Martian fan deposits: Insights on depositional processes and origin from mass balance survey

Earth and Planetary Science Letters 533 (2020) 116049

Contents lists available at ScienceDirect

Earth and Planetary Science Letters www.elsevier.com/locate/epsl

Martian fan deposits: Insights on depositional processes and origin from mass balance survey David A. Vaz a,b,∗ , Gaetano Di Achille a , Brian M. Hynek c,d , William Nelson c,d , Rebecca M.E. Williams e a

National Institute for Astrophysics (INAF), Astronomical Observatory of Abruzzo, Via Mentore Maggini, 64100, Teramo TE, Italy Centre for Earth and Space Research of the University of Coimbra, Observatório Astronómico da Universidade de Coimbra, Almas de Freire, 3040-004, Coimbra, Portugal c Laboratory for Atmospheric and Space Physics, University of Colorado Boulder, 1234 Innovation Drive, Boulder, CO 80303, USA d Department of Geological Sciences, University of Colorado Boulder, Campus Box 600 UCB, Boulder, CO 80303, USA e Planetary Science Institute, Tucson, AZ, USA b

a r t i c l e

i n f o

Article history: Received 26 October 2018 Received in revised form 18 December 2019 Accepted 21 December 2019 Available online xxxx Editor: W.B. McKinnon Keywords: delta fluvial fan mass balance Mars

a b s t r a c t Fan deposits located at the mouths of Martian valleys have been interpreted as indicators of wet conditions during Mars history. However, the processes, time and amount of water needed to carve the valleys and form the fans are still debated. Here we present a detailed morphometric and mass balance analysis of valleys and fan-shaped deposits using high resolution topography, which provides new insights into past depositional environments. Based on morphometric and volumetric measurements, we found that several Martian fans previously interpreted as deltas can be categorized into two main types. Type I fans are more numerous, relatively well preserved, are associated with smaller and less mature drainage networks and were deposited on steeper gradients. The balance between eroded and deposited materials indicates that this set of fans might have been formed mainly in subaerial settings by depositional processes other than typical deltaic sedimentation (e.g. alluvial, glacial, mass-wasting in general), since a significant loss of sediment during deposition did not occur (we obtained a volume ratio of eroded vs. deposited materials of approximately 1). Therefore, we propose that the identified Type I fans were not deposited in prevailing subaqueous deltaic settings, but mainly in subaerial conditions with perhaps sporadic presence of ephemeral bodies of standing water within the basins. In contrast, Type II fans are less abundant, highly eroded, were deposited on flatter areas and formed downstream of more mature drainage networks with longer, deeper and wider valleys. The mass balance for this set of fans clearly shows that considerable amounts of sediment were not retained in the fans, implying large offshore transport of sediment during their formation (eroded vs. deposited volume ratios between 3 and 10 are probable). This evidence supports a fluvio-deltaic origin for this class of fans, indicating the existence of paleolakes over substantial periods of time. Based on the collected morphometric measurements and mass balance modeling we conclude that only a small percentage of the fans (Type II fans correspond to ∼1/3 of the sampled areas) are consistent with the occurrence of favorable and durable conditions for life, i.e. long-lived integrated fluvial, deltaic and lacustrine environments. Whereas, the majority of fans might have been formed in subaerial settings with significant contributions of alternative processes besides fluvial transport and deposition thus not necessarily requiring the occurrence of extended epochs of clement climatic conditions. © 2020 Elsevier B.V. All rights reserved.

1. Introduction

*

Corresponding author at: National Institute for Astrophysics (INAF), Astronomical Observatory of Abruzzo, Via Mentore Maggini, 64100, Teramo TE, Italy. E-mail address: [email protected] (D.A. Vaz). https://doi.org/10.1016/j.epsl.2019.116049 0012-821X/© 2020 Elsevier B.V. All rights reserved.

Fan-shaped deposits of putative fluvio-lacustrine origin have been identified on Mars (e.g. Cabrol and Grin, 1999; Di Achille and Hynek, 2010a,b; Malin and Edgett, 2003), but their age, the time needed for their deposition and their environmental significance

2

D.A. Vaz et al. / Earth and Planetary Science Letters 533 (2020) 116049

Fig. 1. Data used for the global mass-balance survey and block diagrams of the two generic depositional settings evaluated in this study. a) All the regions are located at mid latitudes and most are close the dichotomy boundary; 13 of the 75 measurements have duplicated volume estimates made with HRSC and CTX data, while coincident HiRISE and CTX coverage is only available for one of the sites. b) Fan mainly deposited in a subaerial setting by alluvial and/or any other mass-wasting process (e.g. landslides, glacial flows, lahar, etc.), resulting on a net ratio of valley volume (VV ) by fan volume (VF ) equal to 1. c) Delta/fan-delta forming in a body of standing water and characterized by a considerable amount of sediment off-shore dispersion, caused by sediment suspension and flocculation (forming plumes) and forming distal deposits farther away from the main apron (Vd represents the volume of the distal deposits); in this case the VV /VF ratio will be higher than 1.

are still debated (e.g. Di Achille et al., 2009, 2006; Di Achille and Vaz, 2017; Hauber et al., 2009, 2013; Kleinhans et al., 2010; Kraal et al., 2008b). Surface runoff and/or groundwater sapping were recognized (Hauber et al., 2009; Irwin et al., 2005; Salese et al., 2019) as the most likely processes that formed the deep valleys and related fan-shaped deposits at their mouths. Alluvial fans have been also identified on Mars by several authors and included in global catalogs (e.g. Kraal et al., 2008a; Moore and Howard, 2005; Morgan et al., 2018). In this study, we specifically focus on putative fan-deltas by referring to the global catalog from Di Achille and Hynek (2010a; 2010b). Indeed, the latter study considered all the possible depositional settings (i.e. alluvial, volcanic, glacial, deltaic) for the cataloged fan-shaped deposits, concluding that the included deposits were most likely of deltaic origin. Moreover, the latter database has been used by many global studies about Martian fan-deltas and is recognized by the research community as a catalog of possible fan-deltas rather than alluvial fans (e.g. Halevy and Head III, 2014; Hynek et al., 2010; Wordsworth, 2016). Therefore, since our survey is primarily based on the Di Achille and Hynek (2010b) database we assume that the deposits analyzed for this study are typical examples of Martian fan-deltas. The aim of this paper is to better understand the sedimentary conditions under which previously interpreted fan-deltas might have formed. Indeed, Malin and Edgett (2003) cautioned about the identification of putative river deltas for deposits showing comparable volume of material with respect to the valley that formed them. According to the authors, these deposits cannot be uniquely interpreted as river deltas since they might have formed also in subaerial settings by alluvial processes, mass wasting and/or gravity driven flows. Particularly, if the deposits were formed by limited fluvial activity and mainly by mass-wasting (landslides, glacial flows, lahar, etc.), the volume of transported sediments should be comparable to that eroded within the valley since the material would concentrate within a sediment apron right at the valley’s

mouth where the energy drops (Fig. 1b). In this case the ratio of valley volume vs. fan volume should be close to unity. That is for instance the case of terrestrial alluvial fans, which have mass balanced ratios between 1.3 and 3.7 (Jolivet et al., 2014; Kiefer et al., 1997; Oguchi, 1998). Contrarily, if the deposits were formed in a deltaic environment, they should be farther spread within the hosting basins and not confined at the river mouth since a significant portion of the sediment will be transported and dispersed farther from the river mouth (Fig. 1c). The base notion is that deltaic processes form proximal deposits (the substantial volume of sediments within the main delta close to the valley mouth) and distal deposits, which can be draped within the basin farther from the main deposits through suspension and flocculation (Wright and Nittrouer, 1995). Buoyant sediment plumes on terrestrial deltas can extend hundreds of kilometers from the river mouth (Wright and Nittrouer, 1995), resulting in the bypass of a large proportion of the suspended sediments (e.g. 70% of the suspended sediments delivered by the northern California rivers bypasses the continental shelf; Sommerfield et al., 2007). Additionally, results from numerical sedimentary modeling under Mars’ reduced-gravity conditions (e.g. Hoke et al., 2014) suggest that the volume of the deposits formed by fluvio-lacustrine deposition could be up to eight times bigger than the respective valley volume, due to the offshore dispersion of finer materials within the basins. In this work, we conducted a morphometric analysis of previously interpreted Martian fan-deltas to quantitatively test the valley/fan volume ratio, and used it as an indicator to discriminate whether they formed mainly under subaerial or subaqueous depositional settings. We performed a detailed mass balance of the valley-fan systems, taking into account other erosional/depositional factors that may have affected the measured volume, such as aeolian valley infill, porosity and post-depositional removal of sediments.

D.A. Vaz et al. / Earth and Planetary Science Letters 533 (2020) 116049

1.1. Background Previous studies found that Martian depositional fans exhibit very different morphologies: 1) extensive deposits composed of meandering distributary channels (Malin and Edgett, 2003; Pondrelli et al., 2011); 2) Gilbert-like fan-deltas (Di Achille et al., 2009, 2007; Ori et al., 2000); 3) stair-stepped fans (Di Achille et al., 2006; Grindrod et al., 2018); 4) complex fans (Ori et al., 2000); and 5) fans denoting some degree of association with periglacial or glacial processes (Di Achille and Hynek, 2010b). A recent morphological global survey (Morgan et al., 2018) suggests the existence of three types of fans: type 1 – alluvial fans; type 2 – stepped deltas; type 3: deltas (sub-divided into branched and un-branched deltas). The evident large morphological variability suggests the existence of different depositional environments (e.g. Di Achille and Vaz, 2017), nevertheless an objective global classification has not yet been completed. Since these morphological differences cannot be univocally categorized, being often based on subjective parameters, we focus our analysis on quantitative aspects, such as the morphometry and volumetric assessment of the valleys and fans. Adler et al. (2019, Table 1) defined the key stratigraphic elements needed to discriminate between deltaic and non-deltaic fan deposits on Mars (e.g. the identification of boulders, levees, topset/foreset slope changes, etc.). However, most of these elements are relatively small scale stratigraphic structures, whose identification is highly dependent on favorable strata exposure and on the availability of high-resolution images and topography (e.g. HiRISE data). In order to overcome such limitations we try to discriminate non-deltaic from deltaic deposits using a mass balance analysis. We consider this an additional method, which may complement future detailed stratigraphic studies. Previous mass balance assessments found that, in certain cases, the volume of deposited materials approximates the volume that was excavated to form the feeding valleys: i.e. the ratio of valley vs. fan volume approximates unity (Grindrod et al., 2018; Palucis et al., 2016). A different result was obtained for the Xanthe Terra region, where valley volumes were reported to be ∼3 times higher than the volume of sediments captured/deposited at the fans (Hauber et al., 2009). Malin and Edgett (2003) proposed that the Eberswalde fan formed by persistent fluvio-deltaic activity, reporting a volume ratio of ∼4. In contrast, other fans with ratios ∼1 where interpreted as forming by non-fluvial processes (mass movement processes were invoked, see Note 22 in Malin and Edgett, 2003). 2. Data and methodologies In the following subsections we describe the methods and data used to perform our analysis. Finally, in subsection 2.4 we describe the mass balance model we use to check the consistency of the inferences on the depositional environments obtained with the volume ratio in terms of age and post-depositional evolution of the fan deposits. 2.1. Remote sensing data Digital terrain models (DTMs) and orthoimages derived from Mars Reconnaissance Orbiter Context Camera (CTX) imagery (Malin et al., 2007) are primarily used to map the extent of the valley networks and associated fan deposits. These datasets are obtained with NASA Ames Stereo Pipeline (ASP) stereogrammetry software (Moratto et al., 2010), after performing a bundle adjustment with USGS ISIS (Integrated Software for Imagers and Spectrometers). Ground control points are derived from daytime infrared THEMIS (Thermal Emission Imaging System) geodetically controlled mosaics (Fergason et al., 2013).

3

Large shadowed areas and correlation errors can result in DTM artifacts. To address this problem we performed a final inspection on QGIS, in order to identify and mask these areas. Long wavelength discrepancies between the CTX DTM and the MOLA (Mars Orbiter Laser Altimeter) elevation profiles are minimized by performing a polynomial fit to the residual elevations. The output spatial resolution is 20 m/pixel for all CTX DTMs and ∼6 m/pixel for the orthoimages. Using the MOLA tracks as reference, we estimate an average vertical discrepancy of −1.2 ± 36.8 m (throughout the text standard deviation values appear after the ± sign) for several test cases (Supplementary Table 1). Comparable results were reported for other areas on Mars (Kim and Muller, 2009; Vaz et al., 2014). We also use the averaged triangulation error (obtained for each DTM pixel with ASP) to assess the uncertainty of volume measurements. To survey the fan deposits we produced a total of 41 CTX DTMs, some consisting of mosaics that combine several stereopairs covering the possible maximum extent of the incised valleys and depositional fans (Supplementary Table 2). Where CTX data suitable for stereogrammetry was not available, we use the High Resolution Stereo Camera (HRSC, Neukum et al., 2004) DTMs (50-100 m/pix) and orthoimages (12.5 m/pixel) available in the Planetary Data System. HiRISE DTMs (1 m/pix) are used for two fans. There is a noticeable difference in spatial resolution and overall quality between the HiRISE, CTX and HRSC datasets (Fig. 1 provides a global view of the datasets we used). For this reason, we compared measured volumes for a set of fans and valleys, using the same semi-automatic measurement technique (Section 3.1). This assessment aims to evaluate if a direct integration of these data sources is meaningful. 2.2. Valley volume measurement For each study site, we use QGIS software and the orthoimages to map a polygon that covers the full extent of the valley network that formed the fan deposit (Fig. 2a). Then, we used the DTM to interpolate a pre-incision surface, using the elevation data located on the outer edge of the polygon regions and a natural neighbor interpolation (Fig. 2d-e). The total volume of eroded material was obtained by Riemann integration of the difference between the pre-incision and present day surfaces (Fig. 2f). We evaluated the volume uncertainty numerically by propagating and integrating the per-pixel triangulation errors derived from ASP (Supplementary Fig. 1). For HRSC and HiRISE data we assumed a constant vertical uncertainty for each pixel that equals half the DTM spatial resolution. Valley network centerlines were also mapped from the images (Fig. 2c). We use them to measure the total length of the valleys as well as to derive other morphometric parameters. Cross section profiles orthogonal to the center lines in conjunction with the marker polygons and DTMs were employed to automatically measure the width of the valley (Fig. 2c), its depth and shape (the spacing between the cross sections corresponds to the DTM pixel size). We evaluated the shape of the valleys by computing the correlation coefficients between the valley cross section and an ideal (with the same maximum depth) V-shaped or rectangular section. These coefficients are used to quantify if the valleys can be better described as U or V-shaped. Summary statistics of the parameters computed for each profile are attached to each valley polygon. Longer valley networks are not totally covered by the high resolution DTMs we used in this work. For those eighteen cases we performed the same complete morphometric analysis, which was previously described, only for the part of the valleys that are located closer to the fans. Then, we extrapolate the eroded volumes for the remaining part of the valley networks assuming two different scenarios: 1) assuming that the non-sampled upstream

4

D.A. Vaz et al. / Earth and Planetary Science Letters 533 (2020) 116049

Fig. 2. Illustration of the techniques used to measure the fan and valley morphometric parameters. a) Valley-fan system located in Camichel crater (2.64◦ N, −51.66◦ E); mapped valley and fan outlines overlaid on the CTX orthoimage. b) CTX DTM. c) Orthogonal profiles derived from the mapped “thalweg” that are used to perform the morphometric characterization of the valleys. d) Elevation of valley shoulders used to interpolate the pre-incision surface. e) Interpolated pre-incision surface. f) Depth of eroded materials, the total eroded volume is obtained by the integration of this dataset. g) Fitted crater rim and concentric profiles used to sample the topography outside of the fan (delta topography obtained with SEDFLUX). h) Deposit base topography. i) Thickness of the deposited materials, the integration of this data produces the total fan volume. j) crater and fan located in the mouth of Tyras Vallis (8.38◦ N, −49.66◦ E). e) Elevations used for the interpolation of the base surface. f) Thickness of the deposited fan, in this case we estimate a total volume of 9.41 ± 0.04 km3 using CTX data. (For interpretation of the colors in the figure(s), the reader is referred to the web version of this article.)

sections maintain the same width and depth (maximum volume estimate), 2) assuming that the width and depth decrease upstream (minimum volume estimate). We compute the maximum volume estimate according to V max = V s + ( W md × D md × L ext ), where V s corresponds to the sampled volume (measured from the DTM) and W md and D md represent the median valley width and depth respectively. We obtained the length of the extrapolated section according to L ext = L t × (1 − L s / L t ), where L s and L t correspond to the sampled length and total length of the valley network. The minimum estimate of eroded volume is obtained according to V min = V s + (( W m − W std ) × ( D m − D std ) × L ext ), where the sampled mean (W m , D m ) and standard deviation (W std , D std ) of the widths and depths are used as proxy to the upstream decrease of the valley dimensions. We adopted the average of the two extrapolation scenarios to compute the total eroded volume, while the volume uncertainty is obtained by summing the range of the two extrapolation scenarios to the uncertainty derived from the DTMs.

Valleys generally form continuous sharp edged landforms which are easily mapped from the imagery and DTMs. However, in some cases the valley network that is connected to the fans seems to dissect older valley systems forming knickpoints. For the Coprates Catena fan, the existence of knickpoints inside the valleys was interpreted as evidence of the existence of more than one dissection phase, and that it is the later erosive phase that forms the main body of the depositional fans (Grindrod et al., 2018). However, we cannot be sure if that same interpretation is valid for all the valleys we mapped. Similar knickpoint morphologies could have been formed simply by changes in bedrock strength or base level changes, causing a headward propagating knickpoint. In order to follow uniform mapping criteria, in cases where knickpoints are present on the valley floors, we only mapped the sections of the valleys lying downstream of the knickpoints (e.g. Supplementary Fig. 2). By adopting this criterion we assumed: 1) that the fans associated with valleys that present prominent knickpoints formed during a later erosive phase; and 2) that the location of the knickpoints approximates the upper erosive edge. As discussed, these

D.A. Vaz et al. / Earth and Planetary Science Letters 533 (2020) 116049

assumptions may not apply to all sites potentially resulting in considerable volume errors for the seven valleys that present a knickpoint. Therefore, we do not include those cases on the main mass balance analysis. But we tested and discuss the effect of including the knickpoint cases on the analysis in Supplementary Fig. 3. 2.3. Fan volume measurement To estimate the volume of sediments deposited within the fans it is necessary to infer the pre-depositional geometry of the basin. In nearly 90% of the surveyed cases the depositional basin corresponds to impact craters or circular depressions of likely impact origin. This justifies the adopted procedure, which aimed to integrate the shape of the crater as a morphological constraint used to interpolate the pre-depositional topography. Once more, the orthoimages were used to delineate polygons that mark the extent of the fans and the rim of the crater, which was digitized in order to perform a least square fitting of a circle (Fig. 2g, j). With this simple parameterization of the basin shape, we can produce concentric profiles that approximate the underlying base topography (Fig. 2g). To achieve this, we sampled the DTM elevations for each profile on a buffer area located 1 km outside of the fans (Fig. 2g, k). The 5% lower elevation quantile on each end of the profiles was used to fit a linear function. This resulted in an initial base elevation prediction which was then tested in order to guarantee that it did not intersect the fan surface. When that occurred, the excess elevation was subtracted from the profile elevations in order to avoid negative fan depths. The total volume of deposited material was obtained by Riemann integration of the difference between the fan upper surface and the computed base surface (Fig. 2i, l). In order to assess fan volume uncertainty we summed: 1) the inherent DTM triangulation uncertainties (addressed in the same manner as explained for the valleys’ volumes, Supplementary Fig. 1a), and 2) the complexity of the underlying topography. We estimated this second component independently for each profile by computing, for the outer buffer regions, the root mean square errors between the actual topography and the predicted base elevations for the same areas. The obtained uncertainty value is then assigned to all pixels located in the inner section of the profile (Supplementary Fig. 1b). This process aims to evaluate the complexity of the topography in the outer buffer areas, using this information to extrapolate the elevation uncertainties to the base surface predictions. For the Subur Vallis case, the boundaries of the fan are uncertain, since two superimposed fans exist (Supplementary Fig. 5). Therefore, here we consider two different fan volume estimates that include: 1) the total volume of the two fans; or 2) only the volume of the younger fan. In the case of the fans, the collected morphometric parameters include: the volume, area, the attitude of the base surface (obtained by least squares fitting of a plane) and summary statistics for the fan thickness, elevation and slope of the fans’ upper surface. A geographic database is used to integrate these parameters with the morphometric parameters computed for the valleys (this database is provided as supplementary material 2). In order to evaluate the accuracy of the method used to measure the volume of deposited materials we used the SEDFLUX numerical model (a 3D model that simulates river transport and delta depositional dynamics and stratigraphy, Hutton and Syvitski, 2008) to generate different deltas, as function of river discharge and crater/basin shape. In addition, post-depositional topographic changes were simulated by adding fractal noise and random small craters to the deltas’ final topographies (see Supplementary Fig. 4 for modeling details and examples).

5

2.4. Mass balance model Our volume measurements do not provide a direct insight into the real volumes at the time of formation of valleys and fans. Deltaic processes are known to generate a loss of material due to offshore deposition (Hoke et al., 2014; Sommerfield et al., 2007; Wright and Nittrouer, 1995) and transverse aeolian ridges (TARs) are ubiquitous covering the floor of valleys, proving that some degree of aeolian infill exists. Fans also present different degrees of erosion, which implies that some unknown percentage of material was removed after deposition, by aeolian or other erosive processes. Another issue that we considered on the mass balance analysis is that we do not know the initial porosity of the dissected terrains, nor the porosity of the fan sediments. We quantitatively integrate these factors on a mass balance equation, using our measurements to constrain the volume gain/loss that these processes could have produced. For this purpose, we selected from each area the measurements obtained from the DTMs with the best spatial resolution. We estimated an average equivalent sand thickness (EST, the sand volume divided by the area covered by aeolian bedforms) for three valleys using HiRISE images. Measuring the wavelength of the TARs using the same mapping techniques described in (Vaz and Silvestro, 2014; Vaz et al., 2017) we estimated the volume of the TARs (V TAR ) and an average EST (see supplementary Fig. 6 for details), which was then multiplied by the total area of the valleys in order to estimate the minimum volume of aeolian infill for each surveyed case. The equality ( V V + V TAR ) × (1 − λ V ) = ( V F + V erod ) × (1 − λ F ) × L is the basis for our mass balance analysis, relating the present day fan and valley volumes (V F and V V respectively), their porosities (λ F and λ V ), the volume of the TARs located on the valleys (V TAR ) and the fan volume that was eroded (V erod ). This last parameter only accounts for the volume of materials which may have been lost after deposition, while a multiplicative coefficient (L) is used to account for the possible sediment loss due to offshore sedimentation of finer sediments. With our volume measurements we performed a parametric study of the putative fan post-depositional erosion rates, also testing the plausibility of syn-sedimentary material dispersion. A least absolute deviation (LAD) method  is used to find the optimal 

σ

VV

VV

+

σV F VF

 

( V V + A V ×EST )×(1−λ V ) ( V F + A F × T erod )×(1−λ F )× L − 1 ×   . In this equation we replace the volume of eroded

parameters that minimize S =

material by the product of the fans area and the uniform thickness of an eroded layer ( A F × T erod ). This facilitates the computation of T erod fan erosion rates according to E F = age . The computed volume measurement uncertainties (σ V V and σ V F ) are considered in the last section of the equation, where the relative uncertainties are used as weighting factors.  By numerically solving argmin T erod ∈R S (λ V ,λ F ,EST , L ) ∀λ V ∈ [0.15,



0.25], ∀λ F ∈ [0.3, 0.4] for subsets of our volume measurements we obtain optimal equivalent thicknesses, corresponding to the unknown amount of materials that was eroded from the fans after deposition. These are then converted to erosion rates by considering a range of possible ages. Here we use the maximum age range (0.29-3.63 Ga) obtained for several Martian fans by Hauber et al. (2013), and we test offshore sediment loss factors (L) between 1 and 10 (Hoke et al., 2014). We assume that the fan materials have porosities typical of unconsolidated sands and gravels (0.3 and 0.4), while the porosity of the materials that are dissected by the valleys is similar to the porosities measured on Apollo lunar impact breccias, 0.15-0.25 (Macke et al., 2011). Supplementary Table 3 summarizes the variables and parameters we use to perform the mass balance analysis. In addition, Supplementary Table 4 and

6

D.A. Vaz et al. / Earth and Planetary Science Letters 533 (2020) 116049

Fig. 3. Comparison with previously published volumes and comparison between the values obtained with CTX, HRSC and HiRISE datasets. This assessment demonstrates that HRSC DTMs produce a systematic underestimation of the valleys’ volumes, which implies that without a bias correction this dataset cannot be used to perform an accurate mass balance survey (Supplementary Fig. 9 further illustrates this problem). a) percentage of error between measured volumes and other published estimates (1 - Kleinhans et al., 2010; 2 - Grindrod et al., 2018; 3 - Malin and Edgett, 2003; 4 - Moore et al., 2003; 5 - Palucis et al., 2016; 6 - Hauber et al., 2013; 7 - Di Achille et al., 2006; 8 Howard et al., 2005), we found that our fan volumes are on average 18% higher than previous estimates (the dashed lines represent the average values for fans and valleys, black and gray respectively); in the case of the valleys, our measurements more than double the previous estimates, although the most remarkable is the wide dispersion of the measurements that highlight the difficulty in the integration of volumes obtained from different studies. b) percentage of error between fan volumes obtained from CTX vs. HRSC DTMs (one HiRISE case is also shown), in the case of the fans (black) there is a reasonable agreement between the measurements, without a clear bias (average difference of 5%) and with a standard deviation that does not surpass 26%; the situation is rather different in the case of the valleys, with higher volumes consistently obtained with CTX data (HRSC volumes are 23% lower on average) which in certain cases can more than double the HRSC volumes.

Supplementary Fig. 7 present and discuss alternative porosity settings and the impact of porosity uncertainty on the results. The described procedure assumes that depositional conditions (e.g. overall depositional setting, runoff, sediment supply or base level change) were constant during the formation of the fans. That is certainly a generalization that does not account for the existence of stepped/terraced fans (Di Achille et al., 2006) or late stage channelization in some of the fans, characteristics that may denote changing depositional settings. Therefore, the adopted mass balance technique only provides an averaged representation of the fans evolution and associated environments. 3. Results 3.1. Accuracy assessment We first discuss the accuracy of the proposed fan volume measurement technique, by analyzing the tests made with the deltas modeled with SEDFLUX (Supplementary Fig. 4). For the pristine topographies we observe an excellent performance of the fan volume measurement technique (Supplementary Fig. 8), with an average volume error of 1 ± 0.5% proving that the method is basically insensitive to the shape of the basin (i.e. the degree of crater infill). On the cases where fractal noise and craters are added to the models the errors increase significantly (19 ± 12% of average error). However, this bias also results in an increase of the uncertainty bars, which always encompass the true volume values (Supplementary Fig. 8b). This analysis shows that the proposed method produces unbiased fan volume estimates and that the uncertainty intervals associated with each measurement are meaningful. Different datasets and techniques have been previously used to perform mass balance assessments of selected Martian valleys and fans, in order to model and draw conclusions regarding their formation mechanisms (Di Achille et al., 2006; Grindrod et al., 2018; Hauber et al., 2013; Howard et al., 2005; Kleinhans et al., 2010; Moore et al., 2003; Palucis et al., 2016). However, volume measurement uncertainties are commonly overlooked, ignoring for instance the possible effects of data quality and spatial resolution.

Therefore, these estimates cannot truly be considered as reference values. Even so, in Fig. 3a we compare our measurements (in some cases made with more than one dataset) with the volumes published in those studies. The most striking point of this assessment is the existence of large differences between the compared datasets. For the fans, our values are on average 18 ± 83% higher than previously published estimates, while for the valleys our values are 123 ± 234% higher. These disparities demonstrate that a simple bibliographic compilation of volumes will not generate a meaningful dataset. The effect of using datasets with different characteristics is also very relevant. In Fig. 3b we compare the measurements made with CTX vs. HRSC or HiRISE data, for fans where more than one dataset is available (Fig. 1a). The differences between HiRISE and CTX volumes are small, not surpassing 6%. A clear systematic bias is not evident for the fans with CTX/HRSC data, where we obtained an average difference of 5 ± 26%. However, there is a systematic underestimation of the valley volumes computed with HRSC data (an average underestimation of 23 ± 19% is obtained, gray dashed line on Fig. 3b). The reason for this bias is related to the coarse resolution and higher smearing/blurriness of the HRSC DTMs, which results in a systematic volume underestimation of the eroded materials (see Supplementary Fig. 9). To our knowledge, this association of increasing volume bias with the DTMs’ spatial resolution has never been discussed or addressed on previous valley volume assessments (cf. Luo et al., 2017), which suggests a possible large underestimation of the volume of eroded and mobilized sediments on Mars if only HRSC DTMs are used. In order to guarantee a better integration of all measurements, our final best estimates for cases where only HRSC data is available include a valley volume bias correction which consists of adding the computed average underestimation (23%) to the measured volumes. The best volume estimates obtained from the datasets with lower spatial resolutions are shown in Fig. 4. Uncertainty intervals are high, on average 84% and 17% of the measured volume for fans and valleys respectively. That is mainly due to the complexity

D.A. Vaz et al. / Earth and Planetary Science Letters 533 (2020) 116049

Fig. 4. Fan and valley volume measurements, the dashed line corresponds to a 1 : 1 volume ratio. Note the clustering around the 1 : 1 line and the existence of another set of observations above this line (see Fig. 5 for a population segmentation). The HRSC cases include the discussed valley volume bias correction.

of the fans’ underlying topography, and in a lower degree to data related uncertainties. In any case, these numbers show that any conclusion drawn from this volumetric survey must acknowledge the limitations of using different datasets and a simple geometric model to approximate the complex base topography and crater shape variability. 3.2. Valley/fan volume ratio Fig. 4 shows that the valley volumes surpass the volumes deposited on the fans for the majority of the 60 fan deposits considered in this study. A kernel density analysis of the computed valley/fan volume ratios highlights a clustering along the 1:1 ratio line. Indeed, Fig. 5 shows that 69% of the cases have valley/fan volume ratios between 1/2 and 2 and that the maximum peak of the distribution corresponds to a ratio of 1.24. Therefore, the studied catalog of fans can be divided into two main categories based on valley/fan volume ratio: Type I population with ratios ranging from 1/2 to 2, and Type II population with ratios higher than 2. Overall, these results suggest a) that there is an overall deficit of materials on the fans and b) that two subpopulations of deposits can be discriminated with the majority (∼70%) of valley/fan systems presenting a volume ratio < 2, peaking close to unity. These inferences are further evaluated in the following section using the mass balance equation introduced in Section 2.4. 3.3. Mass balance: assessing formation settings and fans’ erosion past Given the high measurement uncertainties, a case-by-case analysis is difficult. Instead, we treat our measurements as a noisy representation of a complex reality, making inferences separately for each individualized subpopulation. We estimated an equivalent sand thickness of 0.78 ± 0.4 m for the TARs (Supplementary Fig. 6), a value which is comparable to the 0.7 m obtained with a different technique for TARs located on graben depressions (Vaz et al., 2014). Overall this value only corresponds to ∼1% of the total volume of valley incision. However, this should be considered a minimum value for aeolian infill, since it only corresponds to the visible part of the aeolian deposits; any other buried aeolian deposit is ignored. In Supplementary Fig. 3a-b we evaluate the relevance of including the cases where valley networks present knickpoints. These

7

seven cases do not affect the overall distribution, neither its bimodality, nor the maxima location. However, to be more conservative, and due to the possible high valley volume uncertainties (discussed in Section 2.2), those cases were excluded from the mass balance modeling. We present alternative mass balance results which include the knickpoint cases in Supplementary Fig. 3c-d, and these do not differ significantly from those discussed here, still supporting our main conclusions. To the Type I fan population (measured volume ratios between 1/2 and 2, Fig. 5) we fit the mass balance equation in order to evaluate fan erosion rates and the likelihood of offshore sediment loss during deposition. The optimal solution that minimizes the absolute deviation corresponds to a fan porosity of 0.3, a valley material porosity of 0.25, no offshore loss of sediments (L = 1) and an eroded equivalent thickness of 16 m (Fig. 6a). The erosion rates that would explain the removal of this thickness of sediments are within 4-56 nm/yr (4-154 nm/yr for all tested solutions), a range of values comparable to erosion rates inferred to exist during much of Hesperian and Amazonian times (Fig. 6b). These values are one order of magnitude lower than Noachian erosion rates (Golombek et al., 2006) and two orders of magnitude below current erosion rates of friable (yardang forming) light-toned layered deposits (Urso et al., 2018). Most importantly, for the Type I fans the model shows that relevant (L ≥ 2) offshore sedimentation during the formation of the fans is not possible, since a convergent minimum does not exist for positive erosion rates (Fig. 6a). The model outputs for Type II fans are not sensitive to porosity changes, therefore we use the best fit porosities obtained earlier for Type I cases (porosities of 0.3 and 0.25 for fan and valley materials). For the Type II fans (with volume ratios >2), the mass balance model solutions for L = 1 (no offshore sedimentation) requires the post-depositional erosion of more than 1 km of material from the fans (Fig. 7a). This is an excessive high value, corresponding to five times the current average thicknesses of the fans. In addition, the correspondent high erosion rates surpass the admissible Hesperian-Amazonian erosion rates (Fig. 7b), demonstrating that a non-negligible amount of sediment must have been lost during the deposition of the fans. For the Type II fans, the existence of a unique solution is not evident, since both the offshore sedimentation coefficient (L) and the eroded fan volume (V erod ) contribute to increase the net volume of the fans (the right side expression on the mass balance equation). Even so, if we assume an average Hesperian-Amazonian erosion rate of 100 nm/yr, we see that offshore loss coefficients between 3 and 10 are viable, although with different fan age constrains (Early Hesperian minimum age for L = 3 and a minimum age of 0.63 Ga for L = 10). We have also explored an alternative range of lower porosities (0.01 < λ V < 0.2 and 0.2 < λ F < 0.4); however the overall results are not greatly affected by this different set of input parameters (see Supplementary Table 4 and Supplementary Fig. 7 for detailed discussion). In order to test the best fit parameters we computed average net balanced volume ratios (Fig. 8), which correspond to the ratios of the non-porous modeled initial volumes (a perfect model will have all net ratios equal to 1). We assume a fan erosion rate of 8.2 nm/yr with a corresponding age of 1.96 Ga (the middle of the fan age range estimated by Hauber et al., 2013), and the mentioned optimal parameters (a fan porosity of 0.3, a valley material porosity of 0.25, a valley equivalent sand thickness of 0.78 m and L = 1 or L = 5 for Type I or Type II fans respectively). Fig. 8 also shows the raw volume ratios and identifies the surveyed fans. Overall, this dual mass balance model explains 51% of the variance in the data (R 2 = 0.51, although we obtain R 2 = 0.57 when the knickpoint cases are included on the analysis, Supplementary Fig. 10). Further improvements to this model must consider variable ages

8

D.A. Vaz et al. / Earth and Planetary Science Letters 533 (2020) 116049

Fig. 5. Subpopulation segmentation; note that the maximum density corresponds to a ratio slightly larger than 1, suggesting a deficit of volume on the fans. a) kernel density distribution of the valley/fan volume ratios. b) The Type I subpopulation is shown in black (1/2 < ratio < 2), while the Type II subpopulation (gray) corresponds to valley/fan volume ratios >2.

Fig. 6. Modeling of the eroded equivalent thickness (fan volume divided by their area) for the Type I fans. a) For the tested porosities, solutions ranged from 16 to 45 m, although the optimal solution (the one that minimizes the absolute deviation) corresponded to an equivalent thickness of 16 m, a fan porosity of 0.3 and a valley porosity of 0.25. A valley equivalent sand thickness (EST) of 0.78 m and no offshore sedimentation (L = 1) were tested. Relevant offshore sedimentation cases (L ≥ 2, blue dotted lines) do not produce a convergent solution for positive erosion rates (the two cases were computed using the optimal porosities obtained for L = 1). b) Admissible average erosion rates computed for the interval of time of likely fan formation (Hauber et al., 2013); average erosion rates estimated by others are shown for comparison (Golombek et al., 2006 and references therein; Kite and Mayer, 2017; Urso et al., 2018).

Fig. 7. Possible eroded equivalent thicknesses and coefficients of offshore sedimentation (L) for Type II fans. a) L values near 1 are unlikely, since they require the removal of unrealistic amounts of materials from the fans (>1 km of equivalent thickness). b) In order to comply with predicted Amazonian-Hesperian erosion rates, relevant offshore sediment loss is needed (L values between 3 and 10 are admissible for most of the considered ages).

D.A. Vaz et al. / Earth and Planetary Science Letters 533 (2020) 116049

9

Fig. 8. Valley/fan volume ratios of the measured volumes, net volumes assuming non-offshore sediment loss (L = 1) for Type I fans, and an average degree of sediment loss (L = 5) for Type II set (blue markers). A fan age of 1.96 Ga, a fan erosion rate of 8.2 nm/yr, a fan porosity of 0.3, a valley material porosity of 0.25 and a valley EST of 0.78 m were the parameters used to model the net volume ratios. The 1 : 1 ratio line and the threshold values (1/2 and 2) used to segment the two subpopulations are also shown. Knickpoint cases appear on Supplementary Fig. 10. Table 1 Morphometric comparison of type I and II valley/fan sets. The median (± median absolute deviation) for each parameter and subpopulation type is shown. A Wilcoxon rank-sum test is used to evaluate the null hypothesis (H0 : the distributions’ medians are equal) with a 90% confidence level. Valleys’ morphometric parameters (volume, depth, width and length) are significantly different: type II presents valleys with larger dimensions, while smaller ones correspond to type I subpopulation. The cross-sections of the valleys do not present significant differences. For the fans, the parameter that differs significantly for the two subpopulations is the dip angle of the fans’ base surface, obtained by fitting planes to the computed lower surfaces. This suggests that type I fans deposited in surfaces with a higher slope. Morphological parameter

Type I median

Type II median

Reject H0 (90% conf. level)

p-value

Valley

Volume (km3 ) Depth (m) Width (m) Length (km) Rectangular section coef. V-shaped section coef. Slope (◦ )

1.8 ± 1.1 171 ± 81 1680 ± 329 13.6 ± 6.9 0.21 ± 0.03 0.93 ± 0.03 17.0 ± 2.9

27.9 ± 19.3 270 ± 92 2198 ± 816 44.3 ± 28.0 0.19 ± 0.04 0.94 ± 0.03 13.7 ± 3.3

yes yes yes yes no no no

2.3 × 10−5 5.3 × 10−2 9.7 × 10−3 1.9 × 10−3 0.47 0.22 0.20

Fan

Volume (km3 ) Area (km2 ) Thickness (m) Slope (◦ ) Base dip (◦ )

1.4 ± 1.0 38.2 ± 23.5 49.9 ± 16.2 6.9 ± 2.5 2.8 ± 1.6

3.4 ± 2.0 39.3 ± 26.1 58.6 ± 21.6 6.6 ± 2.0 1.9 ± 1.0

no no no no yes

0.34 0.68 0.55 0.70 2.7 × 10−2

and a larger number of accurate volume measurements. Nonetheless, the described results provide a consistent first-order insight on the different processes that may have formed and shaped the fan deposits on Mars. Some outliers should be expected to exist on the compiled database. For instance, the Hypanis fan (Supplementary Fig. 11b) has a very low ratio which may be related to an underestimation of the eroded volume and/or overestimation of the deposited materials. Wrinkle ridges were recognized around the fan (Adler et al., 2019), which means that some degree of warping may exist, artificially increasing the measured fan volume. In any case, the deltaic origin of this fan was recently proposed (Adler et al., 2019). The high ratio obtained for the Nili Fossae fan (Supplementary Fig. 11a) is also dubious. This is an area where later sedimentation (or volcanic cover-up) occurred on the floor of the graben, therefore reducing the exposure of the fan and consequently the measured volume. Statistically significant differences between the morphometric characteristics of the two proposed subpopulations were also tested (Table 1). Valleys’ morphometric parameters (volume, depth, width and length) are significantly different for each population (10% significance level). Valleys with larger dimensions belong to the Type II set, while the smaller ones correspond to subpopulation I. The coefficients we used to characterize the cross-section shape of the valleys do not present significant differences. In the case of the fans, the only parameter that significantly differs be-

tween the two subpopulations is the dip angle of the fans’ base surface, indicating that Type I fans were deposited on steeper surfaces. 4. Discussion Our results reveal that the compiled Martian fan database can be divided into two different subpopulations, providing support from a quantitative point of view to previous qualitative observations (Di Achille and Vaz, 2017). Type I fans are more abundant, and based on the mass balance analysis we conclude that they formed by processes that do not produce a relevant offshore loss of sediments during deposition (examples of this class of fans can be seen in Fig. 9a, b and Supplementary Fig. 12). Under the assumption that typical fluvio-deltaic processes imply a non-negligible amount of offshore sediment dispersion (Hoke et al., 2014; Sommerfield et al., 2007; Wright and Nittrouer, 1995), this type of fan likely formed with significant contributions of subaerial processes. These fans also present higher basal slopes, and their valleys are significantly smaller (shorter, with smaller widths and depths). An unequivocal interpretation of these morphometric proprieties is hard to establish, but it certainly points to a more immature drainage system where alluvial and/or gravitydriven processes may have been relevant, at least during the early phases of deposition.

10

D.A. Vaz et al. / Earth and Planetary Science Letters 533 (2020) 116049

Fig. 9. Examples of intra-class morphologic variability (more examples are presented on Supplementary Figs. 12 and 13). a) Type I: West of Gale simple fan (−5.46◦ N, 136.13◦ E). b) Type I: stepped fan on Camichel crater (2.64◦ N, −51.66◦ E). c) Type II: Eberswalde lobated fan (−23.85◦ N, −33.56◦ E). d) Upper section of the Subur Vallis fan (11.75◦ N, −52.92◦ E).

Included in the defined Type II set of fans we found some of the best studied Martian fans, whose formation was most likely driven by fluvio-lacustrine processes: Eberswalde (Pondrelli et al., 2011), Jezero (Schon et al., 2012), Magong/Sabrina (Hauber et al., 2009), Nanedi Vallis (Hauber et al., 2009), Shalbatana (Di Achille et al., 2009) and South of Claritas Fossae (Mangold and Ansan, 2006). Our results are also consistent with an overall morphological duality. In fact, among the known Martian deltas only a few (Eberswalde, Jezero, Sabrina Hypanis, Moa Valles) show well-developed stratigraphy with evidence of avulsing channels and multilobate depositional patterns (e.g. Fig. 9c); they are the largest deposits (in terms of surface area and volume) and are found exclusively at the mouth of relatively large valleys with tributaries and alluvial plains that were presumably formed by sustained and persistent surface runoff discharges. These deposits fall in the Type II category and might be strictly interpreted as river-deltas and are typically found as highly eroded deposits without original depositional surfaces. Whereas, the majority of the studied fan deposits (Type I) are small (a few km wide and long) single-lobe with deep (and often steep) fronts resembling terrestrial Gilbert-type fan-deltas (e.g. Fig. 9a and b). The latter fans do not show evident signs of avulsion and multilobate depositional patterns and are typically well preserved showing relatively pristine morphology. The presented validation and comparison assessments show that any volumetric survey of fans and valleys on Mars needs to consider the relatively high measurement uncertainties. Data quality and spatial resolution have a significant impact, but it is the complexity of the fans’ underlying topography and crater/basin shape variability that account for most of the uncertainty. However, since we imbedded the uncertainty estimates on the subsequent mass balance modeling, the robustness of our inferences is certainly enhanced. Morgan et al. (2018) mapped a total of 149 possible deltaic deposits, which means we have sampled nearly

half of the fans with putative deltaic origin on the planet (a total of 60 fans were surveyed). Therefore, the conclusions drawn in this study are based on a substantial sample size, which should be representative of the overall population. Results of the mass balance model show that the postdepositional erosion of Type I fans is consistent with the low erosion rates estimated for Hesperian and Amazonian times on Mars (Golombek et al., 2006; Kite and Mayer, 2017). The computed erosion rates (4-154 nm/yr) are also consistent with the rates estimated for some of the oldest and best preserved alluvial fan surfaces on Earth (∼100 nm/yr), formed on the hyperarid environment of the Atacama desert (Nishiizumi et al., 2005). The mass balance analysis of Type II fans reveals a contrasting evolution, since the admissible Hesperian-Amazonian erosion rates are not enough to explain the high valley/fan volume ratios (examples of this class of fans can be seen in Supplementary Fig. 12c, d and Supplementary Fig. 13). Therefore, a considerable amount of sediment must have been lost during the formation of these fans, as a) predicted by delta formation modeling on Mars-like conditions (Hoke et al., 2014) and b) observed for terrestrial deltas (e.g. Sommerfield et al., 2007; Wright and Nittrouer, 1995). The morphometry of the valleys associated with these fans is also different: they have longer, deeper and wider valleys. This suggests more mature drainage systems, globally resulting in ten times more incised volume, when compared with Type I (Table 1). Therefore, based on the valley volume vs. delta volume ratio, overall geomorphology/sedimentology, and mass balance model, we conclude that Martian fans can be divided into two main categories: a few highly eroded (older? Noachian?), sedimentologically complex/mature, multilobate, and relatively large deposits fed by regionally integrated valley networks (Type II, akin to “river-like” deltas) and many small, well-preserved (younger? PostNoachian?), sedimentologically simple/immature, single-lobe fans

D.A. Vaz et al. / Earth and Planetary Science Letters 533 (2020) 116049

fed by isolated valleys (Type I). Type I fans might have not been exclusively formed by fluvio-lacustrine deposition but under more cold/dry conditions by subaerial depositional mechanisms (e.g. alluvial? lava flows? high viscosity/density mass-wasting? lahar/mud flows? glacial processes? landslides?). The block diagrams shown in Fig. 1b and 1c illustrate the generic depositional environments where the two types of fans may have formed. Finally, although a precise correlation of the defined populations with different formation ages, mineralogical contents, or water availability is beyond the scope of this paper, it is noticeable that a bimodality was inferred to exist also for these attributes (Hauber et al., 2013). In addition, we note the existence of significant intra-class morphologic variability. For instance, Type I fans can present different numbers of steps (Fig. 9a and b), while not all Type II fans show clear signs of meandering lobes (Fig. 9c and d). This probably means that additional subclasses are needed to encompass the disregarded morphologic attributes, although class mixing due to the large volume measurement uncertainties cannot be ruled out. Full CTX stereo coverage for all the deposits reported by Morgan et al. (2018) is not available at the moment. However, new data from CTX and CaSSIS cameras (Thomas et al., 2014) may in the future help to improve the presented dataset. The inclusion of other attributes (e.g. other morphometric parameters, ages and mineralogy) and descriptive parameters may also contribute to better constrain our results and conclusions. 5. Conclusions Using high resolution topography, we present a detailed mass balance analysis of a catalog of putative Martian fan-shaped deposits, providing new insights into their past depositional environments. Two different populations of fans might exist on Mars. Type I fans are more numerous, relatively well preserved, and formed by processes that do not produce a relevant offshore loss of sediments during deposition (valley/fan volume ratio close to unity). They are associated with smaller and less mature drainage networks and were deposited on steeper gradients. The mass balance modeling suggests that this type of fan/valley association formed mainly under subaerial settings. The existence of deep syn-depositional paleolakes in those cases must be carefully reviewed, although the existence of ephemeral confined water bodies (playa-like settings?) on the hosting basin cannot be ruled out. On the other hand, Type II fans present all the characteristic signatures of deltaic processes, with larger and more mature drainage networks, deposition on flatter surfaces and high percentage of sediment bypass. These results suggest that the majority of the considered Martian fan-shaped deposits (Type I) do not necessarily imply the occurrence of favorable nor long-lasting habitable conditions, raising the question of whether basins with Type I fans should be considered ideal landing sites for future in situ exploration of Mars. Finally, our results are in agreement with recent overviews about early Mars climate, suggesting that Mars climate was more variable and transient through time than previously thought and forced mainly by a combination of orbital and external factors (Wordsworth, 2016), and/or that lacustrine environments might have formed by processes that allowed for confined hydrological activity under cold conditions (Bristow et al., 2017). Declaration of competing interest The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

11

Acknowledgements This research was supported by Italian Ministry of University and Research (MIUR) through grant FIRB-RBFR130ICQ. D. Vaz thanks CITEUC’s support (projects UID/Multi/00611/2013 & POCI01-0145-FEDER-006922) and FCT grant CEECIND/02981/2017. R. Williams and B. Hynek acknowledge support from the NASA Grant # NNX15AH46G, through the Solar System Workings program. We thank Lucio Primo Pacinelli for his support during the realization of Fig. 1b-c. We are grateful to two anonymous reviewers and to the editor, whose comments and suggestions helped us to improve the manuscript. Appendix A. Supplementary material Supplementary material related to this article can be found online at https://doi.org/10.1016/j.epsl.2019.116049. References Adler, J.B., Bell, J.F., Fawdon, P., Davis, J., Warner, N.H., Sefton-Nash, E., Harrison, T.N., 2019. Hypotheses for the origin of the Hypanis fan-shaped deposit at the edge of the Chryse escarpment, Mars: is it a delta? Icarus 319, 885–908. https://doi. org/10.1016/j.icarus.2018.05.021. Bristow, T.F., Haberle, R.M., Blake, D.F., Des Marais, D.J., Eigenbrode, J.L., Fairén, A.G., Grotzinger, J.P., Stack, K.M., Mischna, M.A., Rampe, E.B., Siebach, K.L., Sutter, B., Vaniman, D.T., Vasavada, A.R., 2017. Low Hesperian PCO2 constrained from in situ mineralogical analysis at Gale Crater, Mars. Proc. Natl. Acad. Sci. 114, 2166. https://doi.org/10.1073/pnas.1616649114. Cabrol, N.A., Grin, E.A., 1999. Distribution, classification, and ages of Martian impact Crater Lakes. Icarus 142, 160–172. https://doi.org/10.1006/icar.1999.6191. Di Achille, G., Hynek, B.M., 2010a. Ancient ocean on Mars supported by global distribution of deltas and valleys. Nat. Geosci. 3, 459–463. https://doi.org/10.1038/ ngeo891. Di Achille, G., Hynek, B.M., 2010b. Deltas and valley networks on Mars: implications for a global hydrosphere. In: Cabrol, N.A., Grin, E.A. (Eds.), Lakes on Mars. Elsevier, Amsterdam, pp. 223–248. Di Achille, G., Hynek Brian, M., Searls Mindi, L., 2009. Positive identification of lake strandlines in Shalbatana Vallis, Mars. Geophys. Res. Lett. 36. https://doi.org/10. 1029/2009gl038854. Di Achille, G., Marinangeli, L., Ori, G.G., Hauber, E., Gwinner, K., Reiss, D., Neukum, G., 2006. Geological evolution of the Tyras Vallis paleolacustrine system, Mars. J. Geophys. Res., Planets 111. https://doi.org/10.1029/2005je002561. Di Achille, G., Ori, G.G., Reiss, D., 2007. Evidence for late Hesperian lacustrine activity in Shalbatana Vallis, Mars. J. Geophys. Res., Planets 112. https://doi.org/10.1029/ 2006je002858. Di Achille, G., Vaz, D.A., 2017. Hydrological and climatic significance of Martian deltas. In: 4th Conference on Early Mars: Geologic, Hydrologic, and Climatic Evolution and the Implications for Life, LPI Contrib. No. 2014. Flagstaff, Arizona, p. 3041. Fergason, R.L., Lee, E.M., Weller, L., 2013. THEMIS geodetically controlled mosaics of Mars. In: 44th Lunar and Planetary Science Conference, LPI Contribution No. 1719. The Woodlands, Texas, p. 1642. Golombek, M.P., Grant, J.A., Crumpler, L.S., Greeley, R., Arvidson, R.E., Bell, J.F., Weitz, C.M., Sullivan, R., Christensen, P.R., Soderblom, L.A., Squyres, S.W., 2006. Erosion rates at the Mars Exploration Rover landing sites and long-term climate change on Mars. J. Geophys. Res., Planets 111. https://doi.org/10.1029/2006je002754. Grindrod, P.M., Warner, N.H., Hobley, D.E.J., Schwartz, C., Gupta, S., 2018. Stepped fans and facies-equivalent phyllosilicates in Coprates Catena, Mars. Icarus 307, 260–280. https://doi.org/10.1016/j.icarus.2017.10.030. Halevy, I., Head III, J.W., 2014. Episodic warming of early Mars by punctuated volcanism. Nat. Geosci. 7, 865. https://doi.org/10.1038/ngeo2293. https://www. nature.com/articles/ngeo2293#supplementary-information. Hauber, E., Gwinner, K., Kleinhans, M., Reiss, D., Di Achille, G., Ori, G.G., Scholten, F., Marinangeli, L., Jaumann, R., Neukum, G., 2009. Sedimentary deposits in Xanthe Terra: implications for the ancient climate on Mars. Planet. Space Sci. 57, 944–957. https://doi.org/10.1016/j.pss.2008.06.009. Hauber, E., Platz, T., Reiss, D., Le Deit, L., Kleinhans, M.G., Marra, W.A., de Haas, T., Carbonneau, P., 2013. Asynchronous formation of Hesperian and Amazonianaged deltas on Mars and implications for climate. J. Geophys. Res., Planets 118, 1529–1544. https://doi.org/10.1002/jgre.20107. Hoke, M.R.T., Hynek, B.M., Di Achille, G., Hutton, E.W.H., 2014. The effects of sediment supply and concentrations on the formation timescale of Martian deltas. Icarus 228, 1–12. https://doi.org/10.1016/j.icarus.2013.09.017. Howard, A.D., Moore, J.M., Irwin III, R.P., 2005. An intense terminal epoch of widespread fluvial activity on early Mars: 1. Valley network incision and associated deposits. J. Geophys. Res. 110. https://doi.org/10.1029/2005JE002459.

12

D.A. Vaz et al. / Earth and Planetary Science Letters 533 (2020) 116049

Hutton, E.W.H., Syvitski, J.P.M., 2008. Sedflux 2.0: an advanced process-response model that generates three-dimensional stratigraphy. Comput. Geosci. 34, 1319–1337. https://doi.org/10.1016/j.cageo.2008.02.013. Hynek, B.M., Beach, M., Hoke, M.R.T., 2010. Updated global map of Martian valley networks and implications for climate and hydrologic processes. J. Geophys. Res. 115, E09008. https://doi.org/10.1029/2009je003548. Irwin III, R.P., Howard, A.D., Craddock, R.A., Moore, J.M., 2005. An intense terminal epoch of widespread fluvial activity on early Mars: 2. Increased runoff and paleolake development. J. Geophys. Res. 110. https://doi.org/10.1029/2005JE002460. Jolivet, M., Barrier, L., Dominguez, S., Guerit, L., Heilbronn, G., Fu, B., 2014. Unbalanced sediment budgets in the catchment–alluvial fan system of the Kuitun River (northern Tian Shan, China): implications for mass-balance estimates, denudation and sedimentation rates in orogenic systems. Geomorphology 214, 168–182. https://doi.org/10.1016/j.geomorph.2014.01.024. Kiefer, E., Dorr, M., Ibbeken, H., Gotze, H., 1997. Gravity-based mass balance of an alluvial fan giant: the Arcas Fan, Pampa del Tamarugal, Northern Chile. Rev. Geol. Chile 24, 165–185. Kim, J.R., Muller, J.P., 2009. Multi-resolution topographic data extraction from Martian stereo imagery. Planet. Space Sci. 57, 2095. https://doi.org/10.1016/j.pss. 2009.09.024. Kite, E.S., Mayer, D.P., 2017. Mars sedimentary rock erosion rates constrained using crater counts, with applications to organic-matter preservation and to the global dust cycle. Icarus 286, 212–222. https://doi.org/10.1016/j.icarus.2016.10.010. Kleinhans, M.G., van de Kasteele, H.E., Hauber, E., 2010. Palaeoflow reconstruction from fan delta morphology on Mars. Earth Planet. Sci. Lett. 294, 378–392. https://doi.org/10.1016/j.epsl.2009.11.025. Kraal, E.R., Asphaug, E., Moore, J.M., Howard, A., Bredt, A., 2008a. Catalogue of large alluvial fans in Martian impact craters. Icarus 194, 101–110. https://doi.org/10. 1016/j.icarus.2007.09.028. Kraal, E.R., van Dijk, M., Postma, G., Kleinhans, M.G., 2008b. Martian stepped-delta formation by rapid water release. Nature 451, 973. https://doi.org/10.1038/ https://www.nature.com/articles/nature06615#supplementarynature06615. information. Luo, W., Cang, X., Howard, A.D., 2017. New Martian valley network volume estimate consistent with ancient ocean and warm and wet climate. Nat. Commun. 8, 15766. https://doi.org/10.1038/ncomms15766. https://www.nature.com/articles/ ncomms15766#supplementary-information. Macke, R.J., Kiefer, W.S., Britt, D.T., Irving, A.J., Consolmagno, G.J., 2011. Densities, porosities and magnetic susceptibilities of meteoritic lunar samples: early results. In: 42nd Lunar and Planetary Science Conference, LPI Contribution No. 1608. The Woodlands, Texas, p. 1986. Malin, M.C., Bell Iii, J.F., Cantor, B.A., Caplinger, M.A., Calvin, W.M., Clancy, R.T., Edgett, K.S., Edwards, L., Haberle, R.M., James, P.B., Lee, S.W., Ravine, M.A., Thomas, P.C., Wolff, M.J., 2007. Context camera investigation on board the Mars reconnaissance orbiter. J. Geophys. Res. 112, E05S04. Malin, M.C., Edgett, K.S., 2003. Evidence for persistent flow and aqueous sedimentation on early Mars. Science 302, 1931. Mangold, N., Ansan, V., 2006. Detailed study of an hydrological system of valleys, a delta and lakes in the Southwest Thaumasia region, Mars. Icarus 180, 75. Moore, J.M., Howard, A.D., 2005. Large alluvial fans on Mars. J. Geophys. Res., Planets 110. https://doi.org/10.1029/2004je002352. Moore, J.M., Howard, A.D., Dietrich, W.E., Schenk, P.M., 2003. Martian layered fluvial deposits: implications for Noachian climate scenarios. Geophys. Res. Lett. 30. https://doi.org/10.1029/2003gl019002. Moratto, Z.M., Broxton, M.J., Beyer, R.A., Lundy, M., Husmann, K., 2010. Ames Stereo Pipeline, NASA’s open source automated stereogrammetry software. In: 41st LPSC, Abstract 1533, p. 2364.

Morgan, A.M., Wilson, S.A., Howard, A.D., Craddock, R.A., Grant, J.A., 2018. Global distribution of alluvial fans and deltas on Mars. In: 49th Lunar and Planetary Science Conference, LPI Contribution No. 2083. The Woodlands, Texas. Neukum, G., Jaumann, R., Team, H.C.-I., 2004. HRSC: the High Resolution Stereo Camera of Mars Express. ESA Special Publications SP-1240. Nishiizumi, K., Caffee, M.W., Finkel, R.C., Brimhall, G., Mote, T., 2005. Remnants of a fossil alluvial fan landscape of Miocene age in the Atacama Desert of northern Chile using cosmogenic nuclide exposure age dating. Earth Planet. Sci. Lett. 237, 499–507. https://doi.org/10.1016/j.epsl.2005.05.032. Oguchi, T., 1998. Late quaternary sediment budget in alluvial-fan–source-basin systems in Japan. J. Quat. Sci. 12, 381–390. https://doi.org/10.1002/(sici)10991417(199709/10)12:5<381::aid-jqs321>3.0.co;2-#. Ori, G.G., Marinangeli, L., Baliva, A., 2000. Terraces and Gilbert-type deltas in crater lakes in Ismenius Lacus and Memnonia (Mars). J. Geophys. Res., Planets 105, 17629–17641. https://doi.org/10.1029/1999je001219. Palucis, M.C., Dietrich, W.E., Williams, R.M.E., Hayes, A.G., Parker, T., Sumner, D.Y., Mangold, N., Lewis, K., Newsom, H., 2016. Sequence and relative timing of large lakes in Gale crater (Mars) after the formation of Mount Sharp. J. Geophys. Res., Planets 121, 472–496. https://doi.org/10.1002/2015je004905. Pondrelli, M., Rossi, A.P., Platz, T., Ivanov, A., Marinangeli, L., Baliva, A., 2011. Geological, geomorphological, facies and allostratigraphic maps of the Eberswalde fan delta. Planet. Space Sci. 59, 1166–1178. https://doi.org/10.1016/j.pss.2010. 10.009. Salese, F., Pondrelli, M., Neeseman, A., Schmidt, G., Ori, G.G., 2019. Geological evidence of planet-wide groundwater system on Mars. J. Geophys. Res., Planets 124, 374–395. https://doi.org/10.1029/2018je005802. Schon, S.C., Head, J.W., Fassett, C.I., 2012. An overfilled lacustrine system and progradational delta in Jezero crater, Mars: implications for Noachian climate. Planet. Space Sci. 67, 28–45. https://doi.org/10.1016/j.pss.2012.02.003. Sommerfield, C.K., Ogston, A.S., Mullenbach, B.L., Drake, D.E., Alexander, C.R., Nittrouer, C.A., Borgeld, J.C., Wheatcroft, R.A., Leithold, E.L., 2007. Oceanic dispersal and accumulation of river sediment. In: Continental Margin Sedimentation. Thomas, N., Cremonese, G., Banaszkiewicz, M., Bridges, J., Byrne, S., da Deppo, V., Debei, S., El-Maarry, M.R., Hauber, E., Hansen, C.J., Ivanov, A., Kestay, L., Kirk, R., Kuzmin, R., Mangold, N., Marinangeli, L., Markiewicz, W., Massironi, M., McEwen, A.S., Okubo, C., Orleanski, P., Pommerol, A., Wajer, P., Wray, J., 2014. The colour and stereo surface imaging system (CaSSIS) for ESA’s trace gas orbiter. In: Eighth International Conference on Mars, vol. 1791, p. 1067. Urso, A., Chojnacki, M., Vaz, D.A., 2018. Dune-yardang interactions in Becquerel crater, Mars. J. Geophys. Res., Planets 123, 353–368. https://doi.org/10.1002/ 2017je005465. Vaz, D.A., Silvestro, S., 2014. Mapping and characterization of small-scale aeolian structures on Mars: an example from the MSL landing site in Gale Crater. Icarus 230, 151–161. https://doi.org/10.1016/j.icarus.2013.08.007. Vaz, D.A., Silvestro, S., Sarmento, P.T.K., Cardinale, M., 2017. Migrating meter-scale bedforms on Martian dark dunes: are terrestrial aeolian ripples good analogues? Aeolian Res. 26, 101–116. https://doi.org/10.1016/j.aeolia.2016.08.003. Vaz, D.A., Spagnuolo, M.G., Silvestro, S., 2014. Morphometric and geometric characterization of normal faults on Mars. Earth Planet. Sci. Lett. 401, 83–94. https:// doi.org/10.1016/j.epsl.2014.05.022. Wordsworth, R.D., 2016. The climate of early Mars. Annu. Rev. Earth Planet. Sci. 44, 381–408. https://doi.org/10.1146/annurev-earth-060115-012355. Wright, L.D., Nittrouer, C.A., 1995. Dispersal of river sediments in coastal seas: six contrasting cases. Estuaries 18, 494–508. https://doi.org/10.2307/1352367.