Delta Deposits on Mars: A Global Perspective

Deltas have long been considered a constraining element to reconstruct the water level of an ancient ocean that may have once occupied the northern lowlands of Mars, and recently this hypothesis started to be challenged. We investigate this hypothesis and present a global inventory of fan‐shaped features showing typical deltaic traits across the entire Martian surface. For each element, we provide descriptive details and classifications based on morphology, location, and relation with characterizing environmental features. In this catalog of 161 deltas, we identified only six having high potential to constrain an oceanic paleoshoreline. Nonetheless, age and location of these candidates display discrepancies with what was previously suggested from independent data sets about shoreline age and locations. Our analyses hence indicate that deltas alone are insufficient to delineate a globally consistent ancient oceanic shoreline, but they have the potential to locally constrain the water level both in space and time.


Introduction
River deltas are landforms generated by deposition of fluid-carried sediments as the flow enters a slower-moving or stagnant water body. The presence of delta deposits on Mars has been thoroughly demonstrated for decades (e.g., Di Achille Nazari-Sharabian et al., 2020;Ori et al., 2000). Large scale mapping (Di Achille & Hynek, 2010;Di Biase et al., 2013;Hauber et al., 2013;Wilson et al., 2021) highlighted the presence of several delta fans mainly located at the dichotomy boundary. These features raised great interest due to their potential as indicators of water availability, habitable environments, and as favorable sedimentary settings for organic matter preservation and have been targeted as landing sites for robotic exploration (e.g., Goudge et al., 2017;Mangold et al., 2020;Summons et al., 2010). Furthermore, deltaic deposits have been investigated as potential evidence that a large ocean covered up to a third of the planet's surface occupying the northern lowlands of Mars (Di Achille Di Biase et al., 2013;Fawdon et al., 2018;Rivera-Hernández & Palucis, 2019). Overall, whether or not a large water body occupied a large portion of the northern hemisphere of Mars is still largely debated. Observations of possible shorelines from orbital images and valley networks terminating near these shorelines have been proposed as evidence supporting the ocean hypothesis (Clifford and Parker., 2001;Hynek et al., 2010;Ivanov et al., 2017;Perron et al., 2007), while arguments against it involved the hypothesis of a prolonged cold and dry climate that could not have sustained an ocean during the Martian past (Wordsworth et al., 2015) or a global paucity of water compared to the required amount to form such a widespread ocean in the first place (Carr & Head, 2015).
In this work, we compile a global map of deltas and propose an updated catalog (https://doi.org/10.6084/ m9.figshare.14564319) that includes previous confirmed observations and new detections. We especially report geomorphological traits, investigate the temporal and spatial relationships among the deposits, and Abstract Deltas have long been considered a constraining element to reconstruct the water level of an ancient ocean that may have once occupied the northern lowlands of Mars, and recently this hypothesis started to be challenged. We investigate this hypothesis and present a global inventory of fan-shaped features showing typical deltaic traits across the entire Martian surface. For each element, we provide descriptive details and classifications based on morphology, location, and relation with characterizing environmental features. In this catalog of 161 deltas, we identified only six having high potential to constrain an oceanic paleoshoreline. Nonetheless, age and location of these candidates display discrepancies with what was previously suggested from independent data sets about shoreline age and locations. Our analyses hence indicate that deltas alone are insufficient to delineate a globally consistent ancient oceanic shoreline, but they have the potential to locally constrain the water level both in space and time.
compare them to the surrounding environment, to assess the relevance and potential of deltas for the reconstruction of past climate conditions on Mars.

Data and Methods
For the morphologic interpretations, we used a Context Camera (CTX; ∼6 m/pixel) global mosaic (Dickson et al., 2018;Malin et al., 2007) while the topographic information was derived from the MOLA (∼460 m/ pixel; Smith et al., 2001;Zuber et al., 1992) colorized elevation model to have an informative global coverage at the highest resolution available. For localized elevation data, where available, we also relied on HRSC Digital Elevation Models providing a vertical resolution of ∼12 m and a regular grid spacing of 50-100 m (Gwinner et al., 2016;Scholten et al., 2005). Observations, mapping, and measurements were made using the QGis software, while for age determination we used the crater size-frequency distributions of impact craters (Hartmann & Neukum, 2001).
We used geomorphological characteristics to perform a global search and mapping of deltas across the Martian surface. Depositional features were interpreted as delta fans when they appeared as fan-shaped deposits fed by a channel, characterized by either a stack of steep foresets of decreasing radius, or by a single steep foreset, either flat-topped or branched (De Villiers et al., 2009, 2013Hauber et al., 2013). Through the identification of these characterizing traits, delta fans were distinguished from alluvial fans visually and through the generation of DEM-derived cross sections. In fact, alluvial fans are defined, differently from deltas, as gently sloping semi-conical landforms associated with the discharge of water and sediments from a confined channel into an unconfined terrain characterized by the absence of a standing body of water (e.g., Moore & Howard, 2005;Morgan et al., 2014). For each observed delta, we collected the following data: (1) location coordinates, (2) a mean elevation value of the deposit's front (i.e., the upper slope break of the frontal scarp).
(3) length and type of the feeding channel (Alemanno et al., 2018), (4) type of basin, which hosts the delta: closed basin = craters with intact rims; lake with outlet = confined low-lying regions with both inlet valleys and an outlet valley (craters with breached rimes, Goudge et al., 2012); and open basin = deltas not contained within craters or deltas formed at the mouth of tributaries opening into outflow channels (Di Achille & Hynek, 2010), (5) flow direction (derived by measuring the direction that connects the apex to the front of the delta), (6) delta morphology categorized with relation to the basin water level according to De Villiers et al., (2013): stepped delta (water level rising); prograding delta (water level constant); and incised/telescoping delta (water level dropping).

Observations
Across the whole planet, we identified a global population of 161 deltas (Figure 1a), among which 68 were not reported in previous catalogs that provided delta locations (Di Achille Head, 2007, 2008;Hauber et al., 2013;Irwin et al., 2015;Palucis et al., 2016;Wilson et al., 2021). The elevation range where these deposits are found spans from a minimum of −4,950 m to a maximum of 2,889 m with a mean elevation value of −1,623 m and a median value of −1,933 m. The overall elevation distribution displays a peak abundance of deltas between about −1,900 and −2,700 m that includes around one third of the whole population ( Figure 1b).
Both deltas, channels, and open-lake basin populations are predominantly abundant between 60°S and 30°N (Alemanno et al., 2018;Fassett & Head, 2008). According to the channel length distribution (Alemanno et al., 2018), deltas are mostly located at the tip of channels shorter than 30 km and quickly diminish in frequency at growing lengths ( Figure 1c). The length of the feeding channel connected to the presence of a delta deposit has been suggested to be: (a) a possible proxy for the duration of the aqueous activity in the channel-delta system and (b) proportional to the age of the delta (Hauber et al., 2013). Short channel lengths are usually associated with pristine morphologies and the absence of mineral alteration suggesting short-lived aqueous processes, specifically, the latter relationship links older deltas near Chryse Planitia (>3 Ga) to longer valleys, while younger deltas are usually fed by shorter valleys (Hauber et al., 2013).
These relationships might be explained by the decreasing availability of liquid water with time in a growing cryosphere, where older deltas formed in a relatively warmer environment when water could be derived over longer time spans or across spatially more extended source regions, both leading to larger and longer feeding valleys (Hauber et al., 2013). The global overview on delta deposits thus highlights that a large DE TOFFOLI ET AL.

10.1029/2021GL094271
3 of 10 portion of the delta population have the potential to be quite young (Hauber et al., 2013). This constitutes a critical aspect that needs to be considered in the future in order to draw further conclusions concerning water availability throughout Martian history.
The previous global delta catalog from Di Achille and Hynek (2010) classified deltas based on the type of basin where the deposit is found. In order to create a comparable updated data set, we used their classification of closed and open basins to distinguish deltas enclosed into craters or not. We additionally highlighted deltas contained in craters with breached rims by separating them into an extra class that we characterized according to the definition of an open basin lake by Goudge et al. (2012). Across our inventory, we found 120 deltas enclosed within endorheic (craters, i.e., closed basins) or exorheic (lake with outlet) basins, while only 41 are hosted within the outflow channel networks or directly within the northern plains (open basins class). On Mars, the delta morphology is primarily affected by the water level behavior. We updated the delta catalog by classifying each deposit based on its morphology: stepped or retrograding deltas formed during water level rise; prograding deltas representing constant water level; and incised deltas, partially destroyed by erosion, indicating water level fall (De Villiers et al., 2013). However, along with water level behavior, sediment supply also plays a role in the definition of delta morphologies. De Villiers et al. (2013) conducted experiments with a fluctuating sediment supply to simulate a real case scenario (e.g., initial collapse of the upstream crater rim and decreasing bank erosion due to feeder channel widening and diminishing flow velocity). On Mars, sediment fluctuation is unknown, and thus, some degree of uncertainty remains especially for: prograding deltas that might also form during water level rise when an adequate sediment supply is present; and in large basins, where water level fluctuation may be very slow compared to the small basin scenario used in the experiments. We assigned each delta to a morphological class, which is linked to the most recent water level change recorded by the deposit and detected 47 stepped deltas, 82 prograding deltas, and 32 incised deltas. By coupling the morphological and the basin classifications, we find that all stepped deltas are found enclosed in endorheic basins with the only exception of five cases where the deltas are located into craters with breached rims. Conversely, all incised deltas are located in open basins or lakes with an outlet counting only four exceptions that can be found in closed basins. We investigated the feeding channel length and elevation distributions for these three morphology-based subgroups but found no significant deviation from the global trends. We also measured the orientation of deltas and found no preferential trends, which is different from what was previously reported by Morgan et al. (2018), who observed for stepped deltas and alluvial fans, a prominent N-S trend that they interpreted as an indicator of control by solar insolation. Since no alluvial fans were included in the present work, it might be hypothesized that the recorded trend was mainly ascribable to alluvial fans instead of deltas.

Discussions
Widespread evidence points to the possible existence of an ancient Martian ocean (e.g., Baker et al., 1991;Carr & Head, 2003;Clifford & Parker, 2001;Head et al., 1999;Parker et al., 1989Parker et al., , 1993, and deltas have been discussed as possible proxies to reconstruct the paleo water level ( We first selected a subpopulation of deltas located at the margin of the northern plains without any evident topographical barrier. Thus, we kept only deposits that might have formed at the edge of the same large ocean-like water body. By performing this downselection, we reduced the population to 29 deltas (Figure 2a). All of the identified deltas are located along the dichotomy boundary between 15°S and 45°N with an elevation range between −3,941 and −933 m (mean = −2,318 m; median = −2,306 m; mode = −2,440 m).
We always refer to the entire elevation range since the topographical distribution of the deposits is never statistically a normal distribution; hence, providing just typical statistical values such as mean and standard deviations lead to the omission of meaningful data (cf., Figure S1). If this subgroup of deltas was associated with the same water body that putatively occupied the northern hemisphere, the topography around the deposits should progressively descend toward the lowlands (mostly in the north) devoid of topographical obstructions. We thus carried out a local-scale analysis of the elevation trend at each delta location investigating the surrounding environment up to a distance from the delta where obstructing morphologies were DE TOFFOLI ET AL.

10.1029/2021GL094271
present. The analysis was concluded when the contour lines showed a consistent progressive topographical smooth decrease toward the north. Through this analysis, we observed that 23 out of 29 deltas are enclosed in topographic depressions. Additionally, the elevation of these basins' rims often coincides with the elevation of the included delta front. Accordingly, there is no evidence that precludes the possibility that such depressions might have acted as local ponds at the time of delta formation. Although it cannot be excluded, DE TOFFOLI ET AL.
10.1029/2021GL094271 5 of 10 no presence of a standing body of water extending across the northern plains is required to explain the existence of these deltas. Only six deltas, hence, appear to have had the potential to be located directly at the margin of a northern ocean. Nevertheless, the widespread presence of deltas across the Martian surface, and thus of hosting water-filled basins, suggests that the water input, both from the surface and the subsurface, must have exceeded evaporation and infiltration leading the basins to fill and be stable enough to allow the generation of deltas at the inlet mouth. Especially, the mere presence of deltas that are not hosted in local topographic depressions but linked to the large basin that might have possibly hosted an ancient ocean suggests that at some point the Martian climatic conditions could have sustained a significantly extended body of water. For this to happen, it might be necessary to invoke the presence of a global water cycle, including a groundwater flow, which might also have reached the surface in low elevated areas, as for example, in the northern lowlands (Fassett & Head, 2008).
Despite the fact that closed basins in proximity to an ocean shoreline may present some subsurface hydrological connection to the ocean and thus display a similar water level, the uncertainties in the selection of this further subgroup of deltas are too large to avoid an arbitrary sampling. Hence, we only investigated the six deltas that, from the observation of the present-day topography, appear not to be enclosed into craters or other topographic depressions, and thus, have higher probabilities to have formed at the margin of an ancient Martian ocean.
(1) Hypanis Valles delta: the Hypanis delta is a complex deposit located at 11°N-45°W that recorded several water-level stages between about −2,500 and −3,000 m on a period spanning from the Late Noachian to the Early Hesperian (for an extensive description refer to Fawdon et al., 2018;Figure 2f) (2) de Vaucouleurs Crater delta: this is a highly degraded deposit located at 15°S-171°E (Figure 2b). The feeding channel cuts a Middle Noachian unit and terminates on a Hesperian and Noachian Transition unit (Parker et al., 2010;Tanaka et al., 2014). The de Vaucouleurs crater rims are still visible on the southwestern side, but the whole crater appears to be pervasively affected, or completely disrupted, by younger chaos terrain formation that degraded also the delta deposit. The delta front today is located at −1,700 m, but due to the chaos formation it cannot be excluded that the front might have been displaced or eroded up to the present-day configuration. The current topography shows no barrier between the delta and the northern plains, nevertheless, it needs to be acknowledged that there is no evidence precluding the possibility that the delta formation took place mainly before the chaos formation, while water was more likely to be contained within the crater rims.
(3) Chryse Planitia delta: This delta (Figure 2c) is located at 36°N-52°W and displays multiple incised fans from an elevation of about −3,100 m to −3,280 m. The delta activity is likely to have lasted during a period between the Late Noachian and the Early Hesperian; in fact, the latest unit that the channel-delta system cut is dated Late Noachian (Tanaka et al., 2014), while the delta tip appears to be cut by a system of grabens, which on an area slightly to the south, are estimated to be around Early Hesperian in age (Nouvel et al., 2006). Nonetheless, age estimates performed on this channel network suggest this terminal area to be 2.43 Ga old (Salese et al., 2016). Due to the presence of faults, it must be taken into account that a certain extent of displacement from its original topographical location might have occurred.
(4) Acidalia Planitia delta: This deposit (Figure 2d) is located 480 km north from the Chryse planitia delta at 44°N-56°W. It is a prograding delta located at −2,203 m on a Middle Noachian unit (Tanaka et al., 2014). (5) Memnonia deltas: at this location two deltas are found at the boundary between the same two units at around 50 km apart (Figure 2e). Besides being spatially extremely close (∼ 5°S-147°W), these two features display similar characteristics, and the delta fronts are located at about (a) −2,525 m and (b) −2,306 m, respectively. One of the two deposits appears to be of outstanding relevance since part of the delta fan is covered by a younger unit that allows us to determine both a minimum and a maximum age for this delta. According to the global Martian map of Tanaka et al. (2014), the unit cut by the feeding channel, thus older than the delta formation, belongs to the Amazonian and Hesperian Volcanic units (AHv) while the one deposited on top of the delta is estimated to be a Late Hesperian transition unit (lHt). For a higher degree of detail, we performed CTX-derived impact crater statistic (chronology functions: Michael, 2013 andHartmann &; production function: Ivanov et al., 2001) on the underlying and overlapping units on 445 and 339 craters on areas of 2.7 × 10 6 km 2 and 1.8 × 10 6 km 2 , respectively, constraining the delta emplacement between 3.6 and 3.3 Ga (Figure 3a). We estimate that this time interval accounts for both the Memnonia deltas. Our observations suggest that the two deltaic deposits correspond to two stages of water level drop within the same basin. The two deltas in this area display channel incision on the delta plain and can be classified as prograding or incised deltas (uncertainties are due to the partial obliteration) thus excluding the hypothesis of a water level rise. This portion of the basin is elongated and deepens moving from SE toward NW and finally opens toward the northern plains without showing any interposed barrier on the present-day topography. The elevations of the two delta fronts approximate two levels of well visible terraces and filled coves along the basin boundary suggesting a progressive water level drop pushing the coastline northwestward until the runoff from the highlands stopped. In fact, no evidence of channels or fluvial activity is recorded on the lHt unit that filled the basin around 3.3 billion years ago (cf., Figure S2).
To test the hypothesis that these deltas once developed at the margin of the same ocean-like water body, we compared the elevation range and formation time (Figure 3b). If an ocean ever occupied the northern regions of Mars, after its optimum, it must have progressively retreated up to a degree of depletion compatible with the present-day situation. Indeed, two subsequent extensive shoreline contacts, Arabia and Deuteronilus, have been observed (Clifford & Parker, 2001;Perron et al., 2007) and their evolution has been analyzed taking into consideration the topographical displacement that likely occurred during the Tharsis emplacement (Citron et al., 2018) and related isostatic rebound (Citron et al., 2021). Shorelines' location and reliability remain, however, controversial. Hypothesized shoreline maps published through time (Carr & Head, 2003;Clifford & Parker, 2001;Ivanov et al., 2017;Parker et al., 1989Parker et al., , 1993Perron et al., 2007;Webb, 2004) show inconsistencies and wide displacements, especially concerning the Arabia shoreline reconstruction; hence, caution needs to be taken in using these data and drawing conclusions about the Martian climate history (Sholes et al., 2021). Overall, the Arabia shoreline is suggested to have formed shortly after the early Tharsis formation around 4 Gyr ago (Citron et al., 2018;Clifford & Parker, 2001), while the younger Deuteronilus shoreline is dated about 3.6 Gyr ago (Ivanov et al., 2017)  Therefore, we did not rely on any specific scenario and used present-day topography for the analyses, since the Martian water drainage is dominated by local or regional slopes rather than by large-scale elevations (Irwin et al., 2011). Nonetheless, whether a standing body of water occupied the northern plains, deltas should have then recorded the shoreline repositioning through time. Thus, expected water level fluctuations might have led to continuous delta formation and erosion up to a final stage of retreat that led to the present-day waterless condition. This latest phase has accordingly the highest chances to be preserved in the geological record, hence expecting younger deltas to be found at progressively lower elevations. However, no age-elevation correlation among our six deltas can be conclusively identified due to the large time spans of potential delta activity (Figure 3b). To further test this hypothesis, we also compared deltas and shoreline positions at each delta site since a certain degree of agreement would be expected. With the only exception of the de Vaucouleurs crater delta, which matches the Arabia shoreline position and timing, the other delta deposits show discrepancies. The Memnonia deltas represent a good example highlighting clear contradiction between the two data sets. We constrained the Memnonia deltas' formation between 3.6 and 3.3 Gyr ago, thus expecting at least partial matching with the coeval or even slightly older Deuteronilus shoreline location. Differently, the Memnonia delta location is better represented by the Arabia shoreline, which is, however, dated early Noachian, while the Deuteronilus shoreline appears to sit several latitudinal degrees northward (Citron et al., 2018 and references therein). Nonetheless, from the analysis of knickpoints in Martian channels in this area, it has been suggested that the Arabia shoreline was reoccupied by the ocean coastline during the late Hesperian due to a significant water level rise that might have occurred between 3.6 and 3.4 Ga (Duran et al., 2019). This interpretation of the ocean evolution shows a good agreement with the observed delta location and age we report, but may collide with the morphology of Memnonia deltas, which seems to indicate water level decrease. Nevertheless, these potential contradictory observations can be reconciled hypothesizing that these deltas represent the result of a further water drop occurring between 3.4 Ga (Duran et al., 2019) and 3.3 Ga (lHt unit age), which necessarily must have followed the water level rise to reach the present-day conditions. Despite that, within the population of candidate delta fans documented in this study, we systematically find stepped deltas (i.e., the delta morphology associated with a rising water level) exclusively associated with closed basins suggesting either unfavorable conditions for the formation of this kind of morphology, a complete global obliteration of these features in association with the northern sea level rise or a progressive continuous drop in the ocean level.

Conclusions
The global inventory of deltas presented herein aims to provide a large-scale overview on the delta deposit distribution and characteristics. We seek to identify the relationship between these features and a hypothesized ancient northern ocean. Accordingly, we focused on those deposits marginally to the basin that would have hosted a northern ocean (29 deltas out of 161). However, only six of them have a high potential of being representative of an oceanic paleo-shoreline. These six deltas are located at different elevations, but no spatial-temporal relationship could be found between their elevations and the time intervals when they had the potential to be active, as would have been expected if they followed a consistently rising or falling water level. Hence, delta information is insufficient to determine a global water level behavior. We looked for similarities in timing and distribution among deltas and shorelines to investigate potential connections. However, we revealed discrepancies between these two data sets making it hard to fully reconcile previously adopted interpretations to the currently analyzed landforms.
Overall, there is no compelling evidence that would make deltas alone a constraining proxy for a global oceanic water level reconstruction on Mars, as previously suggested. Deltas cannot be considered sufficient evidence to identify one shoreline or a consistent shoreline evolution trend, but they may be representative of shoreline fluctuation over time if they develop at the margin of the same body of water. This result challenges the paradigm of the single global water level perspective derived from the previous delta analysis on a planet scale (Di Achille . More modeling and observations are needed to tackle and explain the discrepancy observed in this study between delta deposits and proposed shorelines. A reconciling interpretation might in fact provide further clues and constraints for the Martian climatic evolution.