water interaction is that the basins investigated here do not represent an accurate subset of the volcanically resurfaced open-basin lakes identified byGoudge et al. [2012]. In this scenario, the majority of volcanically resurfaced open-basin lakes do contain evidence of lava-water interaction, and we have simply analyzed 30 basins without such evidence. We do not believe this scenario is probable, as the basins analyzed in this study are spread across a wide range of geographic locations (Figure 1 and Table 1), have a wide range of model ages (Figure 8 and Table 4), and were chosen to be morphologically representative of the volcanically resurfaced open-basin lakes identified byGoudge et al. [2012]. Therefore, we consider the 30 basins to be a representative sample of the previously identified volcanically resurfaced open-basin lakes [Goudge et al., 2012].

4.4.2. Erosion of Evidence of Lava-Water Interaction

[57] Another possible explanation for the lack of features indicating lava-water interaction is that such interaction did in fact occur at these basins, however the geologic evidence has simply been eroded and obscured or removed. As we have examined the basins for five separate features that might indicate lava-water interaction, we examine the possibility of erosion for each geologic feature.

[58] Looking at a specific example of a terrestrial lava delta, lava flow into the Pacific Ocean from Kilauea Volcano in Hawai'i resulted in a lava delta that was approximately 3 km in length and 500 m in width [Mattox and Mangan, 1997]. This lava delta size is only slightly smaller than some of the observed deltas and alluvial fans on Mars [e.g., Ori et al., 2000; Fassett and Head, 2005; Irwin et al., 2005; Di Achille and Hynek, 2010], suggesting that such features would have remained preserved on the Martian surface if they had formed at the time of volcanic resurfacing. If the open-basin lakes studied here contained appreciable amounts of standing water at the time of volcanic resurfacing and were flooded by exterior lava flows, a lava delta should have formed [Moore et al., 1973; Mattox and Mangan, 1997], and would have been likely to remain preserved on the surface of Mars; however, no such features are observed, suggesting that they never formed at these sites.

[59] Similar to lava deltas, terrestrial tuyas are very large features, with lengths ranging from ∼2–15 km, widths ranging from ∼1–10 km, and heights ranging from ∼200–1000 m [Smellie, 2007], and so we expect it is unlikely that such large features would have been eroded. Additionally, features interpreted as tuyas have been observed on Mars [e.g., Chapman and Tanaka, 2001; Ghatan and Head, 2002], and it has been noted that these features have typically larger dimensions than terrestrial tuyas, with widths ranging from ∼30–120 km and heights ranging from ∼700–1800 m [Ghatan and Head, 2002]. Furthermore, it was hypothesized by Ghatan and Head [2002]that the tuyas they studied were associated with the emplacement of Early Hesperian-age volcanic ridged plains. This gives further evidence that such large features should not have been eroded subsequent to the time of emplacement of the volcanic resurfacing units studied in this work, and thus should still be visible if they had initially formed.

[60] Finally, we consider littoral cones, pseudocraters, and maar craters together, due to their approximately similar morphology. Terrestrial littoral cones are large constructional features with diameters of hundreds of meters, and typical heights of 20–80 m, with initial heights of up to 100 m [Moore and Ault, 1965; Fisher, 1968; Jurado-Chichay et al., 1996]. Pseudocraters on the Earth are typically tens to hundreds of meters in diameter, while observed examples on Mars are ∼30–1000 m in diameter [Greeley and Fagents, 2001; Fagents et al., 2002; Fagents and Thordarson, 2007]. Fagents et al. [2002] have also shown that pseudocraters on Mars are >25–60 m in height, based on individual MOLA tracks. Maar craters observed on Earth have the morphology of low, broad craters, with diameters ranging in size from ∼100–2000 m, depths of ∼10–200 m, and raised rims, composed of pyroclastic ejecta, of tens to ∼100 m in height [Lorenz, 1973, 1986].

[61] The comparatively small sizes of these three features makes it more difficult to state with certainty that they would remain on the surface if they had been formed in lava units in the Early Hesperian; however, their large heights compared to rim heights of impact craters of comparable sizes [Pike, 1974; Craddock et al., 1997] make it easier to preserve littoral cones, pseudocraters, and maar craters than impact craters. Additionally, pseudocraters occur in dense clusters [Thorarinsson, 1953; Frey et al., 1979; Hamilton et al., 2010b], which would again make it easier to preserve some indication of such morphologies than stochastically emplaced, individual impact craters.

[62] While it is widely reported that small crater removal is a significant process acting on the surface of Mars [e.g., Hartmann, 1971; Carr, 1992; Hartmann and Neukum, 2001; Fassett and Head, 2008b; Smith et al., 2008], Smith et al. [2008]have shown through modeling of crater obliteration processes that the time-averaged crater obliteration rate at Gusev crater is 4.72 ± 2.58 nm a−1, consistent with measured erosion rates in situ [Golombek et al., 2006; Smith et al., 2008]. The volcanically resurfaced Gusev crater paleolake [Grin and Cabrol, 1997; Cabrol and Grin, 1999, 2001; Squyres et al., 2004] is one of the open-basin lakes we have examined (our Basin 17;Table 1), and so this crater obliteration rate may be a reasonable estimate for the crater obliteration at the 30 sites analyzed in this work. This crater obliteration rate can then be used to calculate the total crater obliteration since the time of emplacement of the volcanic resurfacing units at each of the open-basin lakes studied here (Table 7), that is the maximum depth of an impact crater or maar crater, or the total height of a littoral cone or pseudocrater, that could have been completely erased from the surface since the time of emplacement of the volcanic resurfacing unit [Smith et al., 2008]. It should be noted that crater obliteration rates account for both the crater erosion rate and the crater infilling rate [Smith et al., 2008].

Table 7. Calculated Total Crater Obliteration Heightsa
Basin #Total Obliteration Height (m)Lower Error BarUpper Error Bar
  • a

    Table lists the calculated total crater obliteration height at the 30 analyzed basin sites since the time of emplacement of the volcanic resurfacing units. The values are calculated using the crater obliteration rate of 4.72 ± 2.58 nm a−1 derived from Smith et al. [2008] at the Gusev crater paleolake, and the model ages presented for each basin in Table 4. Table lists the basin number (from Table 1) and the calculated total obliteration height with error, in meters, since the time of emplacement of the volcanic resurfacing unit, based on the Neukum model ages of emplacement [Ivanov, 2001]. Also listed are the average and maximum crater obliteration values, with error, in meters.

114.689.568.14
216.909.259.24
316.438.988.98
416.338.988.94
517.289.459.45
617.189.399.39
716.579.089.06
817.239.439.42
912.228.187.37
1016.199.508.87
1115.1010.298.34
1217.659.659.65
1317.239.429.42
1416.809.199.19
1516.809.199.19
1616.289.458.92
1716.288.918.91
1817.429.539.52
198.876.036.03
2017.469.559.55
2117.099.359.34
2216.999.319.29
2317.949.819.81
2416.619.189.09
2516.809.199.19
2617.659.659.65
2717.519.589.57
2816.949.279.26
2915.1010.298.34
3017.429.529.52
    
Average16.371.701.65
Maximum17.949.819.81

[63] The results of the total crater obliteration calculations (Table 7) show an average obliteration height of 16.37 (+1.65/−1.70) m based on the model ages of emplacement calculated from the Neukum production function [Ivanov, 2001], with an absolute maximum crater obliteration height of 17.94 ± 9.81 m. These heights are clearly less than the observed minimum of 25–60 m for Martian pseudocrater heights [Fagents et al., 2002], and on the very lowest end for the range of heights of littoral cones [Moore and Ault, 1965; Fisher, 1968; Jurado-Chichay et al., 1996] and depths of maar crater [Lorenz, 1973, 1986]. Therefore, it seems probable that if pseudocraters, littoral cones or maar craters had formed at these sites, at least some morphologic evidence would remain in the basin. This argument is strengthened for pseudocraters and littoral cones by the hypothesis that crater obliteration is primarily achieved by infilling of impact craters [Hartmann, 1971; Carr, 1992], which would affect the largely constructional pseudocraters [Thorarinsson, 1953; Greeley and Fagents, 2001] and littoral cones [Moore and Ault, 1965; Fisher, 1968; Jurado-Chichay et al., 1996] much less than it does impact craters. While it is possible that some small-scale features indicative of lava-water interaction were formed at some of these sites and subsequently eroded, it seems unlikely that post-depositional erosion can wholly explain the lack of observed geologic features indicative of lava-water interaction.

4.4.3. Burial of Evidence of Lava-Water Interaction by Younger Lava Flows

[64] One final possible explanation for the lack of observed features related to lava-water interaction is that the resurfacing lava units observed in the open-basin lakes are simply the youngest flows, and they bury previously existing flows that exhibit geologic features caused by lava-water interaction. In the case of lava deltas and tuyas, this seems an unlikely explanation due to the large size of such features [e.g.,Mattox and Mangan, 1997; Smellie, 2007]. However, a lava flow that is approximately tens to hundreds of meters thick could potentially bury any maar craters, littoral cones or fields of pseudocraters that had formed [Moore and Ault, 1965; Fisher, 1968; Lorenz, 1973, 1986; Jurado-Chichay et al., 1996; Fagents et al., 2002]. Such a scenario where littoral cones, maar craters, or pseudocraters formed but were buried with younger lava flows is possible, but testing such a hypothesis is very difficult, due to the fact that the evidence for lava-water interaction would be completely buried if the hypothesis were correct.

[65] Irrespective of this possibility, we interpret the visible volcanic units studied in this work to have been emplaced when the paleolake basins were largely free of both standing and pore water based on the lack of surficial features indicative of lava-water interaction. Although it is possible that these surficial volcanic resurfacing units bury older volcanic resurfacing units that contain evidence for lava-water interaction, none of the observations presented here support such a hypothesis.

4.5. Temporal Overlap of Valley Network Activity and Volcanic Resurfacing

[66] In order to assess the null hypothesis that the populations of basin resurfacing ages and ages of cessation of valley network activity [Fassett and Head, 2008b] are sample subsets of the same distribution of ages, we performed Mann-Whitney U statistical significance tests — a commonly used nonparametric, two-sided statistical significance test for independent populations [e.g.,Wackerly et al., 2002] — on the two populations of ages. Mann-Whitney U tests were performed using both the entire population of basin resurfacing ages and valley network cessation ages, and on a subset of both populations including only those basins with model ages for resurfacing >3.6 Ga and only the 26 oldest valley network cessation ages ofFassett and Head [2008b], which are those thought to be related to open-basin lake activity [Fassett and Head, 2008a, 2008b]. These tests both reject the null hypothesis that the basin resurfacing ages and the valley network cessation ages [Fassett and Head, 2008b] are subset populations drawn from the same distribution of ages at the 95% confidence level.

[67] We therefore conclude that these two populations of ages represent two distinct processes and that their ages do not form part of the same age frequency distribution, although there is some statistical overlap between the emplacement ages for the studied volcanic resurfacing units and the timing of the end of global valley network activity [Fassett and Head, 2008b] (Figure 8). The geologic evidence outlined above for the lack of features indicative of lava-water interaction supports the hypothesis that the basins were largely devoid of water at the time of at least the latest episode of resurfacing. This conclusion suggests that the emplacement of Late Noachian-Hesperian-aged volcanic smooth plains on the floors of these open-basin lakes was not coeval with appreciable amounts of lacustrine or fluvial activity within the paleolake basins, as such a co-occurrence would be likely to have been recorded in the geologic record in one or more of the ways described above. This conclusion supports a climate scenario whereby there was at least some phase of drying at these open-basin lake sites following the end of the valley network activity associated with the individual basins, but prior to at least the surficial episode of volcanic resurfacing.

5. Conclusions

[68] Open-Basin Lake Resurfacing:Detailed analysis of the morphology, physical properties and mineral composition of the resurfacing units on the floors of 30 open-basin lakes indicates that volcanic resurfacing is the process responsible for the emplacement of the smooth floor units. These results are consistent with previous morphologic studies [Goldspiel and Squyres, 1991; Fassett and Head, 2008a; Goudge et al., 2012], as well as detailed in situ analyses at the Gusev crater paleolake [e.g., Squyres et al., 2004], that indicate volcanic resurfacing of open-basin lakes has been an important process in the history of Martian paleolakes.

[69] Additionally, the fact that the basins remain as topographic lows on the surface of Mars [Fassett and Head, 2008a] indicates that it is unlikely these open-basin lakes and their associated valley networks formed via flowing lava, and not flowing water, which has been suggested by previous workers [e.g.,Leverington and Maxwell, 2004; Leverington, 2006]. Such a process would have left ‘high-lava marks’ and evidence for hundreds of meters of subsidence of a solidified volcanic thermal boundary layer, neither of which are observed.

[70] Mineral Signatures of Resurfacing Material:The VNIR reflectance properties of the basin floors measured with CRISM and OMEGA hyperspectral data show absorptions diagnostic of mafic minerals. Varying combinations of olivine, high-calcium pyroxene and low-calcium pyroxene are identified based on crystal field absorptions in the 1–2μm region [King and Ridley, 1987; Adams, 1974; Burns, 1993]. A composition dominated by olivine and high-calcium pyroxene is the most commonly identified among the basins. This mineral assemblage is consistent with the composition of previously identified Hesperian-aged volcanic smooth plains, and distinct from Noachian-aged units, which typically display absorptions indicative of low-calcium pyroxene rich material [Mustard et al., 2005; Baratoux et al., 2007; Rogers and Christensen, 2007; Poulet et al., 2009b; Salvatore et al., 2010; Skok et al., 2010]. Additionally, the correspondence of reflectance properties within individual basins between CRISM and OMEGA data suggests that the resurfacing units are largely compositionally homogeneous.

[71] Resurfacing Ages and the Transition from Valley Network Activity to Smooth Plains:Model ages of emplacement for the volcanic resurfacing units indicate that the resurfacing of the open-basin lakes began near the Noachian-Hesperian boundary, was concentrated in the Hesperian, and continued into the Early Amazonian. Several model ages for basin volcanic resurfacing exhibit statistical overlap with ages of cessation for global valley network activity [Fassett and Head, 2008b]. Despite this overlap, we found no geologic evidence for lava-water interaction at these sites, supporting the interpretation that the basins were largely devoid of both surface water and water in the pore space of surficial sediments at the time of at least the latest episode of volcanic resurfacing. This conclusion implies that the volcanic resurfacing of these specific paleolake basins was not contemporaneous with the fluvial activity that carved the inlet and outlet valley networks and filled the basins with water, and that there must have been some period of desiccation at these sites subsequent to the end of valley network activity but prior to volcanic resurfacing.

Acknowledgments

[72] We are very grateful for the fine work of the NASA MRO project team and the excellent job by the CRISM Science Operations Center (SOC). This work was supported by NASA through a subcontract from the Applied Physics Lab at Johns Hopkins University to JFM. We also gratefully acknowledge support from the Mars Data Analysis Program (MDAP-NNX09A146G) and the NASA-ESA Mars Express High Resolution Camera (HRSC) Team activity (JPL 1237163), both to JWH. We thank David Baratoux, Jacob Bleacher and an anonymous reviewer for detailed and thoughtful reviews and comments that substantially improved the quality of the manuscript. Special thanks are also extended to J. L. Dickson and W. J. Fripp for help processing image and spectral data, and to B. L. Ehlmann and M. R. Smith for helpful discussions.