Ancient Geodynamics and Global-Scale Hydrology on Mars

See allHide authors and affiliations

Science  30 Mar 2001:
Vol. 291, Issue 5513, pp. 2587-2591
DOI: 10.1126/science.1058701


Loading of the lithosphere of Mars by the Tharsis rise explains much of the global shape and long-wavelength gravity field of the planet, including a ring of negative gravity anomalies and a topographic trough around Tharsis, as well as gravity anomaly and topographic highs centered in Arabia Terra and extending northward toward Utopia. The Tharsis-induced trough and antipodal high were largely in place by the end of the Noachian Epoch and exerted control on the location and orientation of valley networks. The release of carbon dioxide and water accompanying the emplacement of ∼3 × 108 cubic kilometers of Tharsis magmas may have sustained a warmer climate than at present, enabling the formation of ancient valley networks and fluvial landscape denudation in and adjacent to the large-scale trough.

The western hemisphere of Mars is dominated by the Tharsis rise, a broad elevated (∼10 km) region extending over 30 million square kilometers. Tharsis is the locus of large-scale volcanism and pervasive fracturing that resulted from the loading of the lithosphere, or outer elastic shell, by voluminous extrusive and intrusive magmatic deposits (1–3). Here we use recently acquired gravity (4) and topography (5) data from the Mars Global Surveyor (MGS) spacecraft (6) to determine the effect of the mass load of Tharsis on the shape and gravity field of the rest of the planet. We test the hypothesis that the deformational response to the Tharsis load is responsible for the topographic trough and the heretofore unexplained ring of negative gravity anomalies (Figs. 1 and2A) (7) that surround the Tharsis rise, as well as for the major gravity and topographic highs that are antipodal to Tharsis. We examine the influence that Tharsis may have had on the timing, orientation, and location of fluvial features on the planet. Because of the enormous mass of the Tharsis load, understanding the global history of Mars requires understanding the role that Tharsis played in that history; the formation of Tharsis may have been an exceptional phenomenon in the evolution of the terrestrial planets.

Figure 1

Gravity anomaly image (4) draped over a three-dimensional (3D) view of topography (5) centered on Tharsis at 260°E longitude. This and all subsequent figures are in sinusoidal projection. The gravity anomaly image has been saturated at ±300 milligals (mGal) and is expanded to l= 60. The prominent topographic feature in the center is the Tharsis rise and its volcanic constructs; the approximate boundary to Tharsis is shown as a dashed line. Valles Marineris extend eastward from Tharsis. Negative portions of the gravity anomaly field form a ring around Tharsis. A region antipodal to Tharsis (seen at the right and left sides of the map near 80°E longitude) is centered on Arabia Terra, which is both a topographic and a gravitational high.

Figure 2

(A) Observed and (B) modeled negative gravity anomaly ring around Tharsis, centered at 260°E longitude. The irregular boundary around Tharsis in the model results from defining the load boundary with a quarter-degree topographic grid (10). (C) Observed and (D) modeled gravity anomalies draped over a 3D view of observed topography centered on Arabia Terra at 80°E longitude. The observed gravity anomaly over Tharsis is shown in the model image. Different scales for observed and modeled fields result from an overprediction of model gravity. For (A) through (D), gravity fields are expanded to l = 10 independently inside and outside of the load boundary.

To test the hypothesis that Mars displays a global deformational response to Tharsis loading, we use a spherical harmonic model (8) of the loading of a spherical elastic shell (9) and isolate the spatially variable Tharsis topography (Fig. 3A) as the only load on the planet (10). The degree of compensation (11) of the load is about 95% (12), an assumption that produces consistency between predicted and observed topography but overpredicts the magnitudes of the associated gravity anomalies. Because we are interested primarily in the spatial correlation of models to observed fields, rather than model amplitudes, the mismatch of predicted and observed gravity is secondary to this discussion (13).

Figure 3

Observed martian topography displayed for (A) the Tharsis and (B) the anti-Tharsis hemispheres compared with modeled topography (tol = 120) for (C) the Tharsis and (D) the anti-Tharsis hemispheres. For the model, actual topography is shown in the Tharsis region. All figures are draped over a 3D view of shaded relief.

The locations of the observed and modeled ring of negative gravity anomalies (Fig. 2, A and B) around Tharsis, expanded to spherical harmonic order and degree 10 (l = 10), are consistent and include several of the intermediate-wavelength features within the ring (such as relative lows to the northwest, northeast, and east, and the relative high to the south). The model also predicts a topographic trough (here termed the Tharsis trough) around Tharsis (Fig. 3C). Such a topographic depression does surround Tharsis over at least 270° of azimuth (Fig. 3A). To the east of Tharsis, this trough extends northward from the Argyre impact basin, through Chryse and Acidalia Planitiae, to the North Polar basin. To the northwest of Tharsis, Arcadia and Amazonis Planitiae comprise the depression, whereas southwest of Tharsis, the depression becomes less evident (14). The fact that the ring of negative gravity anomalies persists to the southwest of Tharsis indicates that more than the surface topography is required to explain the gravity data here; one possibility is that the trough has been filled with sediments that are lower in density than average crustal material.

Antipodal to Tharsis, the model predicts a topographic high (here termed the Arabia bulge) over the elevated Arabia Terra (Fig. 3D); the predicted high also extends over the Utopia basin to the north. The planetary topographic dichotomy of a northern lowland and a southern upland (5) is not an element of a Tharsis loading model; however, the anomalously broad western rim of the Utopia basin (Fig. 3B) may be an expression of uplift resulting from Tharsis loading. The Hellas impact basin is also outside the realm of the model, yet both the modeled and observed topography show a downward slope in Hellas rim topography (5) toward the South Pole. This slope, in the direction of Tharsis, suggests that a portion of the Hellas rim underwent vertical motion during the formation of the trough created by the Tharsis load. In both the observed and modeled gravity anomaly fields (l = 10), there is a gradient across Hellas and a high over Arabia Terra that extends northward to the Utopia basin, mirroring features in the long-wavelength topography (Fig. 2, C and D). The modeled gravity clearly does not account for the smaller scale, positive gravity anomalies at Elysium and the central Utopia basin, which reflect subsurface structures characterized, respectively, by volcanically thickened crust and a combination of crustal thinning during basin formation and infill of the basin depression (2). However, we conclude that the long-wavelength, nonhydrostatic gravity field of Mars is explained simply by the Tharsis load and the resulting global deformation of the lithosphere. Further, the shape of Mars is determined by these two quantities plus the northward pole-to-pole slope that formed in earliest martian history (5).

Extensional structures radial to the Tharsis rise and compressional structures generally concentric to the rise constitute the majority of the tectonic features in the Tharsis region (15). About half of these features are Noachian (16) in age, suggesting that tectonic activity peaked early and decreased with time (17). The positions and orientations of both types of structures are matched by elastic shell loading models (18) constrained by current gravity and topography fields (3). Successful models are able to predict strain levels comparable to those observed in Noachian structures (19) and require that the extent of the load in the Noachian be comparable to that at present. Thus, the overall Tharsis load must have been largely in place by the Late Noachian. Because they are the direct response to the Tharsis load, the Tharsis trough and Arabia bulge must also have existed since Noachian time.

The development of the Tharsis trough and Arabia bulge thus should have influenced the location and orientation of martian valley networks and outflow channels. Valley networks are the most common drainage systems on Mars (20). Their similarity to terrestrial river systems suggests that the genesis of valley networks involved fluvial erosion, although the style of this erosion (such as surface runoff, groundwater discharge, or sapping) remains controversial. Valley network systems are confined mainly to the southern highlands on Noachian terrain and display variability in the number of tributaries, stream order, and planimetric form (21). Additionally, nearly all martian outflow channels (20) originate in or flow into the Tharsis trough (Fig. 3A).

We tested for control of valley network orientations by Tharsis loading. Gradient directions of both modeled [augmented with the pole-to-pole slope (5)] and observed topography were calculated at the downstream ends of valley network trunks. The level of agreement of topographic gradients with valley network downstream azimuths (22) was evaluated by forming the dot products, cos δ, of unit directional vector pairs (23). Histograms of cos δ distributions for the modeled (Fig. 4A) and observed (Fig. 4B) topography (l = 10) support significant Tharsis influence on long-wavelength topography and valley network azimuths. Because of the multiscale morphology of valley network systems, any assessment of azimuth will depend on baseline length. Our estimates of azimuth are based on the long-baseline orientation of the trunk valley. As such, they are most sensitive to slopes with length scales comparable to the size of the basin drained by the valley network system, and not to slopes with shorter length scales (24). For only a subset (∼60%) of valley network systems—where local topography is not dominated by short length scales—do we find that valley network azimuths agree with the gradient directions of the observed topography. Thus, it is appropriate to consider the relative success of modeled topography as compared to observed topography in predicting valley network azimuths. Using the two histogram modes (0.8 ≤ cos δ ≤ 1.0), the model-to-topography ratio is 100% for an expansion limit of l = 10 (∼1000 km minimum length scale) and exceeds 70% at l = 180 (∼50 km minimum length scale) (Fig. 4C). We conclude that the model (Fig. 4D) does a satisfactory job of explaining valley network azimuths and that many of these systems must have formed after a significant fraction of the Tharsis load was in place.

Figure 4

(A) Histogram of cos δ [dot product of model gradient (l = 10) and valley network azimuth unit vectors] distribution. (B) Histogram as in (A), but for observed topography. The P value for the null hypothesis of a uniform distribution is less than 10−16for both histograms. (C) The model-to-topography ratio (%) for the modes of the histograms as a function of spherical harmonic expansion limit l. (D) Black arrows correspond to valley networks in (A) for cos δ ≥ 0.8. The background color image is the model response to Tharsis loading (l = 10) that has been saturated at ±3 km. Central longitude is 0°, and observed topography is shown in the Tharsis region. Pole-to-pole slope has been included in the topographic model, which has been referenced to the observed geoid.

Valley networks were examined in detail (25) in Margaritifer Sinus, a region on the flank of the Arabia bulge and in the Tharsis trough (Fig. 3B). Most formed on regions of relatively high topographic gradient on the flanks of the trough. The majority (∼85%) of observed valley networks here likely formed in Late Noachian time, between ∼4.3 to 3.85 billion years ago (Ga) and ∼3.8 to 3.50 Ga (26), although the possibility exists that earlier valley networks in this region were destroyed by a high impact flux or alternative erosion mechanisms. Because many of these valley network orientations are controlled by Tharsis-induced slopes, the Tharsis load must be largely Noachian in age, which is consistent with inferences made earlier. Superposition and sequence relationships indicate that the valley networks whose azimuths are not explained by the model are nevertheless contemporaneous with the Tharsis-controlled valley networks (27). The formation of valley networks in Margaritifer Sinus is intimately associated with a Late Noachian, large-scale erosion event on the flanks of the Tharsis trough that stripped at least 1.5 × 106 km3 of material from this area, leaving behind numerous mesas of Early and Middle Noachian terrain (25). This same process may have operated in adjacent areas of northwestern Arabia Terra, potentially removing an additional 3 × 106 km3 of the upper crust. Moreover, a global episode of high erosion rates at the end of the Noachian Epoch has been postulated on the basis of crater size-frequency statistics and morphology (28). Thus, there is evidence that the Late Noachian was a time of unusually high landscape modification rates on Mars. The most likely erosional agent is water at or near the surface, which is consistent with a warmer, wetter environment than at present (29).

The release of CO2 and H2O to the martian atmosphere from magma erupted and intruded during Tharsis formation may have affected the Noachian climate. For the 100-km-thick elastic lithosphere we assume (2), the observed topography plus the volume of material contained within the depression resulting from Tharsis loading amounts to ∼3 × 108 km3of igneous material, which is equivalent to a 2-km-thick global layer (30). For a magmatic CO2 content of 0.65 weight percent (wt %) [which is consistent with Hawaiian basaltic lavas (31)] and an H2O content of 2 wt % (32), the total release of gases from Tharsis magmas could produce the integrated equivalent of a 1.5-bar CO2atmosphere and a 120-m-thick global layer of water. These quantities of volatiles are sufficient to warm the atmosphere to the point at which liquid water is stable at the surface (33). The accumulation of atmospheric CO2 may have made the latter part of the Noachian the most favorable time for this condition.

As the emplacement rate of Tharsis volcanic material declined at the end of the Noachian, CO2 and H2O would have been removed from the atmosphere by a combination of impact ejection to space, stripping by the solar wind, thermal escape, and the formation of carbonate minerals on the surface or within the crust (34, 35). Although their relative importance cannot be determined, there is evidence that each of these processes has occurred, and it is possible that, combined, they could have brought clement conditions to an end by removing substantial quantities of atmosphere in less than a few hundred million years (35). With several events occurring at the end of the Noachian, such as the decline in impact rate, initiation of sputtering as the internal magnetic field disappears (35), waning of Tharsis volcanism, and a decrease in surface erosion rate, it is difficult to determine which were causally related and which were coincidental. Nevertheless, it is possible that during the Noachian Epoch, the structural and magmatic events associated with Tharsis evolution were the sine qua non that linked fluvial, geodynamical, and climate activity.


View Abstract

Stay Connected to Science

Navigate This Article