Figuring out the topographic signature of early Martian oceans

Our work depends on three datasets: (1) international digital elevation fashions and bathymetric information for each Earth and Mars; (2) maps of fluvial options on Earth (main international rivers and deltas) and Mars (valley networks, fluvial ridge programs, outlet canyons and interpreted deltas), together with maps of oceanic options on Earth (continental shelf, shelf break and ocean flooring) and interpreted shorelines on Mars; and (3) outcomes of elevation, slope, curvature and panorama classifications for every cell on each Earth and Mars. We describe the info and description our strategies for every dataset under and briefly clarify the flowchart in Supplementary Fig. 1.

Global digital elevation information and bathymetric information

Earth

Three international digital elevation fashions combine each land and ocean surfaces at completely different resolutions and ranges of consistency: (1) the ETOPO Global Relief Model with a common common decision of about 1.85 km per pixel⁶³; (2) the SRTM30_PLUS Estimated Topography with a decision of round 1 km per pixel⁶⁴; and (3) the General Bathymetric Chart of the Oceans (GEBCO) with a decision of roughly 500  m (ref. ⁶⁵). For our global-scale topographic evaluation, we used the ETOPO1 Global Relief Model due to its constant pixel decision throughout each terrestrial and oceanic areas. ETOPO1 offers a uniform 1 arcmin decision (about 1.85 km per pixel), integrating satellite tv for pc altimetry, shipboard echo-sounding and terrestrial measurements right into a cohesive dataset⁶³. This consistency is essential for research requiring seamless information throughout completely different terrains, making certain that each land and ocean topography are represented with the identical degree of element. By distinction, the SRTM30_PLUS Global Bathymetry and Topography dataset, whereas providing greater decision for land areas (30 arcsec), lacks uniformity because it focuses totally on terrestrial areas and offers much less detailed protection for the oceans⁶⁴. Moreover, the GEBCO dataset, though detailed for ocean bathymetry, doesn’t provide the identical constant pixel decision for land topography, resulting in potential discrepancies when integrating land and ocean information⁶⁵. Therefore, ETOPO1 was chosen to make sure uniform decision and complete protection throughout each terrestrial and oceanic environments, addressing the necessity for constant pixel information in our evaluation.

Mars

We used the worldwide Mars Orbiter Laser Altimeter (MOLA) gridded topography, which presents a pixel decision of 463 m per pixel⁶⁶. This dataset is derived from greater than 600 million measurements masking the complete Martian floor. These measurements had been meticulously adjusted to make sure consistency, offering a uniform pixel decision throughout the complete terrain of Mars^67,68.

Data resampling

We resampled each digital elevation fashions to a number of resolutions—2.5 km, 5 km and 10 km—for a number of key causes: (1) Resampling the topographic information of Earth and Mars to a uniform decision is important to use uniform analytical strategies and allow direct comparability of topographic options throughout each planetary surfaces. (2) These particular resolutions had been chosen to deliberately exclude fine-scale landforms on Mars, as these are usually youthful in age⁶⁹. Our research, nonetheless, focuses on older, broader-scale topographic options that present insights into historic floor processes. The resampling was performed utilizing the ‘Resample’ instrument in ArcGIS^70,71, through which we utilized the ‘nearest neighbour’ choice to protect the precise elevation values, minimizing important interpolation and smoothing, and thereby making certain the integrity of the unique information^72,73.

Maps of fluvial and oceanic options on each Earth and Mars

To establish a tough search zone on Mars for the transition from panorama to seascape, we used maps of the key world rivers and deltas on Earth^40,41. These sometimes point out the place the terrestrial panorama ends and the oceanic zone begins. However, this assumption holds provided that the rivers and deltas had been energetic concurrently. Changes in sea ranges may alter this relationship, however the maps nonetheless present a helpful approximation of the extent of the transition zone. Moreover, we used an intensive dataset mapping seafloor geomorphic characteristic⁴², which not solely helps to outline the zone but in addition presents insights into how oceanic geomorphic options evolve spatially. We targeted on the important thing geomorphic options that outline the transition: the continental shelf, the shelf-break slope, the continental slope, the continental rise and the important thing ocean flooring landforms (abyssal and hadal zones).

We relied on international mapping of valley networks^44,74,75, outlet canyons⁴⁴, depositional rivers (fluvial ridges)⁴⁵ and interpreted deltas^{4,10,14,15,16,17,76}. Furthermore, we used maps of topographic contacts, beforehand interpreted as shorelines^10,22,77, to evaluate how the water-formed panorama functioned.

Given the talk over whether or not Martian deltas shaped in open or closed basin programs, we selected to compile the accessible delta datasets after which apply particular filtering standards^{4,10,14,15,16,17,76,78,79}. We chosen deltas that (1) are open to downstream move and situated alongside the dichotomy boundary and/or (2) exhibit complicated stacking patterns interpreted as proof of formation inside both regressive or transgressive depositional environments. This filtering resulted in a set of 48 deltas (Supplementary Table 1 and Supplementary Fig. 4). We additional examined these deltas and labeled them into two classes based mostly on their morphology: single-lobate deltas and stacked deltaic programs. To additional cross-validate our compilation, we calculated the elevations of all channels and lobes—together with these preserved as ridges and interpreted as erosional remnants of historic deltas—throughout the largest deltaic programs in Aeolis Dorsa and Hypanis^14,15,16,17, to seize elevation modifications doubtlessly related to previous sea-level fluctuations (Fig. 5e,f and Extended Data Fig. 7b,c).

Setting a search zone for landscape-to-seascape transitions

We transformed the shapefiles of terrestrial rivers, deltas and Martian rivers, deltas and proposed shorelines into factors in ArcGIS Pro⁷⁰. We then calculated the elevation of every level and plotted the outcomes to look at the place the continental zone transitions to the oceanic zone. For oceanic geomorphic options (polygons), we used the zonal statistics instrument in ArcGIS Pro to calculate the variety of pixels inside every polygon, complete space and the tenth–ninetieth percentile elevation values for every zone.

We extracted elevation information for the key international terrestrial rivers (195,022 information factors), international deltas (10,848 information factors), the continental shelf (14,820,634 information factors), the continental slope (7,606,463 information factors) and the important thing ocean flooring landforms, together with the continental rise (12,144,045 information factors), abyssal plain (116,749,407 information factors) and hadal zone (1,238,491 information factors) on Earth to establish the higher and decrease bounds of the continental shelf. On Mars, the evaluation included valley networks (3,294,322 information factors), depositional rivers (16,515 information factors), outlet canyons (248,865 information factors), deltas (48 information factors), and the Arabia (10,192 information factors) and Deuteronilus (42,900 information factors) shorelines, to determine the higher sure of the potential Martian shelf. On Earth, river deltas prograde throughout and relaxation atop continental cabinets, and the transition from these deltaic deposits to the deep ocean sometimes takes place throughout the higher 2.5 km under sea degree. We, due to this fact, use this depth interval to outline the search window for a possible shelf on Mars.

Raster to factors of elevation, slope and curvature

To receive elevation at every level of the resampled raster information, we used the ArcGIS ‘Add Surface Information’ instrument to pattern elevation values from grids at resolutions of two.5 km, 5 km and 10 km (refs. ^70,71; Supplementary Figs. 2 and three). For every level, the z-value is derived from its x–y coordinates on the underlying floor.

To calculate the steepness of every cell on each terrestrial and Martian surfaces, represented as raster grids, we used the ‘Slope’ instrument in ArcMap⁷¹. Slope (levels) was calculated in ArcMap (Spatial Analyst) utilizing the Slope instrument (3 × 3 neighbourhood), which estimates ∂z/∂x and ∂z/∂y utilizing a finite-difference gradient and computes ({S}^{^circ }=arctan (sqrt{({partial z/partial x)}^{2}+{(partial z/partial y)}^{2})})instances 57.29578).

Curvature reveals the form of the slope, indicating whether or not it’s convex (that’s, ridges and plateaus) or concave-up surfaces (that’s, valleys and depressions), which is especially helpful for figuring out transitions between panorama and seascape. We calculated curvature utilizing the ‘Curvature’ perform in ArcMap⁷¹, which inserts a aircraft to the 9 surrounding cells in a 3 × 3 window to find out floor curvature. The major output offers cell-by-cell curvature values (second by-product of elevation). The optimistic values point out upwardly convex surfaces, damaging values point out upwardly concave surfaces, and values close to zero symbolize flat or almost planar areas. In ArcGIS, curvature values are reported in models of 1 hundredth (1/100) of the DEM z-unit (right here, metres; z-factor = 1). To deal with concave and convex areas equally when it comes to magnitude, damaging values of each planets are multiplied by −1, permitting them to be plotted alongside optimistic values with completely different colors (Fig. 2).

Landscape classification

To map the transition zone between continental and oceanic landscapes on each Earth and Mars, we used the ‘Geomorphons’ instrument⁵³. This algorithm makes use of the idea of Local Ternary Patterns (LTP) to analyse terrain options by evaluating the elevation of every pixel with its neighbouring pixels⁵⁴. Instead of a easy binary comparability, LTP classifies variations into three classes: values which are (1) much like the centre pixel, (2) considerably greater or (3) considerably decrease. This strategy reduces the affect of noise and offers a extra nuanced illustration of terrain, capturing refined variations in pixel elevation. The Geomorphons instrument classifies every cell of an enter raster into frequent landforms, together with flat areas, ridges, shoulders, spurs, slopes, pits, footslopes, hollows and peaks.

On Earth, the transition sometimes happens on a comparatively flat floor, often called the continental shelf²⁹. To detect these flat surfaces, we used the Geomorphons instrument with a selected ‘flat terrain angle threshold’. The tectonic system of Earth, pushed by energetic plate tectonics, differs considerably from Mars, which lacks substantial tectonic exercise. This absence of tectonism on Mars leads to longer topographic wavelengths in contrast with Earth⁵². As a outcome, we utilized completely different flat-terrain angle thresholds for the 2 planets.

For Earth, we performed 40 Geomorphons experiments (Supplementary Table 2) to (1) classify the floor of Earth into shelf cells and non-shelf cells by evaluating the detected flat cells to the mapped continental shelf; (2) decide the flat angle threshold that absolutely detects the terrestrial continental shelf; and (3) set up a spread of shelf space detection at every angle threshold (Extended Data Figs. 2–4 and Supplementary Table 2), which can be used to evaluate the proportion and accuracy of shelf detection on Mars. In every experiment, we utilized a special flat-terrain angle threshold and located that the continental shelf was absolutely detected at an angle of 1.22° (Supplementary Table 2). However, growing the flat angle threshold led to the detection of each shelf and non-shelf areas. For occasion, we detected 100% of the shelf (32,308,476 km²), however it additionally recognized 238,719,226 km² of non-shelf areas, leading to a precision of 12%. We, due to this fact, used the experiments to set an accuracy vary for every flat angle threshold, which might be used to detect the Martian shelf (Extended Data Fig. 3 and Supplementary Table 2).

For Mars, no definitive maps of oceanic options exist, aside from two long-debated proposed shorelines^10,22,77. To map potential shelf-oceanic zones, we targeted on a area within the northern lowlands (Fig. 2c) characterised by a definite flat zone in contrast with its environment. This area additionally preserves 48 deltaic programs, a few of that are related to interpreted submarine-channel belts which are thought to have shaped alongside an historic oceanic margin^{4,10,14,15,16,17} (Supplementary Table 1). Moreover, it incorporates quite a few valley community termini and fluvial depositional ridges, each of that are believed to symbolize the endpoints of fluvial programs. The area additionally preserves the 2 debated shorelines. This zone is well-defined by elevation, starting from −1,800 m to −3,800 m, with a definite median slope of 0.31° at a grid decision of 5 km. We ran the instrument at this flat angle threshold and located that just about the complete northern lowland was marked as a comparatively flat floor. To refine the outcomes, we mixed our topographic evaluation (elevation, slope and curvature) with key morphological options (valley community termini, fluvial ridges, deltas and the 2 proposed shorelines). This allowed us to spatially constrain the flat floor zone between −1,800 m and −3,800 m to areas coinciding with geomorphic indicators of a landscape-to-seascape transition, principally situated between about 30 °S and about 70 °N. On Earth, a flat angle threshold of 0.31° would detect almost 69–71% of the continental shelf space, giving us confidence that the detected floor on Mars corresponds to an analogous transition (Supplementary Table 2).

Statistical evaluation

To take a look at whether or not floor steepness differs between elevation zones, we first computed median slope and curvature values at 200 m elevation intervals on each Earth and Mars. The ensuing Martian profiles present an intermediate-elevation, low-slope, low-curvature interval between −1,800 m and −3,800 m, bounded by greater slopes and curvature at elevations >−1,800 m and decrease slopes and curvature at elevations <−3,800 m (Fig. 2 and Extended Data Fig. 1). On this foundation, we outlined three elevation bands: >−1,800 m, −1,800 m to −3,800 m, and <−3,800 m. We then utilized a Kruskal–Wallis H take a look at⁸⁰ to those three elevation bands to quantify whether or not their slope distributions differ, with out assuming normality. The take a look at confirmed a extremely important distinction in median slopes (H = 27.50, P = 1.07 × 10⁻⁶; Extended Data Fig. 1), indicating that the slope populations of the three elevation zones are statistically distinct. The Kruskal–Wallis take a look at is used right here to evaluate solely the distinctness of elevation zones outlined from the slope–curvature–elevation relationship, to not find the breaks themselves.

Limitations

Our outcomes are topic to a number of limitations. One is the potential alteration of topography resulting from true polar wander and the emplacement of the Tharsis volcanic province, which most likely brought on uplift close to Tharsis and subsidence farther away^11,12. However, analysing the floor as geomorphic domains helps mitigate this, as a result of this deformation would have an effect on broad areas somewhat than the precise elevation ranges of landform mosaics. A second supply of uncertainty is the isostatic response to ocean unloading, which on Earth can modify elevations by a number of hundred metres following ocean retreat⁸¹. However, current estimates for Mars counsel that isostatic rebound most likely ranged from a number of tens to simply greater than 100 m (ref. ¹²). Yet, the roughly 2 km elevation span of our detected shelf-like zone exceeds anticipated rebound estimates and stays in line with depositional options. A 3rd supply of uncertainty is long-term burial, exhumation and erosion⁴⁶. Although these processes might have launched regional variability, they’re unlikely to change the broader topographic patterns we establish on the international scale. A fourth supply of limitation is the erosion and sediment redistribution alongside the dichotomy boundary by Hesperian-aged outflow floods^13,48, which most likely deposited substantial volumes of sediment alongside the northern dichotomy, notably in Chryse Planitia. These outflow occasions most likely contributed to domestically flattening the floor there. However, equally flat, low-slope surfaces are additionally current at different key websites, comparable to Aeolis Dorsa—wealthy in stacked deposits of various origins and interpretations, together with fluvial and deltaic deposits, and presumably even submarine deposits^{15,16,17,82,83,84,85,86}—and alongside the remaining segments of the proposed shelf. This broader distribution, along with impartial proof for sea-level modifications recorded by deltaic deposits at Hypanis¹⁴, means that though Hesperian-aged outflow floods helped flatten the floor in Chryse Planitia, they had been most likely not the first reason behind floor flattening throughout the northern lowlands.