RESUMO
The large volume of high-resolution images acquired by the Mars Reconnaissance Orbiter has opened a new frontier for developing automated approaches to detecting landforms on the surface of Mars. However, most landform classifiers focus on crater detection, which represents only one of many geological landforms of scientific interest. In this work, we use Convolutional Neural Networks (ConvNets) to detect both volcanic rootless cones and transverse aeolian ridges. Our system, named MarsNet, consists of five networks, each of which is trained to detect landforms of different sizes. We compare our detection algorithm with a widely used method for image recognition, Support Vector Machines (SVMs) using Histogram of Oriented Gradients (HOG) features. We show that ConvNets can detect a wide range of landforms and has better accuracy and recall in testing data than traditional classifiers based on SVMs.
RESUMO
Morphological evidence for ancient channelized flows (fluvial and fluvial-like landforms) exists on the surfaces of all of the inner planets and on some of the satellites of the Solar System. In some cases, the relevant fluid flows are related to a planetary evolution that involves the global cycling of a volatile component (water for Earth and Mars; methane for Saturn's moon Titan). In other cases, as on Mercury, Venus, Earth's moon, and Jupiter's moon Io, the flows were of highly fluid lava. The discovery, in 1972, of what are now known to be fluvial channels and valleys on Mars sparked a major controversy over the role of water in shaping the surface of that planet. The recognition of the fluvial character of these features has opened unresolved fundamental questions about the geological history of water on Mars, including the presence of an ancient ocean and the operation of a hydrological cycle during the earliest phases of planetary history. Other fundamental questions posed by fluvial and fluvial-like features on planetary bodies include the possible erosive action of large-scale outpourings of very fluid lavas, such as those that may have produced the remarkable canali forms on Venus; the ability of exotic fluids, such as methane, to create fluvial-like landforms, as observed on Saturn's moon, Titan; and the nature of sedimentation and erosion under different conditions of planetary surface gravity. Planetary fluvial geomorphology also illustrates fundamental epistemological and methodological issues, including the role of analogy in geomorphological/geological inquiry.
RESUMO
Volcanic rootless cones are the products of thermohydraulic explosions involving rapid heat transfer from active lava (fuel) to external sources of water (coolant). Rootless eruptions are attributed to molten fuel-coolant interactions (MFCIs), but previous studies have not performed systematic investigations of rootless tephrostratigraphy and grain-size distributions to establish a baseline for evaluating relationships between environmental factors, MFCI efficiency, fragmentation, and patterns of tephra dispersal. This study examines a 13.55-m-thick vertical section through an archetypal rootless tephra sequence, which includes a rhythmic succession of 28 bed pairs. Each bed pair is interpreted to be the result of a discrete explosion cycle, with fine-grained basal material emplaced dominantly as tephra fall during an energetic opening phase, followed by the deposition of coarser-grained material mainly as ballistic ejecta during a weaker coda phase. Nine additional layers are interleaved throughout the stratigraphy and are interpreted to be dilute pyroclastic density current (PDC) deposits. Overall, the stratigraphy divides into four units: unit 1 contains the largest number of sediment-rich PDC deposits, units 2 and 3 are dominated by a rhythmic succession of bed pairs, and unit 4 includes welded layers. This pattern is consistent with a general decrease in MFCI efficiency due to the depletion of locally available coolant (i.e., groundwater or wet sediments). Changing conduit/vent geometries, mixing conditions, coolant and melt temperatures, and/or coolant impurities may also have affected MFCI efficiency, but the rhythmic nature of the bed pairs implies a periodic explosion process, which can be explained by temporary increases in the water-to-lava mass ratio during cycles of groundwater recharge.
RESUMO
A new mapping-based study of the 900-km-long Mangala Valles outflow system was motivated by the availability of new high-resolution images and continued debates about the roles of water and lava in outflow channels on Mars. This study uses photogeologic analysis, geomorphic surface mapping, cratering statistics, and relative stratigraphy. Results show that Mangala Valles underwent at least two episodes of fluvial activity and at least three episodes of volcanic activity during the Late Amazonian. The occurrence of scoured bedrock at the base of the mapped stratigraphy, in addition to evidence provided by crater retention ages, suggests that fluvial activity preceded the deposition of two of the volcanic units. Crater counts performed at 30 locations throughout the area have allowed us to construct the following timeline: (1) formation of Noachian Highlands and possible initial flooding event(s) before ~1 Ga, (2) emplacement of Tharsis lava flows in the valley from ~700 to 1000 Ma, (3) a megaflooding event at ~700-800 Ma sourced from Mangala Fossa, (4) valley fill by a sequence of lava flows sourced from Mangala Fossa ~400-500 Ma, (5) another megaflooding event from ~400 Ma, (6) a final phase of volcanism sourced from Mangala Fossa ~300-350 Ma, and (7) emplacement of eolian sedimentary deposits in the northern portion of the valley ~300 Ma. These results are consistent with alternating episodes of aqueous flooding and volcanism in the valles. This pattern of geologic activity is similar to that of other outflow systems, such as Kasei Valles, suggesting that there is a recurring, and perhaps coupled, nature of these processes on Mars.