Despite the rapid adoption of robotics and machine learning in industry, their application to scientific studies remains under-explored. Combining industry-driven advances with scientific exploration provides new perspectives and a greater understanding of the planet and its environmental processes. Focusing on rock detection, mapping, and dynamics analysis, I present technical approaches and scientific results of developing robotics and machine learning technologies for geomorphology and seismic hazard analysis. I demonstrate an interdisciplinary research direction to push the frontiers of both robotics and geosciences, with potential translational contributions to commercial applications for hazard monitoring and prospecting. To understand the effects of rocky fault scarp development on rock trait distributions, I present a data-processing pipeline that utilizes unpiloted aerial vehicles (UAVs) and deep learning to segment densely distributed rocks in several orders of magnitude. Quantification and correlation analysis of rock trait distributions demonstrate a statistical approach for geomorphology studies. Fragile geological features such as precariously balanced rocks (PBRs) provide upper-bound ground motion constraints for hazard analysis. I develop an offboard method and onboard method as complementary to each other for PBR searching and mapping. Using deep learning, the offboard method segments PBRs in point clouds reconstructed from UAV surveys. The onboard method equips a UAV with edge-computing devices and stereo cameras, enabling onboard machine learning for real-time PBR search, detection, and mapping during surveillance. The offboard method provides an efficient solution to find PBR candidates in existing point clouds, which is useful for field reconnaissance. The onboard method emphasizes mapping individual PBRs for their complete visible surface features, such as basal contacts with pedestals–critical geometry to analyze fragility. After PBRs are mapped, I investigate PBR dynamics by building a virtual shake robot (VSR) that simulates ground motions to test PBR overturning. The VSR demonstrates that ground motion directions and niches are important factors determining PBR fragility, which were rarely considered in previous studies. The VSR also enables PBR large-displacement studies by tracking a toppled-PBR trajectory, presenting novel methods of rockfall hazard zoning. I build a real mini shake robot providing a reverse method to validate simulation experiments in the VSR.

Mountain landscapes reflect competition between tectonic processes acting to build topography and erosive processes acting to wear it down. In temperate mountain landscapes, bedrock rivers are the primary erosional agent, setting both the pace of landscape evolution and form of the surrounding topography. Theory predicts that river steepness is sensitive to climatic, tectonic, and lithologic factors, which dictate the rates and mechanics of erosional processes. Thus, encoded into topography is an archive of information about forces driving landscape evolution. Decoding this archive, however, is fraught and climate presents a particularly challenging conundrum: despite decades of research describing theoretically how climate should affect topography, unambiguous natural examples from tectonically active landscapes where variations in climate demonstrably influence topography are elusive. In this dissertation, I first present a theoretical framework describing how the spatially varied nature of orographic rainfall patterns, which are ubiquitous features of mountain climates, complicate expectations about how climate should influence river steepness and erosion. I then apply some of these ideas to the northern-central Andes. By analyzing river profiles spanning more than 1500 km across Peru and Bolivia, I show that the regional orographic rainfall pattern this landscape experiences systematically influences fluvial erosional efficiency and thus topography. I also show how common simplifying assumptions built into conventional topographic analysis techniques may introduce biases that undermine detection of climatic signatures in landscapes where climate, tectonics, and lithology all covary – a common condition in mountain landscapes where these techniques are often used. I continue by coupling this analysis with published erosion rates and a new dataset of 25 cosmogenic 10Be catchment average erosion rates. Once the influence of climate is accounted for, functional relationships emerge among channel steepness, erosion rate, and lithology. I then use these functional relationships to produce a calibrated erosion rate map that spans over 300 km of the southern Peruvian Andes. These results demonstrate that accounting for the effects of climate significantly enhances the ability to decode channel steepness patterns. Along with this comes the potential to better understand rates and patterns of tectonic processes, and identify seismic hazards associated with tectonic activity using topography.

Fourier Transform Infrared (FTIR) Spectrometers are used to determine surface composition of celestial bodies such as asteroids or planets by collection of infrared spectral data. However, degraded performance for shorter wavelengths may exist when the target does not fill the field of view and may be off-axis. Further study of the optical implications of such use cases would inform future design. This research project aims to develop a systematic method of rapid prototyping in order to progressively simulate optical conditions to characterize the off-axis implications in FTIR spectrometers regarding effects on spectral data measured. With such findings, FTIR spectrometers may be developed to effectively accommodate a larger field of view beyond the current state-of-the-art without increasing the corresponding package size of aft optics such as interferometer assemblies. Throughput may be further increased than current limitations or smaller aft optics systems may be designed with the same throughput. Specific use cases which would otherwise result in degradation of spectral data could potentially be accommodated, all effectively increasing capability of the current technology. With this intent, a preliminary test setup has been developed and initial results were collected.

A wide range of types of activity in mid-latitude Martian gullies has been observed over the last decade (Malin et al., 2006; Harrison et al., 2009, 2015; Diniega et al., 2010; Dundas et al., 2010, 2012, 2015, 2017) with some activity constrained temporally to occur in the coldest times of year (winter and spring; Harrison et al., 2009; Diniega et al., 2010; Dundas et al., 2010, 2012, 2015, 2017), suggesting that surficial frosts that form seasonally and diurnally might play a key role in this present-day activity. Frost formation is highly dependent on two key factors: (1) surface temperature and (2) the atmospheric partial pressure of the condensable gas (Kieffer, 1968). The Martian atmosphere is primarily composed of CO2and CO2 frost formation is not diffusion-limited (unlike H2O). Hence, for temperatures less than the local frost point of CO2, (~ 148 K at a surface pressure of 610 Pa) frost is always present (Piqueux et al., 2016). Typically, these frosts are dominated volumetrically by CO2, although small amounts of H2O frosts are also present, and typically precede CO2 frost deposition (due to water’s higher condensation temperature (Schorghofer and Edgett, 2006)). Here we use the Thermal Emission Imaging System (THEMIS) and the Thermal Emission Spectrometer (TES) onboard Mars Odyssey and Mars Global Surveyor, respectively, to explore the global spatial and temporal variation of temperatures conducive to CO2 and H2O frost formation on Mars, and assess their distribution with gully landforms. CO2 frost temperatures are observed at all latitudes and are strongly correlated with dusty, low thermal inertia regions near the equator. Modeling results suggest that frost formation is restricted to the surface due to near-surface radiative effects. About 49 % of all gullies lie within THEMIS frost framelets. In terms of active gullies, 54 % of active gullies lie within THEMIS framelets, with 14.3% in the north and 54% in the south.
Relatively small amounts of H2O frost (~ 10–100 μm) are also likely to form diurnally and seasonally. The global H2O frost point distribution follows water vapor column abundance closely, with a weak correlation with local surface pressure. There is a strong hemispherical dependence on the frost point temperature—with the northern hemisphere having a higher frost point (in general) than the southern hemisphere—likely due to elevation differences. Unlike the distribution of CO2 frost temperatures, there is little to no correlation with surface thermophysical properties (thermal inertia, albedo, etc.). Modeling suggests H2O frosts can briefly attain melting point temperatures for a few hours if present under thin layers of dust, and can perhaps play a role in present-day equatorial mass-wasting events (eg. McEwen et al., 2018).
Based on seasonal constraints on gully activity timing, preliminary field studies, frost presence from visible imagery, spectral data and thermal data (this work), it is likely that most present-day activity can be explained by frosts (primarily CO2, and possibly H2O). We predict that the conditions necessary for significant present-day activity include formation of sufficient amounts of frost (> ~20 cm/year) within loose, unconsolidated sediments (I < ~ 350) on available slopes. However, whether or not present-day gully activity is representative of gully formation as a whole is still open to debate, and the details on CO2 frost-induced gully formation mechanisms remain unresolved.