Dark, windblown (eolian) sand on Mars has produced significant geologic effects throughout Martian history. Although local and regional sand sources have been identified, a primary origin, or genesis, for Martian sand has not been demonstrated. This knowledge gap was recently heightened by the discovery of widespread sand motion, implying breakdown of grains to sub-sand sizes. To address the question of sand genesis, we investigated the source(s) of sand in Aeolis Dorsa (AD), the westernmost Medusae Fossae Formation, using comparisons to sand potentially sourced from multiple regions, each connoting a different sand genesis. Our methods included comparison of (1) AD sand mineralogies with those of possible sand source features, and (2) mapped AD sand deposits and inferred emplacement directions with modeled sand deposit locations and transport pathways. The results point to a time-transgressive unit, interpreted as pyroclastic, as a source of dark sand. High-resolution images of this unit reveal outcrops with dark sand weathering out of lithified bedrock. Given the extent of interpreted pyroclastic deposits on Mars, this sand genesis mechanism is likely widespread today and operated throughout Martian history. Whereas this work identified olivine-rich sand, a range of original pyroclastic lithologies would account for the mineralogic variability of dune fields on Mars. These findings can be tested through analyses of other pyroclastic deposits and potentially by data from the NASA Curiosity rover in nearby Gale crater.

Windblown (or eolian) sand has been a pervasive influence on Mars. Loose and lithified sand deposits and sand-eroded forms are observed globally and locally (Diniega et al., 2021), evidencing wind-driven sand over geologic time.

The primary origin(s)—or genesis—of Martian sand is(are) unknown (Diniega et al., 2021). This knowledge gap in our understanding of Martian source-to-sink sedimentology (Grotzinger and Milliken, 2012; Kocurek and Ewing, 2012) is highlighted by detection of widespread dune movement today (Diniega et al., 2021), during which the more energetic saltating grain impacts are inferred to break down to sub-sand sizes (Sagan et al., 1977). Whereas local and regional sources for sand have been inferred, the primary mechanism(s) that originated the sand-sized grains remains(remain) a key area of inquiry.

Multiple mechanisms might create sand on Mars, including glacial grinding, chemical precipitation, fluvial and lacustrine deposition, and volcanism (Greeley and Iversen, 1985). Volcaniclastic deposits, including products of explosive volcanism and weathering of effusive lavas, are consistent with the low albedo and mafic signature of Martian dunes and have analogs on Earth (Edgett and Lancaster, 1993).

We tested for the source(s) of sand in the Aeolis Dorsa (AD) region (0°–8°S, 147.5°E–156°E; Fig. 1C), the westernmost part of the Medusae Fossae Formation (MFF; Fig. 1A; Greeley and Guest, 1987; Tanaka et al., 2014). The AD region, named for the numerous ridges (dorsa) of inverted fluvial deposits (Burr et al., 2021), shows erosional and depositional eolian landforms that testify to extensive wind transport of sand (Fig. S1 in the Supplemental Material1). Yardangs, formed via erosion by wind-driven sand, are pervasive. Dark sand deposits are visible in some yardang troughs, adjacent to topographic features, and in Aeolis Chaos (Fig. 1C), an ~500-m-deep depression adjacent to the highlands. Geologic mapping (Burr et al., 2021) of the AD region and its regional geologic context enable testing of four source regions for AD sand (Fig. 1B). The Cerberus Plains region, northeast of the AD study area, is composed of effusive lavas crosscut by the Cerberus Fossae, where sand ripples (Roberts et al., 2012) and extensive wind streaks (Greeley and Iversen, 1985) attest to sand production. The Elysium Mons edifice north of AD has effusive lava and explosive sedimentary deposits (Tanaka et al., 2014), both of which could break down to produce epiclastic and pyroclastic sand, respectively. The southern highlands host many dune fields (Tirsch et al., 2011), from which sand could be transported northward into the AD study area. Last, the AD sand could originate from locales within the AD region, where the Aeolis and Zephyria Plana units, coinciding with MFF (Fig. 1C), share the interpretation of a pyroclastic deposit (Tanaka et al., 2014; Burr et al., 2021), generically modeled to include sand-sized sediments (Wilson and Head, 1994). To discover the primary origin of this sand, we evaluated each of these four potential sand source regions as a source of AD sand.

This evaluation involved first mapping sand deposits in Aeolis Dorsa. Grain sizes for these dark deposits were estimated from nighttime infrared measurements, yielding sand sizes (Table S1). We then collected mineralogical data both from AD sand and from sand source features (e.g., craters, talus slopes; Supplemental Dataset 1) in the three other potential sand source regions for comparison, expecting AD sand source(s) to show the greatest mineralogic similarity to AD sand. Data from the Compact Reconnaissance Imaging Spectrometer for Mars (CRISM) were used to make these mineralogical identifications.

We also conducted potential sand flux modeling. Spatial continuity between those time-integrated fluxes, or potential sand transport vectors, indicated possible sand transport pathways, which were compared to sand-mobilizing wind directions inferred from bed-form morphologies. Areas of convergent or near-zero potential sand transport were inferred to indicate sites of sand deposition, and these sites were compared to dark sand locations from mapping. More information is provided in the Supplemental Material.

Mineralogies from the four potential sand source regions showed different compositions (Fig. 2). The potential sand source features in the highlands showed the most variability, including pyroxene and hydrated phases (Table S1; Fig. S2). The spectra from the Cerberus Plains showed a mixed mafic signature. The spectra from the Elysium Mons region are indicative of pyroxene, olivine, and/or glass. Spectra from the dark (non-dusty) sands within the AD Plana units, corresponding to MFF deposits, showed more olivine enrichment, and the aggregate AD sand composition was statistically separable from those of the other three potential source regions (Supplemental Dataset 1).

Potential sand flux modeling showed limited potential sand pathways into the AD study area (Fig. 3). Sand from Elysium Mons and the Cerberus Plains would encounter transverse flows that would block transport into the AD region, whereas flux northward from the southern highlands would largely be captured by the Aeolis Chaos depression.

Analyses within the study area (Fig. 1C) provided additional insight into AD sand sources. For the Aeolis Chaos sands (Fig. S3C), modeling showed northward potential transport over the southern edge of the depression. Within the depression, potential sand flux vectors and wind directions inferred from morphologies both indicated southward flow, substantiated by subtle southward ripple motion (S1 Animation). Sand mineralogies throughout the Aeolis Chaos were consistent with a highlands-like sand composition. On these bases, we interpreted Aeolis Chaos sand to be predominately sourced from the highlands and reworked by winds within the depression.

In eastern Aeolis Planum (Fig. S3D), dark sand is visible among yardangs, suggesting that accelerated winds keep that sand freer of dust. To the west, the brighter land surface and a dusty CRISM spectrum provide evidence of dust, whereas limited potential sand flux and scour marks around one side of knobs (cf. Bishop, 2011) indicate underlying sand (Fig. S1B). Thus, we infer that sand in AD is more extensive than is surficially apparent, but its mineralogy is unclear in these locations due to dust.

On central Zephyria Planum (Fig. S3E), sand locations and potential transport directions imply sourcing from Zephyria Planum itself. The most extensive dark sand is visible within an ~28 × 1.5 km linear trough (Fig. S1C) along a unit contact. Despite high potential sand transport over Zephyria Planum, sand is found only within this trough and as isolated interyardang deposits ~85 km to the northwest (Fig. S3F). Whereas mineral identifications were limited and nondiagnostic, any potential sand transport directions would permit sand only from the planum. Sand could also be sourced from abrasion of the walls of the troughs or yardangs, consistent with its topographic confinement.

In southern Zephyria Planum (Fig. S3G), dark sand is visible within yardang troughs (Fig. S1A). The orientations of the potential sand transport vectors in this location (approximately transverse to the yardangs) likely do not reflect the actual transport direction, as the best available input topographic data for the modeling did not resolve the yardangs. In contrast to sand elsewhere in the study area, the mineralogical identifications in southern Zephyria Planum reflected a uniquely olivine-rich signature (Figs. 1C and 2), implying minimal mixing and/or transport from a local source.

Local sources for the Zephyria Planum sand require a mechanism for sand genesis. The MFF, of which Zephyria Planum is a part, has been interpreted as an ignimbrite (Mandt et al., 2008) potentially sourced from Apollinaris Patera (Kerber et al., 2011), and terrestrial basaltic ignimbrites exhibit sand-rich layers tens of centimeters thick (Fisher et al., 1993; Scarpati et al., 2015; Valentine et al., 2019). Sand layers are also documented for terrestrial pyroclastic deposits that are pumiceous (de Vleeschouwer et al., 2005). Based on these terrestrial examples of endogenous sand layers in pyroclastic deposits, we examined high-resolution images on all plana units (Fig. 1C) for dark sandy layers (see Methods in the Supplemental Material). This examination yielded 31 examples of dark strata with submeter thickness above dark slopes with characteristics of gravitational sand deposition (Figs. 4A4C; Supplemental Dataset 2; see the supplemental Methods). An olivine-rich mineralogy for one such outcrop (Fig. 4D), in contrast to the regional mixed mafic signatures, argues against entrainment of olivine sand grains from the Martian surface during pyroclastic flow. The presence of native olivine as sand-sized phenocrysts in terrestrial basaltic ignimbrites (Clemens et al., 2011; Martí et al., 2017) also supports the inference that the olivine-rich sand is native to the Zephyria Planum bedrock.

Where sand genesis from strata was not detected, such as along the sand-rich unit contact on Zephyria Planum (Fig. S1C), eolian abrasion may be liberating sand that is more homogeneously distributed within the outcrop, as documented for lithics and crystals in the terrestrial Campo Piedra Pomez ignimbrite (de Silva et al., 2013). Other possible explanations for sand in this location are its transport from the topographically higher planum unit to the southwest, consistent with the secondary northward component of the modeled potential sand transport, or formation as a lag deposit, consistent with scattered 1-km-scale dark sand deposits (Fig. S1C) on the planum. Any of these three possibilities implies that sand originated from the Zephyria Planum bedrock. The suggestion of sand distributed within the Zephyria Planum bedrock and the identification of discrete sand strata both show how the light-toned MFF could be a source for dark sand.

The vast MFF, the westernmost extent of which maps to Zephyria Planum bedrock (Fig. 1C), is interpreted as a pyroclastic deposit on the basis of erosional morphology, draping of underlying topography, density, compositional information, and multiple radar data sets (Brož et al., 2021). Given the distribution of pyroclastic deposits on Mars (Kerber et al., 2012; Tanaka et al., 2014), the discovery of a pyroclastic source for the Zephyria Planum sand provides a mechanism for the wide-spread distribution of Martian sand from local sources. Explosive deposits with sand-sized grains are widespread on Mars (Wilson and Head, 1994; Kerber et al., 2012; Brož et al., 2021). Though fine sands will fall very close to the vent (Kerber et al., 2012), the potential for saltative sand transport at moderate wind speeds (Sullivan and Kok, 2017; Andreotti et al., 2021) allows for further distribution. Pyroclastic deposits have been inferred globally (Tanaka et al., 2014), identified regionally (Mandon et al., 2020, and references therein), and detected in situ (McCoy et al., 2008). The variety of mafic minerals in terrestrial basaltic ignimbrites (Clemens et al., 2011; Martí et al., 2017) provides a pyroclastic avenue for formation of the mixed mafic compositions of dune fields on Mars (Tirsch et al., 2011).

This mechanism of sand liberation from a pyroclastic deposit, detected to be operating in the present-day Amazonian climate on Mars, also likely operated over much of Martian history. A decline in the generation of granular material from impact and volcanic processes (Grotzinger and Milliken, 2012) would imply that sand on Mars today is predominately recycled (e.g., Edgett et al., 2020). However, with high grain-impact velocities during the low-density atmospheric conditions of the past ~3.0 b.y. (Kok et al., 2012), sand genesis by eolian abrasion of pyroclastic bedrock could have been substantial throughout that epoch.

Olivine phenocrysts in terrestrial basaltic ignimbrites are modeled to have formed during a low-pressure (<180 MPa) period in a shallow magma chamber (Clemens et al., 2011). Consistent with this low-pressure olivine formation, Martian magma has been modeled to result from near-surface partial melting (Schumacher and Breuer, 2007). The confinement of source rocks spectrally dominated by olivine to the early Noachian and the early Hesperian on Mars suggests a mid- to late Noachian mantle cooling that minimized olivine crystallization (Ody et al., 2013). The emplacement of the MFF was temporally extended, with the early Hesperian–age deposits younging to the east (Tanaka et al., 2014). Sand from other MFF members (Fig. 1A), or more broadly from other explosive deposits (Brož et al., 2021), by providing primary volcanogenic mineralogies of decreasing age, enables testing of the magmatic evolution on Mars.

Sediment maturation, by which composition evolves over time, has broad utility on Earth but uncertain application to Mars (Diniega et al., 2021). Olivine has been suggested to characterize mature volcanogenic sand, based on in situ data from Icelandic sand sheets (Mangold et al., 2011) and the “El Dorado” ripple field in Gusev crater (Sullivan et al., 2008). Results from this work indicate that olivine can also characterize highly immature (locally to regionally sourced) sediments.

The findings of this work are testable in other globally distributed pyroclastic deposits (Tanaka et al., 2014; Mandon et al., 2020, and references therein; Brož et al., 2021). The variably olivine-enriched eolian sands in nearby Gale crater are likely not sourced from the olivine-poor Stimson eolian sandstone (Rampe et al., 2018). Sand on Aeolis Mons most likely comes from the upper mound, a possible MFF outlier (Thomson et al., 2011).

This work was initiated at the University of Tennessee (Knoxville, Tennessee, USA), on the traditional territory of the Tsalagi (Cherokee), the Tsoyaha (Yuchi, Muscogee Creek), and other Native peoples. The work was completed at Northern Arizona University (Flagstaff, Arizona), which sits on homelands sacred to the Hohokam Diné, Hopi, Western Apache, and other Native peoples. We honor the past, present, and future generations of these tribes on their ancestral lands. This work was supported by NASA grant NNX16AL47G through the Mars Data Analysis Program, and its presentation was improved through helpful reviews by Briony Horgan and Jim Zimbelman.

1Supplemental Material. (1) Methods, Figures S1–S3, and captions for additional supplemental information, (2) Animation S1, (3) the ArcGIS geodatabase from this work, (4) CRISM browse products, (5) the ArcGIS database with the locations of grain size derivations, and (6) two datasets. Please visit https://doi.org/10.1130/GEOL.S.19611708 to access the supplemental material, and contact editing@geosociety.org with any questions.
Gold Open Access: This paper is published under the terms of the CC-BY license.