An alluvial fan is a semi-conical form that occurs when water-transported material emerges from an upland onto a lowland. Criteria for the development of alluvial fans include 1) topographic setting where an upland is adjacent to a lowland, 2) sufficient sediment supply in the drainage basin and 3) sufficient fluid to transport that sediment out of the drainage basin (Blair and McPherson, 2009). On Earth, arid zone alluvial fans typically build up as successive sheets and lobes of material are transported associated with high magnitude, short duration release of water from rain storms or rapid snowmelt. Morphological characteristics of the terrestrial alluvial fan system include the drainage basin, feeder channel, apex, incised channel, distributary channels, intersection point, and active depositional lobe. The drainage basin is composed of a branching network of tributary channels. The highest order stream, the feeder channel, leads to the most proximal portion of the fan, the apex. In cases where the feeder channel has cut into a pre-existing fan surface to form an incised channel, the intersection point marks the confluence between the channel floor and fan surface. At this point, flows laterally expand and sediment aggradation occurs downslope in an area termed the active depositional lobe. Entrenched or incised channels in fans are due to changing conditions such as increased precipitation, tectonic uplift or a decrease in sediment supply in the headward regions. Terrestrial alluvial fan research has focused on elucidating the roles of climatic, hydrologic, tectonic and lithologic factors controlling fan development (e.g., Beaty, 1990; Bull, 1977; Bull, 1991; Harvey, 1987; Harvey, 1997; Stainstreet and McCarthy, 1993; Lehmkuhl and Owen, 2024).

Terrestrial alluvial fan aggradation can be broadly defined into two end member processes: 1) sediment-gravity processes wherein large volumes of sedimentary material, including interstitial fluid, if any, are transported downslope under the influence of gravity as a direct result of the reduction of ground stability or resisting forces, and 2) fluid-gravity (water-flow) processes that result from precipitation or snowmelt-fed surface runoff that transports sediment downslope (e.g., Blair and McPherson, 2009; Ventra and Clarke, 2018). Fan morphology differentiates between these two end-member depositional processes. Identifying fan aggradation process provides constraints on the amount of fluid and timescale of fan formation. Debris flows, a sediment-gravity process, emplace large volumes of sediment with relatively little fluid component (47%–77% sediment concentration by volume; Costa, 1988) in short periods of time. In contrast, alluvial fans constructed primarily of fluvial processes (sediment concentration in water flows <20% by volume; hyperconcentrated flows 20%–40%; Costa, 1988) build fans incrementally by small amounts typically associated with low occurrence, high magnitude storms. Godt and Coe (2003) identify three types of initiation processes for debris flows: (1) soil slips and (2) overland flow concentrated in steep bedrock-lined channels (i.e., “firehose”), and 3) mobilization of eroded material from steep, non-vegetated hillslopes by a system. Morphostratigraphy of alluvial fans is typically mapped in aerial images, a technique that is used to link geomprohology to formation mechanisms and climate (e.g., Stock, 2013; Ventra and Clarke, 2018; Lehmkuhl and Römer, W., 2022; Lehmkuhl and Owen, 2024). In this study, we draw especially on recognized martian analog sites with arid to hyperarid climates during fan formation (e.g., Harvey, 1997; Morgan et al., 2014; Ritter et al., 2018; Woor et al., 2023).

Although it has long been recognized that Mars has undergone dramatic climate change (e.g., Sagan et al., 1973), considerable ambiguity and disagreement remains. The uncertainty surrounding Mars’ climate is often framed as a debate between two opposing models: a “warm and wet” climate versus a “cold and wet” climate (Bishop et al., 2018; Fairén, 2010; Head and Marchant, 2014; Wordsworth, 2016). With global temperatures consistently above the freezing point of water, a “warm and wet” climate permits the sustained presence of liquid water (Ramirez and Craddock, 2018). Key evidence supporting this model includes widespread valley networks and lacustrine deposits (Fassett and Head, 2008; Hynek et al., 2010), which suggest long-term surface water activity (Craddock and Howard, 2002). In contrast, the “cold and wet” climate scenario is defined by surface temperatures remaining below freezing for most of Mars’ history, with brief intervals of warming triggered by sporadic events such as meteor impacts or volcanic activity (Ramirez and Craddock, 2018; Segura et al., 2002; 2012; Adams et al., 2025). The crux of the paradox is the inconsistency between theoretical and observational evidence (Hynek, 2016): climate models are better able to replicate cold-wet conditions (e.g., Wordsworth et al., 2015; Wordsworth et al., 2016), whereas several studies based on the geologic evidence support warm-wet conditions beyond the Noachian-Hesperian boundary (∼3.7 Ga) albeit with poor timing constraints and the possibility of regional rather than global conditions (e.g., Davis et al., 2016; Davis et al., 2019; Kite et al., 2017; Morgan et al., 2022; Kite and Conway, 2024; Holo et al., 2021). Recently, Adams et al. (2025) performed modeling that demonstrated that episodic warm, wet conditions could occur into the Hesperian due to H₂ release from water-rock interactions supplemented by transient volcanic activity.

There is widespread recognition that a better handle on the timing and duration of surface aqueous conditions post-Noachian is needed to meaningfully advance understanding of the martian climate evolution (e.g., Kereszturi, 2012; Changela et al., 2021). On Mars there are few opportunities to interrogate the climate conditions during the post-Noachian epoch, due to the paucity of the martian geologic deposits available to assess via orbital remote sensing data. Understanding the paleo-hydrologic conditions and timescales of aqueous activity based on the formation of alluvial fans at sites around the globe contributes to a better understanding of the history of water on Mars and is a critical component to identifying past climate conditions and habitable regions.

Despite evidence of fluvial erosion in the form of valleys and outflow channels, sedimentary deposits associated with such landforms are relatively rare on Mars. The existence of depositional basins was hypothesized from Viking images (e.g., Goldspiel and Squyres, 1991), and several studies noted degradation of landforms such as craters (e.g., Craddock and Howard, 2002). Many of the proposed sites for fan deposits based on knobs and mesas located near the termini of valleys that breach crater walls in Viking images (Cabrol and Grin, 1999; 2001) do not show evidence of deposition in the higher resolution data (e.g., note 22 in Malin and Edgett, 2003; Irwin et al., 2005). However multiple examples of fan-shaped depositional landforms were recognized in martian crater basins with higher resolution images and elevation data in the post-Mars Global Surveyor (MGS) era (e.g., Malin and Edgett, 2003; Moore et al., 2003; Moore and Howard, 2005; Irwin et al., 2005 and references therein; Fassett and Head, 2005; Williams and Malin, 2008; Di Achille and Hynek, 2010a; Di Achille and Hynek, 2010b).

Recently, Wilson et al. (2021) published an inventory of martian fans that built on earlier catalogs (e.g., Moore and Howard, 2005; Kraal et al., 2008). Wilson et al. (2021) conducted a comprehensive, global survey of Mars using high resolution image data (ConTeXt Camera, CTX; 6 m/pix) to map fan-shaped deposits within impact craters. They subdivided these landforms into categories based on their morphological attributes and interpreted origin. Ultimately, they identified 890 alluvial fans located within 206 craters (Figure 1). In their study, alluvial fans were differentiated from fans with ambiguous origin, and scarp-fronted fans that are inferred to be deltaic deposits (e.g., Di Achille et al., 2006; Di Achille et al., 2010a; Di Achille et al., 2010b; Goudge et al., 2017), although post-deposition erosion of an alluvial fan is acknowledged as an alternative explanation. Whilst most (70%) alluvial fans apparently formed contemporaneously with valley network activity, a considerable fraction (30%) of alluvial fans formed well after the era of valley networks spanning from Early Hesperian to Middle Amazonian (Morgan et al., 2022).

Data for this project is publicly released and archived at the Planetary Data System (PDS) Geoscience Node. Data was co-registered using ESRI ArcGIS software, as well as JMARS (Java Mission-planning and Analysis for Remote Sensing) (Christensen et al., 2009). Image datasets examined are the Mars Odyssey Thermal Emission Imaging System (THEMIS, Christensen et al., 2009) images from the infrared (IR, 100 m/pix) and visible (VIS, 18 m/pix) cameras, Mars Express High Resolution Stereo Imager (HRSC, up to 2.3 m/pix; Jaumann et al., 2007), Mars Reconnaissance Orbiter (MRO) Context Imager (CTX, 6 m/pix; Malin et al., 2007; global mosaic by Dickson et al. (2024), MRO High Resolution Imaging Science Experiment (HiRISE, up to 25 cm/pix; McEwen et al., 2007), and Mars Orbiter Camera (MOC, 1.5–12 m/pix; Malin and Edgett, 2003) images.

Topographic analysis used several datasets, including from the Mars Global Surveyor Mars Orbiter Laser Altimeter (MOLA, Zuber et al., 1992) supplemented with high-resolution stereo-derived topography from HRSC (50–75 m/pixel). The HRSC DEM (Jaumann et al., 2007) is produced by the HRSC team and available from the HRSCView website. Higher resolution DEMs were generated using the NASA Ames Stereo Pipeline (Beyer et al., 2018) for CTX following work by Shean et al. (2011). Resulting products with CTX (∼10 m/pixel; ∼10 m vertical precision) are archived at DEMs https://Doi.org/10.5066/P9I1QO1U. Where available and publicly released, we also examined HiRISE (∼1 m/pix. ∼0.2 m vertical precision).

THEMIS-derived quantitative thermal inertia from the global mosaics is used to characterize material properties of fan surfaces (Fergason et al., 2006; Christensen and fergason, 2013). Material properties of the surface are reflected in the thermal inertia, which is sensitive to particle size, porosity and compaction (e.g., Fergason et al., 2006; Edwards et al., 2009). Thermal inertia (TI) is defined as (kρc)^1/2, where k is the bulk thermal conductivity, ρ is the bulk density, and c is the specific heat of the material (Kieffer et al., 1977). Higher thermal inertia values are associated with mechanically strong substrates, such as well-indurated rocks. Lower thermal inertia values correspond to mechanically weak surfaces, such as friable or weakly consolidated rocks and/or dominantly fine-to-medium sand-sized clasts.

Martian alluvial fans are an inherently closed sedimentological and hydrological system (source to sink). As depositional features in close proximity with a sediment source, alluvial fans preserve a record of the prevailing climate during the transport and deposition of sediment within their stratigraphy. There is a well-established history of utilizing alluvial fans to decipher past climatic conditions on Earth (e.g., Havrey and Wells, 1994; Havrey and Wells, 2003; Harvey, 2003; Haug et al., 2010; Salcher et al., 2010). In this study, the martian alluvial fan morphology is characterized through mapping, based on terrestrial analogs.

This study seeks to reconstruct the aqueous history on Mars through detailed morphological analysis of understudied alluvial fans, which may have been former habitable environments. Following a visual inspection of high resolution images for all intracrater alluvial fans in Wilson et al. (2021), twenty-seven sites (13%) were identified for further detailed study (Figure 1; Supplementary Table S1). Selected sites had noteworthy attributes pertinent to the study objectives with possibly insightful aspects related to flow process and/or stratigraphic context. High resolution images and digital elevation models (DEMs) from the CTX and HiRISE instruments were used to differentiate sediment-gravity flow deposits from fluid-gravity deposits, as well as the superposition relationships between deposits, to develop an evolutionary sequence.

Maximum age for the alluvial fans is based on the crater-dated geologic unit that underlies the fan-hosting crater in the Tanaka et al. (2014) global geologic map of Mars, as reported in Wilson et al. (2021) (Supplementary Table S1). Six age groups are as follows: 1) Amazonian-to Hesperian-age impact craters (impact unit), 2) Hesperian (includes Early and Late Hesperian), 3) Hesperian to Noachian (transition unit), and 4) Noachian (includes Early, Middle, and Late Noachian). In some cases, additional information is available that refines the alluvial fan age, such as crater counts on the fan surface (Supplementary Table S2). Although the presence of embedded craters in alluvial fan stratigraphy has been observed (Kite et al., 2017), none of the selected sites have reported embedded craters.

One of the greatest concentrations of alluvial fans on Mars is in southwestern Tyrrhenna Terra (Moore and Howard, 2003). Anderson et al. (2022) reported on variations in fan morphology in five intracrater basins. These findings are expanded here to discuss details of fan building process by illustrating deposit morphology and superposition relationships. Finally, an example from the northern hemisphere is presented with similar attributes, and a noteworthy difference.

Figure 2 illustrates the sharp transition from a chute-dominated upper fan, to ridges and lobes downslope. On the upper fan, linear chutes are ∼250 m wide and radiate from the fan apex, reflecting shifting flow paths. Deposits at the termination of chutes have a lobate form, strongly suggesting debris flow processes. Subjacent to these lobes are radiating ridges. Ridges are 50–100 m wide and exhibit subtle relief of up to a few meters. The ridges exhibit low sinuosity and branch downslope consistent with the planimetric pattern of distributary channels. The ridges are interpreted as the erosional remnants of fluvial flows (e.g., inverted channels). The clear superposition between upper lobes and distal ridges is shown in Figure 3, marking an unconformity. This fan deposit pattern is observed in several (∼40%) alluvial fans regionally (Anderson et al., 2022), and observed in the global survey of this study although not tabulated. In a second example, Figure 2E shows the fan feeder channel transitions to a chute with three stacked lobe deposits.

For fans with composite deposits there is often a sharp thermal boundary between the upper and lower fan (Figures 2B, D). In Figures 2C, D; Supplementary Figure S1, the thermal inertia values for the ridged portion of the fan is elevated relative to the upper fan with lobes (difference of ∼50 Jm^-2K⁻¹ s^-1/2). This trend is interpreted to reflect a combination of lithified material and erosional removal of fine grain particles.

Many alluvial fans on Mars have ridges in their distal portions. Figure 4 illustrates an example in northern Arabia Terra. The terraced fan deposit superposes thin ridges radiating downslope (Figure 4B). In contrast to the fans in Tyrrhenna Terra, there are stacked ridges on the crater floor that record multiple fluvial events (Figure 4C). However, the association of these crater floor ridges to the alluvial fans is unclear, as they are spatially separated landforms.

Through the systematic review of the larger alluvial fan sites, a few exemplary locations demonstrate a surprising link between surface texture and flow process. In a Tempe Terra crater, Figure 5 shows the continuity relationships of the deposit from alcove mouth to the crater floor where the pitted deposit terminates in an escarpment. The deposit drapes over a topographic bench, an indication of the material properties of the flow. High viscosity and material strength are required for the flow to coat the escarpment. A sedimentary origin is preferred due to the origin of the flow from a crater rim alcove, and the lack of volcanic source. Some pits on the fan surface are likely impact craters, such as the double pit in Figure 5B. However, the spatial distribution of the circular pits is problematic for an impact crater interpretation for the pitted texture of the fan. The high density and size configuration (concentrated in areas where the flow bends around obstacles) is atypical for a cratered terrain. The observational evidence supports the interpretation of a mudflow origin for pitted terrain texture.

In Xanthe Terra, a pitted texture is present on a thin fan deposit (Figure 6). The flow extends nearly 30 km from the two source alcoves on the crater rim. The flow appears to be a contiguous deposit, with no identifiable lobes or superposition relationships. This suggests that a single triggering event mobilized sediment from both alcoves simultaneously. The pitted texture is associated with the distal portions of the deposit. The crenulated margin (Figure 6B) indicates high material strength of the flow deposit.

Variations in alluvial fan morphology may represent different periods of fan formation. In some cases, alluvial fans transition from chutes and lobes to ridges (channels in inverted relief) at their distal ends (e.g., Figure 2; Anderson et al., 2022). This trend was previously interpreted as preferential erosion exploiting downfan fining of deposits and leaving the upper fan protected by coarser deposits (e.g., a single-phase of fan deposition, Figure 7A; DiBiase et al., 2013). Alternatively, composite fan morphology could result from burial of an older, denuded fan surface (inverted channels across entire fan) by a later stage of alluvial fan activity (chutes and lobes). This sequential fan formation is the multi-phase model (Figure 7B), the favored interpretation in this work. The superposition relationships reported here (Section 4.1), signifies an unconformity (depositional hiatus associated with deflation) that reflects a time sequence change in depositional process: fluvial flows to debris flow.

Additional evidence in support of the multi-phase model is the distinct thermal boundary between fan morphology types (Figure 2; Supplementary Figure S1). The observed sharp thermal contact correlated to fan morphologic types is consistent with material property variations associated with fluvial and debris-flow deposits. In contrast, the single-phase model should produce a more subtle thermal signature. Although a purely fluvially-formed alluvial fan would be expected to have a decreasing thermal inertia signature downfan due to the downslope fining of clasts, the particle size effects would be less pronounced under the single-phase formation hypothesis due to removal of sand-size particles during deflation. With differential erosion removing distal fines, the thermal inertia signature across the fan surface would be uniform or gradational downfan, reflecting the transition from coarser material armoring the proximal fan and indurated materials (corresponding to inverted channels) on the distal fan. (However, it must be acknowledged that interpretation of thermal inertia data is non-unique. Furthermore, Mondro et al. (2024) demonstrate that almost all martian fans (99%) exhibit no discernible spatial pattern in surface thermal inertia, and the homogenous values are consistent with a fines-dominated surface due to sand from depositional, widespread mantling and/or in situ weathering.)

The commonality of this specific type of alluvial fan superposition relationship (composite alluvial fan) in multiple crater basins may reflect changes in water availability. The pervasiveness (but not ubiquitousness) of this unconformity across the globe suggests that climate conditions may be responsible. Fluvial flows may be generated via precipitation patterns. Lesser water volumes are needed for debris flows, which could be seasonally triggered by meltwater. Importantly, both flow processes require clement climate conditions for surface water flow.

Commonly circular depressions on Mars are interpreted almost exclusively as impact craters. However, puzzling superposition relationships have been noted in areas where the older buried terrain exhibits a significantly lower crater density than the younger, topmost strata (Malin and Edgett, 2000; for an illustration see Figure 3 in Edgett and Sarkar, 2021). One explanation for this paradox is different material properties of the terrain, such that the impact crater record is not preserved (e.g., erased by erosion). In this paper, examples of fan deposits with pitted terrain (Figures 5, 6) are evidence that flow processes can be involved in the generation of circular depressions, and this alternative explanation could be considered in evaluating cratered landscapes.

Subsurface volatile release is a mechanism to generate pitted terrain. Migration of buoyant materials through unconsolidated, fluid-saturated layers produces soft-sediment deformation structures. For example, gas bubble migrations through mud deposits at Lake Powell, Utah form cavities tens of meters in diameter (Sherrod et al., 2016; Miller et al., 2018), and larger structures (hundreds of meters) are documented in marine settings (e.g., Cole et al., 2000; Barry et al., 2012). In modern fluvial and pond settings, pressurized water exits vertically through sand and/or clay deposits to create decimeter-scale pits or pockmarks (e.g., Holzer and Clark, 1993; Guhman and Pederson, 1992; Draganits and Janda, 2003).

These examples are noteworthy in the configuration of the cavities with space between pits or pit clusters, a spatial distribution that differs from periglacial space-filling pitted terrain. On Mars, sublimation ice loss driven pitted pattern (e.g., Ramsdale et al., 2018) are larger scale surface textures that typically blanket the landscape and are distinct from the pits shown here that are confined to the fan surface (Figures 5, 6).

A related mechanism to soft-sediment hemispherical cavities described above is air escape from a sedimentary flow that could generate pitted terrain. Apparently not previously documented, this paper presents a terrestrial analog for the martian pitted textures on a mudflow deposit in the Atacama Desert of northern Chile. Although the Chilean pits are much smaller in scale, they have many morphological and distribution attributes in common with the martian pits illustrated in Figures 5, 6.

Along the banks of the Quebrada de Guatacondo, centimeter-scale circular-to-elliptical pits are present in a mudflow deposit on the canyon walls but absent in the channel (bypass flow; Figure 8). In the Quebrada de Guatacondo mudflow deposit, hemispherical cavities decrease in size upslope (Figure 8B). Measurements of a representative maximum pit show the size change: near the channel margin (18 × 16 × 6 cm) to high on the slope (6.5 × 5 × 2 cm). The largest pits are where flow is thickest, near break in slope by the channel. Cavities are cross-cut by polygonal cracks indicating they formed prior to desiccation of the deposit (Figure 8B). Some pits have raised rims (Figure 8D).

The mudflow was viscous and turbulent, resulting in a deposit that is draped on the steep canyon walls (superelevation). Morgan et al. (2014) estimated 1–2 m flow thickness based on run-up distance. The vigorous flow trapped air that escaped prior to the flow drying out. If the mechanism is the same for martian pitted terrain (e.g., highly viscous mudflow with air bubble release), it places constraints on the climate conditions present during the flow.

Recently, several authors have interpreted ridges on crater floors as due to glacio-fluvial processes (e.g., Zhang et al., 2023; Grau Galofre et al., 2024). One concentration is on nineteen crater floors in southeastern Terra Sabea (Gullikson et al., 2023) where dominantly radial ridges and a few circumferential ridges are associated with proglacial paleolakes (Boatwright and Head, 2021; Boatwright and Head, 2022). In the northern hemisphere, a few examples of radial ridges linked to ice-related landforms such as viscous flow features or moraines (Gallagher and Balme, 2015; Butcher et al., 2017; Butcher et al., 2021). The present study expands the locations where this landform type is recognized. Figure 4 illustrates potential examples of glacio-fluvial landforms (Williams et al., 2024). The circumferential ridges in Figure 4C (yellow arrows) are visually similar to the flat-crested ridges in Terra Sabea (e.g., see Figure 18 in Boatwright and Head, 2022). In addition, Boatwright and Head (2022) suggested that ridge stacking in Terra Sabea could be inverted proglacial fluvial channels associated with episodic generation of glacial meltwater, a possible explanation for the configuration in Figure 4C.

Wilson et al. (2016) acknowledged the ambiguity in interpreting the fan-emanating radial ridges in Figure 4B as either candidate eskers or inverted channel (e.g., Figure 7). Distinguishing an esker origin from inverted channel is challenging (Boatwright and Head, 2022), especially as both landforms could co-exist as illustrated in Chukhung crater by Butcher et al. (2021). Although an exclusively inverted channel origin for the ridges shown in Figure 4 is a plausible alternative interpretation, a lava-capped inverted channel is specifically disfavored due to lack of supporting contextual evidence and the relative uncommon occurrence of such features. Volcanic rock-capped inverted channels are relatively rare on Earth constituting <25% of the inverted channel catalog, and these examples do not exhibit stacked ridges as shown in Figure 4C (Zaki et al., 2021). Likewise, few martian inverted channels are inferred to be lava capped. One example is Mangala Valles (Keske et al., 2015) and is distinguished from the features in Figure 4 in several important ways: a direct link to a volcanic source region, surface textures consistent with lava flows and the significantly larger scale of the landform (valley filling).

The proposition of combined glacier and fluvial landforms in the ridges of this northern Arabia Terra crater is bolstered by regional evidence for glaciation. There are ice-rich deposits in nearby Deuteronilus Mensae (Berman et al., 2015) with morphologies that extend to this crater (Williams et al., 2024). Also, the recognition of mid-latitude fresh shallow valleys, including ones adjacent in the so-called “Heart Lake” system, formed from snowmelt-based hydrology that occurred during the Hesperian-Amazonian transition (Wilson et al., 2016). Consistent with that timing, Berman and Williams (2025) report crater counts on the ejecta blanket yield a crater formation age of ∼3.4 Gyr in the middle Hesperian, but note considerably younger crater retention ages for the fan deposits due to modification, ∼110 and 370 Ma in the early Amazonian. Further study of this complex site is needed to understand the association of alluvial fans and the ridges in terms of relative timing of fan aggradation and potential glaciation.