Technical Comments

Comment on "Athabasca Valles, Mars: A Lava-Draped Channel System"

See allHide authors and affiliations

Science  20 Jun 2008:
Vol. 320, Issue 5883, pp. 1588
DOI: 10.1126/science.1154849


Jaeger et al. (Reports, 21 September 2007, p. 1709) presented images of the Athabasca Valles channel system on Mars and asserted that the observed deposits are composed of thin, fluid lavas. However, all the features they described are secondary and postdate the surface by many millions of years, as documented by structural relationships with small, young impact craters.

Jaeger et al. (1) presented images of Athabasca Valles, Mars, acquired by the High-Resolution Imaging Science Experiment (HiRISE) camera, which they contend resolve the “ice or lava” debate that has surrounded the deposits in this region for three decades (24). Their case rests on observations of ring-mound landforms (RMLs) and polygonal terrain, both of which they interpret to be primary volcanic landforms formed during deposition. This interpretation is part of a wider general theory of a volcanic Cerberus plain based on morphology (5, 6), numerical modeling (7), and implied association with young martian meteorites (8). We have demonstrated that the RMLs and polygonal terrain central to this interpretation are both postdepositional in origin [cf. (9, 10)].

Impact craters postdate the surface they occupy. Hence, any landform that superposes an impact crater must itself be postdepositional and postdate impact: a sign of relative age in all but the most contrived scenarios (11). Figure 1 shows that both RMLs and polygons superpose impact craters. This simple observation exposes the missing step in the authors' argument, these landforms and the substrates separated by time (and in all probability a great deal of it, as judged by the fraction of craters affected). Any explanation that does not acknowledge this superposition relation lacks geological validity. Hence, the assertion by Jaeger et al. that the RMLs formed explosively during lava flow emplacement and that polygons formed where lava cracked under the weight of RMLs is inconsistent with the observations, supposed cause and effect separated by an indeterminate period of time documented by the formation of large numbers of impact craters.

Fig. 1.

RMLs and polygons across Athabasca Valles and Amazonis Planitia. North to top, scale bars 100 m. (A) HiRISE PSP_002292_1875 (28 cm per pixel); two RMLs superposing a small impact crater. (Note undeformed RMLs.) (B) HiRISE PSP_002661_1895 (30 cm per pixel); large RML superposed by an oblique impact crater. (Note destruction of right-hand wall of RML and distension of left-hand wall by high-velocity ejecta.) (C) Context of (A). (D) HiRISE PSP_002371_1890 (28 cm per pixel); impact crater truncated by an RML [same image as shown in figures 3C and 4A in (1)]. (E) Context of (B). Note the regular polygonal pattern of the substrate, which suggests that polygons formed before RMLs (otherwise the distinctive sculpture would not form where RMLs are absent). (F and G) MOC R23-00683 and R14-01144 (3.5 m and 4.7 m per pixel); impact craters superposed by polygonal sculpture cutting rims. (H) MOC S09-02331 (3.19 m per pixel); composite image showing polygonal sculpture superposing impact craters [cf. (E)]. (I) RML from same MOC image (S09-02331), 2 km south of (H) [cf. 1J]. (J) Figure 3 from (1) showing typology of polygons.

Impact craters like those observed at Athabasca Valles are created by hypervelocity (>1 km s–1) impact, an inherently nonconservative event that vaporizes the impactor and obliterates preexisting surface features out to 1.6 crater radii (12). Lower-velocity impactors excavate a crater only slightly larger than the projectile itself and would have obliterated these landforms on contact. In both cases, the survival of preexisting landforms within the crater rim (as for the RMLs of Fig. 1, A and D, and Fig. 2, A and C) is not possible (9). This raises two questions, one stratigraphic, one energetic: How can impact craters get “under” primary-formed polygonal sculpture in solid rock, and how do RMLs within the rim survive impact?

Fig. 2.

Stereo red/green anaglyphs of Fig. 1A, B, and D in Athabasca Valles. (A) PSP_002292_1875 and PSP_002147_1875. (B) PSP_002661_1895 and PSP_003924_1895. (C) PSP_002371_1890 and PSP_001540_1890. Context views directly to right (white boxes).

To clarify the true effects of such impact, we show an RML superposed by an impact crater (Figs. 1B and 2B). This RML is clearly deformed by the impact because its eastern wall is completely obliterated at the impact point, and even the wall opposite is shifted outward. Contrast this with Figs. 1A and 2A, where the RMLs within the rim are not deformed, even though the crater is larger (indicating more energetic impact), and Figs. 1D and 2C, where the crater is one-third covered by an RML. These craters are all secondary craters from the primary impact Zunil, 800 km to the east (13), and formed within minutes of each other.

The suggestion by Jaeger et al. that the HiRISE images do not bear out the superposition relations described by (9) seems to reflect a misunderstanding about how such superposition should be expressed, and its importance. That some RMLs superpose impact craters whereas others are superposed by impact craters is not surprising: the RMLs are interpreted to be intrusive, permafrost features in a variety of stages of growth and decay (9, 14). If some are cut by impact craters, then such RMLs precede cratering, but only one need cut a crater rim to establish the postdepositional origin of the whole. That other supposedly primary landforms in these deposits (i.e., the polygons) superpose thousands of impact craters (10) only serves to reinforce this observation.

Where the authors consider the relationship of the RMLs to the substrate, they illustrate, perhaps inadvertently, how each RML is located within a six-sided polygon [figure 3 in (1) and Fig. 1J). They draw attention to the form of these polygons but fail to relate this to structure, a regional view (Fig. 1, F to H) showing that these too superpose craters, close to 100% of craters in Fig. 1H postdated this way. At resolution (3.19 m/pixel), one might argue that these craters simply appear superposed by the polygons, with age having erased all proximal-medial ejecta. However, the largest crater in Fig. 1H (diameter = 400 m, at far right), with polygonal sculpture continuous with that in the substrate reaching right up to the crater rim (height = ∼30 m), suggests otherwise.

It is difficult to argue against the fact that many millions of years have passed between deposition and the formation of polygons and RMLs. Neither is there any reason to suppose that this distinctive landform assemblage is secondary here and primary elsewhere in these deposits.

The authors suggest that previous evidence for superposition of the crater (9, 10) in Fig. 1A [figure S7B in (1)] is based on the circularity of the crater. However, it is the fact that the landform is formed within the crater rim that is crucial, not the shape of the crater. Jaeger et al. further suggest that ejecta seen on top of the RML show that the crater is younger. However, surface material would remain on top of an intrusive landform. Thus, their observation is not discriminatory of process.

Although the authors make a number of interesting morphological comparisons and rightly point out that the surface roughness, thermal inertia, and lack of a hydrogen signal appear consistent with volcanism, each of these observations has multiple possible explanations. They state that HiRISE allows “divergent hypotheses to be tested” but focus on those observations that support their interpretation while ignoring or dismissing those that do not.

It is a geological truism that we objectively record the nature of the stratigraphic record before considering the processes that were responsible for what we observe (15). The pitfalls of not doing so, of taking a “genetic” approach in which theory drives observation rather than the other way around, are misinterpretation of geological history and hypotheses that are never properly tested. A stratigraphic approach proceeds free of all but the most basic of hypotheses: that geological events can be ordered in space and time. Superposition tells us the order in which events happened. However much individual or collective landform similarities impress us, if interpretation is not consistent with stratigraphy, then it did not happen that way and may not have happened at all. These are not lavas.

References and Notes

View Abstract

Navigate This Article