Tomasz F. Stepinski

Machine Detection of Martian Impact Craters From Digital Topography Data

Research on automatic identification of impact craters on Mars and other planetary bodies has concentrated on detecting them from imagery data. We present a novel approach to crater detection that utilizes digital topography data instead of images. Craters are delineated by topographic curvature. Thresholding maps of curvature transforms topographic data into a binary image, from which craters are identified using a combination of segmentation and detection algorithms. We apply our method to a large and technically demanding test site and compare the results to the existing catalog of manually identified craters. Our algorithm finds many small craters not listed in the manual catalog, but it fails to detect heavily degraded craters. A detailed quality assessment of the algorithm is presented. The topography-based crater-detection algorithm offers a relatively simple and ready-to-use tool for identification and characterization of fresh impact craters with an adequate performance for the immediate application to Martian geomorphology

Automated classification of landforms on Mars

We propose a numerical method for classification and characterization of landforms on Mars. The method provides an alternative to manual geomorphic mapping of the Martian surface. Digital elevation data is used to calculate several topographic attributes for each pixel in a landscape. Unsupervised classification, based on the self-organizing map technique, divides all pixels into mutually exclusive and exhaustive landform classes on the basis of similarity between attribute vectors. The results are displayed as a thematic map of landforms and statistics of attributes are used to assign semantic meaning to the classes. This method is used to produce a geomorphic map of the Terra Cimmeria region on Mars. We assess the quality of the automated classification and discuss differences between results of automated and manual mappings. Potential applications of our method, including crater counting, landscape feature search, and large scale quantitative comparisons of Martian surface morphology, are identified and evaluated.

Morphology of drainage basins as an indicator of climate on early Mars

[1] We show that drainage basin morphology correlates with climate. Computational analysis of the 46 basins extracted from the western slopes of the Andes reveals the existence of four different basin morphologic classes. This purely geomorphic partition correlates with division of the same basins on the grounds of climate types. Basins are compared using circularity functions as their formal representations. Self-organizing maps and dendrograms are employed to provide basin classification. One class of basin morphologies corresponds to sites in the arid Atacama Desert, and the other class corresponds to sites in the Atacama exhibiting groundwater sapping landforms. Using the same technique, we study a larger sample of 94 basins that, in addition to the Andean basins, includes other terrestrial basins and 26 basins from Martian sites that show prominent valley networks. The classification of this larger set shows that morphologies of Martian and terrestrial basins bifurcate at the root of the dendrogram, forming two separate domains of basin morphologies. The similarity map reveals that, of all the terrestrial basins, the Atacama Desert basins are morphologically closest to the Martian basins. Extrapolating the terrestrial morphology-climate linkage to Mars points to formation of valley networks in a hyperarid climatic environment. We submit that the Atacama Desert provides the best possible terrestrial morphologic analog to valley network sites on Mars. We discuss climatic and hydrologic particularities of the Atacama Desert and hypothesize that a similar environment existed on early Mars.

Automatic mapping of valley networks on Mars

Martian valley networks bear some resemblance to terrestrial drainage systems, but their precise origin remains an active research topic. A limited number of valley networks have been manually mapped from images, but the vast majority remains unmapped because standard drainage mapping algorithms are inapplicable to valleys that are poorly organized and lack spatial integration. In this paper, we present a novel drainage delineation algorithm specially designed for mapping the valley networks from digital elevation data. It first identifies landforms characterized by convex tangential curvature, and then uses a series of image processing operations to separate valleys from other features having a convex form. The final map is produced by reconnecting all valley segments along drainage directions. Eight test sites on Mars are selected and manually mapped for valley networks. The algorithm is applied to the test sites and delineated networks are compared to mapped networks using a series of quantitative quality factors. We have found a good agreement between delineated and mapped networks. In the process of comparing manual and delineated networks some shortcomings of manual mapping became apparent. We argue that delineated networks are indeed of better quality than the networks manually mapped from images. Although the algorithm has been developed to study Martian surface, it may also be relevant to terrestrial geomorphology.

The υ Andromedae System: Models and Stability

Radial velocity observations of the F8 V star υ Andromedae taken at Lick and at Whipple Observatories have revealed evidence of three periodicities in the line-of-sight velocity of the star. These periodicities have been interpreted as evidence for at least three low-mass companions (LMCs) revolving around υ Andromedae. The mass and orbital parameters inferred for these companions raise questions about the dynamical stability of the system. We report here results from our independent analysis of the published radial velocity data, as well as new unpublished data taken at Lick Observatory. Our results confirm the finding of three periods in the data. Our best fits to the data, on the assumption that these periods arise from the gravitational perturbations of companions in Keplerian orbits, are also generally in agreement but with some differences from the earlier findings. We find that the available data do not constrain well the orbital eccentricity of the middle companion in a three-companion model of the data. We also find that in order for our best-fit model to the Lick data to be dynamically stable over the lifetime of the star (~2 billion years), the system must have a mean inclination to the plane of the sky greater than 13°. The corresponding minimum inclination for the best fit to the Whipple data set is 19°. These values imply that the maximum mass for the outer companion can be no greater than about 20 Jupiter masses. Our analysis of the stability of the putative systems also places constraints on the relative inclinations of the orbital planes of the companions. We comment on global versus local (i.e., method of steepest descent) means of finding best-fit orbits from radial velocity data sets.

Finding regional co-location patterns for sets of continuous variables in spatial datasets

This paper proposes a novel framework for mining regional co-location patterns with respect to sets of continuous variables in spatial datasets. The goal is to identify regions in which multiple continuous variables with values from the wings of their statistical distribution are co-located. A co-location mining framework is introduced that operates in the continuous domain and which views regional co-location mining as a clustering problem in which an externally given fitness function has to be maximized. Interestingness of co-location patterns is assessed using products of z-scores of the relevant continuous variables. The proposed framework is evaluated by a domain expert in a case study that analyzes Arsenic contamination in Texas water wells centering on regional co-location patterns. Our approach is able to identify known and unknown regional co-location patterns, and different sets of algorithm parameters lead to the characterization of Arsenic distribution at different scales. Moreover, inconsistent colocation sets are found for regions in South Texas and West Texas that can be clearly attributed to geological differences in the two regions, emphasizing the need for regional co-location mining techniques. Moreover, a novel, prototype-based region discovery algorithm named CLEVER is introduced that uses randomized hill climbing, and searches a variable number of clusters and larger neighborhood sizes.

Diversity of planetary systems from evolution of solids in protoplanetary disks

We have developed and applied a model designed to track simultaneously the evolution of gas and solids in protoplanetary disks from an early stage, when all solids are in the dust form, to the stage when most solids are in the form of a planetesimal swarm. The model is computationally ecient and allows for a global, comprehensive approach to the evolution of solid particles due to gas-solid coupling, coagulation, sedimentation, and evaporation/condensation. The co-evolution of gas and solids is calculated for 10 7 yr for several evolution regimes and starting from a comprehensive domain of initial conditions. The output of a single evolutionary run is a spatial distribution of mass locked in a planetesimal swarm. Because swarms mass distribution is related to the architecture of a nascent planetary system, diversity of swarms is taken as a proxy for a diversity of planetary systems. We have found that disks with low values of specic angular momentum are bled out of solids and do not form planetary systems. Disks with high and intermediate values of specic angular momentum form diverse planetary systems. Solar-like planetary systems form from disks with initial masses0.02 M and angular momenta3 10 52 gc m 2 s 1 . Planets more massive than Jupiter can form at locations as close as 1 AU from

Subkilometer crater discovery with boosting and transfer learning

Counting craters in remotely sensed images is the only tool that provides relative dating of remote planetary surfaces. Surveying craters requires counting a large amount of small subkilometer craters, which calls for highly efficient automatic crater detection. In this article, we present an integrated framework on autodetection of subkilometer craters with boosting and transfer learning. The framework contains three key components. First, we utilize mathematical morphology to efficiently identify crater candidates, the regions of an image that can potentially contain craters. Only those regions occupying relatively small portions of the original image are the subjects of further processing. Second, we extract and select image texture features, in combination with supervised boosting ensemble learning algorithms, to accurately classify crater candidates into craters and noncraters. Third, we integrate transfer learning into boosting, to enhance detection performance in the regions where surface morphology differs from what is characterized by the training set. Our framework is evaluated on a large test image of 37,500 × 56,250 m2 on Mars, which exhibits a heavily cratered Martian terrain characterized by nonuniform surface morphology. Empirical studies demonstrate that the proposed crater detection framework can achieve an F1 score above 0.85, a significant improvement over the other crater detection algorithms.

Generation of dynamo magnetic fields in the primordial solar nebula

Abstract We study the problem of the existence of dynamo-generated magnetic fields in the primordial solar nebula. The combined action of Keplerian rotation and helical convection enables an αω dynamo to generate large-scale magnetic fields in parts of the nebula where levels of electrical conductivity are high enough to provide coupling between the gas and the magnetic field. The aim of this paper is to identify those regions of the nebula where an αω dynamo is able to maintain a magnetic field for a relatively long period of time. We calculate the electrical conductivity of nebular gas and subsequently the radial distribution of the local dynamo number for two specific nebular models—a viscous accretion disk model, and the quiescent minimum mass nebula. Our calculations show that magnetic fields can be easily generated and maintained by an αω dynamo in the inner and outer parts of the nebula; however, in the middle region of the nebula, between, say, 2 and 5 AU, the existence of dynamo-generated fields is questionable.

Formation of giant planets in disks with different metallicities

We present the first results from simulations of processes leading to planet formation in protoplanetary disks with different metallicities. For a given metallicity, we construct a two-dimensional grid of disk models with different initial masses and radii (M0, R0). For each disk, we follow the evolution of gas and solids from an early evolutionary stage, when all solids are in the form of small dust grains, to the stage when most solids have condensed into planetesimals. Then, based on the core accretion - gas capture scenario, we estimate the planet-bearing capability of the environment defined by the final planetesimal swarm and the still evolving gaseous component of the disk. We define the probability of planet-formation, Pp, as the normalized fractional area in the (M0 ,l ogR0) plane populated by disks that have formed planets inside 5 AU. With such a definition, and under the assumption that the population of planets discovered at R 5 AU, our results agree fairly well with the observed dependence between the probability that a star harbors a planet and the stars metal content. The agreement holds for the disk viscosity parameter α ranging from 10 −3 to 10 −2 , and it becomes much poorer when the redistribution of solids relative to the gas is not allowed for during the evolution of model disks.