Search

Data analysis methods have scarcely kept pace with the rapid increase in Earth observations, spurring the development of novel algorithms, storage methods, and computational techniques. For scientists interested in Mars, the problem is always the same: there is simultaneously never enough of the right data and an overwhelming amount of data in total. Finding sufficient data needles in a haystack to test a hypothesis requires hours of manual data screening, and more needles and hay are added constantly. To date, the vast majority of Martian research has been focused on either one-off local/regional studies or on hugely time-consuming manual global studies. Machine learning in its numerous forms can be helpful for future such work. Machine learning has the potential to help map and classify a large variety of both features and properties on the surface of Mars and to aid in the planning and execution of future missions. Here, we outline the current extent of machine learning as applied to Mars, summarize why machine learning should be an important tool for planetary geomorphology in particular, and suggest numerous research avenues and funding priorities for future efforts. We conclude that: (1) moving toward methods that require less human input (i.e., self- or semi-supervised) is an important paradigm shift for Martian applications, (2) new robust methods using generative adversarial networks to generate synthetic high-resolution digital terrain models represent an exciting new avenue for Martian geomorphologists, (3) more effort and money must be directed toward developing standardized datasets and benchmark tests, and (4) the community needs a large-scale, generalized, and programmatically accessible geographic information system (GIS). [ABSTRACT FROM AUTHOR]

Erosion rates derived using dendrogeomorphology have been used to quantify slope degradation in many localities globally. However, with the exception of the western United States, most of these estimates are derived from short-lived trees whose lifetimes may not adequately reflect the complete range of slope processes which can include erosion, deposition, impacts of extreme events and even long-term hiatuses. Erosion rate estimates at a given site using standard techniques therefore reflect censored local point erosion estimates rather than long-term rates. We applied a modified dendrogeomorphic approach to rapidly estimate erosion rates from dbh/age relationships to assess the difference between short and long-term rates and found that the mean short-term rate was ~ 0.13 cm/yr with high variability, while the uncensored long-term rate was ~ 0.06 cm/yr. The results indicate that rates calculated from short-lived trees, while possibly appropriate for local short-term point estimates of erosion, are highly variable and may overestimate regional long-term rates by > 50%. While these findings do not invalidate the use of dendrogeomorphology to estimate erosion rates they do suggest that care must be taken to select older trees that incorporate a range of slope histories in order to best approximate regional long-term rates. [ABSTRACT FROM AUTHOR]

Transverse aeolian ridges (TARs) are unusual bedforms on the surface of Mars. TARs are common but sparse on Mars; TAR fields are small, rarely continuous, and scattered, making manual mapping impractical. There have been many efforts to automatically classify the Martian surface, but they have never explicitly located TARs successfully. Here, we present a simple adaptation of the off-the-shelf neural network RetinaNet that is designed to identify the presence of TARs at a 50-m scale. Once trained, the network was able to identify TARs with high precision (92.9%). Our model also shows promising results for applications to other surficial features like ripples and polygonal terrain. In the future, we hope to apply this model more broadly and generate a large database of TAR distributions on Mars. [ABSTRACT FROM AUTHOR]

Convergent tectonics and glaciomarine sedimentary processes in the northern Gulf of Alaska have produced a variety of troughs, canyons, channels, gullies, and other drainage related features over a range of spatial scales. To evaluate the effects of glaciomarine and fault processes on observed drainage morphology we measured regional drainage feature cross-profile shape characteristics across the area. Our results show overlap between U-shaped feature cross-profile locations, ice sheet extents, and instability zones on the shallow shelf and slope that were previously inferred by observations of potential mass-movement structures such as slumps, slides, and debris flows. U-shaped profiles were also observed in areas of recent slope progradation such as the flank of the proximal deep-sea Surveyor Fan. We used principal component analysis (PCA) and geographically weighted regression (GWR) to relate drainage cross-profile morphology to proxy variables for ice sheets, faults, and instability zones. This method enables quantitative evaluation of the effects on cross-profile shape explainable by each variable as indicated by the magnitude, distribution, and variation of parameter coefficients both at the local and regional scale. Our multiple variable GWR model can explain most of the regional variability in cross-profile shape with the contribution of each parameter varying throughout the study area. Parameters associated with ice sheet extent have the strongest overall relationship with drainage shape, while the distance from an instability zone is the strongest local relationship. The fault proxy parameter shows relatively strong correlations with cross-profile shape near the Transition Fault, and the narrowest zone of morphologic influence overall. • U-shaped features often correspond with ice sheet extents or zones of instability. • Faults, ice sheets, and instability zones can mostly explain cross-profile shape. • Ice sheet extents had the overall strongest relationship with cross-profile shape. • More focused shape influence for fault zones, and a broader influence for ice sheets. [ABSTRACT FROM AUTHOR]