Tectonically active orogens typically produce deeply incised river valleys.These deep river valleys create steep valley walls which increase the likelihood of slope failure. As landslide deposits fill valleys, rivers can either mobilize the deposits or incise new valleys around the landslide toes. Here, I investigate Arroyo Hondo, a river in central California’s actively uplifting Diablo Range, a part of the Pacific Coast Ranges of Northern California. The Oak Ridge earthflow is a slow-moving landslide that has been failing into Arroyo Hondo for at least the last century, translating blocks of Franciscan Mélange down into the steep-walled valley below. Through remote sensing, field measurements, and hydraulic modeling, I show that Arroyo Hondo has steepened its profile due to boulder input at the landslide toe so that it is capable of mobilizing the largest grain size fraction (D99) of the deposit. These results show that landslide-impacted river reaches, despite receiving extremely coarse material from hillslopes, can evolve to move the coarse sediment supply, like other alluvial rivers. This point is underscored by the fact that river slopes have relaxed below inactive landslides in the field area, suggesting diffusion over time from boulder transport.

When an initial landslide does not fully evacuate all the failed material, remaining deposits may be susceptible to recurring landsliding in the future. These types of consecutive landsliding systems present significant natural hazards and play an important role in landscape evolution, however they are difficult to incorporate into hazard assessments and geomorphic analyses because they violate a commonly held assumption of stationarity. Illuminating the dominant processes that control the behavior and triggering conditions of recurring landslides is critical for improving these assessments.

Landslide movement in the simplest form occurs when the balance of shear stress driving material downslope exceeds the shear strength within a hillslope. In the regime of Coulomb friction, this can be articulated as: τ=(σ-p)tanφ+c Where τ is shear stress, σ is normal stress, p is pore-water pressure, φ is friction angle, and c is effective cohesion. For rainfall driven landslides, infiltration of rainwater into the subsurface increases pore-water pressure within the landslide body, thereby reducing effective normal stresses (σ-p) and shifting the balance towards landslide movement. Field observations suggest that pore-water pressure rise (and consequently landslide deformation) is controlled by the details of how rainfall infiltrates through the vadose zone, i.e., the unsaturated region below the ground surface and above the water table.

In this dissertation, I present three studies combining field observations and numerical data to explore the influence of vadose zone processes on recurring landslides. In the first chapter, I synthesize field observations of a large, slow-moving landslide near San Jose, California, USA to explore how lithologically controlled weathering patterns influence earthflow hydrology and deformation. I find that the landslide has a thin weathered zone, comparable to other landslides in the same geologic formation, which results in a small amount of dynamic storage. I propose that this storage capacity constrains the possible range of stress within the landslide body and consequently limits landslide movement as excess water is shed via springs and saturation overland flow once the water table reaches the surface. In the second chapter, I use dynamically downscaled climate model projections to model variably saturated groundwater conditions over 150 years at this same landslide. Using field data from chapter 1, I link the groundwater water data to an empirical model of landslide movement to explore how multiyear precipitation variability will affect future landslide movement. I find that future increases in precipitation variability may lead to a decrease in slow-landslide movement in California, particularly in areas where the amount of possible recharge is precipitation limited. Declining landslide movement is due in part to greater precipitation intensity which diverts rainfall to runoff rather than recharge. Additionally, the impact of wet years on landslide movement is limited by storage constraints and the impact of dry years may propagate forward for multiple years due to the importance of antecedent saturation state. Finally, in the third chapter, I use numerical experiments to investigate the dominant controls on the reactivation of coastal bluff landslides along Puget Sound, Washington, USA. I find that the changes in hydraulic conductivity and strength from root reinforcement have the greatest impact on the likelihood of landslide reactivation, followed by changes in the properties of the soil water retention curve. Collectively, these studies advance our understanding of vadose zone processes within landslides and the resulting effects on landslide movement.

The persistent movement of deep-seated, slow-moving landslides is a common phenomenon with important scientific and practical implications. These landslides, often referred to as earthflows, can dominate erosion in mountainous landscapes, control the long-term evolution of catchment geometry, and cause progressive damage to infrastructure and property over time. The role of earthflows in shaping landscapes is amplified by their longevity. Historical observations demonstrate that they are commonly active for decades or centuries, while geologic and geochronologic evidence further suggests that some may persist in intermittently-active states for millennia. Slow, persistent motion over these timescales requires complementary mechanisms to both repeatedly reactivate previously failed materials and to limit velocity during periods of activity. More than 30 years of robust scientific investigations into the nature of these mechanisms has yielded tremendous insights. These include the sensitivity of earthflows to rainfall-driven changes in pore-water pressure, the existence of internal feedbacks that limit earthflow velocity after the onset of motion, and the recognition that external factors, like climate and base-level, help set the periodicity of intermittent earthflow movement. A common goal of all these investigations has been to generate data and models that will allow for the prediction of future earthflow behavior, which remains challenging despite many advances. Predicting the velocity of slow landslides over multi-decadal timescales requires an understanding of the potential drivers of earthflow motion, analytical frameworks and models to connect those drivers to kinematic response, and historical records of movement with sufficient spatial and temporal resolution to test model predictions. Predicting velocity on shorter timescales, such as the onset of motion in a given wet season, requires high-temporal resolution, time-series measurements of the hydro-mechanical properties of the earthflow in question. And lastly, scaling up predictions of future velocity at one earthflow to many will require the use of high-spatial resolution, remotely-sensed imagery capable of monitoring large regions with repeat measurements. This dissertation aims to contribute to each of these requirements through an extended study of Oak Ridge earthflow, a long-lived, slow-moving landslide in California’s northern Diablo Range. In Chapter 1, we used a time series of aerial imagery to assemble an 80-year kinematic history for Oak Ridge earthflow. We find that spatial patterns of earthflow velocity were controlled largely by the slope of the underlying failure plane, whereas temporal patterns were governed largely by climate-driven changes in surface moisture balance (PDSI) at the annual-decadal scale. Declines in sediment supply acted as a secondary control on temporal velocity variations over our study period, however, the influence of this driver likely grows at longer timescales. In Chapter 2, we use 10 years of high-spatial resolution, multispectral satellite imagery to assess the long-accepted, but relatively untested, hypothesis that hydro-mechanical isolation of earthflows from surrounding, stable slopes contributes to their ability to fail persistently over time. In Chapter 3, we use of 2.5 years of high-temporal resolution records of rainfall, pore-water pressure, and displacement to argue that the initial onset of earthflow acceleration in response to seasonal rainfall is likely to be more closely related to the timescales of vadose zone processes than pore-water pressure diffusion through saturated soil, as has been previously suggested. Collectively, these studies provide new observations and approaches towards understanding earthflow behavior at multiple scales.

Although sea level has fluctuated repeatedly over the Pleistocene by up to 120 meters, how (or whether) this cyclic base level forcing impacts the development of bedrock river profiles in tectonically active settings is poorly understood. A major reason for this uncertainty is that bedrock river channels in unglaciated locations are typically buried at their mouths under sediment resulting from the Holocene marine transgression, making direct observations impossible. Here I present a novel approach to constraining the influence of cyclical sea-level forcing on the development of bedrock river profiles in the Oregon Coast Range.

Using a 1 m LiDAR DEM, I estimate the depth to the buried bedrock longitudinal profile using measurements of valley width, elevation, wall slope, and present day river width for the Smith River and its tributaries where the river is currently buried by Holocene fluvial and marine sediments. Assuming the current river width is spatially and temporally constant, I calculate the depth to bedrock based on a trapezoidal geometry. In an effort to reduce the noise in our signal, I calculate linear regressions for valley and river width and hold the valley wall slope constant at its average value (~ 30 degrees) for all our calculations.

Several important observations stem from this analysis. First, the bedrock profile of the Smith River projects to the same elevation (~ -110 m) as the bench cut into the continental shelf by the sea-level low stand of the last glacial maximum, suggesting the bedrock river is currently graded to the low-stand elevation. Second, almost all of the tributary bedrock profiles plot tens of meters above the mainstem bedrock profile. One key exception is the significantly larger North Fork of the Smith River, which falls directly on the projected bedrock curve for the Smith. These observations suggest that sea-level low stands force incision of the entire bedrock river network, but that smaller tributaries, which are buried and/or flooded during highstands, cannot keep pace with the mainstem during these periods of brief down-cutting. Our results imply that difference in erosion rates between tributaries and the mainstem results in the formation of fluvial hanging valleys that are buried during marine transgressions.

The potential that sinuous bedrock river canyons archive information about past climatic and tectonic forcing of landscapes has intrigued geomorphologists for over 100 years, in fact they were one of the first landforms to inspire the use of landscapes to constrain geologic history. Yet because bedrock river erosion typically requires bedload impacts and abrasion, and because bedload transport is not tightly coupled with the small deviations in fluid shear stress that cause meandering in lowland alluvial rivers, it was never clear how or if the fluid dynamics understood to drive meandering in alluvial rivers could be applied to rivers confined to bedrock canyons. Without a theory for the process of active meandering in bedrock rivers, the assumption that bedrock sinuosity must be inherited from an antecedent alluvial state has remained pervasive despite evidence to the contrary.

In this thesis, exploration of well constrained field examples enables me to formulate two general requirements for a bedrock river to meander: 1. bank-rock must become susceptible to fluid erosion, and 2. mass wasting of steep bedrock cut banks cannot produce persistent talus to protect the bank from continued lateral erosion. With detailed field exploration, experimentation, and mapping, laboratory rock mechanics testing, weathering experiments, topographic analysis and literature review of mineralogy, we show that where meandering in bedrock occurs, weathering of bank rock satisfies these two conditions. In the mudstone lithology of Pescadero Creek, CA, we show that slaking (fracture due to wetting and drying) of exposed bedrock along cut banks both makes this rock transportable by fluid flow, and makes colluvium easily transportable. For channels in the basaltic lavas of the Kohala Peninsula, Hawai’i chemical weathering can satisfy these same two conditions. This chemical weathering has been shown to be closely controlled by precipitation, and I show a significant decrease in channel sinuosity across the peninsula’s strong orographic precipitation gradient from wet to arid regions. In both these examples, meandering is sensitive to climate. This understanding of how bedrock rivers meander provides a mechanistic foundation for climatic interpretations of sinuous bedrock river canyons (including their associated strath terraces).

In addition, I demonstrate that active meandering rearranges drainage networks and has far reaching consequences for landscape evolution and sediment routing. Specifically, I show that 1) growing meander bends can capture tributaries and cause large knickpoints, and 2) meander neck cutoff events cause terraces which isolate adjoining tributaries from the active mainstem, resulting in long lasting tributary convexities, and sediment storage. These autogenic consequences of active meandering mimic tectonic and climatic transience in landscapes over timescales similar to glacial/interglacial cycles (105-106 years), and also influence sediment transport and storage, which often plays a critical role for aquatic ecosystems.

Bed surface grain size and mobility are fundamental characteristics of a river. The size of coarse sediment on the bed of a river controls channel slope, shapes channel plan form morphology, and provides the architecture for aquatic habitat. Despite its importance to river morphology and habitat, the controls on bed sur- face grain size are imperfectly understood. Varying sediment supply, armoring of the bed surface, roughness effects, and patchiness complicate predictions. Because of its importance, understanding the controls on bed surface grain size is essential despite these complications.

The goal of this dissertation is to further understanding of the controls on the size and mobility of surface grains in gravel bed rivers. The first study, Chapter 2, focuses on predicting reach-averaged grain size at the drainage basin scale. I build on previously proposed methods for grain size prediction, and demonstrate the effectiveness of a new method for predicting grain size in Scott Creek, a small gravel-bedded drainage in the Santa Cruz Mountains, California. This work is then applied to salmon spawning habitat prediction. As a part of that project, I show that grain size is much smaller in Scott Creek than the grain size predicted based on threshold channel theory. This finding motivates the second study in my dissertation. In Chapter 3, I show that there are significant regional trends in the ratio of bankfull Shields stress to critical Shields stress (τbf / τc) across North America. These trends are explained by differences in sediment supply. In Chapter 4, I consider the implications of these regional trends on the mobility of bed surface grains, and the potential implications for benthic habitat.

Interpreting spatial patterns in rates of fluvial incision from river channel elevation long profile data requires an assumption that tectonic uplift rate governs river channel slope. However, application of the most mechanistically explicit description of river incision [Sklar and Dietrich,2004] suggests that sediment flux and sediment grain size, not rock uplift rate, control river channel slopes in many settings. Because it is usually difficult to independently constrain sediment supply, tectonic interpretations of river elevation long profiles are necessarily uncertain. Here we exploit a natural experiment in Boulder Creek, a ~ 30 km^2 drainage in the Santa Cruz Mountains, CA USA in order isolate the affect of grain size and relative sediment supply on river channel slope in an actively uplifting landscape along a restraining bend in the San Andreas Fault.

A single prominent knickpoint exists near the midpoint of Boulder Creek, separating a 6.8 km long region of low slope(~0.8%) from a steeper(~2.6%) 4.8 km reach along the lower portion of the channel . Mapping and field observations reveal that this knickpoint does not coincide with any lithologic or tectonic boundaries; the channel cuts weak sedimentary rock for it length. In addition, longer wavelength changes in rates of rock uplift due to the bend in the San Andreas fault near Boulder Creek are negligible over the relative small size of Boulder Creek's catchment. Instead the knickpoint coincides with the location of the first tributary that taps a source of resistant, granitic sediment that is not found in the upstream reaches of Boulder Creek. Field observations indicate that coarse granitic bedload is sourced by debris flows and introduced by a series of tributaries draining into the steep lower reaches of Boulder Creek. The knickpoint marks a transition in median grain size from ~2cm upstream of the knickpoint compared to an average of ~18cm downstream of the knickpoint. Additionally, upstream of the knickpoint , Boulder Creek is characterized by potholes and sculpted bedrock, consistent with sediment-starved conditions.

The observation that bedrock channel slope changes are not well correlated with patterns in rock uplift supports Sklar and Dietrich's (2006) theoretical result that modest rates of rock uplift do not significantly influence river profile slopes. Based on this result and the clear correlation of channel slope and sediment supply along Boulder Creek, we chose to ignore rock uplift rate and instead explore the relative roles of grain size and sediment flux in influencing profile slopes along Boulder Creek. Using field surveys of grain size and high flow depth, we calculate that ~10% of the slope above the knickpoint and ~30% of the slope below the knickpoint is related to maintenance of the channel at the threshold for sediment motion. This implies that ~90% of the slope above the knickpoint and ~70% of the slope below the knickpoint is due to the excess stress that is required to move the coarse sediment load. This would imply that other factors including sediment supply and bed roughness can be significantly more important than thresholds of coarse sediment motion for setting channel slope in mixed bedrock-alluvial systems.

Virtually all mechanistic predictions of landscape evolution are underpinned by predictions of sediment transport rates. Sediment transport via bedload is the primary way by which coarse sediment is transported through landscapes, and thus, accurate prediction of bedload transport represents a fundamental topic in the study of earth surface processes. Specifically, predictions of coarse sediment transport are strongly dependent on the selection of the threshold for motion, or the forcing at which sediment transport begins. However, it is well known that the threshold for motion varies in space and time, and in response to a number of processes. This dissertation aims to identify and quantify the effects of biology and prior flow history on the onset of coarse particle motion through a combination of mathematical modeling, novel physical experiments, and continuous bedload transport measurements in the field. Chapter 2 develops and tests a mechanistic model for the entrainment of coarse sediment by attached kelp, highlighting the significance of biology in setting coastal sediment transport rates. In Chapter 3, we explore the effects of below-threshold, antecedent flow history on gravel bed structure, grain protrusion, and bedload transport through a series of flume experiments. We find that bedload flux, and by extension, the onset of motion, is extremely sensitive to the duration of prior below-threshold flow. This reduction in bedload transport rates is related to the reduction in the population of high protruding grains, related to the pivoting of precariously placed, highly mobile particles into nearby pockets. In Chapter 4, we carry out a systematic analysis of the temporal evolution of the threshold for motion in a natural channel though the analysis of a unique record of continuous bedload transport measurements. We observe that variations in the threshold for motion are significantly linked to previous flow magnitude, highlighting that the onset of motion is a history-dependent quantity. Collectively, these studies provide new observations and approaches towards predicting the temporal and spatial evolution of the threshold for motion, providing a basis for improved bedload transport models.

The topography of volcanic arcs reflects a combination of magmatic ascent processes that deform the overlying crust, volcanic eruptions that blanket the surface in lava and ash, and geomorphic processes that shape these deposits into the observable landscape and consequently record an erosional archive of surface deformation. A central goal of geomorphology is to utilize the form of landscapes to infer the processes that shape them, yet comparatively little work has been done to quantitatively understand the geomorphology of volcanic landscapes. The Central Volcanic Zone of the South American Andes (CVZ, 15-28 degrees S) represents an ideal location for understanding the processes that shape the surface of volcanic arcs; the arid climate and well-constrained geochronology of volcanic deposits provide a geomorphic archive going back 11 Myrs, and the wealth of available geophysical imaging data provides a high-resolution glimpse into the subsurface. In this thesis I utilize geormophic and geophysical data to explore fundamental questions in volcanic landscape evolution at varying spatial scales. To begin, I examine the isostatic uplift response to crustal intrusion of mantle-derived melt, resulting in growth of the largest known active magma reservoir on Earth (the Altiplano-Puna Magma Body). From topography I show that it is possible to calculate the contribution of mantle-derived melt to crustal thickening, and I compare these results to independent seismic estimates that agree well with our data. I then use the record of volcanic deposits from the Altiplano-Puna Magma Body to estimate the flux of mantle melt into the crust, and show that predicted uplift rates are comparable to the surface uplift associated with convective lithospheric removal. Next, I explore how the topography of individual volcanic systems can help discern the style of melt ascent from deeper crustal magma reservoirs to shallow ones. Using two actively deforming volcanoes in the CVZ, Uturuncu volcano and the Lazufre volcanic complex, I use river profiles, lake shorelines, and lava flow deflections to infer each volcano's surface deformation record into the geomorphic past. Uturuncu volcano shows little signs of permanent paleo-deformation, suggesting transient ascent of magma over millenial timescales, while Lazufre's long wavelength deformation suggests continued accumulation of magma since > 0.3 Ma. Finally, I utilize the unique topography of a windswept ignimbrite erupted from the Altiplano-Puna Magma body to better understand how wind abrasion can incise bedrock canyons, a fundamental geomorphic process in arid landscapes on both Earth and Mars. Using a natural experiment within the 4 Ma Puripicar ignimbrite on the western slope of the Central Andes, we quantify the relative contribution of wind and water erosion in propagating bedrock canyons and show that wind can incise canyons an order of magnitude faster than rivers, streamlining their profiles in the process. Thus, in these chapters I illustrate how geomorphology can be a useful tool for quantifying subsurface magmatic processes, while volcanic landscapes themselves can also provide unique opportunities for understanding fundamental planetary geomorphic processes not often observed elsewhere on Earth.

Rivers are dynamic features of Earth’s surface that contain information about both past and present landscape processes. Shifts in the balances of sediment and flow in a watershed are often reflected in the character of the riverbed, channel geometry, or patterns of sediment transport, making rivers an invaluable tool for investigating changes in the overall sediment transport regime. Through three related studies, this work explores the various ways in which rivers not only respond to changing watershed conditions, but also reflect the broader climatic and lithologic character of a landscape.

This dissertation begins with a case study of the San Lorenzo River in California, in which the causes and timing of river floodplain abandonment are investigated through a combination of field measurements of channel geometry, theoretical estimates of channel depth, and radiocarbon dating of charcoal. Results suggest that significant channel incision can be attributed to historical practices associated with clear-cut logging, underscoring the long-lasting legacy of land use on landscapes and river networks.

Broader relationships between sediment transport, river characteristics, and environmental factors are explored in the following two studies. Chapter 2 uses an aridity index in tandem with suspended sediment rating curves for 71 rivers across the United States to establish a compelling relationship between climate and sediment transport efficiency, and explores the influence of vegetation density and flow variability on sediment supply and riverbed grain size sorting. Chapter 3 follows up on these results by directly examining the link between suspended sediment rating curves and the organization of the riverbed. This study strengthens the interpretations made in Chapter 2 by confirming the likely significance of riverbed grain size and armoring on sediment transport efficiency across climates.

By pairing a case study with both novel and far-reaching geomorphic analysis, this research contributes to a deeper understanding of the mechanisms governing fluvial sediment transport and river form. The specific findings and new applications of old methods provide helpful resources to researchers and policymakers seeking to mitigate the effects of climate change and environmental degradation on river ecosystems and adjacent communities.