Topographic Asymmetry and Climate Controls on Landscape Evolution ARCHNES MASSACHUSETTS INSTITUTE OF TECHNOLOGY by SEP 2 8 2015 Paul William Richardson LIBRARIES B.S., University of Washington (2009) Submitte I in partial fulfillment of the requirements for the degree of Doctor of Philosophy at the MASSACHUSETTS INSTITUTE OF TECHNOLOGY September 2015 Paul Richardson, MMXV. All rights reserved. The author hereby grants to MIT permission to reproduce and distribute publicly paper and electronic copies of this thesis document in whole or in part in any medium now known or hereafter created. Author........ Signature redacted Massachusetts Institute of Technology August 31, 2015 Certified by.... Signature redacted .... .................... J. Taylor Perron Associate Professor, Massachusetts Institute of Technology Thesis Supervisor Acre ted bi p .... Signature redacted .... .... ... .... ... Robert van der ilst Department Head and Schlumberger Professor of Earth Sciences Earth, Atmospheric, and Planetary Sciences, Massachusetts Institute of Technology 2 Topographic Asymmetry and Climate Controls on Landscape Evolution by Paul William Richardson Submitted to the Massachusetts Institute of Technology on August 31, 2015, in partial fulfillment of the requirements for the degree of Doctor of Philosophy Abstract Landscapes are expected to evolve differently under the influence of different climate conditions. However, the relationship between landscape evolution and climate is not well understood. I investigate the relationship between landscape evolution and climate by using natural experiments in which climate varies with slope aspect (geographic orientation) and causes differences in landscape form, such as steeper equator- or polefacing slopes. In order to understand which mechanisms are responsible for the development of this topographic asymmetry, I adapted a numerical landscape evolution model to include different asymmetry-forming mechanisms such as aspect-induced variations in soil creep intensity, regolith strength, and runoff, and also lateral channel migration. Numerical experiments reveal topographic signatures associated with each of these mechanisms that can be compared with field sites that exhibit asymmetry. I used these numerical model results, along with estimates of field-saturated hydraulic conductivity, soil strength, evidence of stream capture and channel beheadings, and erosion rates determined from cosmogenic radionuclides to determine which asymmetryforming mechanisms are likely responsible for the topographic asymmetry at Gabilan Mesa, a landscape in the central California Coast Ranges. I find that aspect-dependent differences in runoff are most likely responsible for the bulk of the asymmetry at Gabilan Mesa, but lateral channel migration has contributed to the asymmetry in some locations. To further investigate climate's influence on landscape evolution, I compiled new and previously published estimates of slope-dependent soil transport efficiency across a range of climates. I find that soil transport efficiency increases with mean annual precipitation and the aridity index, a measure that describes water availability for plants. I also find that soil transport efficiency varies with lithology and that different measurement techniques can bias estimates of the soil transport coefficient. Thesis Supervisor: J. Taylor Perron Title: Associate Professor of Geology 3 Table of Contents A cknow ledgem ents................................................................................ 5 C hapter 1. Introduction.............................................................................. 7 Chapter 2. Modeling the formation of topographic asymmetry by lateral channel migration and aspect-dependent erosional processes...........................................11 Chapter 3. Unraveling the mysteries of an asymmetric landscape.........................67 Chapter 4. The influence of climate on hillslope sediment transport efficiency..........113 C hapter 5. C onclusion ............................................................................ Referen ces...........................................................................................14 A p p en dix ............................................................................................ 4 139 5 159 Acknowledgements First and foremost, I would like to acknowledge Lisa Richardson, who has been supportive throughout this process and has found much time to pursue new hobbies and interest while I completed my thesis. Secondly, I would like to thank my family and friends who encouraged me to persevere when results did not come as quickly as one might like. There are also a number of other people that I owe significant gratitude to including my thesis advisor Taylor Perron and my thesis committee members: David McGee, Dara Entekhabi, Noah Snyder, and Jim Kirchner. They all supplied valuable input. I would also like to thank Scott Miller who offered me considerable help with the topographic asymmetry analysis and very useful feedback throughout my graduate studies. I thank Naomi Schurr for helping compile some of the data and for making some of the topographic measurements in Chapter 4. I would also like to thank those who helped me in the field or helped prepare equipment for the field including David de Jong, Michael Sori, and Peter Polivka. Furthermore, I would also like to thank all of the funding agencies that supported this work, including the National Defense Science and Engineering Graduate Fellowship through the Department of Defense. I would also like to thank the Geological Society of America for funding some of my travel and research, and the National Science Foundation for a grant to Taylor Perron that provided some of the funding for this research. 5 A man's interest in a single bluebird is worth more than a complete but dry list of the fauna andflora of a town. -Henry David Thoreau 6 Chapter 1. Introduction Despite many studies addressing how landscapes respond to differences in climate [e.g., Riebe et al., 2004; Perron et al., 2008; Dixon et al., 2009; Moon et al., 2011; Owen et al., 2011; Champagnacet al., 2012; Ferrieret al., 2013b], many questions still remain about how climate influences landscape evolution. In particular, questions remain about how hillslope form, erosion rates, and sediment transport efficiency vary under different climate conditions. Some geomorphologists have successfully quantified relationships between precipitation and erosion rates at specific sites [Moon et al., 2011; Ferrieret al., 2013a], but a general relationship between these factors and climate does not exist [Portengaand Bierman, 2011]. The difficulty in quantifying the relationship between climate and erosion is likely due in part to differences in rock type, tectonic processes, vegetation, and the degree of chemical versus physical weathering at different sites. One way to make progress on this question is by studying natural experiments in which climate varies, but other factors remain constant. An ideal natural experiment for studying the effects of small climatic variations has been carried out in many landscapes. Variations in insolation, the amount of sunlight that strikes a surface, can generate microclimates - small variations in temperature, soil 7 moisture, humidity, wind and other measurable climate factors - on slopes that face different directions [e.g., Branson and Shown, 1990; Yetemen et al., 2010; Anderson et al., 2012]. These differences in microclimate can lead to the development of topographic asymmetry, and several mechanisms - including aspect-induced variations in soil creep intensity, regolith strength, and runoff, and also lateral channel migration - have been proposed to explain the asymmetry [e.g., Melton, 1960; Kane, 1970; Dohrenwend, 1978; Yetemen et al., 2010; West et al., 2013; McGuire et al., 2014]. Topographic asymmetry is the phenomenon that slopes with different aspect exhibit different topographic characteristics, such as the occurrence of north-facing slopes that are steeper than southfacing slopes. The specific mechanisms that control topographic asymmetry are not well understood, but two leading hypotheses have emerged to explain the occurrence of topographic asymmetry in semi-arid landscapes when lithological or structural explanations cannot be invoked [Kane, 1970; Dohrenwend, 1978]. One hypothesis relies on aspect-dependent differences in erosion such as differences in regolith strength, soil creep rates, or runoff. According to the other hypothesis, aspect-dependent differences are not enough, and the occurrence of lateral channel migration, possibly driven by differences in sediment flux rates on opposing slopes due to the microclimates, leads to oversteepening of the undercut slope. However, the long-term topographic consequences of the different mechanisms have not been fully explored, and identifying which mechanism or mechanisms are responsible for the topographic asymmetry in a given landscape remains a challenge. In Chapter 2, I present a set of landscape evolution modeling experiments that test how landscapes respond to microclimatic differences and other possible 8 asymmetry-forming mechanisms. I explore the relationship between different topographic characteristics that develop asymmetry and erosion rate patterns to determine if unique signatures exist that may help pinpoint which asymmetry-forming mechanisms are occurring in real landscapes. In Chapter 3, I use these predictions to help determine which asymmetry-forming mechanisms are most responsible for the topographic asymmetry at Gabilan Mesa, a landscape in the central California Coast Ranges that exhibits very high topographic asymmetry. At Gabilan Mesa, significant differences in hillslope morphology are well-documented and correlate with modern differences in aspect-controlled microclimates and vegetation. I use a combination of techniques including terrain analysis, measurements of field-saturated hydraulic conductivity and soil shear strength, numerical modeling, and erosion rate measurements to test which asymmetry-forming mechanisms are responsible for the topographic asymmetry at Gabilan Mesa. Another outstanding question about climate and landscape evolution is how hillslope sediment transport efficiency varies with climate conditions. Fernandes and Dietrich [1997] suggested that the long response time of hillslopes to perturbations in topography or soil transport makes it unlikely that hillslopes were capable of reaching an equilibrium shape during the Quaternary, and that hillslopes may have constantly modified their form as climate fluctuated between glacial and interglacial periods. Numerous studies point toward climate and vegetation influencing sediment transport [e.g., Fernandesand Dietrich, 1997; Roering, 2004; Hughes et al., 2009; Hurst et al., 2013a; McGuire et al., 2014; West et al., 2014]. However, how climate influences sediment transport efficiency is not well understood, but is fundamental in influencing 9 global sediment flux rates. Hillslope sediment flux rates, which are dominantly controlled by hillslope gradient and disturbance mechanisms that influence the sediment transport efficiency [McKean et al., 1993; Roering et al., 2002], are not easily predicted without specific measurements for a particular landscape. This information generally requires field measurements [e.g., Almond et al., 2008; Jungers et al., 2009], estimates of erosion rates [Perronet al., 2012; Hurst et al., 2013a], or knowledge of the age and initial topography of hillslopes [e.g., Colman and Watson, 1983; McGuire et al., 2014]. In Chapter 4, I compile previously published estimates of hillslope soil transport efficiency, add a new set of estimates using laser altimetry and erosion rates estimated from cosmogenic nuclides, and analyze the combined dataset to determine how hillslope soil transport efficiency varies with climate and other factors. 10 Chapter 2. Modeling the formation of topographic asymmetry by lateral channel migration and aspectdependent erosional processes 11 Abstract Some landscapes exhibit the intriguing characteristic that slopes facing a certain direction are systematically steeper than slopes facing other directions, even where there is no bias introduced by bedrock structure. This topographic asymmetry, which is particularly common in semi-arid regions, has inspired a variety of explanations, but most rely in some way on the presence of different microclimates on opposing slopes. In the simplest scenario, the microclimatic differences lead to differences in erosion rates on opposing slopes and topographic asymmetry develops. In contrast, some geomorphologists have argued that lateral channel migration and the corresponding steepening of undercut slopes are the dominant cause of topographic asymmetry in some landscapes, but this hypothesis has received less attention. To examine these two proposed origins of asymmetric topography, I adapted a numerical landscape evolution model to include lateral channel migration as well as aspect-induced variations in soil creep intensity, regolith strength, and runoff. I compared the characteristics of asymmetric ridges and valleys along with the spatial and temporal patterns of erosion rates produced by each mechanism and found that the model with lateral channel migration produces a unique ridgetop Laplacian and erosion rate signature relative to the other models. This result provides a way to test if lateral channel migration is the dominant cause of topographic asymmetry in a given landscape. To further investigate the dynamics of lateral channel migration, I developed a simple model of hillslope profile evolution in which sediment fluxes from opposing slopes control lateral channel migration rate and direction by deflecting the channel. I find that topographic asymmetry in this system can be self-sustaining and that the degree of asymmetry depends on two non-dimensional numbers that express the relative magnitudes of river incision and soil creep and the relative rates of lateral base level migration and river channel response. 12 2.1. Introduction 2.1.1. Motivation An outstanding problem in the study of Earth's surface is determining how climate influences erosion rates and the long-term evolution of landscapes. Currently, there is no clear consensus on the relationship between erosion rates and climate variables such as precipitation and temperature [von Blanckenburg, 2006], although some geomorphologists have successfully quantified relationships between precipitation and erosion rates at specific sites [Moon et al., 2011; Ferrieret al., 2013a; 2013b]. The difficulty in quantifying the relationship between climate and erosion is likely due in part to differences in rock type, tectonic processes, vegetation, and the degree of chemical versus physical weathering at different sites. One way to make progress on this question is by studying natural experiments in which climate varies, but other factors remain constant. An ideal natural experiment for studying the effects of small climatic variations has been carried out in many landscapes. Variations in insolation, the amount of sunlight that strikes a surface, can generate microclimates - small variations in temperature, soil moisture, humidity, wind and other measurable climate factors - on slopes that face different directions [e.g., Branson and Shown, 1990; Yetemen et al., 2010; Anderson et al., 2012]. These differences in the microclimates may lead to the development of topographic asymmetry, which is the phenomenon that slopes of different aspect exhibit different topographic characteristics (Figure 2.1). Several mechanisms, including lateral channel migration and aspect-induced variations in soil creep intensity, regolith strength, and runoff, have been proposed to explain the asymmetry [e.g., Melton, 1960; Kane, 13 1970; Dohrenwend, 1978; Yetemen et al., 2010; West et al., 2013; McGuire et al., 2014]. However, the long-term topographic consequences of the different mechanisms have not been fully explored, and identifying which mechanism or mechanisms are responsible for the topographic asymmetry remains a challenge at many sites. W4 Figure 2.1. Aerial photograph of Gabilan Mesa, CA. The landscape exhibits a high degree of topographic asymmetry, and there are large differences in microclimates and vegetation on slopes with opposing aspects. Oak trees are primarily present on northfacing slopes and rarely grow on south-facing slopes. 2.1.2. Hypothesized origins of asymmetric topography The systematic occurrence of steeper pole-facing slopes and gentler equator-facing slopes is documented in a variety of landscapes around the world [Bass, 1929; French, 1971; Dohrenwend, 1978; Churchill, 1982; Cerda et al., 1995; Wende, 1995; Siegmund and Kevin, 2000; Burnett et al., 2008; Poulos et al., 2012]. The opposite scenariosteeper equator-facing slopes and gentler pole-facing slopes-has also been observed to 14 occur, but has been reported less often [Gilbert, 1884; Davis, 1895; Fuller, 1914; Glock, 1932; Churchill, 1982; Burnett et al., 2008; Poulos et al., 2012]. Numerous explanations have been proposed for the systematic presence of asymmetric hillslopes, and the debate over their origin dates back almost to the beginning of the formal study of Earth surface processes. Many early observers attributed asymmetric hillslopes to the Coriolis effect summarized in Baer's Law, which they believed caused preferential erosion of the right stream bank in the northern hemisphere [Gilbert, 1884; Davis, 1895; Fuller, 1914; Einstein, 1926; Glock, 1932]. Later workers found that the steepest valley slopes were not always along the right stream bank, and instead found that the steepest slopes were more often the pole-facing slopes [Reed, 1927; Bass, 1929; Fairchild,1932; Emery, 1947]. Asymmetry has also been attributed to differences in eolian deposition and erosion or evapotranspiration driven by regional prevailing winds [Reed, 1927; Bass, 1929; Fairchild, 1932; Emery, 1947], but wind is unlikely to be the main cause of asymmetry because the direction of the prevailing winds does not always match the direction of asymmetry [Powell, 1874; Emery, 1947; Dohrenwend, 1978]. In some regions, topographic asymmetry can be attributed to regional stratigraphic dips or other aspects of bedrock structure or lithology [Powell, 1874; Bass, 1929; Emery, 1947; Melton, 1960; Dohrenwend, 1978], but in areas without such effects, other explanations are necessary. In the absence of tectonic and bedrock structural controls, topographic asymmetry has been attributed to microclimates [Hack and Goodlett, 1960; Dohrenwend, 1978; Istanbulluoglu et al., 2008; Yetemen et al., 2010; Anderson et al., 2012; Poulos et al., 2012]. 15 / The observed correlation between topographic asymmetry and slope aspect-polefacing slopes, which receive less sunlight, are generally steeper-implies that insolation influences the physical and chemical processes that shape landscapes [Poulos et al., 2012]. Recently, geomorphologists have focused on the role of aspect-dependent differences in hillslope b) Aspect-dependent erosional efficiency a) Lateral channel migration Ridgeline migrates towards the equator Pole Ridgeline migrates towards the pole 4- - q, I / I / I I % / / / / / .. / ~ I % / IN. / / / *~ I, /j / -. / / I / / / / / I ., I - qN ~I%*~ / / - -- I I / I I N. N. / / I / Baselevel migrates towards the equator / Figure 2.2. Diagrams showing evolution of asymmetry for two hypothesized asymmetryforming mechanisms. The aspect-dependent erosional efficiency scenario is diagramed in (a) and the lateral channel migration scenario is diagramed in (b). The diagrams show the time evolution of a simplified hillslope profile for hillslopes that develop steeper polefacing slopes. The solid brown lines show the initial and final hillslope profiles, and the dashed red lines show intermediate profiles. In (b), the lengths of the red arrows represent the relative magnitudes of sediment flux from opposing slopes (q,) and the brown blocks represent the relative magnitudes of sediment transported to the main channel. 16 processes that may lead to differences in the efficiency of erosion processes between pole-facing and equator-facing slopes (Figure 2.2). If soil transport or channel incision is more efficient on one side of the divide, the divide will be displaced toward the less efficient side until the difference in slopes compensates for the difference in efficiency. Possible mechanisms causing such a difference in erosional efficiency include (1) reduced runoff, and therefore slower channel incision, on more vegetated slopes [Hack and Goodlett, 1960; Kane, 1970; Wende, 1995; Istanbulluogluet al., 2008; Yetemen et al., 2010] due to either more rapid infiltration [Emery, 1947; Hack and Goodlett, 1960; Kane, 1970] or increased evapotranspiration; (2) higher regolith strength, and therefore slower channel incision, on more vegetated slopes [Emery, 1947; Ollier and Thomasson, 1957; Yetemen et al., 2015] [Emery, 1947; Ollier and Thomasson, 1957]; or (3) asymmetry in soil creep rates due to differences in bioturbation rates [Perronand Hamon, 2012; West et al., 2013; McGuire et al., 2014], frost-generated crack growth [Anderson et al., 2012], or solifluction [Ollier and Thomasson, 1957]. Another hypothesis is that asymmetric aggradation of sediments in a valley bottom forces a river flowing through the valley to migrate away from the side of the valley that experiences faster aggradation, leading to undercutting of the opposite bank and steepening of the adjacent hillslope [Bass, 1929; Melton, 1960; Dohrenwend, 1978] (Figure 2.2). In the northern hemisphere, once the initial undercutting of the north-facing slope occurs, the increased length of the south-facing slope should increase the sediment flux to the north bank of the channel. This may lead to a positive feedback where lateral channel migration is maintained by the difference in sediment flux and aggradation due to the different slope lengths [Wende, 1995]. However, if undercutting persists and the 17 whole ridgeline migrates, erosion rates must remain asymmetric. One possible reason why an initial difference in aggradation may occur is because of the presence of different microclimates. For a site in the central California Coast Ranges, Dohrenwend [1978] suggested that microclimatic differences initially led to a slightly higher erosion rate on south-facing slopes. He suggested that this difference in erosion rates caused higher aggradation on the north bank of channels and southward lateral channel migration, but that the higher erosion rate alone on the south-facing slope was not enough for the development of the topographic asymmetry. Dohrenwend [1978] posited that microclimate-driven lateral channel migration is the dominant mechanism responsible for the high degree of topographic asymmetry witnessed at his study site in the central California Coast Ranges. Furthermore, he concluded that microclimate-driven lateral channel migration is a general process that leads to the development of topographic asymmetry in other semi-arid environments. After investigating asymmetric topography in the Laramie Range, WY, Melton [1960] suggested that most cases of microclimateinduced asymmetry are attributable to lateral channel migration. Wende [1995] concluded that the asymmetric valleys of the Tertiary Hills of Lower Bavaria, Germany were not necessarily formed by microclimates as had been previously suggested for that region. Instead, Wende suggested that the asymmetry might be due to other factors such as non-microclimate-driven lateral channel migration or the initial development of the drainage network. McGuire and coworkers [2014] explored the topographic consequences of different asymmetry-forming mechanisms on cinder cones in the western United States, which generally have gentler south-facing slopes. They found that the asymmetry was likely due 18 to more efficient colluvial transport on the south-facing slopes. Cinder cones offer a unique opportunity to examine the role of aspect-dependent erosional processes where base level effects such as lateral channel migration can easily be ignored. However, base level effects cannot easily be ruled out in many landscapes. Other geomorphologists have also incorporated aspect-dependent erosional processes into landscape evolution models for sites where base level effects are more challenging to rule out [Anderson et al., 2012; Yetemen et al., 2015]. In these cases, they were not able to include the potential effects of lateral channel migration. Microclimates clearly influence erosional processes in some landscapes, but the degree to which lateral channel migration influences the development of topographic asymmetry is not well understood. 2.1.2. Implications for landscape evolution If differences in sediment transport efficiency on opposing slopes are enough to explain the development of topographic asymmetry, then relatively small differences in climate-differences that currently exist on opposing slopes in some regions-can lead to significant differences in hillslope evolution and form. This is especially true for landscapes where the climatic conditions are at a tipping point and a small change in climate can substantially impact the type of vegetation that is capable of growing. Landscapes may undergo considerable drainage network reorganization if lateral channel migration occurs. Valleys and hillslopes are expected to migrate across the landscape if lateral channel migration is occurring, since opposing slopes experience different erosion rates. Where migration rates vary, drainage capture may occur instead of continuous migration of neighboring valleys. Dohrenwend [1978] identified multiple 19 sites of drainage capture in the central California Coat Ranges and suggested that they were due to continuous microclimate-driven lateral channel migration. Whether or not lateral channel migration can occur is likely dependent on additional factors in addition to the presence of microclimates. If lateral channel migration is the dominant mechanism controlling topographic asymmetry then hillslope erosional processes may be less sensitive to microclimates, and therefore changes in climate. Different asymmetric landscapes may be influenced by different asymmetryforming mechanisms. If so, some asymmetric landscapes may be undergoing drainage reorganization associated with topographic asymmetry whereas others are not. Understanding whether or not differences in erosional efficiency or sediment transport efficiency are significant enough to explain observed topographic asymmetry or if lateral channel migration is required to explain asymmetry at specific sites is critical for understanding how sensitive erosional processes are to changes in climate. 2.1.3. Purpose and outline The purpose of this study is to explore the topographic characteristics of asymmetric landscapes formed by different mechanisms to determine if topographic signatures can be used to distinguish the mechanisms from one another. In section 2.2, I modify a landscape evolution model to incorporate a range of possible asymmetryforming mechanisms, including lateral channel migration, aspect-dependent soil creep, aspect-dependent regolith strength and aspect-dependent runoff. In section 2.3, I carry out a series of model experiments in which I use numerical models to create hillslopes with varying degrees of asymmetry. In section 2.4, I investigate the results of a one- 20 dimensional hillslope model with lateral channel migration in which the topographic asymmetry is driven by differences in sediment flux on opposing slopes instead of having a fixed lateral channel migration rate. In section 2.5, I discuss the results and implications of the numerical modeling experiments. Based on the results from the experiments, I propose two tests to determine if lateral channel migration is responsible for the formation of asymmetric topography and explain how these tests could be applied to high-resolution topographic data and erosion rate estimates determined from cosmogenic radionuclides (CRNs). 2.2. Model 2.2.1. Model description I adapt a numerical landscape evolution model (LEM) to incorporate hypothesized mechanisms for generating topographic asymmetry. I do not attempt to explicitly include hydrology, vegetation, or energetics into the model or parameterize the model for a specific landscape. I modify a commonly used landscape evolution governing equation, DV 2 z+ E az at A' Vz s 2 DV z - K(A' Vz -9)+ E A"n Vz" >6, (2.1) where z is elevation, D is soil transport efficiency, K is the fluvial incision coefficient and depends on bedrock lithology, precipitation, and channel morphology in addition to other 21 factors [Perronet al., 2008], A is the cumulative drainage area, 0, is the fluvial incision threshold, and E is the uplift rate or boundary lowering rate [Howard, 1994; Perron et al., 2008]. DV2 z describes soil creep for the scenario in which soil flux is linearly for the to the hillslope gradient. K(A" Vz proportional -Q,) describes channel incision - scenario in which the channel incision rate is proportional to stream power [Seidl et al., 1994] or shear stress [Howard and Kerby, 1983]. Stream discharge is approximated by A and they are related by a power law [Knighton, 1998]. I use the Peclet number (Pe) to describe the magnitude of fluvial incision relative to soil creep, K Pe= - where t("+I-" 0 t, 2 - 6LC)(2.2) is relief and [ is slope length. Pe has been used to capture the competition between soil creep and channel incision in small catchments in soil-mantled landscapes [Perronet al., 2009; 2012]. For a particular landscape, many of the parameters in equation (2.1) are often considered constant [Howard, 1994; Tucker and Bras, 1998; Roering et al., 1999; Perronet al., 2009; 2012] while f and ( depend on the scale of the hillslope feature being analyzed. K, D, 0, m, and n can vary significantly among landscapes with different rock types, tectonic settings or climates [Perronet al., 2012]. Pe ~ 100 is representative of a hillslope with small 1 "-order valleys [Perronet al., 2012]. Pe ~ 250 is representative of the transition from Ist-order to 2"d-order valleys and hillslopes with Pe ~ 1000 usually have well developed 2"d-order valleys. 22 I adapted the Tadpole model developed by Perron and coworkers [2008, 2009, 2012] to include aspect-dependent erosional mechanisms and lateral channel migration. Tadpole is a numerical finite-difference LEM capable of modeling landscapes dominated by fluvial incision and soil creep. Tadpole solves equation (2.1) on a rectangular grid of N, x Ny points with grid spacing of Ax in the x-direction and Ay in the y-direction. I set Ax = Ay for all of the experiments. For all of the experiments, the grid represents a ridge bounded by two straight channels that correspond to the y-boundaries, such that NxAx represents the slope width along the bounding channels and NAy12 is the length of the slope on either side of the divide. The y-boundary conditions are periodic while the xboundary conditions are fixed for all of the model runs. Perron and coworkers [2009, 2012] set 0, = 0 m for most of their model experiments because they were interested in how the competition between soil creep and channel incision controls landscape form, although Perron and coworkers [2009] did explore some of the effects on valley spacing when 0, > 0 m. I explore the topographic consequences for both O = 0 m and for 0, > 0 m. 2.2.2 Models with aspect-dependent processes I incorporate aspect dependence into the LEM by incorporating a weighting function into the governing equation that modifies the efficiencies of processes, including soil creep, regolith strength, and runoff, according to the degree to which the slope is facing the sun. 23 2.2.2.1 Weighting function for aspect-dependent processes To incorporate the effects of insolation into a landscape evolution model, parameters in the model associated with insolation-dependent processes were weighted according to #- the vertical angle between the surface normal and the sun. The sun altitude angle can be parameterized for a specific latitude. cos(#) is a good proxy for the direct solar radiation that reaches a hillslope (Figure 2.3). I parameterized the model with a sun altitude angle of 700, which minimizes the variance between cos(#) and annually averaged solar radiation for a landscape at a latitude of 36'. This is the latitude of the landscape in Figure 1, where both lateral channel migration and erosional efficiencies have been suggested as asymmetry-forming mechanisms [Kane, 1970; Dohrenwend, 1978]. I developed a simple weighting function that is suitable for modifying the magnitudes of individual terms in equation (2.1). The form of the weighting function is 16 cos(#b) osG c+ ) 1 CO cos(#) cos(#,,dg,) (2.3) = 1-1 cos() cos(#-dge where /ridge N cos() <cos(#,,d,,) d is the slope-normal vector of the ridgeline (always 90' from horizontal) and 6 is the magnitude of the weighting function (Figure 2.4). If 6 > 0, w is higher on the polefacing slope. If 6 < 0, o is higher on the equator-facing slope. Increasing the magnitude of 6 causes larger differences in co on opposing slopes. 24 350 I I I I I I 300- A 250 200300 150 -2 - 'Z P250 200 C 100 150 50- 0500 m 0 0 0.1 0.2 0.3 0.5 cos(#) 0.4 0.6 0.7 N 100 0.8 0.9 1 Figure 2.3. Mean annual solar radiation for daylight hours against cos(O) for the portion of Gabilan Mesa, CA, shown in the inset. For clarity, only 10% of data (selected randomly) are plotted. Inset is a shaded relief map overlaid with a color map of mean solar radiation (W m 2 ) for daylight hours. I chose the form of the weighting function so that w = 1 at the ridgeline. I chose a weighting function that is similar to that of Petroff and coworkers [2012], but I modified its form so that w is always greater than zero. Petroff and coworkers explored a limited range of weighting magnitudes, so they did not encounter o < 0. If w is not normalized relative to the ridgeline, then as 5 increases, the coefficient describing the aspectdependent mechanism increases or decreases on both slopes, but at different rates. I formulated equation (2.3) so that as 5 increases, a increases on one slope while decreasing on the other slope instead of increasing or decreasing on both slopes at 25 different rates. Defining the weighting function in this manner also avoids dramatic variations in average erosion rate as 6 varies. a) | b) Pole 1.5 10 10 0 0.2 0.6 0.4 0.8 0.5 1 cos((,)) Figure 2.4. (a) Example of the weighting function, w, for aspect-dependent process rate coefficients. cos(#)= 1 occurs when the slope-normal vector points directly at the sun. Red line shows o for 6 = 10 and the blue line shows o for 6 = -10. The dashed line is at o always equals 1. (b) Perspective view of a landscape produced with the LEM showing value of w for 6 = 10. Horizontal tick interval is 100 m; vertical tick interval is 50 m. 0ridge where For sites near the equator, differences in solar radiation are minimal on pole-facing and equator-facing slopes. At very high latitudes, sun angles are low and differences in solar radiation on opposing slopes are large. However, this does not imply that high latitudes exhibit the largest sensitivity in microclimates. Often semi-arid environments found at mid-latitudes exhibit the most striking aspect-dependent differences in vegetation. This is because semi-arid landscapes are often near a tipping point where pole-facing slopes have adequate soil moisture that is available for vegetation while equator-facing slopes have limited soil moisture available to vegetation [Branson and Shown, 1990; Kutiel, 1992; Istanbulluoglu et al., 2008]. 26 2.2.2.2 Aspect-dependent infiltration and runoff Infiltration and runoff are not calculated directly in the model. Instead, I weighted A, which effectively changes the predicted stream discharge at each point in the landscape. For idealized scenarios, such as during a storm when evapotranspiration is insignificant, runoff is simply the difference between precipitation and infiltration. Normally, discharge is determined by calculating the upslope contributing area and assuming that all parts of that area contribute an equal flux of water. For the aspectdependent runoff LEM, instead of each upslope cell having a value of 1, the cell is weighted according to co and the weighted grid cells are then summed in the typical manner. I introduced aspect-dependence to the relationship between volume discharge (Q,) and A so that Q, = okA" and both k and a are determined empirically [Leopold and Maddock, 1953; Knighton, 1998]. If volume discharge is conserved and precipitation is spatially uniform, which is a reasonable assumption at the hillslope scale, a =1 and the governing equation can be written as az DV 2 z+E (wA)" Vz at DV 2 z - K((wA)" Vz -6) + E (wA)" Vz s (2.4) In order to introduce aspect dependence to the fluvial incision term, McGuire and coworkers [2014] weighted K. However, weighting A instead of K allows the non-local effects of aspect-dependent infiltration, or runoff in this case, to be integrated across the drainage basin. 27 2.2.2.3 Aspect-dependent regolith strength In semi-arid landscapes, incision occurs in ephemeral channels during large storm events that are capable of removing transportable material and vegetation from the channel [Tucker et al., 2006]. In between storm events, I assume bedrock material in the channels is converted to regolith. In this case, regolith refers to weathered material above the bedrock, including soil, and exists on both the hillslopes and in the channel. Regolith strength is partially reflected in the value of Oc [ProsserandDietrich, 1995], but may also influence D and K. I assume that Oc is comparable on both the hillslope and in the channel. Differences in vegetation type and density due to different microclimates on opposing slopes may cause regolith strength to vary with aspect [Yetemen et al., 2015]. For the aspect-dependent regolith strength experiments, I focus on modifying Oe in equation (2.1). Oc is weighted by w and the governing equation is az at DV 2 z+ E A"' Vz w6 -= 2 > Vz A"' E DV z - K(A' Vz -wO)+ (2.5) (2 2.2.2.4 Aspect-dependent soil creep To model the effect of aspect-dependent soil creep, I modified the LEM to include aspect-dependent D. When D varies in space and time according to w, D becomes D(O), and the diffusion term in the governing equation becomes V - D(w) Vz . I investigate the simplest case: D(w) = coD. The form of the modified governing equation is 28 az at V -wDVz+ E A" Vz -= V -o)DVz -K(A"' 1Vz1 -0 C)+E sO Vz(2.6) A"' jVzj > OC Since o changes relatively slowly as the landscape evolves, D(w) can be incorporated into the existing numerical scheme and solved with the Crank-Nicolson method in a similar fashion as if D were constant. 2.2.3 Lateral channel migration Like most previously published LEMs, my model does not explicitly include channels [Howard, 1994; Tucker and Bras, 2000; Moon et al., 2015; Yetemen et al., 2015]. This is in part due to the difficulty of coupling channel bank evolution with hillslope processes, the lack of adequate process laws for the evolution of channel crosssections, and the difficulty of representing relatively narrow channels in grids that span entire landscapes. These issues make it similarly challenging to incorporate lateral channel migration into a LEM. Instead of attempting to model the migration of discrete channels across a landscape, I consider the y boundaries of the model grid to represent straight channels with a fixed spacing equal to NyAy bounding a single hillslope (e.g., Figure 2.4) and perform the simulation in the reference frame of these migrating , to the governing equation so channels. I introduce a lateral channel migration term, y ay that 29 V-DVz+E-y A"' Vz " s ay where y (L T-') is the lateral channel migration rate. The lateral channel migration term is an advection term that shifts the model topography in the positive y-direction, which mimics the effect of channels undercutting slopes that face in the positive y-direction and migrating away from slopes that face in the negative y-direction. The addition of this term leads to competition between the lateral channel migration term, which tends to make these opposing slopes more asymmetric, and the fluvial incision term, which tends to even out the opposing slopes. I describe this competition with a dimensionless value that I refer to as the Migration number, M = Y(2.8) C where C= KA" Vz fl-I , the wave celerity of the fluvial incision term [Whipple and Tucker, 1999]. 2.2.4 Model experiments I investigate the degree of topographic asymmetry that develops for the different LEMs by executing a series of model runs with different parameters. For the aspectdependent efficiency runs, I vary Pe and the weighting parameter 6. For the LEM with lateral channel migration, I vary Pe and the lateral channel migration rate y. By varying the asymmetry-forming mechanism and Pe, I am able to explore the degree of 30 topographic asymmetry that develops for different regions of parameter space, as well as the other topographic and erosional characteristics of the asymmetric landscapes produced by each mechanism. 2.2.4.1. Aspect-dependent efficiency runs In order to explore how topographic asymmetry develops due to differences in aspect-dependent efficiency mechanisms, I run a series of models in which I vary 6 and Pe for the aspect-dependent runoff LEM, aspect-dependent regolith strength LEM, and the aspect-dependent soil creep LEM. I produce hillslopes with different Pe by varying t (by changing N,) in equation (2.2) and estimate the required Pe using equation (2.3). As asymmetry develops, different values of Pe develop on opposing slopes due to differences in f and co. Since ( is not known a priori, I estimate the parameter value for 0, = 0 m to produce the required Pe. If O is low and ( is reasonable, the difference in the actual Pe is small. I calculate the actual Pe a posteriori,once ( is known, for all of the model analyses. I run each model for 10 Myr, with a time step that guaranteed kinematic waves travel no more than Ax or Ay during one time step using the parameters listed in Table 1.1. I solve the advection term using an explicit, forward-time, upwind differencing technique and use the Crank-Nicolson method, which is unconditionally stable, to solve the diffusion term. The stability condition is modified for each aspect-dependent efficiency model to guarantee that no part of the landscape is unstable. All of the modeled hillslopes are oriented so that the side of the ridge that faces in the negative y direction abuts the equator-facing slope and the upper boundary abuts the pole-facing slope. 31 Table 2.1. Tadpole Model Parameters (unless otherwise noted) Parameter (units) Value K(m- 2 m yr-1) ix10-4 D (m2 yr-) 0.02 n 0.5 1 Oe (m 2 m E (m yr-') 1 e-4 Ax, Ay (in) 5 N, Ny 200, 100 Sun angle 700 Pe 100-1000 6* -500-500 50-1500 7** (m Myr~') *aspect-dependent efficiency models **lateral channel migration models 2.2.4.2. Lateral channel migration runs To explore how topographic asymmetry develops due to lateral channel migration, I run a series of models where I vary y and Pe. Similar to the aspect-dependent efficiency models, I produce models with different Pe by varying f (by changing Ny) in equation (2.2). For this set of model experiments, y is fixed for each model run. I calculate M by using y from the respective run and estimate C using the half-width of the hillslope and Hack's law, which relates slope length to drainage area by A = kat h where ka and h are empirically derived, to estimate a representative A. For Hack's law, I use h and ka from Table 2.2. In my experiments, n =1, so C is independent of Vz . I ran each model for 10 Myr, which guaranteed that the model reached a steady form, and with a time step that 32 guaranteed stability for the advection term and did not exceed 1000 yr. The model parameters are summarized in Table 2.1. 2.2.4.3 Definition and measurements of asymmetry Geomorphologists have described topographic asymmetry in many different ways. When multiple hillslopes or valleys exhibit topographic asymmetry in a landscape, multiple terms have been used, including valley asymmetry [Bass, 1929; Emery, 1947; Dohrenwend, 1978], hillslope asymmetry [Poulos et al., 2012], slope asymmetry [Kreslavsky and Head, 2003] and topographic asymmetry [McGuire et al., 2014] to describe the same characteristics. A review of the literature reveals that valley asymmetry is the most popular term, likely because most geomorphologists historically witnessed asymmetry in-person from the bottom of valleys instead of along ridgelines. The weakness of these terms is that they do not describe a specific characteristic that exhibits asymmetry and are defined differently by each author. I choose to refer to the asymmetry as topographic asymmetry as it is sufficiently vague as to not point to a single characteristic or mechanism while also being descriptive enough to characterize the phenomenon. I define topographic asymmetry to include all topographic characteristics that exhibit aspect-dependent asymmetry. To describe the asymmetry of a single characteristic, I define specific metrics. Emery [1947] developed a simple method for reporting the magnitude of topographic asymmetry for opposing slopes. He calculated a single value, referred to as an asymmetry index, which is the ratio of the north-facing hillslope gradient to the southfacing hillslope gradient. Poulos and coworkers [2012] defined a slightly modified 33 asymmetry index (IN-s) as the logarithm of the ratio of the mean gradient of north-facing pixels to the mean gradient of south-facing pixels within a window. This measure is well suited to measuring the topographic asymmetry of large regions, but often does not reflect the significant differences in slope length that occur across a valley. This is because the longer slope can be more deeply incised and therefore comparably steep to the opposing slope on average, even though significant differences in slope lengths exist. I define an asymmetry metric, the bulk slope asymmetry (BSA), that is similar to the slope gradient asymmetry metric used by Emery [1947]: BSA& 1 0lo 2 (2.9) Pt/ SO where S is the hillslope relief divided by the horizontal slope length, pf refers to polefacing slopes and efrefers to equator-facing slopes. I chose to quantify bulk slope asymmetry in this manner because the measurement effectively compares opposing slope lengths normalized by the relief of the hillslope and therefore reflects the visual impression of asymmetry witnessed by an observer of a landscape. I also define an additional metric of asymmetry, the erosion rate asymmetry (ERA), that is used to compare the difference in erosion rate on opposing slopes: ERAN_, = log 2 34 E r Eef (2.10) where E is the erosion rate. If soil creep dominates the morphology near the ridgeline, V 2 zR , the ridgetop Laplacian, is related to the ridgeline erosion rate by D [Perronet al., 2009], the soil transport coefficient, so that (2.11) E = -DV 2 z I developed an additional asymmetry metric that can serve as a proxy for ERApfef and may be useful if estimates of erosion rates on opposing asymmetric slopes are not available, but suitable topographic data does exist. I define ridgetop Laplacian asymmetry (RLA) as RLA_% = log2 V z (2.12) V_1 RC2 where V2 zR is the ridgetop Laplacian. If a landscape is eroding at steady state, V 2 zR is expected to be constant where soil creep dominates and channel incision does not occur. However, if the erosion rate is not constant across the hillslope, differences in exist. To estimate V2Z V 2ZR may , I plot V2z against A"' Vz " where A is drainage area and m and n are semi-empirical exponents that can be parameterized for a particular landscape [Perronet al., 2012]. I bin V 2 z into 10 bins spaced logarithmically in A' Vz n, calculate the median in each bin, and assign V2 ZR as the most negative of these median values. In 35 this case, V 2 zR is a proxy measurement of the fastest erosion rate on the soil creepdominated portion of the hillslope. 2.3.2 Model Results 2.3.2.1 Aspect-dependent efficiency results In this section, I present the results for each of the aspect-dependent efficiency model runs with 0, = 1 m and show how topographic asymmetry varies for different values of 5 and Pe. For most of the model results, I present results in terms of o instead of 6. For all of the aspect-dependent efficiency models, I compare the topographic asymmetry that develops against the ratio of the mean o on equator-facing slopes, Wef, with the mean w on pole-facing slopes, opf. I do this because 0 is a function of 6 and the hillslope morphology, particularly relief, and best accounts for the magnitude of the asymmetric forcing. For the aspect-dependent runoff LEM, pole-facing slopes become steeper and equator-facing slopes become gentler when (o is higher on the equator-facing slope relative to the pole-facing slope (Figure 2.5). This occurs because the magnitude of channel incision increases on the equator-facing slope relative to the pole-facing slope. The opposite occurs when a) is lower on the equator-facing slope relative to the polefacing slope (Figure 2.5). 36 Pole a) 0.06 0.04 l> BSAfCef S 10 0 * 0 S I 0 0 0 0 0 0 Sa S 0 0 0 S 10- S S 0 0 Pol I- 200 0 10 0 100 AO.5IzI (M) 10 I -1 b) 0.06[ 0 S ~ 0.04 -3 0 p 0.02 F 3 10 2 600 400 800 1000 Pe P 0.02 0 10 1 10 0 AO,5IzI 10 (M) Figure 2.5. Model results for the aspect-dependent runoff LEM showing BSApfet in color. I exclude results for 6 = 500 and 6 = -500 because they produce IBSApte I> 3. Profiles to the right of the scale bar show schematic examples of hillslope profiles with BSApfr = 2. Model results of the ridgetop Laplacian signature are shown in (a) and (b) for models with BSApfeI z 2 and BSApfe/ z -2. In (a) and (b), lower values of A 045 Vz are near the ridge while higher values are in the valley. Dark grey points are from the equator-facing slope and light grey points are from the pole-facing slope. White circles are the binned medians. The solid line is fit through the binned medians for the pole-facing data and the dashed line is fit through the binned medians of the equator-facing data. Insets are perspective views of the final model landscapes. Horizontal tick interval is 100 m; vertical tick interval is 10 m. For regions of the hillslope where A 0 5. VzI <~1 m, there are no measurable differences in V 2 z on equator-facing and pole-facing slopes (Figure 2.5a and 2.5b). At steady state, RLApfef and ERApej both approximate zero for the aspect-dependent runoff model runs. 37 Pole a) 0.06 :0 0 .04 BSApfef 104 102 Pole S 0 0 S S 0 S 0 100 0 0 0 0 0 0 0 S ea S 0 10 0.02 3 r S 12 ob eb I 0 0 600 400 800 100 10- 1 AO 5IVzI 0 10 (M) S 0 S 2 0I O b) 0.06 -3 200 0 0 0 S 0 0 0.04 - 1000 Pe Ii 0.02 [ 0 10 100 10 A0 |IVz (m) Figure 2.6. Model results for the aspect-dependent fluvial incision threshold LEM showing BSAp1eLin color. Profiles to the right of the scale bar show schematic examples of hillslope profiles with BSApfe[ = 2. Model results of the ridgetop Laplacian signature are shown in (a) and (b) for models with BSApfe,~ 2 and BSApfe 1 ~-2. In (a) and (b), dark grey points are from the equator-facing slope and light grey points are from the pole-facing slope. White circles are the binned medians. The solid line is fit through the binned medians for the pole-facing data and the dashed line is fit through the binned medians of the equator-facing data. Inset is of shaded relief map. Horizontal tick interval is 100 m; vertical tick interval is 10 m. Unlike the aspect-dependent runoff model, higher values of 0 on the equator-facing slope lead to a decrease in slope length of the equator-facing slope for the aspectdependent regolith strength LEM (Figure 2.6). This occurs because an increase in 060 leads to less channel incision. Slopes that experience a decrease in ow6 due to the 05 weighting function have V 2z that are less negative at low values of A . Vz relative to the slope where w6e increases due to the weighting function (Figure 2.6a and Figure 2.6b). 38 Even though some differences in the behavior of V2z exist on opposing slopes, no RLApfef was discernable for any of the aspect-dependent regolith strength model runs using my current V2 zR measuring technique because neither slope produced more locations with negative V 2 z. In addition, ERApjefalso did not vary significantly from zero when steady state was reached. Unlike the aspect-dependent runoff and regolith LEMs, an increase in Pe for the aspect-dependent soil creep LEM does not necessarily lead to a consistent style, or even sign, of topographic asymmetry (Figure 2.7). For Pe ~ 250, BSApjej does not develop for small differences in o (Figure 2.7a). This is likely due to the increase in erosion rate on the interfluves balancing the increase in channel filling. As Pe increases, the importance of channel incision as an erosional mechanism also increases. If soil creep efficiency on one slope is increased by the weighting function then it can partially fill the channels and limit the effectiveness of channel incision, causing the slope that experiences higher soil creep efficiency to become shorter (Figure 2.7d and 2.7e). If soil creep efficiency continues to increase to the point that the channels are entirely filled, then the slope that experiences higher soil creep efficiency can become longer relative to the opposing slope (Figure 2.7c and 2.7f). Because D is effectively changing on each slope for the aspectdependent soil creep model, V 2zR cannot serve as a proxy for the erosion rate like it does for the other LEMs. This leads to asymmetry developing in V 2zR even though there is no difference in the erosion rate on opposing slopes (Figure 2.7b). The aspect-dependent soil creep LEM did not produce landscapes that appear realistic and often have large RLApjej (>3). This is because for large BSApfef to develop, the longer slope must have 39 BSAPf f a) 102 1.5 0 d c) 0.5 100 -0.5 2 * 0 -1.5 - 10- 200 400 Pe 600 800 1000 d) RLAf e b) 102 I 0 00 10 0 e) 8 e) 20 * 10- 0 -5 200 A 600 400 800 1000 Pe Figure 2.7. (a) Model results for the aspectdependent soil creep LEM showing BSApfef in color. Examples of the model results are shown in 1-4. (b) Model results for the aspect-dependent soil creep LEM showing RLAp-ef. Perspective views of model topography for (c) 6 = 500 and BSAp-eJ = 0.9, (d) 6 = 50 and BSApfef = -0.4, (e) 5 = -25 and BSAp-ef= 0.3, and (f) 6 = -500 and BSApjef= -1.1. Horizontal tick interval is 100 m; vertical tick interval is 10 m. 40 much high soil creep efficiency and no channel incision while the opposing slope was heavily dissected (Figure 2.7c and 2.7f). None of the aspect-dependent LEMs produced significant erosion rate asymmetry when the model runs reached a steady form. However, while topographic asymmetry was developing and the ridgeline was being actively offset, erosion rate asymmetry did exist. This occurred because the lengthening slope erodes more slowly as it becomes longer and shallower, and conversely, the shortening slope erodes more rapidly as it becomes shorter and steeper. Eventually, equilibrium is reached, topographic asymmetry is maintained, and the erosion rates on equator- and pole-facing slopes are equal. For some of the highly asymmetric scenarios, I observed low-order valleys nested in the main tributaries on the longer slope that continue to migrate even though the main ridgeline has stopped migrating. The internal migration is driven by differences in w that occur on equator- and pole-facing slopes in the nested valleys. These discrepancies in wo that occur in small valleys are visible in Figure 2.4b. These variations in w do cause some local variations in erosion rate, but do not significantly affect the mean erosion rate of the whole equator- or pole-facing slope. 2.3.2.2 Lateral channel migration results For the lateral channel migration LEM, significant differences in slope length develop on the equator-facing and pole-facing slope and depend on M (the ratio of the lateral channel migration rate to wave celerity of the fluvial incision term) and Pe. Model runs with wider hillslopes, and therefore higher Pe, developed higher topographic 41 BSAIkf a) 0.15, 3 I 0.05 0 * -0.05 Ii I I I Id U 0 I d) -2 0 200 400 Pe 800 600 Pole 0.06 F -1 -0.1 -0.15 Pole 12 0. 1 -0.04 1000 0.02 RLAjfeO b) 015 2 S I 0 0.5 10- 0.1 0.05 100 AO 5IVz (m) 10 05 e) - -0.05 0.06 r -1.5 - -2 0 200 600 400 800 0.04 1000 Pe L>0.02 ERA Pftf c) 0.I5r 0. I 4 0 0.05 II 0 -0.05 -V 3 I I I S I I I 0 10-1 4 AO 5 10 ( -0.15 101 1VzI (in) -3 -0.1 -4 -0.15 0 200 600 400 800 1000 Pe Figure 2.8. Model results for hillslopes produced with the lateral channel migration LEM indicating (a) BSApe, (b) RLA,., and (c) ERAptejwith color map. Profiles to the right of the scale bar show schematic examples of hillslope profiles with BSAp, = 2. Model results of the ridgetop Laplacian signature are shown in (d) and (e) for models with BSAp, -2. In (d) and (e), dark grey points are from the equator-facing slope ~ 2 and B and light grey points are from the pole-facing slope. White circles are the binned medians. The solid line is fit through the binned medians for the pole-facing data and the dashed line is fit through the binned medians of the equator-facing data. Inset is of shaded relief map. Horizontal tick interval is 100 m; vertical tick interval is 10 m. 42 asymmetry for the same M (Figure 2.8). The undercut slope developed more negative V 2 z than the aggrading slope and a dip in the binned V 2 z can be observed near Aohvzl ~ 1 m where the most negative values of V2 z occur (Figure 2.8d and Figure 2.8e). The erosion rate near the ridge is not uniform. The spine of the divide erodes at the same rate as E, but varies on the rest of the hillslope. On the undercut slope, the erosion rate increases with distance from the divide and is highest along the steepest pitch of the creep-dominated zone. On the aggrading slope, the erosion rate is lower than E. The sustained difference in erosion rates on the undercut and aggrading slope causes sustained migration of the hillslope. Even though sustained differences in erosion rate occur, the hillslope does reach a steady form where the ridgeline and the channel migrate at the same rate and maintain an asymmetric profile. 2.3.2.3 Model predictions and comparisons All of the models that I tested are capable of producing topographic asymmetry. For the aspect-dependent LEMs, the spatial pattern of erosion rates followed a similar history as each model evolved from an initial condition to a steady state. Initially, the slope with the higher erosional potential eroded faster, causing the divide to migrate towards the slower eroding slope until steady state was reached. In stark contrast, the lateral channel migration LEM predicts that differences in erosion rate are maintained on opposing slopes and that the hillslope reaches a steady form that continually migrates. Both the lateral channel migration LEM and the aspect-dependent soil creep LEM are capable of producing hillslopes with nonzero RLApf..ef(Figure 2.9b). For the lateral channel migration 43 LEM, this occurred because of differences in the erosion rate on equator- and pole-facing because of slopes. For the aspect-dependent soil creep LEM, nonzero RLApfefoccurred differences in D on equator- and pole-facing slopes and not because of differences in the erosion rate. However, in cases for which the aspect-dependent soil creep LEM produced hillslopes with high BSApf-e, other characteristics of the topography were unrealistic, such as steep, heavily dissected slopes opposing shallow, completely undissected slopes (Figure 2.7c and 2.7d). The lateral channel migration LEM was the only model that produced asymmetric erosion rates (nonzero ERApfg) (Figure 2.9a). In addition, the lateral channel migration LEM was the only model that produced RLAp-ej significantly different from zero and high BSApf-ef while also producing realistic topographic characteristics. a) 3 SOcreep 2 1 b) ** ,* ,* 3 -00 CP 0 * A runoff * regolith strength * lateral channel migration S Ck* ,.~00 a o6 0 0 0 * * **, -3 * *** 00 0 0 3 0 * -2 -2 -2 -1 0 1 2 -3 3 -2 -l 0 1 2 BSApfef BSAp.ej soil Figure 2.9. Asymmetry signatures for hillslopes produced with the aspect-dependent creep LEM, the aspect-dependent runoff LEM, the aspect-dependent regolith strength (a) LEM, and the lateral channel migration LEM. O, = 1 m for all of the model runs. Erosion rate asymmetry against bulk slope asymmetry. (b) Ridgetop Laplacian asymmetry against bulk slope asymmetry. Refer to legend in (a) for symbol definitions. 44 3 2.3.2.4 Effect of fluvial incision threshold The RLApfef and ERApfef signaturesare valuable for distinguishing the results of the aspect-dependent LEMs from the results of the lateral channel migration LEM. Since QC determines where fluvial incision occurs on the landscape and can influence the morphology near the ridgeline, RLApf< may be sensitive to different values of OC. To determine how QC effects the asymmetry signatures, I duplicated the modeling experiments for the aspect-dependent runoff LEM, aspect-dependent soil creep LEM, and the lateral channel migration LEM, but set 0, = 0 m instead of 6c = 1 m (Figure 2.10). I excluded the aspect-dependent regolith strength LEM since it requires Oc > 0 m for topographic asymmetry to develop. 0 creep A Runoff 2 - * lateral channel migration 1 o0 00 0 0 0 0 0 ~00 -1 -e o -2- 0 0 -3 0 -2 -1 0 0 1 2 3 BSApfef Figure 2.10. Asymmetry signatures for hillslopes produced with the aspect-dependent soil creep LEM, the aspect-dependent runoff LEM, the aspect-dependent regolith strength LEM, and the lateral channel migration LEM. 0, = 0 m for all of the model runs. I also performed an additional set of lateral channel migration LEM runs with 0, = 2 m and explore how 0, may influence RLA~pf and ERA~pf- (Figure 2.11). I did not explore 45 scenarios for the aspect-dependent runoff LEM or the aspect-dependent regolith strength LEM for O, > 1 m because the relationship between BSApjej and RLApfef is unlikely to differ significantly from the OC = 1 m scenario. This is because as QC increases, the creep-dominated zone near the ridgeline should grow larger and lead to a larger zone of uniform V2 zR if the other parameters remain constant. I only expect the RLApfef measurements to change when the soil creep-dominated zone is small, which occurs for lower values of 0. , not higher. The results for the aspect-dependent soil creep LEM are similar to the results when = I m as they both exhibit significant differences in V2 zR between equator- and polefacing slopes when the hillslope is asymmetric (Figure 2.9b and 2.10). The differences in V 2ZR are significantly less pronounced on the equator- and pole-facing slopes for the lateral channel migration runs for Oc = 0 m. This occurs because the entire hillslope, including the ridgeline, experiences fluvial incision when Oe = 0 m, and this mutes the highest magnitude V2 z that would develop if soil creep alone were responsible for responding to the higher erosion rate on the ridges of the undercut slope. For high BSApfef, the aspect-dependent runoff LEM did produce small differences in V2 zR for 6hc 1 m (Figure 2.10). This occurred because the higher magnitude of fluvial incision on the lengthening slope influenced V 2 z ,, not because there was a difference in erosion rates. 46 a) 3- * Om 1000 AOL=Im 2 * b) O=0m A= I M 900 *0,=2m=2m 700 600 300 2000 -3- 0 -3 2 BSANteI -2 0 3 BSApfef Figure 2.11. Effect of fluvial incision threshold on characteristics of asymmetric topography. (a) Plot of ERApf_, against BSApg for hillslopes produced with the lateral channel migration LEM for different values of 0c. (b) RLApp,/against BSA p/e/ for hillslopes produced with the lateral channel migration LEMs for different values of 0'. Color indicates Pe and is the same color scale as (a). For the lateral channel migration LEM, runs with lower Pe or higher 0, produce a steeper trend between ERApfef and BSApf-ef and also RLApfej and BSApfef (Figure 2.11). This may occur for two reasons. First, channels near the ridgeline are less effective at migrating the ridgeline away from the undercut slope for model runs with higher O. If the ridgeline is not able to migrate as efficiently, higher BSApf.ef will develop as the slope is undercut. Second, as 0, increases or Pe decreases, bigger differences in V 2z R exist because the creep-dominated portion of the ridgeline becomes larger. The most significant differences in erosion rate occur on the steeper portions of the hillslope away from the ridgeline. Model runs with higher values of 0, or lower values of Pe have expanded creep-dominated zones that include steeper portions of the landscape. 47 2.4 1 -D model of lateral channel migration In the previous model experiments, I explored the behavior of a 2-D LEM that included lateral channel migration that occurred at a constant rate. Given the hypothesis that lateral channel migration is driven by asymmetric sediment fluxes to channels from adjacent hillslopes, it is possible that there are feedbacks between lateral channel migration and the asymmetric erosion rates it produces. In this section I use a 1 -D (profile) model to explore how topographic asymmetry develops when lateral channel migration varies with time and is set by the difference in the sediment flux on equatorand pole-facing slopes. In one set of experiments, I modify the lateral channel migration rule so that the migration rate is determined by the difference in sediment flux from opposing slopes instead of occurring at a fixed rate and explore the influence of initial conditions on asymmetry. In a second set of experiments, I investigate how model parameters D, K, y, and slope length influence asymmetry. 2.4.1 1-D lateral channel migration model framework The 1 -D model consists of a topographic profile of a ridge bounded on either end by a migrating channel, analogous to a transect in the y-direction of the 2-D model. The profile is subject to both fluvial channel incision and soil creep. The lateral channel migration rate is modified so that y=K (EIL -EL) 48 (12) where Kcm (L) is the lateral migration constant, E is the mean erosion rate for the slope denoted by the subscript, and L is the horizontal slope length from the main divide to the channel. The sediment flux at each boundary is calculated by summing the eroded volume per unit width on equator- and pole-facing slopes at each time step and dividing by the length of the time step. The area that is advected across the grid boundaries is also considered in the flux calculation so that mass is conserved across the domain. In this scenario, a model with perfectly symmetrical initial topography and erosion rates will never develop asymmetry, so the model needs to be seeded by an additional asymmetryforming mechanism or by the appropriate initial conditions in order for a discrepancy in sediment flux to occur and asymmetry to develop. 2.4.2 Model experiments I ran a series of 1 -D models to determine if models with the new lateral channel migration rule can initiate and sustain lateral channel migration in response to an initial background slope or asymmetry. I also explored how M, K, D, L, and y influence asymmetry. For these experiments, I ran each model with the parameters listed in Table 2.2 until a steady form was reached. I chose a ridge half-width of 500 m such that Pe = 250, which corresponds approximately to the transition from l't-order to 2"d-order valleys [Perron et al., 2008]. 49 Table 2.2. l-D Model Parameters* Parameter Value K (mi-2"' yr-t) 1x10 -5 D (m2 yr-1) 0.01 h** 1.67 ka** (m2-h) 6.69 m 0.5 n I E (m yr-) 2x10- L (m) 1000 Ax (m) 2 *Unless otherwise noted **Hack [1957] 2.4.2.1 Model experiments to explore the influence of initial conditions I ran two experiments to explore how either initial hillslope asymmetry or a background slope may influence the development of asymmetry. In the first experiment, I varied the initial BSApfef from -3 to 3 in increments of 0.25 and Kc, from 2.5x 10-3 to 4x 10-3 m- 1 in increments of 1 x 104 m-1 to explore the final degree of BSApfef that develops. The initial form of the topography was of a hillslope with linear slopes and initial relief of 0.05L. In the second experiment, I varied the background slope by seeding the model runs with a tilted initial surface. I varied the background slope from -0.015 to . 0.015 in increments of 0.002 and Ke, from 0 to 4x10-3 m-1 in increments of 2x10-4 m~ 1 For each of these runs, I measured the final BSApef and compared it with the background slope. 50 2.4.2.2 Model experiments to explore the influence of M, K, and y on asymmetry. I ran two series of experiments. For both of these experiments, I use a fixed lateral channel migration rate, but the results should be consistent regardless of which lateral channel migration rule is used. For the first set of experiments, I varied L, D, or K and y. I varied L, D, and K so that Pe ranges from 50 to 500. I used equation (2.8) and varied y for each of the model runs with a different L, D or K so that M ranges from 0 to 0.15 in increments of 0.01. For the second experiment, I varied y to determine how BSApjej responds to different values of y and K for Pe = 250. In order to maintain Pe = 250, I varied D according to equation (2.8). I ran models with three different values of K: M-1 . Ix10-5, 5x10-5, and 10x10 2.4.3 1-D model results 2.4.3.1 Influence of initial conditions For models with Ki, > 2.8x 10-3 m~ 1, topographic asymmetry always developed when some initial asymmetry was present (Figure 2.12). The degree of asymmetry that develops is almost entirely dependent on Kcm and not the initial degree of asymmetry (Figure 2.12a). The time required to reach a steady-state form varied significantly for the different model runs. I define the time to reach steady-form as the time required for the model to reach a state for which the maximum change in the elevation at any point is less than 1 x 10-9 m y-1. It was necessary to require such a slow rate because model runs with low initial BSApfef and low Kcm develop asymmetry extremely slowly. If the definition is less strict, the hillslope will not reach the final degree of topographic asymmetry that 51 BSA5fef a) 3 2 3. 44 Pole I 0 -1 -3 2.5 -3 Initial 2 1 0 -I -2 3 BSAp-ef BSAfef b) A 3 3.5 3 M 2.5 6 0 .1 1.5 -1 -2 0.5 -3 -0.02 -0.015 -0.01 -0.005 0 0.005 0.01 0.015 0.02 Background slope (dz/dx) Figure 2.12. (a) Phase diagram of Klm against initial BSA ,lfor the 1-D aggradationdriven lateral channel migration model. Pe = 250 for all models. Profiles to the right of the scale bar show examples of hillslope profiles. Each grid cell represents an individual model solution. (b) Phase diagram of Kicm against background slope gradient for the 1 -D aggradation-driven lateral channel migration model. Pe = 250 for all models. Profiles to the right of the scale bar show examples of hillslope profiles. Each grid cell represents an individual model solution. would otherwise develop if the model ran longer. I verified that this definition worked by running the model for significantly longer and verifying that the form did not change significantly. The model with Kcm = 2.8x 10-3 m~ 1 and initial BSApfef= 0.25 required 173.4 52 Myr to reach a steady form while the model with the same Kem and initial BSApfef= 0.75 only required 9.9 Myr. The model with Kem = 4x 10- m- and initial BSApff = 0.25 required 8.3 Myr to reach a steady-state form while the model with the same Kic, and initial BSApfg= 3 only required 2.4 Myr. Models with high Klem and initial BSApfef that is similar to the final BSApjej reached a steady form the fastest. If a background slope is initially present, asymmetry develops, but is relatively small (Figure 2.12b). This is because the slopes must develop different gradients and slope lengths on each side of the divide to maintain a uniform erosion rate. When Kicm > 0 mI, higher asymmetry develops and the difference in slope lengths due to the background slope causes a difference in sediment flux on the equator- and pole-facing slopes that drives lateral channel migration. 2.4.3.2 Exploring the influence of M, K, and y on asymmetry For the first set of experiments (Figure 2.13a) in which I changed L, D, or K and y, the model results for a given value of M, the Migration number, produced comparable results for BSApff. The degree of BSApfef that develops is determined by the rate at which the bounding channel migrates and the rate at which the hillslope can respond to the migration. As asymmetry develops, the undercut slope becomes steeper and shorter while the aggrading slope becomes longer and gentler. Even though the undercut slope steepens, it is not capable of matching the sediment flux of the aggrading slope. Even though the sediment fluxes in the model are not balanced on opposing slopes, the hillslopes do reach a steady form. The rate at which the slope can respond to 53 Pe -JU a) 450 400 350 2- -300 250 200 150 100 0.15 0.1 0.05 0 M b) 6 * K=lx 5 - yrI K=5x10-5 yrI K K = I0xI0-5 yr 1 5 - - 4 - 2 T 100 200 300 400 500 600 700 800 900 1000 y (m Myr-1) Figure 2.13. (a) Plot of BSA,/,against M for the 1-D fixed-rate lateral channel migration model. For these results, K and y were varied to produce models with different Pe and M. M. Model 1 However, changing D or L also produce the same trend between BSA ,/against results with BSApjIe < 3.5 are included in the figure. (b) Plot of BSApe against y for the 1D fixed-rate lateral channel migration model. I varied y for three different values of K to explore the dependence of BSAple 1 on K. 54 over-steepening is a function of D and C (= KA" Vz ). This is why both Pe, which includes D and some components of C, and M, which includes C and y, must both be considered to predict BSApf-ef (Figure 2.13a). As D and C increase, the rate at which the hillslope can respond to undercutting increases and the ridgeline is able to migrate faster, leading to a decrease in BSApjej. For the fixed-y modeling scenario, M is known a priori. For the aggradation-driven scenario, y, and therefore M, are emergent values. For the second set of model experiments (Figure 2.13b), model runs with lower values of K developed higher BSAp.f I include the results in Figure 2.13b to show the strong control that K has on BSApjej and to show how BSApgf develops for different values of y. Asymmetry plateaus in the case of K= 1 x 10 m-' because the oversteepened slope becomes a cliff and cannot be further steepened. This degree of asymmetry is unrealistic and the inclusion of non-linear soil creep would limit the BSApfef that develops in these scenarios. Whether or not y is reasonable depends significantly on C. 2.5. Discussion 2.5.1 Controls on topographic asymmetry The degree of topographic asymmetry that develops depends on many factors. For the aspect-dependent efficiency LEMs, these factors are summarized by two dimensionless values that emerge from equation (2.1) and subsequent modifications: Pe and co, / of (Figure 2.5, 2.6, and 2.7). Changes in Pe can represent changes in slope length (n ), soil transport efficiency (D), or rock strength and fluvial incision efficiency (K). wo / w, represents the magnitude of the asymmetry-forming mechanism. If 55 a/ / WP, is comparable across a landscape, differences in basin area and orientation may explain much of the variability in topographic asymmetry witnessed for a single landscape. The aspect-dependent soil creep LEM produced hillslopes with both signs of BSApf ef for positive values of 6 and negative values of 6 (Figure 2.7). If the governing equation is modified so that soil creep does not occur in the channels, I expect that the sign of BSApfef would not change as Pe is increased. I use the existing form of equation (2.1) because it has been validated for some landscapes [Perronet al., 2009; 2012]. The aspect-dependent soil creep LEM predicts that slopes experiencing the same microclimatic effects may develop different signs of BSApf-ef depending on Pe, which correlates with f for a particular landscape. McGuire and coworkers' [2014] numerical modeling experiments of asymmetric cinder cones in the American West suggests that the asymmetry at their field sites develop due to higher soil creep rates on south-facing slopes. Our results match their findings when Pe <~400. However, our results suggest that this result is not universal and that cinder cones experiencing the same microclimatic conditions may develop the opposite sign of asymmetry if their slopes are longer and Pe > ~400 (Figure 2.7a). When Pe > 400, the sign of asymmetry that develops depends on the magnitude of co,, / op/. For the lateral channel migration LEM, the degree of asymmetry that develops can also be summarized by two dimensionless factors: Pe and M (Figure 2.8). For the aggradation-driven lateral channel migration scenario, M is an emergent property determined by the efficiency of lateral channel migration (represented by Ken), C, and Pe. M emerges due to the equilibrium that is reached when the difference in sediment 56 flux on equator- and pole-facing slopes is at a maximum and an increase in y no longer leads to a further increase in the difference in sediment fluxes. For both the aggradation-driven and fixed-rate lateral channel migration models, M and Pe are dimensionless numbers that can be used to predict the final BSA pj.. that develops (Figure 2.8 and Figure 2.13a). It should be possible to estimate the value of y required to produce a particular value of BSApjej if Pe and C are known from high-resolution topographic data [Perronet al., 2012]. 2.5.2 Model and weighting function One limitation of incorporating a weighting function into a LEM to characterize aspect-dependent differences is the challenge of reproducing the correct spatial scale of microclimates. There is likely a minimum scale for which microclimates can effectively cause differences in erosional processes. Xu and coworkers [2004] examined the correlation coefficients between landscape characteristics (e.g., elevation, gradient) and microclimate proxies (e.g., air temperature, soil temperature, and soil moisture) at different length scales along a transect in the southeastern Missouri Ozarks and found that landscape characteristics measured at scales below 100 meters do not correlate well with microclimate proxies while landscape characteristics measured at scales between 100-500 meters exhibit some correlation. Landscape characteristics measured at scales above 500 meters generally correlated the best with microclimate proxies. These scales are on par with the scale of many hillslopes. Of the three aspect-dependent LEMs that I test, the aspect-dependent runoff LEM best captures the spatial scale of microclimates since co is integrated over the upstream drainage area at each point on the landscape. For 57 all of the aspect-dependent LEMs, o is determined locally and the calculation does not take into account the surrounding microclimatic environment. For example, locations in low-order valleys on equator-facing and pole-facing slopes may have identical # even though the microclimatic conditions for the equator-facing and pole-facing slopes are quite different. This may lead to an over or under estimation of w depending on the surrounding microclimate conditions. In this study, I did not investigate the topographic signatures of multiple asymmetry-forming mechanisms acting simultaneously even though there is some evidence that multiple mechanisms may act in unison to cause topographic asymmetry. Hack and Goodlett [1960] suggested that differences in both soil creep rates and runoff led to the development of topographic asymmetry at their field site near the headwaters of the Shenandoah River, VA. Other geomorphologists have also suggested that multiple asymmetry-forming mechanisms may act in conjunction to cause asymmetry [Walker, 1948; Churchill, 1981; Siegmund andKevin, 2000]. It would be possible to combine multiple asymmetry-forming mechanisms in the LEM, but this was beyond the scope of this study. I modeled soil creep with a linear sediment flux law, which is a good descriptor for soil creep in many soil-mantled landscapes [Hanks et al., 1984b; McKean et al., 1993; Rosenbloom and Anderson, 1994]. However, as hillslopes approach a critical gradient (~0.6), non-linear sediment flux is a better approximation of soil creep [Roering et al., 1999]. For the 2-d LEM, I was careful to produce models that did not have gradients > 0.6. In regards to topographic asymmetry, the sediment flux law likely 58 influences the degree of BSApfejthat develops when steep gradients exist. A non-linear sediment flux law for soil creep may be necessary for future modeling scenarios. I limited my investigation to the evolution of a hillslope divide and the corresponding slopes on either side of the divide. I focused on hillslope signatures due to the difficulty of incorporating lateral channel migration into a larger scale LEM. This causes the divide and the valley to be strongly coupled. In real landscapes, neighboring valleys do not necessarily migrate at the same rate, which may lead to differences in the rate of ridgeline and channel migration. Larger scale topographic signatures of lateral channel migration may develop due to differences in ridgeline and valley migration rates, boundary conditions or drainage geometry. I was not able to investigate these signatures due to the limitations of incorporating lateral channel migration into a LEM which could allow multiple channels to migrate. 2.5.3 Self-sustained lateral channel migration and microclimates In my 1 -D model, a range of solutions exists for which hillslopes maintain a constant form and migrate at a constant rate while maintaining differences in sediment flux rates. The extent to which this actually occurs in nature likely depends on local conditions such as drainage geometry and the regularity of basin size, because basins with different sizes will develop different lateral channel migration rates for the same efficiency of undercutting (represented by Kcm). Smith and Bretherton [1972] developed a simple geometric model to explore the stability and form of aggradation-driven lateral channel migration, but in their experiments, the hillslope did not reach a steady form. This likely occurred because they did not consider the feedback between multiple 59 migrating ridgelines and valleys or allow for the migration of an entire ridgeline. They only considered the feedbacks between an aggrading slope and the opposing undercut slope, which led to a scenario where the undercut slope was neither able to balance the sediment flux on the aggrading slope nor migrate away from the aggrading slope. If they had allowed the entire undercut slope, including the ridgeline, to migrate, their hillslope model should also have been able to reach a steady form. My 1 -D lateral channel migration model runs with initial topographic asymmetry or background slope developed topographic asymmetry due to the differences in slope lengths and the corresponding difference in sediment flux. This suggests that the presence of microclimatic variability may not be necessary for topographic asymmetry to develop. However, for some of these scenarios, it is important to consider the amount of time required to reach a steady form. Some of the model runs that started with low initial topographic asymmetry required over 100 Myr to reach a steady form, and the asymmetry develops slowly throughout the model run. It is unrealistic that any natural landscape will reach a steady form or develop much asymmetry at all if so much time is required. The time required for the model runs with higher values of Kc, were much more realistic, but still could require on the order of 10 Myr to reach a steady form. 2.5.6 Application to field sites The lateral channel migration models predict a unique relationship between BSApfj. and ERApfef (Figure 2.9a). Cosmogenic radionuclide-derived erosion rates could be used to test if lateral channel migration is occurring in landscapes with topographic 60 asymmetry. The lateral channel migration LEM clearly predicts that sustained differences should exist, with the undercut slope eroding faster than the aggrading slope. The lateral channel migration LEM also predicts a unique relationship between BSApjefand RLApfef that can often be distinguished from the results of the other LEMs (Figure 2.9b). This asymmetric signature could also be used to test whether or not lateral channel migration is occurring at a field site and if the degree of lateral channel migration is sufficient to explain the degree of topographic asymmetry that is present. For the lateral channel migration LEM, model runs with higher values of 6, are more easily distinguished from the aspect-dependent runoff and aspect-dependent regolith strength results than model runs with low at a field site is low, it may QC. If QC be difficult to use the RLApfef against BSApfef signature to identify lateral channel migration because the degree of RLApfef that develops may be insufficient. The aspectdependent soil creep LEM also produces hillslopes with high values of RLAp.ej, but the slope between BSAp.ef and RLApfpej is much steeper than the slope for all of the lateral channel migration LEM scenarios that I explored. In addition, the aspect-dependent soil creep LEM produced unrealistic topographic characteristics for models with high RLApfef and only moderate BSApe{. If an additional asymmetry signature was used, for example one that would measure asymmetry in drainage density, it should be easy to distinguish the topographic characteristics produced by the aspect-dependent soil creep LEM and the lateral channel migration LEM. My 1 -D hillslope modeling results suggest that initial differences in slope length alone are capable of producing topographic asymmetry, but the fact that the sign of asymmetry is latitude-dependent implies that microclimates do play an important role in 61 controlling the sign of asymmetry [Parsons, 1988; Poulos et al., 2012]. Our results corroborate the work of Wende [1995] that it may be important to consider the role of tilting or initial basin geometry in the formation of asymmetric landscapes if lateral channel migration is responsible for development of topographic asymmetry. Landscapes that exhibit high asymmetry, such as Gabilan Mesa, CA, commonly form in poorlyconsolidated, gently dipping sediments [Dohrenwend, 1978]. In some cases, tilting alone may be enough to cause topographic asymmetry to develop [Garciaand Mahan, 2012]. Dohrenwend [1978] postulated that rock strength is an important factor influencing the degree of topographic asymmetry that develops in landscapes where LCM causes the asymmetry. In particular, lateral channel migration would occur less efficiently and at a slower rate in valleys with hard bedrock. Rock hardness likely influences Kcn, and softer rocks are better described by high values of Kicm and more efficient lateral channel migration for a particular difference in sediment flux on opposing slopes. Channels with strong bedrock are less prone to migrate laterally [Montgomery, 2004; Limaye and Lamb, 2014]. In landscapes with weaker rocks, lateral channel migration may occur more easily, which would be reflected in a higher Kcm in the aggradation-driven hillslope model. All else being equal, models with higher K1,, have higher values of M and develop higher BSApfef. The aspect-dependent LEMs should not exhibit the same sensitivity to bedrock channel strength. No study has been carried out to investigate the occurrence and degree of asymmetry in landscapes with different bedrock strength. The dynamics between Kem and K for different rock strengths may be critical for determining if more or less topographic asymmetry develops for weaker rocks. If landscapes with stronger bedrock develop less asymmetry, this may be evidence that lateral channel migration is occurring 62 at many sites, as it is the only asymmetry-causing mechanism that is sensitive to rock strength. If aggradation-driven lateral channel migration is a major cause of topographic asymmetry and only differences in sediment fluxes are necessary to cause the asymmetry, it is reasonable to ask why topographic asymmetry does not develop in all landscapes. The simplest reason is that the conditions necessary for lateral channel migration are probably not present in all landscapes. Dohrenwend [1978] suggested that some basins at Gabilan Mesa developed little or no asymmetry because their channel gradients were sufficiently high that significant aggradation did not occur, and thus there was insufficient sediment to deflect channels. An updated explanation would rely on stream power or shear stress instead of channel gradient as a predictor of the channel sediment transport capacity. In addition, some valleys may experience insufficient sediment flux from the hillslopes to develop significant aggradation-driven asymmetry, independent of the channel's transport capacity. My 1 -D modeling experiments (Figure 2.12a) provide an additional explanation for why asymmetry may not develop. Even though the models reach almost identical BSApfef independent of the initial BSApfef, the time required for the model profile to reach a steady form can be exceedingly long because the asymmetry develops slowly. The models that started from conditions of low initial BSApfef and had low values of Kc, developed topographic asymmetry very slowly and can require > 100 Myr to reach a steady form. If this accurately reflects the time required for landscapes to develop pronounced asymmetry, it is understandable why asymmetry is not more widespread. 63 2.6. Conclusion I incorporated microclimate-dependent erosional mechanisms and lateral channel migration into a landscape evolution model to investigate the origins of asymmetric topography. The model with lateral channel migration predicts a unique topographic signature that the ridgetop Laplacian of the undercut slope should be higher than the ridgetop Laplacian on the south-facing slope. This topographic signature is distinguished easily from the other models for many scenarios. The lateral channel migration model also predicts that the steeper slope in asymmetric landscapes will be eroding at a significantly higher rate than the opposing, shallower slope while the other models do not. The aspect-dependent runoff model and aspect-dependent regolith strength model produce comparable signatures between bulk slope asymmetry and ridgetop Laplacian asymmetry. The aspect-dependent soil creep model produces topography with relatively high ridgetop Laplacian asymmetry and low bulk slope asymmetry and is not capable of producing topography with high values of the P6clet number, which describes the competition between channel incision and soil creep, with realistic topographic characteristics. The aspect-dependent soil creep model also predicts that different signs of bulk slope asymmetry develop for different values of Pe. Multiple mechanisms may act in real landscapes, making it difficult to decipher which mechanism is responsible for causing the topographic asymmetry. Results from a 1 -D model with aggradation-driven lateral channel migration suggest that topographic asymmetry may develop or be sustained in landscapes without variability in microclimates. 64 2.6. Acknowledgements I would like to thank Scott Miller for useful discussions that helped direct some of the modeling efforts. I would also like to acknowledge the Department of Defense for funding through a National Defense Science and Engineering Graduate Fellowship. 65 66 Chapter 3. Unraveling the mysteries of an asymmetric landscape 67 Abstract North-facing slopes in semi-arid regions of the northern hemisphere are commonly steeper than south-facing slopes. The most common modern explanations for this topographic asymmetry ultimately invoke aspect-related microclimate. However, the specific mechanisms that generate the asymmetry are not well understood. I investigated the potential causes of topographic asymmetry at Gabilan Mesa, CA, a site that experiences large differences in microclimates and has highly asymmetric landforms with north-facing slopes that are considerably steeper than south-facing slopes. Two different hypotheses have been suggested to explain the asymmetry at Gabilan Mesa. For one hypothesis, different microclimates on opposing slopes are responsible for causing differences in erosional efficiency, which directly leads to the topographic asymmetry. For the other hypothesis, differences in microclimates alone are not enough to cause the asymmetry to develop. Instead, lateral channel migration, driven by differences in sediment flux on the opposing slopes, and the corresponding undercutting causes the topographic asymmetry. I also considered the role of initial tilting of the mesa in causing the asymmetry. I carry out numerical modeling experiments, complete terrain analysis, and make field measurements to test these different hypotheses. I considered two aspectdependent efficiency mechanisms: aspect-dependent runoff and aspect-dependent regolith strength. If aspect-dependent runoff causes the asymmetry, runoff should be higher on south-facing slopes. I estimated field-saturated hydraulic conductivity at two different sites with varying degrees of asymmetry and found that field-saturated hydraulic conductivity is considerably higher on south-facing slopes in a highly asymmetric basin. This is consistent with the expectation if aspect-dependent runoff is responsible for the asymmetry. If aspect-dependent regolith strength causes the asymmetry, the north-facing slopes should have higher soil shear strength. I measured soil shear strength on northfacing and south-facing slopes and found that soil shear strength was significantly higher on the south-facing slope, which is inconsistent with aspect-dependent regolith causing the topographic asymmetry. If lateral channel migration is responsible for the asymmetry, associated stream captures and channel beheadings may occur. I identified the locations of several stream captures and channel beheadings, but did not identify them in most of the asymmetric valleys. I also tested erosion rate and topographic predictions made by the numerical landscape evolution models that incorporate the different asymmetry-forming mechanisms against cosmogenic radionuclide-derived erosion rates and topographic characteristics at Gabilan Mesa. The aspect-dependent runoff model and the aspectdependent regolith model best reproduce the signature between erosion rate asymmetry and topographic asymmetry when compared to the lateral channel migration model. When considered with the field measurements of soil strength and field-saturated hydraulic conductivity, aspect-dependent runoff is the most likely mechanism responsible for the bulk of the topographic asymmetry at Gabilan Mesa. However, it is not possible to rule out the role of tilting in influencing the initial development of asymmetry in some basins. Furthermore, the erosion rate analysis and the physical evidence suggests that lateral channel migration is occurring at some locations, but taken in conjunction with the numerical modeling predictions and erosion rates, lateral channel migration is most likely only intensifying topographic asymmetry locally and not fundamentally responsible for its development. 68 3.1. Introduction 3.1.1 Motivation Although much effort has been made to understand the role of climate in landscape evolution, many fundamental questions remain [Molnar, 2004; Molnar et al., 2006]. Some studies have found a relationship between erosion rate and precipitation rate and show that erosion rate increases with increased precipitation rate [Moon et al., 2011; Ferrieret al., 2013a]. However, even the relationship between something as fundamental as precipitation and erosion rate is not easily predicted in many landscapes, with some geomorphologists arguing that erosion rates may increase with aridity due to differences in the frequency and magnitude of storms [Molnar, 2001; Molnar et al., 2006] or a decrease in cohesive vegetative groundcover [Bull, 1997]. One way of addressing how landscapes evolve under different climates is by studying how landscapes respond to differences in microclimates, which can vary on hillslopes that face different directions with respect to the sun. Understanding how microclimatic variability at the hillslope scale influences landscape evolution offers a unique opportunity to isolate climatic variability from other variables that influence landscape evolution, such as lithology and tectonic uplift. Some landscapes that exhibit microclimatic differences also exhibit significant differences in topographic characteristics that appear to also vary with aspect. Some geomorphologists have suggested that the different microclimates are directly responsible for the differing topographic characteristics [e.g., [Emery, 1947; Melton, 1960; Dohrenwend, 1978; Churchill, 1982; Burnett et al., 2008; McGuire et al., 2014]. I seek to address some of the potential mechanisms that may be sensitive to the differences in microclimate and responsible for the development of topographic asymmetry. 69 Understanding how landscapes respond to different microclimates should help illuminate the relationship between larger-scale climates and landscape evolution. 3.1.2 Background Numerous studies have been carried out to address how microclimates may influence the development of asymmetric topography, in which characteristics such as slope gradient differ on hillslopes with different aspects (Figure 3.1). The occurrence of steeper pole-facing slopes and gentler equator-facing slopes has been documented for different landscapes around the world [Bass, 1929; French, 1971; Dohrenwend, 1978; Churchill, 1982; Cerdli et al., 1995; Wende, 1995; Siegmund and Kevin, 2000; Burnett et al., 2008; Poulos et al., 2012]. In regions where the topographic asymmetry cannot be attributed to regional stratigraphic dips or other aspects of bedrock structure or lithology, geomorphologists have often relied on the presence of microclimates to explain the topographic asymmetry [Hack and Goodlett, 1960; Dohrenwend, 1978; Istanbulluogluet al., 2008; Yetemen et al., 2010; Anderson et al., 2012; Poulos et al., 2012]. Two leading hypotheses have emerged to explain the occurrence of microclimatically induced topographic asymmetry. One hypothesis states that small differences in microclimate cause differences in the efficiency of erosional processes on slopes with different aspects. The more efficiently eroding slope erodes faster until the resulting slope asymmetry compensates for the difference in erosional efficiency. Possible mechanisms causing such a difference in erosional efficiency include: (1) 70 Figure 3.1. Image of Gabilan Mesa, CA (36.917' N, 120.760' W). Striking topographic asymmetry and large differences in vegetation on opposing slopes are plainly visible. North-facing slopes have steeper gradients and denser vegetation than south-facing slopes. Image courtesy of Google Earth. reduced runoff, and therefore slower channel incision, on more vegetated slopes [Hack and Goodlett, 1960; Kane, 1970; Wende, 1995; Istanbulluoglu et al., 2008; Yetemen et al., 2010] due to either more rapid infiltration [Emery, 1947; Hack and Goodlett, 1960; Kane, 1970] or increased evapotranspiration; (2) stronger regolith, and therefore slower channel incision, on more vegetated slopes[Emery, 1947; Ollier and Thomasson, 1957; Yetemen et al., 2015] or (3) asymmetry in soil creep rates due to differences in bioturbation rates [Perronand Hamon, 2012; West et al., 2013; McGuire et al., 2014], frost-generated crack growth [Anderson et al., 2012], or solifluction [Ollier and Thomasson, 1957]. According to the hypothesis that depends on lateral channel migration, the difference in erosional efficiency is not sufficient to create the observed topographic asymmetry. Instead, sediment aggradation at the foot of the more quickly eroding slope 71 forces lateral channel migration and undercutting of the opposing slope [Bass, 1929; Melton, 1960; Dohrenwend, 1978], which steepens the undercut slope and reduces the gradient of the opposing slope. Melton [1960] investigated asymmetric topography in the Laramie Range, WY and suggested that most cases of microclimate-induced asymmetry are attributable to lateral channel migration. Dohrenwend [1978] also came to a similar conclusion for a landscape with a high degree of topographic asymmetry in a semi-arid region of California. Wende [1995] examined asymmetric valleys of the Tertiary Hills of Lower Bavaria, Germany and, contrary to former conclusions, suggested that microclimates may not be required for their formation. Instead, Wende suggested that factors not relating to microclimates such as lateral channel migration that was driven by asymmetry in initial development of the drainage network may be responsible. In Chapter 2, I showed that lateral channel migration may be a self-sustaining process and does not necessarily require the presence of microclimates if an asymmetry in sediment flux to opposite banks of a channel occurs for other reasons such as differing slope lengths across a valley. There are conflicting hypotheses about the origins of topographic asymmetry, with some geomorphologists appealing to lateral channel migration, which may or may not be driven by microclimatic differences, and others suggesting that differences in erosional efficiency on opposing hillslopes experiencing different microclimates is enough to cause the asymmetry. Understanding which mechanism is dominantly responsible requires a detailed examination of a field site where either mechanism may be active. In the following section, I describe such a field site. 72 3.1.3 Study site: Gabilan Mesa, California 3.1.3.1 Site Description Gabilan Mesa is a ~2500 km 2 rectangular region in central California that is bordered to the southwest by the Salinas River and to the northeast by the San Andreas Fault (Figure 3.2). The area is considered a mesa because of the elevated, roughly planar surface defined by concordant ridge tops that stretch from the Salinas River towards the San Andreas Fault. Evidence of landslides is uncommon [Perronet al., 2009]. Portions of Gabilan Mesa remain undissected; these relict surfaces are most common in the northern portion of the mesa while the southern portion of the mesa is generally dissected into ridge-and-valley topography characterized by sharply concave-up valleys and smooth, concave-down hilltops. Many first- and second-order valleys lack active channels, and instead appear as colluvium-mantled hollows, possibly indicating a change in sediment storage and discharge that accompanied the Pleistocene-Holocene transition [Reneau et al., 1986]. The dominant drainage direction of Gabilan Mesa is to the southwest, and the roughly planar surface defined by ridgetops throughout the mesa dips 1.5' to 2' southwest. The landscape is incised into the Paso Robles formation, which is composed of moderately consolidated conglomerates, sandstones, and siltstones that 73 Legend a) Alluvium Paso Robles Formation mudstone, and limestone) tN (sandstone, conglomerate, Pancho Rico Formation (sandstone, conglomerate, mudstone, diatomite, andporcelaneousrocks) Hames Member of the Monterey Formation (siliceousmudstone, porcelanite, chert, and dolomite) Figure 3.2. (a) Shaded relief map of Gabilan Mesa, CA with lithological map overlay. Outline of California in the lower left of the frame shows the location of Gabilan Mesa marked with a star. (b) The south-facing hillslopes appear more regularly and deeply incised than the north-facing hillslopes (36.069' N, 120.863' W). Image courtesy of Google Earth. Location of image marked as Photo 1 in (a). (c) Some valleys at Gabilan Mesa exhibit evidence of southward lateral channel migration and undercutting (35.913' N, 120.8360 W). This high degree of undercutting is not typical of every basin at Gabilan Mesa. Location of image marked as Photo 2 in (a). 74 were originally deposited as a bajada. The dominant dip of beds in the Paso Robles formation is 1.5' to 3' to the southwest [Durham, 1974]. The Paso Robles formation is Pliocene to Pleistocene in age and is underlain conformably by Pliocene marine deposits. The Paso Robles formation has not been internally deformed by tectonics despite the proximity to the San Andreas Fault and several other minor regional faults [Dohrenwend, 1978]. This is likely due to the strength of the basement rocks, which are primarily composed of granitic intrusive and metamorphic rocks [Durham, 1974]. Gabilan Mesa is classified as a semi-arid, steppe-type climate in the Kppen climate classification [McKnight and Hess, 2008]. The summers are hot and dry, while the winters are mild and wet. The mean annual rainfall in the nearby town of Paso Robles is 374 mm. Interior Live Oaks (Quercus wislizeni) are common on the mesic (moderately moist) north-facing slopes. Both the north-facing slopes and the xeric (dry) south-facing slopes are blanketed with a mixture of grasses that includes both introduced (Bromus, Avena, and Festuca) and native genera (Stipa, Poa, and Aristida) [Kane, 1970]. Evidence from marine sediments, plant fossils, pollen, soil caliche and charcoal suggests that the climate has changed since the Pleistocene, which was likely wetter and cooler relative to present conditions [Johnson, 1977]. 3.1.3.2 Proposed origins of the asymmetric topography at Gabilan Mesa At Gabilan Mesa, significant differences in hillslope morphology are welldocumented and correlate with modem differences in aspect-controlled microclimates and vegetation. However, the specific erosional mechanisms that lead to the differences 75 in morphology are not well understood. I seek to determine the mechanism or mechanisms that dominantly control the topographic asymmetry. Residents of the area have long noted the topographic asymmetry, and one popular idea was that that the asymmetry was due to preferential aeolian erosion and deposition due to the dominant wind direction (Figure 3.1). Reed [1927] was the first author to write a scientific paper addressing the topographic asymmetry of Gabilan Mesa. He noted that the dominant wind direction was inconsistent with the orientations of the asymmetric slopes and suggested that, independent of the dominant wind direction, the asymmetry in vegetation caused the differences in aeolian erosion and transport. The north-facing slopes of Gabilan Mesa are generally more heavily vegetated and steeper, and he suggested that the vegetation was able to trap aeolian sediment, causing northfacing slopes to steepen relative to the sparsely vegetated south-facing slopes. This hypothesis considers the topography to effectively behave as a set of large dunes. This idea was later abandoned as more reasonable alternatives were proposed. Kane [1970] came to a different conclusion to explain the valley asymmetry at Gabilan Mesa. In Sarah Canyon, a highly asymmetric valley, Kane interpreted qualitative infiltration rate estimates as evidence that differences in erosional efficiency on slopes with different aspects were responsible for the valley asymmetry. Additionally, Kane did not find any evidence that preferential aggradation was correlated to the presence of asymmetry, and doubted that slope undercutting could lead to significant valley asymmetry in semi-arid environments. In stark contrast, Dohrenwend [1978] argued that systematic undercutting of north-facing slopes was the primary mechanism driving the formation of asymmetric 76 valleys and concurred with Melton's [1960] conclusion that differences in the efficiency of erosion processes are not sufficient to cause valley asymmetry. Dohrenwend [1978] argued that other characteristics of the site's topography support the slope-undercutting hypothesis, including beheaded streams and the capture of southern tributaries that may record the southward migration of north-facing hillslopes. Recently, Garcia and coworkers [2011] suggested that tectonic tilting instead of microclimates controls the asymmetry of Peachtree Valley, a major northwest-southwest trending valley at Gabilan Mesa. 3.1.3.3Approach and outline In Chapter 2, I used a numerical model to make unique topographic and erosion rate predictions for different asymmetry-forming mechanisms, including aspectdependent erosional efficiency and lateral channel migration. These predictions can be tested using a combination of high-resolution topographic data and cosmogenic radionuclide-derived erosion rates. Furthermore, field measurements of soil properties and topographic evidence of drainage network reorganization can be used to test for evidence of the hypothesized mechanisms described in the previous section. I take a multi-faceted approach to determine the possible causes of asymmetry at Gabilan Mesa. I carry out terrain analysis in section 3.2 and identify locations of river capture and beheaded channels. I also address the applicability of ridgetop Laplacian analysis developed in Chapter 2. In section 3.3, I consider the possible role of initial tilting and the resulting background slope at Gabilan Mesa as a possible mechanism that contributes to asymmetry. In section 3.4, I present results for cosmogenic radionuclide- 77 derived erosion rates for opposing north-facing and south-facing slopes at Gabilan Mesa. In section 3.5, I present numerical landscape evolution modeling experiments that include different asymmetry-forcing mechanisms and compare the topographic and erosion rate predictions with the erosion rate estimates and topography of Gabilan Mesa. 3.2 Terrain analysis 3.2.1 Evidence of drainage network reorganization I mapped beheaded valleys and stream captures at Gabilan Mesa from aerial imagery and digital elevation data. I identified 27 previously unidentified stream captures or beheaded valleys (Figure 3.3). I focused on identifying stream captures and beheaded valleys in the Paso Robles Formation. Three large drainage basins in or near my study area-Powell Valley, Portuguese Valley, and Indian Valley-all exhibit physical evidence of lateral channel migration in the form of beheaded channels or stream piracy (Figure 3.2a and 3.3). Near the outlet of Indian Valley where it joins the Salinas River (Figure 3.2a), evidence of westward migration and valley beheadings in the neighboring basin serve as strong evidence of lateral channel migration and undercutting of an east-facing slope. In contrast, Portuguese Valley mostly experiences undercutting of its northwest-facing slope. This lends evidence to the idea that lateral channel migration is an important mechanism for the reorganizing of drainage networks, but does not clearly point towards a microclimatic origin since the channels have migrated in different directions. 78 3.2.2 Metrics for measuring asymmetry I use the asymmetry indices developed in Chapter 2 and modify them specifically for the northern hemisphere. I use the bulk slope asymmetry measurement to quantify the asymmetry in slope length across a valley. Bulk slope asymmetry is defined as BSA = log 2 Sf (3.1) s~f Where S is bulk slope and is calculated as the hillslope relief divided by the slope length. Subscript nfrefers to north-facing while sfrefers to south-facing and denotes the aspect of the slope. I measure BSAnfsf across valleys, as opposed to hillslopes, and split the valleys along the channel, separating the valley into north-facing and south-facing zones. Within each zone, I measure slope length as the horizontal distance from the channel to the ridgeline along a transect that is oriented perpendicular to the average channel direction. I make this measurement at each pixel location along the channel and define slope length for the zone as the mean of the slope length measurements. The relief is calculated as the vertical difference in the elevation of the ridgeline and the channel at each transect measurement and I use the mean of the relief measurements as the valley relief for the zone. I also measure the ridgetop Laplacian asymmetry, defined as RLA,_1S 1 = log 2 V 2z 2 (3.2) " R,s# 79 Figure 3.3. Shaded relief map created from 30m NED data with sites of stream captures or valley beheadings that were identified in this study and in previous studies. Field sites where infiltration rate and soil shear strength measurements were made are also marked. Field Site 1 (FS-1) is located at 36.9180 N, 120.8270 W. Field Site 2 (FS-2) is located at 35.9150 N, 120.7670 W. 80 where V2 zR is the ridgetop Laplacian of the slope denoted by the second subscript (nfor sj). If a ridgeline has responded to base level conditions then RLAnjgcan be used as a proxy for erosion rate asymmetry. If the slopes on opposing sides of a valley are eroding at the same rate then RLAnf-f is 0. If the hillslope has completely responded to the base level lowering rate and RLA nfsf > 0, the south side of the valley (north-facing slope) is expected to be eroding faster than the north side of the valley (south-facing slope). In order to estimate V 2 zR,, I bin V 2 z values into 20 bins spaced logarithmically in A 0 3 5S, calculate the median in each bin, and use the most negative median as V2 zR (Figure 3.6). I exclude bins with less than 25 data points. I use the binned value with the most negative median for two reasons. First, the ridgelines of Gabilan Mesa may be still be experiencing a transient response to the initial incision of the original mesa surface and may have not fully responded to the base level lowering condition. The most negative bin should represent the portion of the hillslope that has responded the most, which is generally on the thinner interfluves or on the portion of the ridgeline furthest from the main divide but still dominated by soil creep. Second, if lateral channel migration is occurring at Gabilan Mesa then differences in V2 zR are expected to occur near the ridgeline and are unrelated to overland flow or channel incision on the hillslope (Chapter 2). I estimate V2 zR within each zone by splitting the valley into north-facing and southfacing regions and estimating V 2 ZR separately on the north-facing and south-facing ridgeline. It is important to isolate the ridgelines for this analysis, so I exclude the valley bottom by mapping channels with drainage area greater than 10,000 m 2 and also excluding pixels within a 50 m radius and that are within 5 m of the elevation of the 81 excluded channel pixels. Where significant alluvial deposits exist, I use a similar technique and exclude all of the sediment in the valley from the analysis. 3.2.3 Ridgetop Laplacian asymmetry I analyzed ridgetop Laplacian asymmetry in 9 zones at Gabilan Mesa where high-resolution topographic data derived from LiDAR (gridded to 1 m) are available (Figure 3.4 and Table 3.1). I chose to analyze zones within the Paso Robles Formation where north-facing and south-facing tributaries are approximately perpendicular to the main channels and span at least a few low-order tributaries. I also measured BSAfg within these zones. The magnitude ofV2 zR varies significantly at Gabilan Mesa, with some ridgelines appearing as if they have not responded fully to the base-lowering rates while other hillslopes appear to have responded. For example, for ridgelines within zone R6, the north-facing and south-facing ridgetops appear relatively gentle and wide and have a low magnitude of V 2 z R (-5.95 0.07x10-3 and -4.67 0.03x103, respectively). This hillslope appears as if it has only partially responded to the boundary lowering rates and still exhibits relict characteristics of its history as a mesa. For zone R2, the ridgelines have responded more significantly and the magnitude of V 2 zR is considerably higher on both the north-facing (-9.72 0.19x 10~3) and south-facing ridgetops (-13.02 0.03 x 10-3). I measured positive RLAgfgf in all analyzed zones except zone RI and R2, which exhibit negative RLAfjg(Table 3.1). Bulk slopes are steeper on south-facing slopes than north-facing slopes in zone RI, while zone R2 exhibits relatively low BSAfg in comparison to the other zones that I analyzed. 82 Figure 3.4. LiDAR-derived shaded relief map showing ridgetop Laplacian asymmetry for each analyzed zone. The site name is listed next to each zone. For reference, the downstream edge of R6 is located at 36.929' N, 120.808' W. Table 3.1. Estimates of the ridgetop Laplacian and bulk slope asymmetry. Uncertainty is reported as I standard error of the mean. Zone Ri 2 V zR,nf (X10- 3 i2 R,sf RLAnjgg BSAnyff -0.25 +/- 0.02 / 0.02 -0.51 +/- 0.02 / 0.02 0.35 +/- 0.01 /0.01 2.30 +/- 0.02 /0.02 1.20 +/- 0.02 /0.02 1.15 +/- 0.01 /0.01 0.56 +/- 0.01 /0.01 1.43 +/- 0.01 /0.01 0.86 +/- 0.01 /0.01 1.77 +/- 0.01 /0.01 -1) (X10-3 0.19 0.13 -8.11 -13.02 -7.68 0.09 0.03 1.41 -7.22 0.18 0.04 -13.29 -5.35 0.08 0.03 -0.42 +/- 0.01 /0.01 1.61 /-0.25 /0.31 0.11 /-0.01 /0.01 0.43 +/- 0.01 /0.01 R6 -5.95 0.07 R7 -11.10 0.09 R8 R9 -12.22 -18.15 0.04 0.11 -4.67 -10.58 -8.71 -10.48 0.03 0.04 0.06 0.03 0.35 0.07 0.49 0.79 -6.82 0.18 -9.72 -23.5 R4 R5 -14.39 R2 R3 83 +/- 0.02 +/- 0.01 +/- 0.01 +/- 0.01 /0.03 /0.01 /0.01 /0.01 3.3. Field measurements I measured infiltration rates and soil shear strength during a field campaign to Gabilan Mesa carried out from December 30, 2012 until January 9, 2013. I made measurements on north-facing and south-facing slopes at two field sites (Figure 3.3). Gabilan Mesa receives the majority of its annual precipitation in the winter months, but no significant storms had occurred yet during that winter. 3.3.1.1 Infiltration measurements To determine if runoff varies on north-facing and south-facing slopes at Gabilan Mesa, I estimated the field-saturated hydraulic conductivity, Ks, on north-facing and south-facing slopes at two field sites following established procedures [Reynolds and Elrick, 1990; Mertens et al., 2002]. Kfs describes the infiltration rate of water due to gravity under a zero-pressure head. Multiple ponding-depth measurements were made in order to independently solve for Kf, without needing to explicitly solve for the matrix flux potential, which describes the infiltration of water due to capillary forces [Reynolds and Elrick, 1990]. I measured the infiltration rates with a single-ring infiltrometer at two ponding depths (5 cm and 10 cm). Once I completed the 5 cm ponding depth measurements, I added water to the infiltrometer to achieve a ponding depth of 10 cm and additional measurements were made to calculate q,, the field-saturated infiltration rate, at the 10 cm ponding depth. At each ponding depth, I measured the length of time that was required for 200 ml of water to infiltrate into the soil repeatedly until infiltration rates were 84 approximately the same for multiple, contiguous measurements. For each site, the best-fit line to the cumulative infiltration rate for the quasi-steady infiltration measurements against time was chosen as q. From Reynolds and Elrick [1990], Kfs can be calculated as Kf = a H2 - H, (3.3) where G=0.316 ( +0.184 a (3.4) is an empirical, dimensionless parameter dependent on d, the depth of insertion of the infiltrometer, and a, the infiltrometer radius. G describes the 3-dimensional flow geometry beneath the infiltrometer. Q, (L 3/T) and Q2 (L 3/T) are the quasi-steady-state, saturated flow rates and H1 and H 2 are the depth of ponding from the two infiltration measurements [Reynolds and Elrick, 1990]. 3.3.1.2 Infiltration rate results My estimates of Ks at Gabilan Mesa range over two orders of magnitude with higher rates generally on north-facing slopes, especially at FS-2. I measured infiltration rates at 25 hillslope locations at FS-1. Thirteen of those locations were on the south-facing side of the valley and 12 were on the north-facing side of the valley. I did not measure a statistically significant difference in Ks on north-facing and south-facing sides of the valley at FS-1. Two of the south-facing measurements produced negative estimates of Ks and are excluded from the analysis. Negative estimates of Ks are nonphysical, but are not uncommon when estimating Ks in the field using two head heights 85 [Elrick and Reynolds, 1992], as I did. Phillips [1985] carried out two numerical experiments for unsaturated flow and showed that heterogeneities in soil conductivity with depth or uncertainty in the infiltration rates could lead to negative estimates of K. Furthermore, increased flow through macropores activated by the increased pressure at the higher head height may lead to negative estimates of Kfs [Wu et al., 1993; Mertens et al., 2002]. At FS-2, I measured infiltration rates at 20 sites. I made 11 measurements on the north-facing side of the valley and 9 measurements on the south-facing side of the valley. Kfs differs significantly on the north- and south-facing sides of the valley at FS-2. The mean Kfs on the south-facing side of the valley was 0.5 the north-facing side of the valley was 7.6 0.1 cm/hr while the mean Kfs on 2.1 cm/hr. Kfs can vary with hillslope gradient [Casanovaet al., 2000], so I regressed Ks against the local slope gradient (Figure 3.5). I measured the local gradient from National Elevation Dataset topography gridded to -10 m/pixel at each location where I estimated Ks. At both FS-I and FS-2, log-transformed Kfs on south- facing slopes exhibited moderate to low dependence on slope gradient (R2 = 0.14, 0.35 and p-value = 0.14, 0.09, respectively). At FS-1, the log-transformed K, on north-facing slopes exhibited moderate dependence on slope gradient (R2 = 0.32, p-value = 0.06) while the log-transformed Ks on the north-facing slopes at FS-2 did not exhibit dependence on slope gradient (R 2 = 0.005, p-value = 0.83). If the two outliers with low Kfs are excluded at FS-2, the north-facing log-transformed Ks at FS-2 does exhibit dependence on gradient (R 2 = 0.44, p-value = 0.05). Outliers are a common result from double-ponded infiltration measurements and may be caused by an increase of flow through macropores that are activated at deeper 86 ponding-depths due to expansion of the saturated bulb [Wu et al., 1993; Mertens et al., 2002]. * FS-1 n-facing o FS-1 s-facing AA & FS-2 n-facing 10- A FS-2 s-facing FS-2n-fin n OA 100 10- 0 0.1 0.2 0.3 0.4 Ivi 0.5 0.6 0.7 0.8 Figure 3.5. K against hillslope gradient. Best-fit lines for the log-transformed K, and hillslope gradients on the north-facing and south-facing hillslopes at the two field sites. 3.3.2.1 Soil shear strength measurements At each field site, I made multiple measurements of soil shear strength to determine if aspect-dependent differences exist in soil shear strength.. I measured shear strength of the soil surface with a shear vane (Humboldt H-4212MH) at the same locations where I estimated Kf at FS-1 and FS-2. Zimbone and coworkers [1996] and Leonard and Richard [2004] used a similar device to measure soil shear strength and found the measurements to be reasonable indicators of soil shear strength. At each site 87 where I measured the soil shear strength, I made 10 measurements and used the mean as the representative value for the site. Measurements were made on bare soil and vegetated soil. For the vegetated soil measurements, I clipped the vegetation down to the surface before making the measurement to limit the entanglement of the shear vane blades with above-surface vegetation. 3.3.2.2 Soil shear strength results At FS-1, the bare soil measurements on the south-facing slope were comparable to the bare soil measurements on the north-facing slopes (Table 3.2). This is also the case for the measurements made in vegetated soils. When the south-facing and north-facing slope measurements are considered together, the vegetated soil exhibited higher soil strength than the bare soil (p-value: 0.02). At FS-2, The differences between bare soil and vegetated soil were less pronounced and not statistically meaningful (p-value: 0.43). However, the vegetated soil strength measurements on the south-facing slope are significantly higher than the vegetated soil strength measurements on the north-facing slopes (p-value: 0.002). I also measured a significant difference in soil shear strength of the bare soils on north-facing and south-facing slopes (p-value: 0.05). I measured a small dependence on soil strength with slope gradient for the vegetated shear strength on north-facing slopes (R2 = 0.26, pvalue= 0.015), with slopes with higher gradients exhibiting slightly lower shear strengths. The south-facing soil shear strength exhibited weak dependence on slope (R 2 = 0.117, pvalue = 0.13). The north-facing and south-facing bare soil shear strength both exhibited 88 weak dependence on slope gradient (R2 = r-squared = 0.1, 0.09, and p-values = 0.17, 0.2, respectively). Table 3.2. Field measurements of field-saturated hydraulic conductivity and soil shear strength. Uncertainties are reported as 1 standard error of the mean. BSAnf-sf Ks (cm/hr) 2 Soil shear strength (kg/cm Solsertenh(k C2 Vegetated soil Bare soil ) Field site FS-1 south-facing north-facing 0.69 0.69 0.03 0.03 2.6 2.7 1.2 (11) 0.5 (12) 0.13 0.14 0.01 (12) 0.02 (11) 0.17 0.17 FS-2 south-facing north-facing 2.01 2.01 0.10 0.10 0.5 0.1 (9) 7.6 2.1 (11) 0.16 0.09 0.03 (9) 0.01 (9) 0.18 0.02 (9) 0.11 0.01 (11) 0.01 (12) 0.02 (12) 3.4 Cosmogenic radionuclide-derived erosion rates I determine cosmogenic radionuclide-derived erosion rates at Gabilan Mesa in order to test the predictions made in Chapter 2 by the landscape evolution models that incorporate different asymmetry-forming mechanisms. 3.4.1 Cosmogenic radionuclide methods Analyses of cosmogenic radionuclides (CRNs) have been used successfully at many sites around the world to determine long-term denudation rates for landscapes [Hancock et al., 1999; Balco and Stone, 2005; Ferrieret al., 2005]. I used CRNs to estimate erosion rates from (1) catchment-averaged sediment samples collected from small, fluvially deposited fans at the mouths of hollows [Grangerand Kirchner, 1996] [Bierman, 1996], and (2) from bedrock samples collected at or just below the soil89 bedrock contact [Heimsath et al., 1997; DiBiase et al., 2010]. The samples were processed and analyzed at the Purdue Rare Isotope Laboratory (PRIME Lab) following the methods developed at PRIME lab for '0 Be processing [Clifton et al., 2005]. I collected two samples in March 2010 and 23 samples in December 2013. For the two samples collected in 2010, I collected quartz pebbles from fluvial fans, which I crushed and processed at PRIME Lab for accelerator mass spectrometry (AMS) analysis. During the 2013 trip, I collected sediment from fans and hollows at 20 sites. Ten of those sites were from small south-facing catchments while the other 10 were from small northfacing catchments. I also collected bedrock samples at three of the sites-two on northfacing slopes and one from a south-facing slope. Pebbles and sediment less than 2 mm in diameter were crushed and recombined for AMS analysis. For the catchment-averaged samples, I used 10 m NED data to calculate a basinaveraged shielding value. For the depth samples, I calculated the shielding factor at the location where the sample was collected. I used the basin-averaged coordinates and elevation to calculate the soil production rate. For the depth samples, I used the coordinates and elevation to calculate the soil production rate and used the slope-normal depth of the sample to calculate the depth-corrected production rate. Due to the young age of the Paso Robles formation, it is necessary to account for the inherited 10 Be concentration, N,,h,, from previous exposure of the sediment. The concentration of 1 0Be in previously-buried sediment that was recently exposed at the surface along road cuts has been determined [Perron,2006] and is used to constrain N,. The total concentration is N,, = N,,, + N,.e where N,0, is the total 10Be concentration measured with an accelerator mass spectrometer and Nrec is the 90 10 Be concentration due to recent exposure of the rock to cosmic rays. I measured N, and used it to estimate N . I used the CRONUS calculator to solve for the long-term denudation rates [Balco et al., 2008]. 3.4.2 Erosion rate asymmetry metric I measured the asymmetry in erosion rates by comparing the paired erosion rates for low-order basins that directly oppose one another on north-facing and south-facing slopes. I define the erosion rate asymmetry as ERAf S = log2 (3.5) Where E is the erosion rate of the slope denoted by the subscript. I use ERAnf-fto describe the difference in the erosion rates at Gabilan Mesa and I test the values against the LEM predictions. 3.4.2 Erosion rate results Catchment-averaged bedrock erosion rates on south-facing slopes range from 46 5 m/Myrs to 87 12 m/Myrs while catchment-averaged bedrock erosion rates on north-facing slopes exhibit a wider range from 42 4 m/Myrs to 266 110 m/Myrs. I report the inheritance-corrected erosion rates and uncorrected erosion rates in Appendix 1. 91 a) 0 2.5 I- o 2 Field Evidence of LCM No field evidence of LCM -< 1.5 1 0.5 I 0 -0.5 F -1.5 ' -1 0.5 1 1.5 BSA/f 2 2.5 Figure 3.6. (a) Erosion rate asymmetry against bulk slope asymmetry for sites shown in (b) at Gabilan Mesa. Uncertainty bars show 1 standard error of the mean. (b) Hillshade created from -10 m NED data showing sites where samples were collected for CRN analysis. Sites 3 and 9 exhibit evidence of lateral channel migration (LCM). The black circles mark the sample locations. Topography along the A-A 'transect is analyzed in section 3.5.4. For reference, the end of the transect at A is located at 35.930' N, 120.7890 W. 92 Two samples that I collected at the bedrock-soil interface are from north-facing slopes (GAB13023 and GAB13024) and had low concentrations of 10Be. Their concentrations of 10Be are similar to a sample that was recently exposed along a road cut that has not experienced modern exposure and production of 10Be [Perronet al., 2012]. Applying an inheritance correction to GAB 13023 and GAB 13024 is difficult because a large uncertainty is introduced due to the low 10 Be concentrations of the samples and the high uncertainty in the inherited concentration that I estimate from the deeply buried samples. The other sample collected at the bedrock-soil interface (GAB 13022) was from a south-facing slope and the erosion rate at that location (53 7 m/Myrs) was comparable to the catchment-averaged erosion rate for the basin (63 6 m/Myrs). I exclude them from the analysis. In three of the valleys (sites 2, 8, 9 in Figure 3.6a), north-facing slopes erode faster than south-facing slopes. In two other valleys (sites 3, 5 in Figure 3.6a), the southfacing side of the valley erodes faster. For the other five erosion rate pairs, the erosion rates on the north-facing and south-facing sides are comparable. In Figure 3.6a, I plot erosion rate asymmetry against bulk slope asymmetry. There is no clear trend between them at Gabilan Mesa. Valleys with low BSAfg (~0.5 to 1) exhibit insignificant differences in erosion rates on north-facing and south-facing slopes. Valleys with higher BSAnfg1 exhibit a much larger range in erosion rate asymmetry. The two sites that exhibit physical evidence of lateral channel migration both have higher north-facing erosion rates relative to south-facing slopes. In addition, site 8 (Figure 3.6b) also has high erosion rate asymmetry, with the north-facing catchment eroding considerably faster (161 43 m/Myrs) than the south-facing catchment (55+6 m/Myrs). At 93 site 5, which has the highest BSAnf.f of the sites where I estimated erosion rates, the south-facing catchment erodes faster (77 11 m/Myrs) than the north-facing catchment (42 4 m/Myrs). 3.5. Landscape evolution model As in Chapter 2, I modified a simple LEM to include asymmetry-forming mechanisms. Here I briefly summarize the model framework from Chapter 2.2. The general form of the governing equation is az DV 2z+ E A"m jVz" s OC(36 - -- A' 1 Vz t at DV z K l +EK Vz"+ E A' Vz >6 2 2V Z (3.6) where z is elevation, D is soil transport efficiency, K is the fluvial incision coefficient [Perronet al., 2008], A is drainage area, 0, is the fluvial incision threshold, and E is the uplift rate or boundary lowering rate [Howard, 1994; Perronet al., 2008]. I assume that the channel incision rates scale with the bed shear stress and not the excess shear stress. The shear stress threshold (K6c) is often subtracted from the shear stress term (KA" Vz l), but there is a lack of field evidence in support of either approach for small drainages. By not subtracting shear stress, the calibration of the model with Oc > 0 is significantly simplified. I discuss the calibration of the fluvial incision threshold in section 2.2. 94 3.5.1 Aspect-dependent LEMs Following the same procedure as in section 2.2.2, I adapted the governing equation in order to model either differences in regolith strength or differences in runoff. Differences in vegetation and microclimates on south-facing and north-facing slopes at Gabilan Mesa may lead to differences in regolith strength. Increased vegetation density on the north-facing slopes may lead to an increase in regolith strength and make initial gullying more difficult relative to the south-facing hillslopes. The vegetation and microclimatic differences may also influence infiltration rates and runoff rates when precipitation exceeds the infiltration capacity. I incorporated a weighting function into the governing equation to account for aspect dependence of regolith strength and runoff. The weighting function is of the form (16(cos(#b) ((co 6 1+ CO= 1-61 o s( s(#) cos(ridge) where - - # is the angle between the surface 1 1 cos(#) cosr(,,dg,) (3.7) cos(#) <cos(n,,) g' normal and the sun, qSdge is the slope-normal vector of the ridgeline (always 900 from horizontal) and 6 is the magnitude of the weighting function. If 6 > 0, o is higher on the south-facing slope. If 6 < 0, w is higher on the north-facing slope. Increasing the magnitude of 6 causes larger differences in w on opposing slopes. I parameterize the sun altitude angle for Gabilan Mesa and use 700, 95 which is the angle that minimizes the variance between cos() and the annually averaged solar radiation. I included aspect-dependent regolith strength in the LEM by weighting 0, by w. This changes the value of A Vz required for incision to start on the landscape. I incorporated aspect-dependent runoff into the LEM by weighting A by w. This effectively changes the relationship between runoff and drainage area and mimics an aspect-dependent infiltration rate. Also, by weighting A instead of modifying K, I allow for non-local effects since A represents the upslope area that drains to a point on the landscape. 3.5.2 Lateral channel migration Following the same approach as in section 2.2.3, I model lateral channel migration by adding a linear advection term that shifts the landscape in the y-direction towards the positive y-boundary, which represents a bounding stream. This effectively mimics a migrating channel undercutting the base of a hillslope. The modified form of the governing equation is DV 2 z+E-y az A'" Vz az _ay at DV 2z - KA'" Vz" +E-y az A'" Vz" >0 ay where y (L T-) is the lateral channel migration rate. I used a fixed lateral channel migration rate for the modeling experiments. In natural landscapes, lateral channel 96 (3. (3.8) migration may be driven by initial tilt, differences in initial slope morphology, or microclimates. In this scenario, I assume that the necessary mechanisms for driving lateral channel migration are active, but I do not explicitly model them. 3.5.3 Model parameterization I calibrated the model from the same section of topography as Perron and coworkers [2009] and used their estimate of K and m. I assume n =1. I recalibrated D using a slightly modified technique and also estimate O. Previous numerical models Table 1. LEM model parameters (M2 yr-') 0.0138 K (MI-2, yr-1) m n E (in yr-) Oc 1.x10~ 4 0.35 1 1.47x10~4 1 (mi 7 ) D of Gabilan Mesa did not include a fluvial incision threshold [Perronet al., 2009; 2012]. I calibrated the model for a non-zero fluvial incision threshold. I did this for two reasons. First, it allows me to test if aspect-dependent differences in regolith strength, which can be described by introducing aspect dependence to O, can explain the asymmetry at Gabilan Mesa. Second, it allows me to better reproduce the Laplacian near the ridgetop. If Oc = 0, overland flow will influence the Laplacian near the ridgetop. Here, I impose a fluvial incision threshold so that overland flow does not contribute to erosion near the ridgeline, which is supported by topographic evidence [Perronet al., 2009]. Past efforts have generally involved estimating 0, from field measurement or by calculating best-fit values of 0, for detachment-limited stream incision and channel 97 profiles [Snyder et al., 2003]. These approaches work well for estimating bedrock river incision thresholds, but here I focus on estimating the threshold that influences the transition from a ridgetop dominated by soil creep to 1 s-order valleys that experience episodic incision and fill from colluvium. I estimate 0, by calibrating the other parameters in the LEM and then choosing a 0, that best reproduces the transition from a constant Laplacian at low A 0.35 VzI to a spatially variable Laplacian that is influenced by overland flow and channel incision (Figure 3.7). - 0.12 - 0.1 0.08 0.06 - 0.04 -0.02 - - 0.02 -0.04 102 100 10-' 035 A 1VZ 101 (M 0 7 ) 10-3 Figure 3.7. Changes in the Laplacian with increasing drainage area and slope. Grey points show LiDAR-derived data from Gabilan Mesa near site 10 in Figure 3.6b. For clarity, I only show a random subsample of 25% of the raw data. Green circles show the binned means of the logtransformed data. Crosses show the binned means of the log-transformed data for the calibrated model. 98 3.5.4 Consideration of tilted initial topography The background tilt at Gabilan Mesa influences the orientations of the major tributaries that drain into the Salinas River. The headwaters of the large drainages are located to the northeast with basin outlets to the southwest. Previous geomorphologists have excluded tilt as a possible explanation for the topographic asymmetry at Gabilan Mesa because the orientation of the asymmetry is generally perpendicular to the main drainages and therefore to the dominant direction of tilt [Reed, 1927; Kane, 1970; Dohrenwend, 1978]. Nonetheless, I investigated the role of tilt because a background slope is present for some of the asymmetric hillslopes at Gabilan Mesa. I measured a background slope from the northwest to the southeast of 0.60-0.65' across a series of highly asymmetric neighboring basins (Figure 3.6b and 3.8c). To explore the role that the background slope of the land surface may have on the development of topographic asymmetry, I ran a series of LEMs calibrated to Gabilan Mesa with a background slope that varied from 0.5' to 5' (Figure 3.8a) and analyzed the developing topography throughout the evolution of the model run. The models with the lowest background slopes developed the highest initial BSAnff. This somewhat counterintuitive result occurred because hillslopes with higher initial background slopes developed lower relief on the north-facing slope, which led to relatively low bulk slope gradient on the north-facing slope and therefore low bulk slope asymmetry. Figure 3.8b shows the evolution of a model landscape calibrated to the topography at Gabilan Mesa with the mean background slope and width (890 m) of the valleys in transect A-A ' (Figure 3.8c). Even though this tilting scenario is capable of 99 initially producing a high degree of topographic asymmetry, the asymmetry is not maintained throughout the model run. Initially, the north-facing slope erodes faster than the south-facing slope, causing the main divide to migrate southward until erosional equilibrium is reached. Once this occurs, typically in less than 1 Myr, the hillslope exhibits insignificant asymmetry. a) 2 1.5 Myrs I b) Initial tilt 0.50 10 20 30 40 50 3 0.8 2 0.6 0.4 0 0.5 0.2 0 0 I 0 1 0.1 1.5 2 0 BSA' Time (Myrs) c) Transect - 400 r 0 1 0.5 - - - - - - Background slope (valley) Background slope (ridgeline) 350 A' 300 0 1000 3000 2000 4000 5000 Distance along transect (m) Figure 3.8. (a) Evolution of topographic asymmetry (measured as bulk slope asymmetry) from 0.1 Myr to 1 Myr for hillslopes with different degrees of initial tilting. (b) Transient results for a model calibrated from Gabilan Mesa topography and an initial background tilt that is estimated from transect A-A'. The model topography is initially asymmetric, but decreases through the model progression. (c) Elevation profile across transect A-A' shown in Figure 3.5b. 100 3.5.5 Landscape evolution modeling of aspect-dependent process efficiency I use LEMs that include aspect-dependent runoff and aspect-dependent regolith strength to explore the transient development of topographic asymmetry. I use the modern erosion rates and relief measured at Gabilan Mesa to estimate a minimum amount of time that Gabilan Mesa has been incising and to begin my transient analysis. The mean bedrock erosion rate for Gabilan Mesa is 78 error of mean) and the mean relief is 80 12 m/Myr (mean standard 5 m (mean : standard error of mean) for the zones with erosion rate pairs. If the long-term mean erosion rate is less than twice the modern erosion rate, then it would require ~0.5 Myr to produce the mean modem relief at Gabilan Mesa. I analyzed the transient model results from 0.5 Myr to 3 Myr, which is the maximum time that is generally required for the modeled hillslopes to reach either steady-state topography in the case of the aspect-dependent LEMs or a steady form for the lateral channel migration LEM. I ran models with widths of 545 m, 1020 m, and 1860 m, which are the minimum, mean and maximum widths of the valleys in which the erosion rate pairs were estimated. I varied 6 in equation (3.7) to reproduce the range of topographic asymmetry that I measured at Gabilan Mesa. For the LEMs with aspect-dependent runoff and aspectdependent regolith strength, the south-facing slope erodes faster than the north-facing slope while the hillslope is responding transiently to the different erosional efficiencies. Eventually, the ridgeline is offset enough that the steeper north-facing slope is able to match the more efficiently eroding south-facing slope. The aspect-dependent regolith strength and aspect-dependent runoff LEM produce model results that generally cannot be distinguished by the relationship between ERAf.g and BSAfnff (Figure 3.8). 101 The aspect-dependent runoff LEM required large differences in the mean co on north-facing and south-facing slopes to produce the degree of asymmetry at Gabilan Mesa. A ratio of the mean co on north-facing to south-facing slopes of-30 was required to reproduce the mean BSAgfgf (1.6) for a hillslope with the mean width (1020 m) of the zones where erosion rates were estimated. A ratio of the mean (o on the north-facing to the south-facing slopes of -200 was required to reproduce the maximum amount of BSAgfgf (2.4) measured at Gabilan Mesa. According to the numerical modeling . .o 2.5 2 S a -U... 1.5 Ib OV UL DRO LEM A A DRS LEM CM LEM A G EM with ackground slope Hbilan Mesa (field ev idence of LCM) G abilan Mesa (no field ev idence of LCM) *U N 1 *U. 0.5 -." 0 *AA -0.5 A 0 *. 0i 0.5 A A A 00 A A 4.A 1 *AO 1.5 40 *A * -11 0 06 D AA 2 2.5 BSA Figure 3.8. Comparison of CRN-derived erosion rate asymmetry results with the results for the aspect-dependent runoff (ADRO) LEM, aspect-dependent regolith strength (ADRS) LEM, and the lateral channel migration (LCM) LEM, and the results from the modeling experiment with background slope. All models are calibrated for Gabilan Mesa. The orange background color shows the general region for the lateral channel migration LEM results. The blue background shows the general region for the combination of the aspect-dependent runoff and the aspect-dependent regolith strength LEM. 102 experiments in Chapter 2, the required difference to produce the same degree of BSAf1 gf should decrease slightly as hillslope width increases. The aspect-dependent regolith strength LEM required a considerably smaller difference in o on the north-facing and south-facing slopes to achieve the same degree of asymmetry. In order to produce the proper sign of asymmetry as Gabilan Mesa, soil strength must increase on north-facing slopes, leading to a decrease in fluvial incision, relative to the south-facing slopes. The aspect-dependent regolith strength LEM required a ratio of mean w on north-facing to south-facing slopes of -5. The aspect-dependent regolith strength LEM required a ratio of the mean o on north-facing to south-facing slope of -10 to reproduce the maximum degree of asymmetry measured at Gabilan Mesa. This suggests that the wider basins at Gabilan Mesa may require smaller differences in co on north-facing and south-facing slopes to produce the same degree of BSA, 1f-f relative to narrower basins. 3.5.6 Landscape evolution modeling of lateral channel migration For the lateral channel migration LEM, I varied the lateral channel migration rate to reproduce the range of topographic asymmetry that I measured at Gabilan Mesa. Similar to the aspect-dependent LEM experiments, I also analyzed the transient results and compared the asymmetry that develops between 0.5 Myr and 3 Myr. Both the transient and the steady-state results predict that north-facing slopes erode faster than south-facing slopes (Figure 3.9). Lateral channel migration rates of -600 m/Myr were necessary to produce BSAffgof 1.6 and lateral channel migration rates of -750 m/Myr were required to produce BSAfjgof 2.4. The rate of lateral channel migration required for 103 asymmetry to develop is strongly dependent on the Migration number (Chapter 2). If K is lower than the estimate made by Perron and coworkers [2012], a slower rate of lateral channel migration would be required to reproduce the same degree of asymmetry. 3.5.7 Comparison of LEM predictions and erosion rates The transient modeling results of a hillslope that formed on an initially tilted surface produced a similar trend between BSAnfjgand ERAngff as the steady-state results created from the lateral channel migration LEM (Figure 3.9). These trends differed significantly from the results of the aspect-dependent LEMs. The steady-state results created from the aspect-dependent LEMs produced no trend between BSAnfjgand ERAnf-f, and produced negative ERAnfjg while the models were transiently responding (Figure 3.8). Two of the ERAfpsgvalues from Gabilan Mesa overlap with the lateral channel migration LEM results within 1 standard error while not overlapping with the results from the aspect-dependent runoff or aspect-dependent regolith strength LEMs (Figure 3.9). Of the two, one of the basins (Site 2) is directly downstream from a southward river capture in Portuguese Valley. Six of the ERAnfsf values made at Gabilan Mesa overlap uniquely within 1 standard error with the results of the aspect-dependent runoff LEM and aspect-dependent regolith strength LEM (Figure 3.9). One of the ERAnggvalues (Site 9) does not overlap with either model and occurs in Powell Valley, which exhibits physical evidence of lateral channel migration. 104 3.6. Discussion Numerical modeling suggests that even a small degree of initial tilt can lead to the development of significant topographic asymmetry for young landscapes. The amount of time that Gabilan Mesa has been incising is not well constrained and I cannot completely rule out the initial influence of tilt on the development of topographic asymmetry, especially where a background tilt is present. Two of the most asymmetric zones that I analyzed (Sites 5 and 9, Fig. 3.6b) have the orientation expected if tilting is responsible for the asymmetry. In addition, a transect across several of the other highly asymmetric basins also exhibits a background slope (Figure 3.8). However, the erosion rate signature does not match the expected signature if tilt is solely responsible for the topographic asymmetry (Figure 3.9), making it unlikely that initial tilting of the landscape is the sole driver of the asymmetry. It is important to consider the transient development of topographic asymmetry in landscapes that have not reached a steady form because the topographic signatures, especially in erosion rate asymmetry, are different than the signatures that develop at steady state. The relationship between BSAgfgf and ERAfpj for the aspect-dependent runoff LEM and the aspect-dependent regolith strength LEM is consistent with 7 of the 10 measured sites at Gabilan Mesa (Figure 3.9), whereas the relationship between BSAgfg and ERAflf for the lateral channel migration model only matches 3 of the 10 values (Figure 3.9). Considering that the results of the aspect-dependent LEMs are a better match at most of the sites, I conclude that lateral channel migration is likely not the dominant mechanism causing the topographic asymmetry at Gabilan Mesa. 105 If a discrepancy in soil shear strength is responsible for the valley asymmetry at Gabilan Mesa, then soil shear strength should be higher on north-facing hillslopes. I find no difference in soil shear strength at Field Site 1; while I find that the south-facing slopes at Field Site 2 exhibit higher soil shear strength, not the north-facing slopes. These findings are not consistent with the hypothesis that soil shear strength controls the topographic asymmetry at Gabilan Mesa. Measurements of field-saturated hydraulic conductivity suggest that significant differences in infiltration rates exist in asymmetric basins. At Field Site 2, Kf, is -15 times higher on the north-facing slope in comparison to the south-facing slope. In order to reproduce the mean BSAnf-s that I measured at Gabilan Mesa for the sites where I made erosion rates, the aspect-dependent runoff model required that the drainage area be weighted so that w is -30 times higher on the south-facing slope relative to the northfacing slope. The differences in Kfs measured in the field cannot be directly compared with the differences in co required in the model because Ks describes the infiltration rate while the numerical model addresses runoff. Kfs of the north-facing slope is sufficiently high (7.6 2.1 cm/hr) that few storms may be intense enough to exceed the infiltration capacity and cause overland flow. K, of the south-facing slope is considerably lower (0.5 0.1 cm/hr), and therefore this slope may experience runoff much more frequently. This threshold effect associated with the magnitude of storms and the recurrence interval of erosive events could lead to an effective difference in runoff on north-facing and southfacing slopes that is much greater than expected if considering infiltration rates alone. This suite of observations suggests that aspect-dependent runoff production is the most likely mechanism that consistently contributes to the asymmetry at Gabilan Mesa. 106 However, the model requires extreme differences in runoff on north-facing and southfacing slopes to replicate the most asymmetric slopes seen at Gabilan Mesa, which suggests that aspect-dependent runoff is probably not exclusively responsible for all of the topographic asymmetry, especially in the most asymmetric basins. Evidence of lateral channel migration, the presence of a background slope, or both, is apparent at each of the most asymmetric sites at Gabilan Mesa, suggesting that the asymmetry may be polygenetic at some locations. Of the three sites where the aspect-dependent efficiency models are not able to match the relationship between BSAgfgf and ERA fs,, physical evidence of lateral channel migration exists at two. At the other site, the absence of physical evidence such as stream capture or the presence of beheaded valleys does not preclude the possibility that lateral channel migration is also happening at that location. Even though lateral channel migration is probably not the dominant cause of asymmetry at Gabilan Mesa, the occurrence of stream captures and beheaded valleys suggests that the drainage network at Gabilan Mesa is experiencing significant reorganization. Furthermore, spatial variability in erosion rates for nearby sites in Portuguese Valley suggests that lateral channel migration occurs episodically. Pulses of lateral channel migration may cause oversteepening to occur more rapidly than it occurs in the model and might require less overall undercutting than the long-term rate that I reported. I excluded aspect-dependent soil creep as a possible asymmetry-forming mechanism because predictions of the asymmetry signatures (Chapter 2) do not match well with the predictions at Gabilan Mesa. In order to match the BSAgfgf at Gabilan Mesa, 107 the south-facing slope must experience much higher soil creep rates than the north-facing slope. Drastically increasing soil creep rates on the south-facing slopes leads to complete infilling and eradication of valleys. At Gabilan Mesa, the south-facing slopes are deeply incised. Thus, the model predictions of topography for aspect-dependent soil creep do not match the topographic characteristics at Gabilan Mesa. Ridgetop Laplacians vary at Gabilan Mesa, and many of the valleys exhibit ridgetop Laplacian asymmetry on opposing sides of the valley. There are two possible factors that may contribute to the RLAgfgf that I measured. One possibility is that lateral channel migration is causing an increase in the erosion rate on the north-facing slope. Another possible explanation is that the RLAgfgf is an artifact of the transient topography. Considering the very low estimates of V 2 zR at some locations and the relict mesa surface visible in some parts of the landscape, I suggest that much of the asymmetry of ridgetop Laplacians at Gabilan Mesa is due to the transient response of the ridgeline and does not reflect lateral channel migration. There are multiple possible explanations for why the north-facing slopes have higher Kfs than south-facing slopes. In a compilation study, Ludwig and coworkers [2005] concluded that in semi-arid environments, vegetation patches often retain more water and have higher infiltration rates and higher biomass production, which may contribute to the higher estimates of Kfs on north-facing slopes. In addition to the direct microclimatic effects, steeper north-facing slopes, which likely experience higher sediment flux locally, may experience increased soil mixing and maintain more pore space than shallower slopes that experience lower sediment flux. Thus, infiltration rates could also be influenced by shallower soil depth on south-facing slopes. I measured shallower soils on 108 the south-facing slopes (Appendix 1), and Reed [1927] observed that bedrock outcrops generally occur on south-facing slopes rather than north-facing slopes, indicating that south-facing slopes generally have thinner soils. 3.7. Conclusions Due to the high degree of topographic asymmetry, Gabilan Mesa is an ideal field site to test which asymmetry-forming mechanisms are responsible for the development of topographic asymmetry in semi-arid environments. I tested numerous asymmetryforming mechanisms including aspect-dependent runoff, aspect-dependent regolith strength, and lateral channel migration. I also investigated the role that initial tiling of the mesa may have had on the development of the topographic asymmetry. I compared topographic and erosion rate signatures predicted from the different models with measurements from Gabilan Mesa. The aspect-dependent runoff and aspect-dependent regolith strength LEMs are best at reproducing the relationship between topographic asymmetry and erosion rate asymmetry at Gabilan Mesa. In order for aspect-dependent soil shear strength to explain the asymmetry at Gabilan, north-facing slopes should have higher soil shear strength. Field measurements of soil shear strength do not support higher soil shear strength on the north-facing slope and actually show that soil shear strength is higher on south-facing slopes at one of the two sites. Field measurements do show that large differences in field-saturated hydraulic conductivity exist at Gabilan and that field-saturated hydraulic conductivity is higher by a factor of ~15 on north-facing slopes. This is consistent with the expectation if aspectdependent runoff is responsible for the asymmetry at Gabilan Mesa. Evidence of stream 109 captures and beheaded valleys is likely to exist in valleys that experience lateral channel migration. I identified new locations of stream captures and beheaded valleys, but they are not present in most of the asymmetric valleys. The aspect-dependent model predicts that differences in aspect-dependent runoff must be very large if aspect-dependent runoff is solely responsible for the asymmetry at Gabilan Mesa. For the most asymmetric basins, additional asymmetry-forming mechanisms may also be present. Many of the most asymmetric basins at Gabilan Mesa have a background slope, suggesting that initial tilting of the mesa may have influenced the development of the drainage basins and caused high topographic asymmetry to develop without the aid of microclimates. Asymmetry due to tilting is short lived (< 1 Myr), but due to the poor age constraints on the history of Gabilan Mesa, it is not possible to completely rule out the influence of tilting in some valleys. When all of the evidence is taken together, it seems likely that multiple mechanisms are acting together to produce the high asymmetry witnessed in some of the basins at Gabilan Mesa, but that aspect-dependent runoff is dominantly responsible for the bulk of the asymmetry. This analysis suggests that lateral channel migration should be carefully reconsidered as a general asymmetry-forming mechanism in semi-arid environments. Acknowledgements I would like to thank David DeJong for help building the infiltrometer, Peter Polivka and Michael Sori for helping with field work, and Scott Miller for help in the field and insightful discussions about topographic asymmetry. I would also like to 110 acknowledge support from the US National Science Foundation Geomorphology and Land Use Dynamics program through award EAR-0951672 to Taylor Perron. 111 112 Chapter 4. The influence of climate on hillslope sediment transport efficiency 113 Abstract Hillslopes compose the majority of Earth's land surface and are responsible for the bulk of sediment produced and delivered to the oceans, lakes and seas. The efficiency at which material can be transported from a hillslope to a channel plays a major role in the sediment budget of rivers. Therefore, understanding hillslope soil transport efficiency is important for both reconstructing past sediment budgets and also for understanding how sediment flux rates may change under different climate conditions. Estimates of soil transport efficiency, which is often described by a soil transport coefficient, or diffusivity (D), have been made for many sites globally. I compile previous estimates of D and also make new estimates at sites where erosion rates have been measured and high-resolution topographic data are available. Estimates of D vary over 3 orders of magnitude. For the logarithmically-transformed data, D exhibits a power-law relationship with a slope less than one with mean annual precipitation and the aridity index, which describes the moisture available to plants. I compare how D varies with the type of vegetation present at the site and find that, not surprisingly, D is lowest for arid sites. However, for sites that exhibit low mean annual precipitation (5 to 25 cm), sites described as desert have significantly lower estimates of D (17 5 cm 2/yr) than sites described as grasslands/scrublands (52 5 cm 2 /yr) or as savannah/lightly forested (70 20 cm 2 /yr). For moderate to high values of mean annual precipitation (50 cm to 150 cm), there is no difference in D for sites categorized as either grasslands/scrublands, savannah/lightly forested, or forested. I also find differences in D for sites with different lithology and where different techniques were used to estimate D. Estimates of D made in unconsolidated sediments are higher than estimates made for igneous/metamorphic rocks. This may be due to a bias introduced by the technique that was used to estimate D. Of my compiled estimates, 23 of the 33 estimates of D made in unconsolidated sediments are based on models of scarp diffusion. I find that estimates of D made with the scarp modeling technique are generally lower than the estimates made with other techniques. Even with the confounding influences of lithology and measurement technique, the compilation reveals an overall trend in which D increases rapidly with increasing moisture among relatively dry sites and less rapidly with increasing moisture at relatively wet sites. This trend suggests that the establishment of life in a landscape substantially accelerates soil creep, whereas differences in biological communities associated with different degrees of moisture have a relatively small effect on creep. 114 4.1. Introduction 4.1.1 Motivation Hillslopes are responsible for producing the vast majority of sediment that is transported by rivers to the ocean. The rate at which sediment is fluxed into streams and rivers plays a pivotal role in influencing a myriad of ecological conditions from the quality of salmon spawning streams [Platts et al., 1989] to the health of marine estuaries [Wolanski et al., 2004]. Fernandes and Dietrich [1997] suggested that the long response time of hillslopes makes it unlikely that hillslopes have reached a spatial or temporal erosional equilibrium during the Quaternary due to fluctuations in climate between glacial and interglacial periods. Therefore, it is likely that hillslopes have constantly modified their form during this period and have experienced different sediment flux rates under different climate conditions. Numerous studies point towards climate and vegetation influencing sediment transport rates [Fernandesand Dietrich, 1997; Roering, 2004; Hughes et al., 2009; Hurst et al., 2013a; McGuire et al., 2014], but the relationship is not well understood. Hillslope sediment flux rates, which are dominantly controlled by hillslope gradient and disturbance mechanisms that influence sediment transport efficiency, are challenging to estimate, and estimates generally require site-specific information such as topographic data in conjunction with field measurements [e.g., Almond et al., 2008; Jungers et al., 2009], estimates of erosion rates [Perronet al., 2012; Hurst et al., 2013a], or knowledge of how and when a landform developed [e.g., Colman and Watson, 1983; McGuire et al., 2014]. 115 Culling [1963] developed a mathematical framework to describe how sediment flux rates relate to landscape topography and influence the form of hillslopes. Geomorphologists subsequently developed techniques for estimating sediment transport rates from landscape characteristics and estimates of soil transport efficiency that could be applied to specific regions [Nash, 1980a; 1980b; Colman and Watson, 1983]. Nash [1984] suggested that if hillslope characteristics were known, such as the topography, local climate, hillslope aspect, and the properties of the material being transported, then sediment transport efficiency, and by extension sediment transport rates, could be accurately estimated for a site without the need of any additional measurements. Estimating sediment transport efficiency has proven more difficult. Numerous studies have been carried out to estimate hillslope sediment transport efficiency since that prediction, but a simple relationship between hillslope sediment transport efficiency and hillslope characteristics has not been established. Instead, studies have suggested more complicated relationships between hillslope sediment transport efficiency, climate, erosion rates and landscape characteristics [e.g, Hughes et al., 2009; Hurst et al., 2013a] 4.1.2 Background Hillslope sediment transport efficiency is influenced by myriad factors including the occurrences of freeze thaw cycles [Anderson et al., 2012], burrowing of mammals [Thomas and Montgomery, 1991; Gabet, 2000; Yoo et al., 2005], tree throw [Roering et al., 2010], and fire frequency [Pierce et al., 2004; Roering and Gerber, 2005]. Soil transport efficiency is often described by a soil transport coefficient (D), a diffusivity-like 116 parameter relating hillslope sediment flux (L 2 T-1) with slope gradient [Culling, 1963], so that q, = -DVz (4.1) where q, is the sediment flux rate and z is elevation of the land surface. When incorporated into a conservation of mass framework, a governing equation for the evolution of hillslope elevation can be derived of the form p, az-DV2Z p, at (4.2) Where t is time, U is bedrock uplift rate, pr is bedrock density, and ps is soil density. If the landscape is in topographic steady state, then D= -P -- (4.3) p,. V2Z There is some empirical support for equation (4.1) [McKean et al., 1993], equation (4.2) and equation (4.3) [Perron et al., 2012; Hurst et al., 2013b]. Modification of equation (4.1) to include soil depth dependence [Heimsath et al., 2005; West et al., 2014] or nonlinear dependence on slope gradient [Roering et al., 1999] also has empirical and theoretical [Furbishet al., 2009] support for some locations. 117 Geomorphologists have noticed that D increases as climate becomes less arid [e.g., Hanks et al., 1984b; Fernandes and Dietrich, 1997; Hurst et al., 2013a]. Hanks [1984] observed that sites with low precipitation, such as shorelines of Lake Bonneville, UT, have low estimates of D while sites with moderate to high precipitation exhibit higher values. Hanks also noted that the estimated value of D for a poorly-consolidated, wave-cut bluff in Emmet County, MI is similar to the value estimated for the Santa Cruz sea cliffs, CA and the Raymond fault [Nash, 1980] in Pasadena, CA even though the bluff in Emmet County receives higher precipitation. Hanks [1984] suggested that the higher vegetation cover may offset the effects of high precipitation. Hughes and coworkers [2009] found that soil transport efficiency at a site with gentle gradients (<30%) in the Charwell Basin, New Zealand likely increased from conditions during the Pleistocene as the landscape transitioned from a scrubland/grassland to a forest during the Holocene. Even though there is some evidence that soil transport efficiency increases as landscapes become less arid, other studies suggest that the relationship between climate and soil transport efficiency may be more complicated due to differences in erosion rate, vegetation, soil depth, temperature and precipitation rates among landscapes. A global compilation that spans a wide range of climates and includes enough sites to discern any trends that may exist despite the multiple influences on D is needed to address how soil transport efficiency varies globally. Estimates of D have been made at many sites, but limited effort has been made to compile the data or determine the trends between D and climate. I carry out such an analysis below. 118 4.1.3 Approach I compiled existing estimates of soil transport efficiency (D) and made new estimates for sites where both high-resolution topographic data (from LiDAR or differential GPS) and erosion rate estimates are available. I then compared the estimates of D against climate proxies, including mean annual precipitation (MAP), an aridity index (Al), and a measure of seasonality. I then investigated the role of the underlying lithology and the measurement technique used to estimate D to determine if they influence D. 4.2. Techniques for estimating D Numerous techniques have been developed to estimate D. I present a short summary of the techniques used to estimate D in the data compilation. 4.2.1 Scarp modeling The first estimates of D were made by modeling the evolution of fault scarps and paleo-shorelines of known ages [Nash, 1980b; Colman and Watson, 1983; Hanks et al., 1984a]. Multiple scarp modeling techniques have been developed [Colman and Watson, 1983; Hanks andAndrews, 1989; Avouac et al., 1993] and produce differing results [Pelletieret al., 2006] depending on the height of the scarp, assumptions about the initial geometry, and whether linear or nonlinear flux laws are used to estimate D [Pelletieret al., 2006]. The simplest solution for the evolution of a fault scarp that forms instantaneously and then evolves gradually due to creep is 119 z(x,t) = a*erf x) +bx (4.4) 2 D where erf(x, t) is the error function, a is half the initial vertical difference in elevation along the scarp, b is the is the pre-existing slope, and x is the distance from the center elevation of the scarp. The function is often evaluated at x=0 and is where the scarp is predicted to experience the highest slope gradient [Hanks, 2000]. More sophisticated numerical approaches have been developed that allow the entire profile of the scarp to be analyzed [Avouac, 1993; Arrowsmith et al., 1998]. Pelletier [2006] found that methods that incorporate the entire profile of the scarp in addition to uncertainty in the initial scarp angle yield the most accurate results. 4.2.2. Laplacian and erosion rate Roering [2002] estimated D for a transient hillslope profile along the Charwell River on the South Island, New Zealand using the hillslope Laplacian and estimated erosion rates along the profile. Geomorphologists [Roering et al., 2007; Perron et al., 2009; Hurst et al., 2012] have used the ridgetop Laplacian and catchment-averaged erosion rates to estimate D in conjunction with equation (4.3) so that D= P U pr V2 ZR 120 (4.5) where V 2 zR is the Laplacian at the ridgeline. An important assumption required for this analysis is that the ridgeline is eroding in steady state, such that the uplift rate U equals the measured erosion rate. However, due to the long response time required for hillslopes to reach steady state and variability in climate through the Quaternary, this assumption is rarely perfectly met [Fernandesand Dietrich, 1997]. Hillslopes are typically the last part of a landscape to respond to changes in channel incision rates or regional tectonics [Furbishand Fagherazzi, 2001]. Nonetheless, evidence exists that ridgetop Laplacians do record changes in channel incision rates, albeit with a delay [Hurst et al., 2013b]. 4.2.3 Relief and erosion rate In addition to the ridgetop Laplacian and erosion rate technique, another relationship has been derived that relates D, topographic characteristics, and erosion rate. Roering and coworkers [2007] derived an analytical solution relating dimensionless relief (R*) and dimensionless erosion rate (E*): R* (1+(E* ln I1+ 1+(E*) 1 (4.6) where R*=E*/4, E* = (-2V2z R LH) / S, , La is the mean hillslope length, and S, is the critical hillslope angle at which downslope sediment fluxes become infinite. Callaghan [2012] used equation (4.5) to modify E*, yielding 121 E 2E(p' p,)LH DS, (47) where E is the erosion rate and can be solved for with cosmogenic radionuclide (CRN) analysis. Callaghan [2012] combined equation (4.6) and equation (4.7) to solve for D for a series of sites along a strong climate gradient along the Chilean coast. 4.2.4 Colluvial flux and slope Hughes and coworkers [2009], in a similar fashion to Reneau and coworkers [1989], estimated the mass of dated colluvium in hollows and used colluvial infilling rates to estimate D. Gabet [2000] estimated D by measuring the sediment flux from ground squirrels at a field site near Santa Barbara, CA and by using sediment traps [Gabet, 2003]. Others [McKean et al., 1993; West et al., 2014] have used meteoric 1Be to determine sediment flux rates in conjunction with slope gradients and equation (4.1) to solve for D. 4.2.5 Landscape evolution modeling Others have estimated D using landscape evolution models (LEMs) and generally utilize error-minimization techniques to tune D so that other characteristics of the landscape are reproduced from a LEM [Petit et al., 2009; Pelletieret al., 2011; McGuire et al., 2014]. Roering and coworkers [1999] estimated D for a field site in the Oregon Coast Range by picking a value of D that minimized the error between predicted erosion rates using a nonlinear flux law and a long-term erosion rate determined by CRNs. 122 4.3. Data Compilation 4.3.1 Compilation of values from the literature Multiple compilations of D have been made by others [e.g., Fernandesand Dietrich, 1997; Hanks, 2000; Hurst et al., 2013a]. I include these estimates and compile additional estimates of D that exist in the literature (Appendix 2). I include all estimates that I found where D was estimated for natural landscapes under modern climatic conditions. If multiple estimates of D were made at the same site during different studies, I include all of the estimates [e.g., Roering et al., 2002; Almond et al., 2008; Hughes et al., 2009]. I exclude an estimate of D made at Gabilan Mesa, CA [Roering et al., 2007] because a newer estimate of D has been made using a better-constrained erosion rate [Perron et al., 2012]. In a few of the studies included in the compilation, the data required to estimate D were reported in the literature, but D was not estimated. In these cases, I use the reported data to estimate D and include it in the compilation. For example, Reneau and Dietrich [1991] estimated colluvial transport rates in the Southern Oregon Coast Range and reported the slope gradient at most of the sites. I used their reported values in conjunction with equation (4.1) to estimate D. 4.3.2 New estimates of D I made nine new estimates of D by using existing high-resolution topographic data and published erosion rates and by solving for D using equation (4.3) (Appendix 3). Seven of the ridgetop Laplacian estimates were made from topographic data created from LiDAR that is publicly available through OpenTopography [Krishnanet al., 2011]. I 123 made the other two ridgetop Laplacian estimates from differential GPS surveys of hillslopes in the Atacama Desert, Chile, provided by Justine Owen [Owen et al., 20111. 150 0 W 120 0 W 90 0 W 60 0 W 30 W 0 30 E 60 0 E 90 0 E 120 E 150 0 E 600 N 3 0 N 00 30 S 600S Figure 4.1. Map showing site locations where estimates of D have been made. Grey circles show locations of previous estimates of D. X's show locations for new estimates of D made in this study. At each site, I measured the ridgetop Laplacian from soil-mantled portions of the landscape upslope from where the erosion rate estimates were made following a technique similar to Perron and coworkers [2012]. I calculated the Laplacian by fitting a center-weighted, second-order polynomial to elevation data within a 15-by-15 meter window and summing the second derivatives in the x- and y-directions. To determine a single value that is representative of the ridgetop, I plotted the Laplacian against the areaslope product and binned the data into 20 logarithmically spaced bins. Starting from the bin with the lowest area-slope product, I include all additional neighboring bins until the magnitude of the binned mean decreases. If more than one erosion rate estimate exists at a site in a suitable location to estimate D, I estimated D for each erosion rate and assigned the mean of these estimates 124 of D as the site D. I estimated the uncertainty in D as either the standard error of the mean of D or the sum in quadrature of the standard errors of individual estimates, whichever is greater. I report the ridgetop Laplacian of each site as the mean of the unique estimates of the ridgetop Laplacian used to calculate D for the site. If there is no published estimate of Pr or ps at the site, I use the commonly-invoked density ratio of pr/ps = 2 [Heimsath et al., 1999; DiBiase et al., 2010; Hurst et al., 2013b]. A few of my new estimates of D differ from previously published values for the same sites. I briefly comment on these discrepancies here. My estimate of D = 147 + 33 cm 2 /yr for the Oregon Coast Ranges is higher than a previously published estimate of 36 16 cm 2 /yr, the estimate made by Roering and coworkers [1999]. The erosion rates [Bierman et al., 2001] that I used to calculate D are slightly higher than the erosion rates Roering and coworkers [1999] use to estimate D. The ridgetop Laplacian where I estimate D is considerably less negative than the ridgetop Laplacian in the catchment where Roering and coworkers [19991 estimated D. I use catchment-averaged erosion rates made by Bierman and coworkers [2001] in the Oregon Coast Range, but they also estimated one hillslope erosion rate of 81 24 m/Myr, which was measured from a sample collected at the base of a 300 m hillslope. If the hillslope erosion rate is used to estimate D instead of the catchment-averaged rate, D is ~80 cm2 /yr. This value is in better agreement with Roering and coworker's [1999] estimate of D, but is still higher by a factor of~2. I used colluvial flux rates and hillslope gradients reported by Reneau and Dietrich [1991] to make an additional estimate of D (51 13 cm 2 /yr) for the Oregon Coast Ranges. This estimate is comparable to the estimate made by Roering and coworkers 125 [1999]. However, Reneau and Dietrich [1991] may have underestimated the sediment mass flux from the hillslopes. They estimated sediment transport rates from the mass of colluvial fill in dated deposits. If some of the hillslope sediment was fluxed through the hollows instead of being stored, the sediment flux rates would be underestimated, which would also lead to an underestimation of D. Reneau and Dietrich [1991] estimated an erosion rate of 70 m/Myr from the mass of colluvial infilling. Bierman [2001] estimated CRN-derived erosion rates of 136 43 m/Myr for the same region using catchment- averaged erosion rates. This suggests that Reneau and Dietrich [1991] may have underestimated the long-term erosion rate for the Oregon Coast Range, which would lead to an underestimate of D, or that the erosion rates derived from CRNs are biased high. Jungers and coworkers [2009] included the necessary data to estimate D for a site in the Great Smoky Mountains, NC. I used the mean slope that they reported (-14 degrees) and the estimated sediment flux rate (65-100 cm 2/yr) to estimate D for their site (331 cm 2 /yr). This is considerably higher than my estimate of D (19 1 cm 2 /yr) made nearby using the ridgetop Laplacian and erosion rate technique. Their estimate assumes plug flow of the 60 cm thick active layer of soil and their estimated soil velocities of 1.1 to 1.7 cm/yr from meteoric and in situ 10Be data. If soil velocity exhibits a linear dependence on depth instead of being constant with depth, then the real sediment flux rates may be considerably lower than their estimate. This would cause a decrease in the estimate of D. Petit and coworkers [2009] used a numerical model to estimate D for the Wasatch Range, UT, and estimated an unprecedentedly high D (1200 cm 2 /yr). Mattson and Bruhn [2001], using a scarp modeling technique, estimated a much lower value of D for the 126 same region of 28 11 cm 2/yr; while I estimated a value of D of 83 15 cm 2 /yr. Lacking any reasonable explanation for the discrepancy, the large range in estimates of D made at the Wasatch Range suggests that one of the estimates is likely incorrect. When considering that Matson and Bruhn [2001] used a better-established technique and that the value estimated by Petit and coworkers [2009] is anomalously high in comparison to other estimates of D in the literature, it seems most likely that the high estimate of D made by Petit and coworkers [2009] is inaccurate. 4.4. Relationship between D and climate proxies I compare D with three different proxies for climate in order to determine which proxy best describes the variability in D. I compare D to mean annual precipitation (MAP), CGIAR-CSI Global-Aridity Index (Al) [Zomer et al., 2008], and seasonality. MAP is calculated from global precipitation data from 1950-2000 and is gridded to 30 arc-seconds (~A kM 2 ) [Humans et al., 2005] and is also used in the Al calculation. Al is defined as MAP divided by mean annual potential evapotranspiration (PET) and is useful as a proxy for the water available to vegetation. PET describes the ability of the atmosphere to remove water through evapotranspiration under idealized assumptions [Allen et al., 1998]. Al is confusingly defined such that landscapes with low Al have high aridity and landscapes with high Al are less arid. I define seasonality as the difference between the maximum and minimum monthly precipitation normalized by the mean monthly precipitation. The monthly precipitation data is derived from the same dataset used to calculate MAP [Hijmans et al., 2005]. 127 Previously published This study 102 ,! Figure 4.2. (a) Plot of D against mean annual precipitation (MAP) with least-square regression line fit to log-transformed data. (b) Plot of D against Aridity Index (AI) with least-square regression line fit to logtransformed data. (c) Plot of D against seasonality where seasonality is defined as the difference in precipitation between the wettest and driest month divided by the mean monthly precipitation. . a) 10 3 0 00 10' 10"0 R2 =0.33 0 10" 10 MAP (cm/yr) 10-1 b) 102 13 102 ) 0 . 0.. 10 10- . R2 = 0.35 0 10-1 10 0_3 AI c) 0 102 10 10. 0 0.5 1 1.5 2.5 2 Seasonality 3 3.5 4 4.5 I found a positive correlation between D and both MAP and Al (Figure 4.2). I did not find a correlation between D and seasonality (p-value of linear regression = 0.17). Least-squares linear regression of log-transformed values of D against log-transformed 2 MAP and Al reveals that Al is a slightly better predictor of D than MAP (R 128 = 0.35 and R2= 0.33, respectively). Both MAP and Al exhibit a power-law scaling relationship with D, with D = 0.14*MAP 58 and D = 1.86*AIo 55 where D is in cm 2 /yr, MAP is in cm/yr and Al is a dimensionless quantity. 4.5. Effect of vegetation I split the estimates of D in the compilation into 4 different vegetation categories: 1) arid/desert, 2) grassland/scrubland, 3) savannah/lightly forested, and 4) forested according to the site descriptions included in the original publications. If modern land use has changed the vegetation cover present at a site, I categorize the site based on the vegetation that existed for the majority of the time interval that the estimate of D reflects. For example, if the land where the measurement was made was recently cleared, but had been previously forested and the technique used to estimate D required CRN-derived erosion rates (> 1 kyr timescale), I categorize the site as forested instead of grassland. If a description of the vegetation was not included, I assigned the category by inspecting available photographs and satellite imagery of the site. For sites that experience MAP between 5 cm and 25 cm, D is lower for the arid/desert sites (17 5 cm 2 /yr, mean standard error) than for the grassland/scrubland category (52 5 cm2 /yr), which in turn is lower than D for the savannah/light forested category (70 20 cm 2/yr) (Figure 3). However, the only statistically significant difference (p = 0.02) is between the means for arid/desert sites and savannah/lightly forested sites for the 5 cm to 25 m range in MAP. For the 50-150 cm range of MAP, the forested sites exhibit a higher mean value of D (101 26 cm 2 /yr) relative to the savannah/lightly forested category (72 13 cm 2 /yr) 129 19 cm 2 /yr). At the 95% significance level, no difference and the scrubland/grassland (83 exists in D between the different vegetation categories for sites with MAP between 50 and 150 cm. In the savannah/lightly forested category, I excluded an outlier of D (1200 cm 2/yr) made by Petit and coworkers [2009] in the Wasatch Mountains. However, even if the outlier is included, a significant difference at the 95% significance level still does not exist between the categories. o o Desert / arid Grassland / scrubl and Savannah / Lightl y for ested Forested to 0 C 102 'vs 0 E 10' 0 100 10- 0 c 0 100 101 102 MAP (cm/yr) Figure 4.3. Plot of D against MAP separated into four different vegetation categories. The light grey patch highlights the MAP zone between 5 cm and 25 cm. The dark grey patch highlights the MAP zone between 50 cm and 150 cm. For the savannah/lightly forested category, estimates of D did not vary significantly at the 95% significance level for an increase in MAP from the 5-25 cm range (70 :20 cm 2/yr) and the 50-150 cm range (72 13 cm 2/yr). D increased by a factor of ~2 between estimates from the grasslands/scrublands category for MAP of 5-25 cm 130 and from the forested category for MAP of 50-150 cm. However, I did not find a difference between the two categories at the 95% significance level. This suggests that once landscapes become moderately vegetated, D may not increase significantly even for landscapes with higher precipitation or increased vegetation cover. 4.6. Effect of lithology D varies over three orders of magnitude even for comparable values of Al or MAP (Figures 4.2 and 4.3). In order to hopefully explain some of the variance in D, I explored the relationship between D and lithology. To complete this analysis, I split 0 10 Igneous / metamorphic 0 - Sedimentary 40 Un consolidated 102 2 10' 0 100 400 10- 3 10~- 10~ 10-2 100 Al Figure 4.4. Plot of D against Al separated into different lithological categories. The grey line is the best fit to the log-transformed data for the igneous/metamorphic category. The black line is the best fit to the log-transformed data for the unconsolidated sediment category. R2 values are listed next to the respective regression line. 131 the estimates of D into three different categories: 1) Unconsolidated sediments, 2) Sedimentary rocks, and 3) igneous and metamorphic rocks. I completed regression analysis of the log-transformed data for each lithology category. The estimates of D for sedimentary rocks exist over a limited range of Al, which makes them poorly suited for the regression analysis. I focus on the results for the unconsolidated sediments and the igneous/metamorphic category. The estimates of D in unconsolidated material generally exhibit lower estimates of D when compared to estimates of D where the underlying material is igneous or metamorphic rock (Figure 4.4). 4.7. Effect of measurement technique To determine if the method used to estimate D biases the estimate of D, I split the estimates of D into the five different categories summarized in section 2. I then performed least-squares linear regression on the log-transformed data and compared the different method categories (Figure 4.5). Estimates of D made from the LEM or colluvial flux category (Figure 4.5d) exhibit a limited range in Al and were excluded from the regression analysis. The scarp modeling method category produced the lowest estimates of D for a particular value of Al (Figure 4.5a). The relief and erosion rate method category produced the highest estimate of D (Figure 4.5c) while the Laplacian and erosion rate method category produced an intermediate estimate (Figure 4.5b). 132 a) 10a o b) to, Scarp modeling * Laplacian and erosion rate * 0 102 0 @0 10 10 1 10D 100 0.32 = c) 1 t10-3 - to, L 10- R2 =0.49 * R2 103 10' 10-2 100 d) 103 0 Relief and erosion rate 102 A LEM A Colluvial flux oQ 10 2 A 101 0 0 A 102 A (p 100 10' E ~tkA 10 100 10" 0.47 R2 = 10 0 10' 10' 10 10 0 - 10' 10 3 0I 102 Al Al e) 10 3 0 Scarp modeling * Laplacian and erosion rate Relief and erosion rate o 10 E 0 S0" 10 102 10 - 10-1 100 Al Figure 4.5. Plots of D against Al for D estimated with (a) the scarp modeling technique, (b) the Laplacian and erosion rate technique, (c) Relief and erosion rate technique, (d) LEM and colluvial flux techniques. The best-fit regression line and the R2 values of the log-transformed data are included in (a)-(c). (e) Plot of D against Al for the three measurement techniques with suitable ranges of Al for comparison. Best-fit regression lines for the scarp modeling technique (black line), Laplacian and erosion rate technique (dark grey line), and the relief and erosion rate technique (light grey line) are also shown. 133 4.8. Discussion and conclusions 4.8.1 Influences on D My compilation shows that D increases with mean annual precipitation and aridity index as a power between 0.5 and 0.6, which effectively means that the magnitude of D levels off (within the range of variability) in wetter landscapes. This sub-linear relationship may indicate that, despite transitions in the dominant creep mechanisms as landscapes become more heavily vegetated, the new transport mechanisms that are activated may not significantly increase the sediment transport efficiency. For example, burrowing mammals may dominate sediment transport in a grassland landscape [Thomas and Montgomery, 1991]. If climate conditions change and the landscape becomes forested, the frequency of animal burrowing may decrease while the occurrence of tree throw increases [Gabet and Mudd, 2010], but this transition may be a tradeoff instead of increasing sediment transport efficiency. The more substantial increase in D with precipitation and aridity index among dry landscapes may reflect a more impactful transition from abiotic creep to biotically mediated transport mechanisms that more efficiently transport sediment. D increases by an average factor of ~4 between arid landscapes and savannah/lightly forested landscapes for sites with low precipitation (MAP of 5-25 cm). As precipitation increases from sites with low precipitation to sites with high precipitation, the differences in D become smaller for the same change in mean annual precipitation. D increases by a factor of -2 between grasslands/scrublands that experience low precipitation (MAP of 5-25 cm) and forested terrain experiencing higher precipitation (MAP of 50-150 cm). However, there is significant variability in D for both of these 134 categories and even though D increases, there is not a significant difference between the two categories. In Charwell Basin, New Zealand, Hughes and coworkers [2002] found a similar result that D increased by a factor of ~2 as forest colonization occurred during the Pleistocene-Holocene transition. This result, like mine, counters the idea that flux rates decrease as landscapes become more forested and experience soil stabilization due to increased soil cohesion from roots. However, my results also show that the average increase in D from grasslands to forests is rather small when considering the three-orderof-magnitude range in D reported in the literature. Multiple factors may influence the generally higher estimates of D in igneous/metamorphic rocks relative to unconsolidated sediments. First, it is possible that the apparent correlation of D with lithology may be an artifact, because lithology in our compilation is correlated with the measurement technique used to estimate D. Of the 25 estimates of D made with the scarp modeling technique, 23 are in unconsolidated sediments. The other 2 are in poorly consolidated sediments. In stark contrast, estimates of D made with the relief and erosion rate technique and the Laplacian and erosion rate technique were primarily made in granitic landscapes. One reason for this is that these techniques use CRN-derived erosion rates to calculate D, which generally require the presence of 2 6Al or 10Be found in quartz. A bias from the measurement technique may be introduced due to the younger and less well-developed soils that may be present on the scarps relative to possibly better developed soils on older hillslopes in the same region. Another reason might be that scarps are often composed of coarse sediment such as alluvial fans or pluvial shorelines. The coarse sediment may inhibit vegetation growth, especially where the soils are young or poorly developed. For wave-cut shorelines of 135 Lake Bonneville, Pelletier [2006] found that soil texture had a weak but significant inverse relationship with D. McKean [1993] estimated a high value of D (360 50 cm 2 /yr) in clay-rich soils at a site in northern California. Particle size likely influences a number of potential transport mechanisms that influence sediment transport efficiency including shrink-swell cycles of clays while also influencing biotic activity. Furthermore, a scarp sampling bias may exist for two reasons. First, the presence of significant vegetation may make it difficult to identify scarps. Second, unlike hillslopes, scarps have a definite lifespan unless repeated offset occurs for scarps of active faults. A bias may exist where estimates of D exist mostly for slower evolving, longer-lived scarps. Hurst and coworkers [2013] compiled a substantial collection of climate, vegetation, and geological data for sites with estimates of D. Hurst and coworkers explored the relationship between climate proxies, D, and lithology. Taking lithology into consideration did not help explain the variability in D at the different sites. However, at a new field site that they consider in that study, they do find that lithology likely influences soil transport efficiency. They found that weak bedrock had higher values of D. However, this is not necessarily inconsistent with our results. Independent of whether the bedrock or underlying material is consolidated, the soils may be poorly developed. The key to high soil transport efficiency may be that sufficient soils exist that are attractive to vegetation and fauna, which are primarily responsible for physically disturbing the soils and causing sediment transport. 136 4.8.2 Towards a general explanation for variations in D Nash [1984] suggested that it may be possible to estimate D from regional information if the controlling mechanisms are known. If this is the case, it may be possible to develop an empirically based model to estimate D. In order to do this, the dominant factors that influence D at different locations must be known. In this study, I identified multiple factors that correlate with D and could therefore be used to estimate D. Syvitski [2007] used categorical estimators in conjunction with topographic characteristics to model global sediment flux. A similar approach could be taken to estimate D. Being able to estimate D globally from climatic, lithological, and other controlling factors in conjunction with topographic data with near global coverage would enable us to better estimate global hillslope sediment fluxes in the past, present and the future. Acknowledgements I would like to thank Naomi Schurr who helped compile some of the data and helped make some of the topographic measurements. I would also like to acknowledge the Department of Defense for funding through a National Defense Science and Engineering Graduate Fellowship. I would like to thank Justine Owen for sharing topographic data that she collected in the Atacama Desert, Chile. 137 138 Chapter 5. Conclusion In the three preceding chapters, I investigated the relationship between climate and landscape evolution. I accomplished this by (1) investigating how erosional mechanisms that may be sensitive to differences in microclimates can generate asymmetric topography, and by (2) compiling estimates of the soil transport coefficient, and then empirically determining important factors, including climate factors, that influence soil transport efficiency. In Chapter 2, I incorporated aspect-dependent erosional mechanisms, such as aspect-dependent regolith strength, aspect-dependent soil creep, and aspect-dependent runoff, into a landscape evolution model (LEM). I also incorporated lateral channel migration into the LEM. The LEM with lateral channel migration predicts a unique relationship between topographic and erosional characteristics for asymmetric landscapes if lateral channel migration is the root cause of the asymmetry. The aspect-dependent runoff and aspect-dependent regolith strength LEMs predict comparable signatures between topographic and erosional characteristics that differ from the lateral channel migration LEM. The aspect-dependent soil creep LEM predicts a more complex response to asymmetry in soil creep than the other aspect-dependent efficiency 139 models. In general, the aspect-dependent soil creep LEM is not capable of producing hillslopes with a high degree of topographic asymmetry while maintaining realistic topographic characteristics. I also explored controls on asymmetry and found that topographic asymmetry increases with an increase in the Peclet number, a value that describes the competition between fluvial incision and soil creep. This occurred for all of the asymmetry-forming LEMs except the aspect-dependent soil creep LEM. The practical application of this knowledge is that for the same magnitude of asymmetry forcing, higher asymmetry is expected to develop in wider basins. For the LEM with lateral channel migration, I found that the Pdclet number and another dimensionless number, the Migration number, control the topographic asymmetry that develops. I also further investigated the controls on topographic asymmetry for the lateral channel migration model with a 1 -D model that includes hillslope and channel processes. In the 1 -D model, I modified the lateral channel migration rule so that the lateral channel migration rate is driven by differences in sediment flux on opposing sides of a valley instead of occurring at a fixed rate. The 1 -D model suggests that topographic asymmetry may develop and be sustained in landscapes without variability in microclimates if differences in sediment flux occur on opposing slopes for other reasons, such as a difference in initial slope length. In Chapter 3, I investigated which asymmetry-forming mechanisms are most likely responsible for the topographic asymmetry at Gabilan Mesa, CA. The aspectdependent runoff and aspect-dependent regolith strength LEMs are best at reproducing the relationship between topographic asymmetry and erosion rate asymmetry. In a particularly asymmetric valley at Gabilan Mesa, field-saturated hydraulic conductivity is 140 significantly higher on north-facing slopes than on the opposing south-facing slopes, suggesting that significant differences in runoff may occur at Gabilan Mesa under current conditions. My measurements of soil shear strength do not support the hypothesis that aspect-dependent differences in regolith strength are controlling the asymmetry at Gabilan Mesa. Physical and erosion rate evidence of lateral channel migration exist in some basins, but not in all of the asymmetric basins. For the most asymmetric basins at Gabilan Mesa, the LEM predicts that differences in aspect-dependent runoff must be quite high if aspect-dependent runoff is solely responsible for the asymmetry. This suggests that multiple asymmetry-forming mechanisms may be acting in unison. Initial tilting of the mesa and the corresponding background slope may have influenced the initial development of the drainages and caused topographic asymmetry to initially develop. However, the pattern in erosion rates at Gabilan Mesa does not match the prediction if tilting is solely responsible for the topographic asymmetry. When the field measurements and observations are considered alongside the numerical modeling experiments and the topographic and erosion rate analysis, aspectdependent runoff seems to be the most likely explanation for the bulk of the asymmetry at Gabilan Mesa. However, the most asymmetric basins may be influenced by additional asymmetry-forming mechanisms such as initial tilting and lateral channel migration. This analysis suggests that aspect-dependent differences in runoff are important for influencing the development of topography at Gabilan Mesa. Aspect-dependent runoff should also be considered as a likely asymmetry-forming mechanism in other semi-arid landscapes that exhibit differences in microclimate and topographic 141 asymmetry. Furthermore, this study casts doubt on the role of lateral channel migration as a dominant asymmetry-forming mechanism and suggests that evidence of lateral channel migration must be carefully weighed, as its presence at some locations does not indicate that it is occurring everywhere. In Chapter 4, I compiled estimates of the soil transport coefficient (D), and also made new estimates at sites where both high-resolution topographic data and erosion rate estimates already exist. I found that D has a power-law relationship with an exponent of -0.5-0.6 with mean annual precipitation and the aridity index, a useful proxy for water availability in soils. D increases markedly between arid landscapes and more vegetated landscapes. Among vegetated landscapes, further increases in D with precipitation and aridity index are relatively minor, especially compared with the observed range of D among landscapes. This suggests that once a landscape is sufficiently vegetated, increases in precipitation do not necessarily lead to significant increases in sediment transport efficiency or significant increases in hillslope erosion rate. Much work remains to be done to unravel the relationship between climate and landscape evolution. By focusing on understanding the relationship between climate and landscape evolution at specific sites and exploiting natural experiments, we can continue to improve our understanding of how landscape evolution is influenced by climate. We can also continue to improve our understanding of climate and landscape evolution by carefully analyzing compilations that enable us to examine a much larger range in climate than is generally possible for regional studies. In regards to topographic asymmetry, there are still many fundamental questions left to address. My numerical modeling experiments suggest that lateral channel migration may be a self-sustaining process in which 142 differences in slope length alone are enough to produce and maintain topographic asymmetry, but does this actually occur in real landscapes? Determining appropriate field sites where this might be the case and testing the topographic and erosional predictions made from numerical modeling experiments should help solve this mystery. Lateral channel migration has been suggested as a dominant cause of asymmetric topography [Dohrenwend, 1978], and Gabilan Mesa was considered a prime example of such a landscape. However, my analysis suggests that lateral channel migration is not the dominant cause of the asymmetry at Gabilan Mesa. My numerical modeling predictions provide a useful framework for testing if lateral channel migration is fundamentally responsible for the asymmetry at other sites. It will be interesting to apply these tests to other field sites and determine if lateral channel migration is a dominant mechanism in any landscapes that exhibit topographic asymmetry. My analysis of the compilation of D suggests that soil transport efficiency depends on climate. Soil transport efficiency is particularly sensitive in arid regions that are on the cusp of being able to support vegetation and fauna. An important result of aspect-dependent runoff causing the topographic asymmetry at Gabilan Mesa is that it serves as valuable evidence that landscapes can respond dramatically to relatively small differences in climate-differences that currently exist on opposing slopes. As is the case for the compilation of D, we once again learn from this natural experiment at Gabilan Mesa that life is important for determining tipping points in landscapes. Landscapes that exist on these tipping points have likely experienced the most significant changes due to climate change in the past and may experience considerable changes due to anthropogenic climate change in the near future. 143 144 References Allen, R. G., L. S. Pereira, D. Raes, and M. Smith (1998), Crop evapotranspirationGuidelines for computing crop water requirements-FAO Irrigation and drainage paper 56, FAO. Almond, P. C., J. J. Roering, M. W. Hughes, F. S. Lutter, and C. Lebouteiller (2008), Climatic and anthropogenic effects on soil transport rates and hillslope evolution, Sediment Dynamics in ChangingEnvironments, Christchurch, New Zealand: IAHS Publication, 325, 417-424. Anderson, R. S., S. P. Anderson, and G. E. Tucker (2012), Rock damage and regolith transport by frost: an example of climate modulation of the geomorphology of the critical zone, Earth Surf Proc. Land, 38(3), 299-316, doi:10.1002/esp.3330. Arrowsmith, J. R., D. D. Rhodes, and D. D. Pollard (1998), Morphologic dating of scarps formed by repeated slip events along the San Andreas Fault, Carrizo Plain, California, J. Geophys. Res., 103, 10141-10160. Avouac, J. P. (1993), Analysis of Scarp Profiles: Evaluation of Errors in Morphologic Dating, J. Geophys. Res., 98(B4), 6745, doi:10.1029/92JB01962. Avouac, J. P., and G. Peltzer (1993), Active tectonics in southern Xinjiang, China: Analysis of terrace riser and normal fault scarp degradation along the Hotan Qira Fault System, J Geophys. Res., 98(B12), 21773-21807, doi:10.1029/93JB02172. Avouac, J. P., P. Tapponnier, M. Bai, H. You, and G. Wang (1993), Active thrusting and folding along the northern Tien Shan and Late Cenozoic rotation of the Tarim relative to Dzungaria and Kazakhstan, J Geophys. Res., 98(B4), 6755-6804, doi: 10. 1029/92JB0 1963. Balco, G., and J. 0. H. Stone (2005), Measuring middle Pleistocene erosion rates with cosmic-ray-produced nuclides in buried alluvial sediment, Fisher Valley, southeastern Utah, Earth Surf Proc. Land, 30(8), 1051-1067, doi:10.1002/esp.1262. 145 Balco, G., J. 0. Stone, N. A. Lifton, and T. J. Dunai (2008), A complete and easily accessible means of calculating surface exposure ages or erosion rates from 1 OBe and 26AI measurements, QuaternaryGeochronology, 3(3), 174-195, doi:10.1016/j.quageo.2007.12.001. Bass, N. W. (1929), The Geology of Cowley County, Kansas, Kansas Geol. Survey Bull. Begin, Z. B. (1992), Application of quantitative morphologic dating to paleo-seismicity of the northwestern Negev, Israel, Isr. J. Earth Sci., 41, 95-103. Bierman, P. (1996), Estimating rates of denudation using cosmogenic isotope abundances in sediment, Earth Surf Proc. Land, 21, 125-139. Bierman, P. R., E. M. Clapp, K. K. Nichols, A. R. Gillespie, and M. W. Caffee (2001), Using Cosmogenic Nuclide Measurements In Sediments To Understand Background Rates Of Erosion And Sediment Transport, in Landscape Erosion and Evolution Modeling, edited by R. S. Harmon and W. W. Doe, pp. 89-115, Springer US, New York. Binnie, S. A., W. M. Phillips, M. A. Summerfield, and L. K. Fifield (2007), Tectonic uplift, threshold hillslopes, and denudation rates in a developing mountain range, Geology, 35(8), 743-5, doi: 10.11 30/G2364 1 A. 1. von Blanckenburg, von, F. (2006), The Control Mechanisms of Erosion and Weathering at Basin Scale From Cosmogenic Nuclides in River Sediment, Earth Planet. Sci. Lett., 242(3-4), 224-239, doi:10.1016/j.epsl.2005.11.017. Bowman, D., and R. Gerson (1986), Morphology of the latest Quaternary surface-faulting in the Gulf of Elat region, eastern Sinai, Tectonophysics, 128(1-2), 97-119, doi: 10.1016/0040-1951(86)90310-0. Branson, F. A., and L. M. Shown (1990), Contrasts of Vegetation, Soils, Microclimates, and Geomorphic Processes Between North- and South-facing Slopes on Green Mountain Near Denver, Colorado, US. Geol. Surv. Water Resour. Invest. Rep., 894094. Bull, W. B. (1997), Discontinuous ephemeral streams, Geomorphology, 19(3-4), 227276, doi:10.1016/SO169-555X(97)00016-0. Burnett, B. N., G. A. Meyer, and L. D. McFadden (2008), Aspect-Related Microclimatic Influences on Slope Forms and Processes, Northeastern Arizona, J. Geophys. Res., 113(F03002), -, doi: 10.1029/2007jf000789. Callaghan, L. E. (2012), Climate and Vegetation Effects on Sediment Transport and Catchment Properties Along an Arid to Humid Climatic Gradient. Ph.D. thesis, Univ. of Edinburgh, Edinburgh. Carretier, S., J. F. Ritz, J. Jackson, and A. Bayasgalan (2002), Morphological dating of 146 cumulative reverse fault scarps: examples from the Gurvan Bogd fault system, Mongolia, Geophys. J Int., 148, 256-277. Casanova, M., I. Messing, and A. Joel (2000), Influence of aspect and slope gradient on hydraulic conductivity measured by tension infiltrometer, Hydrol. Proc., 14, 155164. Cerda, A., A. C. Imeson, and A. Calvo (1995), Fire and aspect induced differences on the erodibility and hydrology of soils at La Costera, Valencia, southeast Spain, Catena, 24, 289-304. Champagnac, J.-D., P. Molnar, C. Sue, and F. Herman (2012), Tectonics, climate, and mountain topography, J. Geophys. Res., 117, B02403, doi:10.1029/2011JB008348. Churchill, R. R. (1981), Aspect-related differences in badlands slope morphology, Ann. Assoc. Am. Geogr., 71. Churchill, R. R. (1982), Aspect-induced differences in hillslope processes, Earth Surf Proc. Land., 7, 171-182. Clifton, T., D. E. Granger, Z. Gilbert, and M. Caffee (2005), Quartz sample preparation for AMS. Colman, S. M., and K. Watson (1983), Ages estimated from a diffusion equation model for scarp degradation, Science, 221(4607), 263-265, doi: 10.11 26/science.221.4607.263. Culling, W. E. H. (1963), Soil creep and the development of hillside slopes, J. Geol., 71, 127-161. Davis, W. M. (1895), Deflection of rivers by the earth's rotation: deflected rivers in Australia, Science, 27, 32-33. DiBiase, R. A., K. X. Whipple, A. M. Heimsath, and W. B. Ouimet (2010), Landscape form and millennial erosion rates in the San Gabriel Mountains, CA, Earth Planet. Sci. Lett., 289(1-2), 134-144, doi: 10. 1016/j.epsl.2009.10.036. Dixon, J. L., A. M. Heimsath, J. Kaste, and R. Amundson (2009), Climate-driven processes of hillslope weathering, Geology, 37(11), 975-978, doi:10.1130/G30045A.1. Dohrenwend, J. C. (1978), Systematic valley asymmetry in the central California Coast Ranges, Geol. Soc. Am. Bull., 89, 891-900. Durham, D. (1974), Geology of the Southern Salinas Valley Area, California,U.S. Geol. Survey Prof. Paper 819. Einstein, A. (1926), The Cause of the Formation of Meanders in the Courses of Rivers 147 and of the So-Called Baer's Law, Die Naturwissenschaften, 14. Elrick, D. E., and W. D. Reynolds (1992), Methods for analyzing constant-head well permeameter data, Soil Sci. Am. J., 56(1), 320-323, doi: 10.2136/sssaj 1992.03615995005600010052x. Emery, K. 0. (1947), Asymmetrical valleys of San Diego County, California, Bull. South. Calf Acad Sci., 46. Enzel, Y., R. Amit, N. Porat, E. Zilberman, and B. J. Harrison (1996), Estimating the ages of fault scarps in the Arava, Israel, Tectonophysics, 253(3-4), 305-317, doi: 10.1016/0040-1951(95)00072-0. Fairchild, H. L. (1932), Earth rotation and river erosion, Science, 76, 423-426. Fernandes, N. F., and W. E. Dietrich (1997), Hillslope evolution by diffusive processes: The timescale for equilibrium adjustments, Water Resour. Res., 33, 1307-1318. Ferrier, K. L., J. T. Perron, S. Mukhopadhyay, M. Rosener, J. D. Stock, K. L. Huppert, and M. Slosberg (2013a), Covariation of climate and long-term erosion rates across a steep rainfall gradient on the Hawaiian island of Kaua'i, Geol. Soc. Am. Bull., 125(78), 1146-1163, doi:10.1130/B30726.1. Ferrier, K. L., J. W. Kirchner, and R. C. Finkel (2005), Erosion rates over millennial and decadal timescales at Caspar Creek and Redwood Creek, Northern California Coast Ranges, Earth Surf Proc. Land, 30(8), 1025-1038, doi:10.1002/esp.1260. Ferrier, K. L., K. L. Huppert, and J. T. Perron (2013b), Climatic control of bedrock river incision, Nature, 496(7444), 206-209, doi: 10.1038/nature 11982. French, H. M. (1971), Slope asymmetry of the Beaufort Plain, Northwest Banks Island, NWT, Canada, Can. J. Earth Sci., 8, 717-731. Fuller, M. L. (1914), Geology of Long Island, U.S. Geological Survey Professional Paper252, 50. Furbish, D. J., and S. Fagherazzi (2001), Stability of creeping soil and implications for hillslope evolution, Water Resour. Res., 37, 2607-2618. Gabet, E. J. (2000), Gopher bioturbation: Field evidence for non-linear hillslope diffusion, Earth Surf Proc. Land., 25, 1419-1428. Gabet, E. J. (2003), Sediment transport by dry ravel, J Geophys. Res., 108(B 1), 2049-8, doi: 10.1029/2001 JBOO 1686. Gabet, E. J., and S. M. Mudd (2010), Bedrock erosion by root fracture and tree throw: A coupled biogeomorphic model to explore the humped soil production function and the persistence of hillslope soils, J Geophys. Res., 115(F4), F04005, 148 doi: 10.1029/2009JF001526. Garcia, A. F., and S. A. Mahan (2012), The influence of upper-crust lithology on topographic development in the central Coast Ranges of California, Geomorphology, 138(1), 243-262, doi:10.1016/j.geomorph.2011.09.009. Gilbert, G. K. (1884), The sufficiency of terrestrial rotation for the deflection of streams, Am. Jour. Sci., 27, 427-432. Glock, W. S. (1932), Meridional deflection of streams due to earth's rotation, Pan-Am. Geologist, 57, 97-100. Granger, D. E., and J. W. Kirchner (1996), Spatially averaged long-term erosion rates measured from in situ-produced cosmogenic nuclides in alluvial sediment, J. Geol., 104, 249-257. Hack, J. T., and J. C. Goodlett (1960), Geomorphology and forest ecology of a mountain region in the Central Appalachians, U.S. GeologicalSurvey ProfessionalPapers, 294-B. Hancock, G. S., R. S. Anderson, 0. A. Chadwick, and R. C. Finkel (1999), Dating fluvial terraces with 1 OBe and 26AI profiles: application to the Wind River, Wyoming, Geomorphology, 27, 41-60. Hanks, T. C. (2000), The age of scarplike landforms from diffusion-equation analysis, in QuaternaryGeochronology methods and applications, edited by J. S. Noller and J. M. Sowers, pp. 313-338, Quaternary Geochronology methods and applications, AGU, Washington, D.C. Hanks, T. C., and D. J. Andrews (1989), Effect of far field slope on morphologic dating of scarplike landforms, J Geophys. Res., 94(B1), 565-573, doi: 10.1029/JB094iBO 1p00565. Hanks, T. C., and R. E. Wallace (1985), Morphological analysis of the Lake Lahontan shoreline and Beachfront fault scarps, Pershing County, Nevada, Bulletin of the Seismological Society ofAmerica, 75(3), 835-846. Hanks, T. C., R. C. Bucknam, K. R. Lajoie, and R. E. Wallace (1984a), Modification of wave-cut and faulting-controlled landforms, J. Geophys. Res., 89(B7), 5771-5790. Hanks, T. C., R. C. Bucknam, K. R. Lajoie, and R. E. Wallace (1984b), Modification of wave cut and faulting -controlled landforms, J. Geophys. Res., 89(B7), 5771-5790, doi:10.1029/JB089iB07p05771. Heimsath, A. M., D. J. Furbish, and W. E. Dietrich (2005), The illusion of diffusion: Field evidence for depth-dependent sediment transport, Geology, 33(12), 949, doi:10.1130/G21868.1. 149 Heimsath, A. M., J. Chappell, W. E. Dietrich, K. Nishiizumi, and R. C. Finkel (2000), Soil production on a retreating escarpment in southeastern Australia, Geology, 28(9), 787-790, doi:10.1 130/0091-7613(2000)28<787:SPOARE>2.0.CO;2. - Heimsath, A. M., W. E Dietrich, K. Nishiizumi, and R. C. Finkel (1999), Cosmogenic nuclides, topography, and the spatial variation of soil depth, Geomorphology, 27(1 2), 151-172, doi:10.1016/S0169-555X(98)00095-6. Heimsath, A. M., W. E. Dietrich, K. Nishiizumi, and R. C. Finkel (1996), Soil production and landscape equilibrium: Hillslope analysis using cosmogenic nuclides in Northern California and coastal Oregon, Eos Trans. AGU, (77), 46. Heimsath, A. M., W. E. Dietrich, K. Nishiizumi, and R. C. Finkel (1997), The soil production function and landscape equilibrium, Nature, 388, 358-361. Hijmans, R. J., S. E. Cameron, J. L. Parra, P. G. Jones, and A. Jarvis (2005), Very high resolution interpolated climate surfaces for global land areas, Int. J. Climatol., 25(15), 1965-1978, doi:10.1002/joc.1276. Howard, A. D. (1994), A detachment-limited model of drainage basin evolution, Water Resour. Res., 30, 2261-2285. Howard, A. D., and G. Kerby (1983), Channel Changes in Badlands, Geol. Soc. Am. Bull., 94, 739-752, doi: 10.1130/0016-7606(1983)94<739:CCIB>2.0.CO;2. Hughes, M. W., P. C. Almond, and J. J. Roering (2009), Increased sediment transport via bioturbation at the last glacial-interglacial transition, Geology, 37(10), 919-922, doi:10.1 130/G30159A.1. Hurst, M. D., S. M. Mudd, K. Yoo, M. Attal, and R. Walcott (2013a), Influence of lithology on hillslope morphology and response to tectonic forcing in the northern Sierra Nevada of California, J. Geophys. Res., 118(2), 832-851, doi: 10.1 002/jgrf.20049. Hurst, M. D., S. M. Mudd, M. Attal, and G. Hilley (2013b), Hillslopes record the growth and decay of landscapes, Science, 341(6148), 868-871, doi:10.1 126/science.1241791. Hurst, M. D., S. M. Mudd, R. Walcott, M. Attal, and K. Yoo (2012), Using hilltop curvature to derive the spatial distribution of erosion rates, J. Geophys. Res., 11 7(F2), F02017, doi:10.1029/2011JF002057. Istanbulluoglu, E., 0. Yetemen, E. R. Vivoni, H. A. Gutidrrez-Jurado, and R. L. Bras (2008), Eco-geomorphic implications of hillslope aspect: Inferences from analysis of landscape morphology in central New Mexico, Geophys. Res. Lett., 35(14), doi: 10.1029/2008GL034477. Johnson, D. L. (1977), The late Quaternary climate of coastal California: Evidence for 150 an ice age refugium, QuaternaryResearch, 8(2), 154-179, doi:10.1016/00335894(77)90043-6. Jungers, M. C., P. R. Bierman, A. Matmon, K. K. Nichols, J. Larsen, and R. C. Finkel (2009), Tracing hillslope sediment production and transport with in situ and meteoric lOBe, J. Geophys. Res., 114, F04020, doi:10.1029/2008JF001086. Kane, P. S. (1970), Asymmetrical valleys: an example from the Salinas Valley, California. M.S. thesis, Univ. of California, Berkeley, Berkeley. Knighton, D. (1998), FluvialForms and Processes:A New Perspective, Hodder Arnold, London. Kreslavsky, M. A., and J. W. Head (2003), North-south topographic slope asymmetry on Mars: Evidence for insolation-related erosion at high obliquity, Geophys. Res. Lett., 30(15), doi:10.1029/2003g1017795. Krishnan, S., C. Crosby, V. Nandigam, M. Phan, C. Cowart, C. Baru, and R. Arrowsmith (2011), OpenTopography: a services orientedarchitecturefor community access to LIDAR topography, ACM, New York, New York, USA. Kutiel, P. (1992), Slope aspect effect on soil and vegetation in a mediterranean ecosystem, Isr. J Bot., 41(4-6), 243-250, doi:10.1080/0021213X.1992.10677231. Leonard, J., and G. Richard (2004), Estimation of runoff critical shear stress for soil erosion from soil shear strength, Catena, 57, 233-249. Leopold, L. B., and T. Maddock (1953), The Hydraulic Geometry of Stream Channels and Some Physiographic Implications, U.S. GeologicalSurvey ProfessionalPaper 252. Limaye, A. B. S., and M. P. Lamb (2014), Numerical simulations of bedrock valley evolution by meandering rivers with variable bank material, J. Geophys. Res., 119(4), 927-950, doi:10.1002/2013JF002997. Matmon, A., P. R. Bierman, J. Larsen, S. Southworth, M. Pavich, R. C. Finkel, and M. W. Caffee (2003), Erosion of an Ancient Mountain Range, The Great Smoky Mountains, North Carolina and Tennessee,, 303(9), 817-855, doi: 10.2475/ajs.303.9.817. Mattson, A., and R. L. Bruhn (2001), Fault slip rates and initiation age based on diffusion equation modeling: Wasatch Fault Zone and eastern Great Basin, J. Geophys. Res., 106(B7), 13739-13750, doi:10.1029/2001JB900003. McGuire, L. A., J. D. Pelletier, and J. J. Roering (2014), Development of topographic asymmetry: Insights from dated cinder cones in the western United States, J Geophys. Res., 119(8), 1725-1750, doi:10.1002/2014JF003081. 151 McKean, J., W. E. Dietrich, R. C. Finkel, and M. W. Caffee (1993), Quantification of soil production and downslope creep rates from cosmogenic 1 0Be accumulations on a hillslope profile, Sediment Dynamics in ChangingEnvironments, 21, 343-346. McKnight, T., and D. Hess (2008), Climate zones and types, in Physical Geography: A LandscapeAppreciation, pp. 217-221, Prentice Hall, Upper Saddle River, NJ. Melton, M. A. (1960), Intravalley variation in slope angles related to microclimates and erosional environment, GSA Bull., 71(2), 133-144. Mertens, J., D. Jacques, J. Vanderborght, and J. Feyen (2002), Characterisation of the field-saturated hydraulic conductivity on a hillslope: In situ single ring pressure infiltrometer measurements, J. Hydrol., 263, 217-229. Molnar, P. (2001), Climate change, flooding in arid environments, and erosion rates, Geology, 29(12), 1071-1074, doi:10.1130/00917613(2001)029<1071:CCFIAE>2.0.CO;2. Molnar, P. (2004), Late Cenozoic increase in accumulation rates of terrestrial sediment: How might climate change have affected erosion rates? Annu. Rev. Earth Planet. Sci., 32, 67-89, doi: 10.1 146/annurev.earth.32.091003.143456. Molnar, P., R. S. Anderson, G. Kier, and J. Rose (2006), Relationships among probability distributions of stream discharges in floods, climate, bed load transport, and river incision, J Geophys. Res., ]1(F2), F02001, doi:10.1029/2005JF000310. Montgomery, D. R. (2004), Observations on the role of lithology in strath terrace formation and bedrock channel width, Am. J. Sci., 304(5), 454-476, doi: 10.2475/ajs.304.5.454. Moon, S., C. P. Chamberlain, K. Blisniuk, N. Levine, D. H. Rood, and G. E. Hilley (2011), Climatic control of denudation in the deglaciated landscape of the Washington Cascades, Nat. Geosci., 4(7), 469-473, doi:10.1038/ngeol 159. Moon, S., E. Shelef, and G. E. Hilley (2015), Recent topographic evolution and erosion of the deglaciated Washington Cascades inferred from a stochastic landscape evolution model, J. Geophys. Res., 120(5), 856-876, doi: 10.1002/2014JF0033 87. Nash, D. (1980a), Forms of bluffs degraded for different lengths of time in emmet county, Michigan, U.S.A, Earth Surf Proc., 5(4), 331-345, doi: 10.1002/esp.3760050405. Nash, D. B. (1980b), Morphologic dating of degraded normal fault scarps, J. Geol., doi: 10.2307/30062456. Nash, D. B. (1984), Morphologic dating of fluvial terrace scarps and fault scarps near West Yellowstone, Montana, Geol. Soc. Am. Bull., 95(12), 1413-1424, doi: 10.1130/0016-7606(1984)95<1413:MDOFTS>2.0.CO;2. 152 Nishiizumi, K., M. ImamuraA, M. W. Caffee, J. R. Southon, R. C. Finkel, and J. McAninch (2007), Absolute calibration of 101Be AMS standards, Nucl. Instrum. Methods Phys. Res., 258, 403-413, doi: 10.101 6/j.nimb.2007.01.297. Niviere, B., and G. Marquis (2000), Evolution of terrace risers along the upper Rhine graben inferred from morphologic dating methods: evidence of climatic and tectonic forcing, Geophys. J. Int., 141(3), 577-594, doi: 10.1046/j.1365-246x.2000.00123.x. Ollier, C. D., and A. J. Thomasson (1957), Asymmetrical valleys of the Chiltern Hills, Geogr. J., 123, 71-80. Owen, J. J., R. Amundson, W. E. Dietrich, K. Nishiizumi, B. Sutter, and G. Chong (2011), The sensitivity of hillslope bedrock erosion to precipitation, Earth Surf Proc. Land, 36(1), 117-135, doi:10.1002/esp.2083. Parsons, A. J. (1988), Hillslopeform, Routledge, New York. Pelletier, J. D. et al. (2011), Calibration and testing of upland hillslope evolution models in a dated landscape: Banco Bonito, New Mexico, J Geophys. Res., 116(F4), F04004-24, doi:10.1029/2011JF001976. Pelletier, J. D., and M. L. Cline (2007), Nonlinear slope-dependent sediment transport in cinder cone evolution, Geology, doi:10.11 30/G23 992A. 1;2figures;DataRepositoryitem2007266. Pelletier, J. D., S. B. DeLong, A. H. Al-Suwaidi, M. Cline, Y. Lewis, J. L. Psillas, and B. Yanites (2006), Evolution of the Bonneville shoreline scarp in west-central Utah: Comparison of scarp-analysis methods and implications for the diffusion model of hillslope evolution, Geomorphology, 74, 257-270, doi: 10.101 6/j.geomorph.2005.08.008. Perron, J. T., and J. L. Hamon (2012), Equilibrium form of horizontally retreating, soilmantled hillslopes: Model development and application to a groundwater sapping landscape, J Geophys. Res., 117(F1), F01027, doi:10.1029/2011JF002139. Perron, J. T., J. W. Kirchner, and W. E. Dietrich (2009), Formation of evenly spaced ridges and valleys, Nature, 460(7254), 502-505, doi:doi: 10.103 8/nature08174. Perron, J. T., P. W. Richardson, K. L. Ferrier, and M. Lap6tre (2012), The root of branching river networks, Nature, 492(7427), 100-103, doi: 10.103 8/nature 11672. Perron, J. T., W. E. Dietrich, and J. W. Kirchner (2008), Controls on the spacing of firstorder valleys, J. Geophys. Res., 113(F4), doi:10.1029/2007JF000977. Petit, C., Y. Gunnell, N. Gonga-Saholiariliva, B. Meyer, and J. Seguinot (2009), Faceted spurs at normal fault scarps: Insights from numerical modeling, J. Geophys. Res., 114(B5), B05403-13, doi:10.1029/2008JB005955. 153 Petroff, A. P., 0. Devauchelle, A. Kudrolli, and D. H. Rothman (2012), Four remarks on the growth of channel networks, Comptes Rendus Geoscience, 344(1), 33-40, doi: 10. 1016/j.crte.2011.12.004. Philip, J. R. (1985), Approximate analysis of the borehole permeameter in unsaturated soil, Water Resour. Res., 21(7), 1025-1033, doi: 10.1029/WRO21 i007pO 1025. Pierce, J. L., G. A. Meyer, and A. J. T. Jull (2004), Fire-induced erosion and millennialscale climate change in northern ponderosa pine forests, Nature, 432(7013), 87-90, doi: 10.1038/nature03058. Pierce, K. L., and S. M. Colman (1986), Effect of height and orientation (microclimate) on geomorphic degradation rates and processes, late-glacial terrace scarps in central Idaho, Geol. Soc. Am. Bull., 97, 869-885. Platts, W. S., R. J. Torquemada, M. L. McHenry, and C. K. Graham (1989), Changes in Salmon spawning and rearing habitat from increased delivery of fine sediment to the South Fork Salmon River, Idaho, Trans. Am. Fish. Soc., 118(3), 274-283, doi: 10.1577/1548-8659(1989)118<0274:CISSAR>2.3.CO;2. Portenga, E. W., and P. R. Bierman (2011), Understanding Earth's eroding surface with 1Be, GSAT, 21(8), 4-10, doi:10.1130/G111A.1. Poulos, M. J., J. L. Pierce, A. N. Flores, and S. G. Benner (2012), Hillslope asymmetry maps reveal widespread, multi-scale organization, Geophys. Res. Lett., 39, L06406, doi: 10.1029/2012GL051283. Powell, J.W., (1874), Remarks on the structural geology of the valley of the Colorado of the west, Phil. Soc. Wash., 1, 48-5 1. Prosser, I. P., and W. E. Dietrich (1995), Field experiments on erosion by overland flow and their implication for a digital terrain model of channel initiation, Water Resour. Res., 31(11), 2867-2876, doi:10.1029/95WR02218. Reed, R. D. (1927), Wind and soil in the Gabilan Mesa, J. Geol., 35, 84-88. Reneau, S. L. (1988), Depositional and Erosional History of Hollows. Ph.D. thesis, Univ. of California, Berkeley, Berkeley. Reneau, S. L., and W. E. Dietrich (1991), Erosion rates in the southern Oregon Coast Range: Evidence for an equilibrium between hillslope erosion and sediment yield, Earth Surf Proc. Land, 16, 307-322. Reneau, S. L., W. E. Dietrich, and R. Dorn (1986), Geomorphic and Paleoclimatic Implications of Latest Pleistocene Radiocarbon Dates From Colluvium-Mantled Hollows, California, Geology, 14, 655-658. Reneau, S. L., W. E. Dietrich, M. Rubin, D. J. Donahue, and A. J. T. Jull (1989), 154 Analysis of hillslope erosion rates using dated colluvial deposits, i. Geol., 97(1), 4563, doi:10.2307/30062184. Reynolds, W. D., and D. E. Elrick (1990), Ponded infiltration from a single ring. I. Analysis of steady flow, Soil Sci. Soc. Am. J, 54, 1233-1241. Riebe, C. S., J. W. Kirchner, and R. C. Finkel (2004), Erosional and climatic effects on long-term chemical weathering rates in granitic landscapes spanning diverse climate regimes *, Earth Planet. Sci. Lett., 224(3-4), 547-562, doi: 10.101 6/j.epsl.2004.05.019. Riggins, S. G., R. S. Anderson, S. P. Anderson, and A. M. Tye (2011), Solving a conundrum of a steady-state hilltop with variable soil depths and production rates, Bodmin Moor, UK, Geomorphology, 128(1-2), 73-84, doi: 10. 1016/j.geomorph.2010.12.023. Roering, J. J. (2004), Constraining climatic controls on hillslope dynamics using a coupled model for the transport of soil and tracers: Application to loess-mantled hillslopes, South Island, New Zealand, J. Geophys. Res., 109(F 1), FO1010-19, doi: 10.1 029/2003JF000034. Roering, J. J., and M. Gerber (2005), Fire and the evolution of steep, soil-mantled landscapes, Geology, 33(5), 349-5, doi: 10.11 30/G21260.1. Roering, J. J., J. Marshall, A. M. Booth, M. Mort, and Q. Jin (2010), Evidence for biotic controls on topography and soil production, Earth Planet. Sci. Lett., 298(1-2), 183190, doi:10.1016/j.epsl.2010.07.040. Roering, J. J., J. T. Perron, and J. W. Kirchner (2007), Functional relationships between denudation and hillslope form and relief, Ear. Plan. Sci. Lett., 264, 245-258. Roering, J. J., J. W. Kirchner, and W. E. Dietrich (1999), Evidence for nonlinear, diffusive sediment transport on hillslopes and implications for landscape morphology,, 35(3), 853-870, doi: 10.1029/1998WR900090. Roering, J. J., P. Almond, P. Tonkin, and J. McKean (2002), Soil transport driven by biological processes over millennial time scales, Geology, 30(12), 1115-1118. Rosenbloom, N. A., and R. S. Anderson (1994), Hillslope and channel evolution in a marine terraced landscape, Santa Cruz, California, J Geophys. Res., 99(B7), 1401314029, doi: 10.1029/94JB00048. Seidl, M. A., W. E. Dietrich, and J. W. Kirchner (1994), Longitudinal profile development into bedrock: an analysis of Hawaiian channels, J Geol., 102(4), 457474, doi:10.2307/30065663. Siegmund, M., and H. Kevin (2000), A study of valley-side slope asymmetry based on the application of GIS analysis: Alexander Island, Antarctica, Antarctic Sci., 12(4), 155 471-476. Small, E. E., R. S. Anderson, and G. S. Hancock (1999), Estimates of the rate of regolith production using and from an alpine hillslope, Geomorphology, 27(1-2), 131-150, doi: 10.1016/S0169-555X(98)00094-4. Snyder, N. P., K. X. Whipple, G. E. Tucker, and D. J. Merritts (2003), Importance of a stochastic distribution of floods and erosion thresholds in the bedrock river incision problem, J Geophys. Res., 108(B2), 2117, doi:10.1029/2001JB001655. Spelz, R. M., J. M. Fletcher, L. A. Owen, and M. W. Caffee (2008), Quaternary alluvialfan development, climate and morphologic dating of fault scarps in Laguna Salada, Baja California, Mexico, Geomorphology, 102(3-4), 578-594, doi: 10.1016/j.geomorph.2008.06.001. Syvitski, J. P. M., and J. D. Milliman (2007), Geology, geography, and humans battle for dominance over the delivery of fluvial sediment to the coastal ocean, J Geol., 115(1), 1-19, doi:10.1086/509246. T. Perron (2006), Thesis Appendix, PhD thesis, University of California, Berkeley. Tapponnier, P. et al. (1990), Active thrusting and folding in the Qilian Shan, and decoupling between upper crust and mantle in northeastern Tibet, Earth Planet. Sci. Lett., 97(3-4), 382-403, doi: 10.1016/0012-821X(90)90053-Z. Thomas, B. A., and D. R. Montgomery (1991), Sediment Transport by borrowing mammals, Marin County, California, Earth Surf Proc. Land, 16, 163-172. Tucker, G. E., and R. L. Bras (2000), A stochastic approach to modeling the role of rainfall variability in drainage basin evolution, Water Resour. Res., 36(7), 19531964. Tucker, G. E., and R. L. Bras (1998), Hillslope processes, drainage density, and landscape morphology, Water Resour. Res., 34(10), 2751-2764. Tucker, G. E., L. Arnold, R. L. Bras, H. Flores, E. Istanbulluoglu, and P. Solyom (2006), Headwater channel dynamics in semiarid rangelands, Colorado high plains, USA, Geol. Soc. Am. Bull., 118(7-8), 959-974, doi:10.1130/B25928.1. Walker, E. H. (1948), Differential erosion on slopes of northern and southern exposure in western Wyoming, Geol. Soc. Am. Bull., 59. Walther, S. C., J. J. Roering, P. C. Almond, and M. W. Hughes (2009), Long-term biogenic soil mixing and transport in a hilly, loess-mantled landscape: Blue Mountains of southeastern Washington, Catena, 79(2), 170-178, doi:10.l01 6/j .catena.2009.08.003. Wende, R. (1995), Drainage and valley asymmetry in the Tertiary Hills of Lower 156 Bavaria, Germany, Geomorphology, 14, 255-265. West, N., E. Kirby, P. Bierman, and B. A. Clarke (2014), Aspect-dependent variations in regolith creep revealed by meteoric 1 0Be, Geology, 42(6), 507-510, doi:10.1 130/G35357.1. West, N., E. Kirby, P. R. Bierman, R. Slingerland, L. Ma, D. Rood, and S. Brantley (2013), Regolith production and transport at the Susquehanna Shale Hills Critical Zone Observatory, Part 2: Insights from meteoric 101Be, J. Geophys. Res., 118(3), 1877-1896, doi:10.1002/jgrf.20121. Whipple, K. X., and G. E. Tucker (1999), Dynamics of the stream-power river incision model: Implications for height limits of mountain ranges, landscape response timescales, and research needs, J Geophys. Res., 104(B8), 17661-17674. Wolanski, E., L. A. Boorman, L. Chicharo, E. Langlois-Saliou, R. Lara, A. J. Plater, R. J. Uncles, and M. Zalewski (2004), Ecohydrology as a new tool for sustainable management of estuaries and coastal waters, Wetl. Ecol. Manage., 12(4), 235-276, doi:10. 1007/s 11273-005-4752-4. Wu, L., J. B. Swan, and J. L. Nieber (1993), Soil-macropore and layer influences on saturated hydraulic conductivity measured with borehole permeameters, Soil Sci. Am. J., 574(4), 917-923. Xu, M., Y. Qi, J. Chen, and B. Song (2004), Scale-dependent relationships between landscape structure and microclimate, Plant Ecol., 173, 3 9-57. Yetemen, 0., E. Istanbulluoglu, and E. R. Vivoni (2010), The implications of geology, soils, and vegetation on landscape morphology: Inferences from semi-arid basins with complex vegetation patterns in Central New Mexico, USA, Geomorphology, 116(3-4), 246-263, doi:10.1016/j.geomorph.2009.11.026. Yetemen, 0., E. Istanbulluoglu, J. H. Flores-Cervantes, E. R. Vivoni, and R. L. Bras (2015), Ecohydrologic role of solar radiation on landscape evolution, Water Resour. Res., 51(2), 1127-1157, doi:10.1002/2014WR016169. Yoo, K., R. Amundson, A. M. Heimsath, and W. E. Dietrich (2005), Process-based model linking pocket gopher (Thomomys bottae) activity to sediment transport and soil thickness, Geology, 33(11), 917, doi: 10.1 130/G2183 1.1. - Zimbone, S. M., A. Vickers, R. P. C. Morgan, and P. Vella (1996), Field Investigations of different techniques for measuring surface soil shear strength, Soil. Technol., 9(1 2), 101-111. Zomer, R. J., A. Trabucco, D. A. Bossio, and L. V. Verchot (2008), Climate change mitigation: A spatial analysis of global land suitability for clean development mechanism afforestation and reforestation, Agric. Ecosyst. Environ., 126(1-2), 6780, doi:10.1016/j.agee.2008.01.014. 157 158 Appendix 159 Appendix 1. Cosmogenic radionuclide-derived erosion rates and related data. All reported uncertainties are one st ndard error of the mean. [IO4ei Sample Site BSA, Aspect Type Location* (*N/*W) ) (m(AMSL) (spallation/muons) [o[hr "BBe' Be Carrier "Be production rate Elevation' fantor (mg) Quartz (g) 1( x0) (atoms/g/yr) GAB 13003 GAB 13004 GABI3005 GAB 13006 GAB13023 0.99 0.99 0.99 0.98 0.99 0.99 0.99 0.99 0.99 0.99 0.99 0.98 1 0.99 0.99 0.99 0.99 0.98 0,98 0.98 1 26.180 24.072 29.020 28.108 23.279 35.247 34.392 33.390 28.979 42.280 32.409 19.808 24.575 43.977 41.996 22.826 39.997 30.265 16.688 29.325 13.743 0.2863 0.2852 0.2821 0.2836 0.2834 0 0 0 0 0 0 0 0 0 0.4 5.69/0.209 5.65/0.208 5.44/0.205 5.23/0.203 5.29/0.203 5.27/0.202 5.24/0.202 5.23/0.202 5.13/0.200 5.08/0.199 5.12/0.200 4.99/0.199 5.44/0.204 5.33/0.203 5.38/0.204 5.34/0.203 5.18/0.201 5.16/0.201 4.79/0,195 4.80/0.196 4.35/0.200 0 0 0 0 0 0 0 0 0 0 Isotoee 8 SiO) 0.15910.068 0.159 0.068 0.159 0.068 0.159 0.068 0.159 0.068 0.159 0.068 0.159 0.068 0.159 0.068 0.159 0.068 0.159 0.068 0.159 0.068 0.159 0.068 for inriheitance - south-facing bedrock-soil interface - north-facing bedrock-soilinterface 35.9128/120.7770 354 0.6 3.98/0.203 0.98 12.257 0.2864 21.4 1.6 0.253 0.027 0.159 0.068 2451 1202 1202 north-facing bedrock-soil interface 35.9119/120.7710 322 1.4 2.75/0.201 0.97 23.316 0.2776 32.0 2.7 0.213 0.022 0.159 0.068 1749 +4185 south-facing north-facing south-facing north-facing +0.08 -0.08 soath-facing north-facing catchment catchment +0.06 ((7 -0.06 +0.03 2.33 -0.03 south-facing north-facing south-facing north-facing catcuanrot catchment catchment catchment 2.02 +0.06 -0.05 south-facing 0.73 +0.03 -0.03 south-facing north-facing catchment catchment catchment catchment 1.36 + 0.05 -0.05 south-facing north-facing 1.66 +0.03 -0.03 south-facing north-facing + 0.03 south-facing north-facing 1.92 6 8 10 8 0.69 - -0.03 north-facing catchment catchment catchment catchment catchment catchment catchment catchment catchment 0 0.2837 0.2879 0.2876 0.2930 0.2890 0.2803 0.2834 0.2759 0.2848 0.2834 0.2842 0.2867 0.2850 0.3493 0.3187 0.2856 interface. 0.159 0.069 0.159 0.068 0.159t0.068 0.159 0.068 0.159 0.068 0.159 0.068 0.159 0.068 0.159*0.068 0.159+0.068 Erosion rate not corrected for inheritance (M/Myrs) 87t12 78 10 64 7 266 110 63 7 45 4 69 9 58 7 77 11 42 4 55 6 64t8 46 5 42 4 55 6 161 43 51 5 72 9 72 10 93 19 53+7 (I/km2/yr) 219 30 196 25 161 19 666 E 276 158 18 113 11 172 22 145 17 192 27 105 10 136 15 160 21 114 12 105 10 137 14 403 t106 128 13 181 23 179 24 232 47 132 18 3 -0.03 +0.05 2.10 -0.05 locations and elevations nre the means for the samnled hasins Reported e from the bedrock-soil thickness of small rnins does not cause inificant attenuation of cosmic ravs. so the depth of surface samples is assumed to he zero. For qamoles collected ratios are standardied accordins to isotonic standard 07KNSTD WNishiibmi et al.. 2007). rates production and statistics counting mass carrier in blank uncertainties incorporate unceetainties 'Rinnk ahumdance wasscaled hvsamnle carrer mass Renorted 'Inherited concentration estimated from "Be concetration of deenhy buried rock that has recently been exposed alona road cuts (Perrone al.- 2012). TIh 112.8 3.8 113.0 3.9 155.8 5.3 58.0 2.2 125.0 4.5 244.4 6.7 167.1 7.4 185.7 8.1 126.7 6.5 297.9 10.2 191.7 7.2 102.9t5.3 178.4 7.5 323.5 8.1 251.5 7.6 63.9 4.7 244.2 6.7 142.2 4.7 110.4i4.3 60.2 5.2 78.1 4.9 (10' atoms/ g SiA) 0.787 0.028 0.053 0.032 0.978 0.034 0.356 0.016 0.975 0.037 1.287 0.036 0.906 0.041 1.039 0.047 0.021 0.044 1.337 0.047 1.[)78 0.042 0.934 0.051 1.299 0.056 1.378 0.035 1.111 0.034 0.488 0.039 1.145 0.032 0.862 0.030 0.832 0.036 0,686t0.069 1.012 0.068 3 2 GAB13011 GAB13024 441 432 385 350 351 348 340 338 316 303 312 294 373 360 372 363 325 334 246 248 319 fi10' atoms/ 35.9287/120.7493 35.9213/120.7481 35.9201/120.7613 35.9157/120.7617 35.9163/120.7739 35.9119/120.7702 35.9086/120.7674 35.9057/120.7645 35.9135/120.7971 35.9084/120.8003 35.9182/120.8102 35.9161/120.8076 35.9279/120.7864 35.9255/120.7845 35.9230/120.7799 35.9190/120.7772 35.9502/120.7928 35.9418/120.7988 35.9202/120.8278 35.9165/120.8255 35.9128/120.7731 1.+0.03 89 GAB13009 GAB13012 GAB 13013 GAB 13014 GABI3015 GAB13016 GAB13017 GAB13018 GAB 13019 GAB 13020 GABI 3021 GAB13028 GAB 13029 GMIO-03 GMIO-04 GAB13022 catchment hnierites? [OBeI Erosion rote corrected Bedrockerrsion rate corrected for c + the denth is mesured as the slone-normal distance from the sample to the soil surface 85 481 +981 -481 + 700 6 1675 (t/km/vr) 174 13 159 12 134 10 367 28 131 10 98 7 141 11 122 10 154 13 92 7 115 9 132 11 100t8 92 6 116t8 270 28 109 8 147 11 144 11 176 21 110+ 10 Bedrock erosion rate not corrected for inheritance (n/Myrs) 69 5 63 5 54 4 147 11 53 4 39 3 56 4 49 4 61 5 37 3 46 3 53 4 40 3 37 3 47 3 108 11 44 3 59 4 5814 70 9 44+4 440 55 176 22 431 52 173 21 Appendix 2. Estimates of the soil transport coefficient (D) made in this study. Site ON Location (lat/lon 0) Bedrock Erosion Ridgetop Laplacian 3 Rate (m/Myrs) (x10 1/rM) Source of erosion rates Source of topographic data 19 + 1 Rate from Portenga and Bierman (2011). Rate originally determined by Matmon et al. (2003). OpenTopography Rate from Willenbring et al. (2013). Rate originally determined by Binnie et al. (2007). OpenTopography 2 -28.2 148 -157.3 + 7.3 175 21 -24.7 0.6 83 15 108+ 17 -30.2 0.6 71 12 37.850/-122.550 102 23 -13.8 0.9 174 21 Drift Creek, OR, USA 44.517 / -123.844 155 30 -22.0 1.6 147 33 Blasingame, CA, USA Atacama Desert, Chile Atacama Desert, Chile 36.954/-119.631 -24.130/-69.990 -29.770/-71.080 -26.9 -24.6 -29.1 0.4 6.0 1.2 22 3 Great Smokey Mountains, NC, USA 35.621714/-83.204 27 San Bernardino Mountains, CA, USA 34.051351/-116.934 1373 Wasatch Mountains, UT, USA 40.892/-111.865 89+9 San Gabriel Mountains, CA, USA 34.3641/-117.992 Tennessee Valley, CA, USA 30 4 1 0 27 3 4.3 2 D (cm /yr) 1+ 1 16 2 Rate from Willenbring et al. (2013). Rate originally determined by Stock et al. (2009). Rate from Willenbring et al. (2013). Rate originally determined by DiBiase et al. (2010) Rate from Portenga and Bierman (2011). Rate originally determined by Heimsath et al. (1997). Rate from Portenga and Bierman (2011). Rate originally determined by Bierman et al. (2001). Dixon et al. (2009) Owen et al. (2011) Owen et al. (2011) OpenTopography OpenTopography OpenTopography OpenTopography OpenTopography J. Owen J. Owen Appendix 3. Compilation of the soil transport Site Loaion Charwell * Sasin, New AhuriiNew Zealand Ahuriri, New Zealand Carrizo Plain. CA, USA coeflicient (D) Latitude (c) Soaree and site information. MAP Longitude(0)&A] l Ss (an/ye) Underlying lithology Description Lithology Technique Technique Vegetation Vegetation category' Description Cntegsry' Description Categnry' 2 Grassland/shrubland (recently cleared), but previously bech 2 Previous vegetation was likely podocarphardwood forest, but recently repalced by pasture grasses Transient hillslope Laplacian and mond ef a]. [2008] -42.450 173-357 30= 0 1.42 11 0.57 Loess underlain by fluvial gravel terraces I Almond er al. [2008] -43.702 172.584 31 4 0.76 68.8 0.70 Thick loess deposits underlain by altered basalt I Almond et al. [2008] -43.702 172.594 70 20 0.76 68.8 0.70 Thick underlain by altered basalt I "'Cs fallout nuclidederived eosion rate (-50 yr timescale) and slope Laplacian 2 Previous vegetation was likely podocarphardwood forest, but recently repalced by pasture grasrs; 2 35.271 -119.827 0.33 46.7 2.24 2 Scarp modeling I Grasses and shrubs 2 Avouac andPeltzer (I993) Avoac et al. (1993) 36.800 80.500 33+ 14 0.03 3.3 2.55 Loose fan gravels Scarp modeling I 44.048 86.790 55,20 0.19 18.4 1.57 Loose fan gravels Scarp modeling 1 Begin (1992) 31.262 34.802 4 +3 0.16 23.3 2.78 Fluvial gravel terraces Scarp modeling I Not vegatated Bowsan and Gesn 31.386 35.361 4 0.07 10.9 2.97 Gravel Scarp modeling I Not vegatated 35.240 4 0.04 6 2.60 Scarp modeling I Not Arrowsith (1998) et al. 86 8 Paso Robles Formation (poorly consolidated gravels and sands) and Holocene alluvial fan timescale from beginning of Holocene to present Transient hillsope Laplacian and timscale ofvoclanic event (-27ka) to present units Region, Xinjiang, China Hotan Tien Shan, China Northern Negev' Israel Lake Lisan, Dead Sea, lsrael oreAaa' Israel Chile (1986) Goss Bowman and (1989) reported in Hank (2000) 30.658 Callaghan (2012) -32.99 -71.42 0.41 55=24 48.2 3.41 Chile Callaghan (2012) -32.98 -71.42 70 + 36 043 51.8 3.41 Granitic Chile Callaghan (2012) -32.98 -71.42 41 + 20 0.43 51.8 3.41 Granitic Chile Callaghan (2012) -32.94 -71.43 46 20 0.34 39.7 3.26 Granitic 58 27 0.45 53.1 3.44 Granitic vcgatated vegatated Relief and erosion rate Mostly herbaceous. Relief and erosion Mostly herbaceous. some ttees 'ate Callaghan (2012) -33.01 Chile Callaghan (2012) -31.12 -71.58 46 + 7 0.14 16.7 3.59 Granrtic Chile Callaghan(2012) -31.12 -71.56 44 13 0.15 16.7 3.59 Granitic Chile Chile Callaghan (2012) Callaghan (2012) and erosion Mostly rate Relief and oen Relief aderosion 3sone Herbaceous, 3 Relief and erosion ate 3 -31.12 -30.55 -71.55 -71.63 49+ 13 68 158 0.16 0.13 18.3 13.8 3.54 Relief and erosion rate 4.09 Granitic Relief and erosion rate Relief and erosion rater Chile Callaghan (2012) -30.55 -71.63 212 +92 0.13 13.8 4.09 Granitic Chile Callaghan (2012) -29.62 -71.20 38 + 13 0.07 7.6 3.47 Granitic Chile Callaghan(2012) -29.62 -71.20 38 + 11 0.07 7.6 3.47 Granitic Chile Callaghn (2012) -29.62 -71.20 35+12 0.07 7.6 3,47 Granitic Chile Callaghan (2012) -29.58 -71.14 20 + 7 0.56 7.3 3.62 Granitic Chile Callaghan(2012) -29.57 -71.16 19+7 0.06 7.4 3.73 Granitic Chile Callaghan (2012) -29.22 -71.18 27 9 0.06 6.5 3,51 Granitic CHiC Callaghan (2012) -29.23 -71.18 14 c 5 0.05 6.5 3.51 Granitic CHl Callaghan (2012) -28.41 -71.05 16 = 7 0.04 4.7 3.57 Granitic CHIle Callaghan (2012) -28.40 -71.06 11 + 5 0.03 4.5 3.73 Granitic Chile Callaghan (2012) -28.39 -7107 15 7 0.03 4.3 3.91 Granitic Chile Callaghan(2012) -28.36 -71.05 18 9 003 4 3.90 Granitic Chile Callaghan (2012) -26.57 -70.44 2 1 0.02 2 4.20 Granitic Chile Callaghan(2012) -26.56 -70.48 3 1 0.02 2.3 4.17 162 Relief and erosion and roson rate Relief rate Mostly herbaceous. 3wineetrees Relief 3 Mostly herbaceous, few trees, sore bare ground 3 Mostly herbaceous, wsoe trees 3 Mostly herbaceous, bare few trees, ground 3 Mostly herbaceous. bare few trees, ground some some Mostly herbaceous, few trees. some bare 3 ground and erosion rate 3 Mostly herbaceous. fewtrees.soanbare ground Relief and erosion 3 Mostly hcrbaccous.few sine bare ground Relief and rate erosion 3 Mostly herbaccous, bare few ground erosion 3 rate Relief and rate rate 3 Relief 3 rate Relief snd erosion rate Relief and erosion rate Relief and erosion rate trees, some ofherbaceous groundcovernand bare ground Mixture of herbaceous groundcover and bare ground Mosrtly bare ground, sereeberbeous and erosion Relief and erosion rate Relief and erosion trees. Mixture ground 3 few trees Mostly herbaceous, f some trees 3 Relief and erosion rate Granitic herbaceous, trees Mostly herbaceous, trees erosion rates / Chile -71.44 Relief 3 trees 3some / M Not Grasses and shrubs cover 3 Bare ground 3 Bare ground 3 Bare ground 3 Bare ground 3 Bare ground 3 3 2 3 Chile Callaghan (2012) -26.56 -70.51 4+12 0.02 2.3 4.17 Granitic Chile Callaghan(2012) -26.59 -70.49 4 '2 0.02 2.3 4.17 Gniti Chile Callaghan (2012) -26.57 -70.56 9 i4 0.02 2.4 4.00 Granitic Chile Callagh n (2012) -40.58 -73.69 58 17 2.23 184 1.42 GUti6 Chile Callaghan(2012) -40.58 -73.60 61 20 2.13 178 1.40 Gr Chile Callaghan (2012) -37.90 -73.28 40 14 1.52 169 1.94 Chile Callaghan (2012) -36.97 -73.12 93045 1.13 123 2.16 Grtic Grntic 2.16 Grt Chile Callaghan (2012) -36.97 -73.12 142165 1.13 123 Chile Callaghan (2012) -35.84 -72.51 66 23 0.80 90.7 2.74 Chile Callaghan(2012) -35.86 -72.48 0.76 85.3 2.76 Chile Callaghan (2012) -34.61 -71.58 116 42 19 + 12 0.60 75.5 3.16 Chile Callaghan (2012) -33.88 -71.50 65 0.33 42.3 3.23 Chile Callagha (2012) -33.90 -71.49 32 0.34 45.2 3.24 Granitic i +14 Graniic Granitic Graniic Graniic Granitic Granitic Graniic Callaghan (2012) -32.94 -71.42 53 23 0.34 39.6 3.27 Callaghan (2012) -32.27 -71.41 75 31 0.24 30.1 3.11 Gr Chile Callaghan(2012) -32.27 -71.40 73238 0.23 30 3.04 G Chile Callaghan (2012) -32.08 -71.42 58,*28 0.20 25.9 3.0) Grtiti= Chile Callaghan(2012) -31.56 -71.42 61+ 16 0.15 18.8 2.81 Chile Callaghan (2012) -31.52 -71.42 16 = 4 0.17 20.8 3.06 Chile Callaghan (2012) -30.52 -71.66 71 0.12 12.6 4.00 Chile Callaghan (2012) -30.53 13 3.97 -30.55 -71.62 84=37 0.13 4 4.11 Chile Callaghan (2012) -30.57 -71.63 200 t 88 0.13 14 4.11 Chile Callaghan (2012) -29.65 7.5 4.00 Chile Callaghan (2012) -29.67 -71.16 Bugd fault system. Mongolia Lane Bonneville. UT. USA Southern Arava Cartir et al. (2002) 44.840 100.303 23 8 19.7 0.07 3 Bare ground Forested 33 33rate 3 3 Relief and erosion re rate3 erosion Relief and 3 3 Relief and 33rate G ii 3 3 rate 3 erosion Forested erosion Relief and erosion 3 Relief and rate 3 3 Forested 3 3 3 3 Relief 3 Relief and erosion and crosion Relief r ecrosion and Graniic itic Graniic iti Forested Forested 3 3 Forested 3 Mostly herbaceous. few 33 3 rt3r at3r 3 3 erosion Relief and 3 Herbaceous 3 Herbaceous 3 3 33rate 3 Relief 33 Relief and erosion and erosion Granitic 3 R ate Gr 3 at3a Relief and erosion rate and crosion rat. Relief and erosion rate 3 Relief 3 andeerosion 33 erosion 3 ic Graniic 3 3 Graniic 3 Graniic Granitic G 3 Graniic 3 3 Relief 3 Relief and Mostly 3 r 3 tiRelief Granitie 3 at3r 33 aderso reic trees Mostly herbaceous. some trecs Mostly herbaceous. some trees Graniic ate Mostly herbaceous. some Mostly herbacau.. some trees 3 Graniic Graniic Graniic -71.11 3 Graniic Granitic Callaghat(2012) Reliefradt Graniic Graniic Chile erosion Relief and erosion 33 Bare ground 3 R .liefand erosion Graniic 0.12 Bare ground trees Gran(i, Chile 74=30 andateerosion rate Graniic Chile -71.66 Relief rate 3 3 Forested Grtiti= Granilti, 29 and erosion 3 Granitic 29 Relief 3 herbaceous Mostly herbaceous. some trees Mostly herbaceous. sine trees Mostly herbaceous. few trees 3 3 Mostly herbaceous, some trees Mostly herbaceous. few Ite. Mostly bare ground 3 Mixture and esion of herbaceous and bae 3 at. 3 Gravel 1 Scarp modeling I Not vegetated I 0.78 Gravel I Scarp modeling I Orsses and shrbs 2 3.10 Sandy gravel I Scarp modeling I Not vegetated I 2.46 Paso Robles Formation (poorly consolidated sands) gravels 2 4 Coastal Coarse alluvial deposits I Scarp modeling I I I scrubland visible in photographs ofsit I Grass and sme sparse vegetation visible in 0.07 7.7 3.74 17 0.18 13.9 3.63 1 0.16 19.9 0.02 3,) groundcover ground 2 Gurvan Valley, Trssts Israel Ranges. CA, USA Colman and (1983) Enzel at4son t al. (1996) Gabt (2000) 33 39.625 -113.211 9f 29.612 34.983 3 34.693 -120,041 74 0.38 49.2 Sediment flux and estimated from gopher burrows and Sage 2 slope grasses and sparse Rtvytnd Fau Hankes r al. (1984) LA. CA. USA Scarp. 34.119 -118.131 160 0.33 46.2 2.60 trac tt .ovettiihl ateite and aerial images Lost River. ID. USA Haks (2000) 44.166 -113.870 10 0.31 28.3 1.06 Alluvial gravels I I Scar modeing Soarpttmotdtling tkt Lahstta. NV. USA Hanks andifilace (1985) 40.152 -117.925 11 0.14 18.8 0.83 Alluvial gravels I Scarp Lake Bonneville. Hanks r al. (1984) UT. USA modeling 2 2 2 satellite images 39.613 -112.299 11 0.24 29.5 I Gravels 0.61 Scarp modeling I Grasscs and shrubs Lower ttrraos 2 are farmed. upper terraces Santa Cruz sea cliffs. Hanks or at (1984) CA. USA 36.984 -122.127 110 0.72 79.8 2.62 2 Mudston am grassland; unlikely that the lower terraces Scarp modeling were 2 ever Forested (Rosenbloom and Anderson. 2004) Low Drum Mtnts.. UT. USA Haks ot at (1984) 39.650 -112.136 11 0.26 32.6 0.59 Alluvial gravels I Scarp modeling I SE Australia -36.605 149.493 40 0.74 86.9 0.69 Gratnodiorite 3 of Heimsathor al. (2005) -36.605 149.493 28 0.74 86.9 0.69 Granodiorite 3 depth-integrated soil production rates Huhso t(09 4.5 Heimsath e (2000) at' and CRN-derived rate SE 2 erosion Sediment flux Nunnock River. Australia and shadescale) Hillslope Laplacian Nunnock River. shrubs (sagehsh Schlerophyll forest (lightly forested) rm and 5 Schleophyll forest (lightly forested) depth-gradient Product Cherwell Basin. New Zealand Feather River. CA. USA htghes tt tO (2009) Hurstt ral (2012) -42.40 39.652 173.357 -121.312 88 1 30 80 1.42 (.01 116 117 Loss underlain 0.57 2.47 163 fluvial gravel Sediment flux from by terraces Granitic 3 deposits and slope ridetp Laplacian and CRN-drivtd erosion rates 2 Mixed conider forest 4 Feather River. CA. USA Hurst et al. (2013) Feather River, CA. USA Hurst e al.(2013) 39.724 39.710 -121.285 -121.262 48 88 18 1.10 33 1.15 113 2.43 ridgetop Laplacian and CRN-derived 3 Metavolanics 2 Mixed conifer forest 4 2 Mi Mixed conif 4 crosion rates 150 Granodiorite 2.30 Laplacian ridgetop and rsion rates 3 sediment flux estimated from s-it velocity . es Related Smokey Great Mountains, NC. USA Jungers ei al. (2009) LakeBonneville. UT. USA MattsonandBruhn (2001) -83.176 331 1.93 70 185 (determined 3 Quartzite 0.33 situ and 5 by in- meteoric 4 Deiduous forest "oBe) and hillslope Wasatch Fault Zone, UT, USA San Francisco 40.489193 Mn ason andBruhn 40.723594 -111.8232455 (2001) (2014). Springville Volcanic McGuire et al. -112.3262747 12 3 0.4097 28 I 0.4203 43.7 49.1 Al Al 0.82 gradient uvial shoreline I deposits A Iluvial 0.95 gravels I 35.390 Scarp modeling -111.570 0.44 40 49.3 1.36 Basaltic cinder cones I 34.190 and aerial 2 Pinyon pine, at lower assumed sagebrush age initial shape and elevation to Ponderos 4 of cinder cone LEM with (2014) 2 Grasses scrubland visilbe in satellite images (2084). east-c1ntral and shrubs Scarp modeling LEM with M uietal Volcanic Field in norhtem Arizona Field in 35.559 -109.570 0.50 50 56.4 1.70 Basaltic cinder cones elevation Ponderosa pin., gasibel oak, alligator assumed initial shape and age S barkjunipr. Douglas fir. Pinyon, sagebrush 4 ofcinder cone Arizona 3 pin. forests at higher and juniper in lower elevations Lodgepole pine. offrey Pandora., Medicine Lake Volcanic Field in northeastem 3McG.I et (2014). al. 41.640 -121.740 0.44 75 45.2 1.46and 4 LEM with assumed initial shape and age pin, sugar pine, white pine: western 4 and white fit at cnderconered of .f~ider on.higher clevation; I California Western juniper East Bay Regional Park. CA. USA AcKean st al. (1993) 37.974 -121.865 50 360 0.34 43.1 2 Marine shale 2.53 Emmet County. MI. USA NAsh (1980a) 45.575 -85.113 120 0.94 77.9 1.00 ohesionless send and gravel morain deposits Drum Mms., UT' Nash (1980b) 39.650 -112.136 4 0.26 32.6 0.59 Alluvial gravels USA flux sediment 2 slose at lower elevations .P" 0 3 and 2 Grse 5 2 hardwoods with white pine and hemlocks. Native Scarp modeling scattered 4 Law shrubs h and (sageb 2 shadescale) and spasm coverage of pine trees visible in 3 photographs of site3 and satellite and aerial Prairie grasses Hebge Lake. MT. USA Nash (1984) 44.701 -111.204 20 + 4 0.72 62.2 0.56 47.637 7.516 14 0.88 73 0.76 Staid and gravel Scarp I modeling Estimate Upper Rhine Graben. Niviee andarquis (2000) Germany Fluvial gravels and coarse sands I swarp from both modeling and estimating sediment at toe of from modeling Lathr p WIlls, NV. USA Pelleier andCln (2007) 36.690 -116.510 39 39 Looswvesicular scoris 0,710.1.43 0.07 80.9 143 numerical using initial and curet shape of cinder lapil cone cone . Age of 4 Spar. I Grasses vegetation I is -77 ka determined from 8 8diometric dating Alluvial shoreline Bonneville. UT. USA Lake Pelletier el al, (2006) 39.400 10 -113.700 0.20 25 0.72 swarps Scarp modeling (mostly sand and/or gravels) and shrubs 2 Measured soil thik1nes and known B nit. lava flow. Valles Cadets. Pelleterel NM. USA Banco al. (20 11) 36.840 -106.590 1 6 0.44 48.2 1.15 P88848 888 pin. age of lava flow to ak lest a nonlinear. R6volit 4 numerical gamble . and mixed scbland LEM and determine the best- conifer 3 forest fit D AI18ghey u P et al. (2012) Peron 39.971 -80161 PA. USA U.8A. Mesa, CA. Perrone , al. (2012) 35.923 -120.826 USA 100 + 8 0.98 105 0.53 124 19 0.18 28.4 2.41 Sandstone 2 Poorly consolidated 2 conglomerate 2 2 Decidious Laplacian 2 Grasses & 08k8 2 Gass&Ok and erosion rates 4Ridgat p and erosion rates Used LEM to parameter inversion estimate D through 8isib in photographs 3 W UsatAh Mt.. UT. Petitel al. (2009) 41.031 USA -111.894 1200 100 0.53 57.5 8.00 Gneiss88 G8s a Monte Carl. 3 and error minimization. Big Lost River Valley, ID. USA Pierce and Colman (1986) 43.809 -113.336 21 61 -20 0,28 28.3 0.98 Carbonate gravels and wads I forest Ridgetop Laplacian Scarp modeling of analytical solution w/error function 4 3 Patchy vegetation with Irmes. mixture grasses3 of sage. and a ofsite. satellite, and image$ aerial South-facing slopes ame shrub desert and north-facing slopes are 2 Praire grassland Intnsel8y sheared TCA.es ee VUSA8y' Reneau (1988) eported in Heimath e thrust sheets greenstoni' gr ywwck sandstone and chert (Franciscan of 37.863 -122.550 50 0.94 94.2 2.42 al. (2005) assemblage) 164 of landslide deposits. Colluvial infilling 2 Coastal grassland scrub and 2 Uoitys*A Southern Coast Range, OR Reneau (1988) reported in Heimsith et al. (2005) 38.047 -122.852 43.788 -123.931 30 0.93 99.1 2.53 + 13 1.7848 168 2 Quartz diorite and grantdierilr Colluvial infilling of 3dBspsits Iandslidr deoits 5 Bhbp pine forest 4 5 Conifer forest 4 Colluvial infilling Reneau ndDietrich (1991) Clecrwater Sincer C Reneaunt al. (1989) WA, USA Bodmin Moor. Comnwall, UK Riggins .1 al. (2011) Sullixce Cemk. OR ' Roering el al. (1999) USA Charwell River' SouthIsland, New Zealand Roering eal (2002) 47.660 -124.000 47 25 4.20 50.508 -4.439 394 7 1.96 43.463 -124.119 36+ 16 200 -42.450 173.357 120 + 80 142 Charwell River. South Island, New Roeringefal (2004) Zealand -42.450 Oullivan Cieek, OR Roeringetal (2007) u Sa 43.463 USA 51 173.357 -124.119 160 + 60 29 14 1.42 2.00 311 114 (68 I6 Tyre Formaion (Ecer graywacke sandstone) 1.72 Silts, sandstones and 0.73 Granite 1.96 0.07 2 sandstone) 2 Sediment flux estimates from dating hollow deposits (-10,000 yr timescale) and hillslope gradient 0 Westetn hemlock Pacific silver fie ferest 4 3 Ridgetop Laplacian and soil production rate 2 Grasses, but previously hazel and rik woodland 4 HillIslope and Lois underlain 168 1-96 Tyre Formation (Eocene graywacke sandstone) Laplacian vegetation-driven creep (-9k yrs) 2 Numerieal modeling by fluvial gravel terraces and Douglas fir, mixed conifer forest timscale of underlain by 0.57 Minimized error between modeled rates and measured erosion rates for non-inear erosion equation erosion - Loess fluvial gravel terraces 116 and hillslope gradient conglomerates Tyee Formation (Eocene graywacke flux oftransient hillslope 4 Podocarp and beech forest Podocarp and beech forest I laplacian erosion rate Douglas fir, mixed Ridgetop and coifer Rosenhloom and Anderson (1994) 36.984 -122.127 WindRiver Ruege. WY. USA Small e al (1999) 43.370 -109.750 Laguna Salads, Baja Califernia, Mexiu Spele, al. (2008) 32.075 -115.383 Qilian Shan. China Tupp9e)ell. (1990) 39.262 99.608 100 12 176 2 0.72 79.8 2.62 Mudstone 1.00 60.3 0.44 Granite and gneiss Numerical model with best-fit D 4 Ridgetop laplacian and erosion rates 0.04 33+ 17 0.11 8.3 2.02 Gravel terraces famed, upper terraces grassland. Likely never forested Non-vegeatated. alpine landscape non-vegetated. but vegetation near active fans and channel bars. Mostly some Finite-slupe end 0.9 0.4 0.3 forest terraces are Lower Santa Cruz. CA, USA 3 profile deinit-lupeca modeling tecbique Mostly non-vegetated. 11.9 Fanglomerates 1.39 bedrock, but blanketed with Basalt BlueMountien WA. USA Walther e al (2009) 46.148 -117.938 0.82 48 7 74.4 which sparse Scrpmsome 2.62 regetation visible in photographs of site Slope ofline between differential erosion rate (from glass age estimate and peak profile of Mazama ash) and differential lIens, controls erosion rate. coniferusnfet curvature. Susquehanna Shale HilIn Critical Observatory. PA. USA Atacama Desert. Decidious .st et al. (2014) 40.667 -77.903 67 + 56 -40 0.95 97.6 Shale 0.50 2 Meteoric "Be and billslope gradient 5 forest on hillslopes and hemlock and pine in vallev (Aaet al., 4 2013) Tbin cud Chile -24.130 -69.990 1.4+ .5 0.01 0.7 3.43 Granitic (Owen 2011) el al.. Ridgetop Laplacian and erosion rates Ridgetop Laplacian and erosion rates Desert I Ridgetop Laplacian and erosion rates Oak grsIand 3 2 Deer Atacama Desert. Chile Tistd -29.770 -71.080 16 2 0.07 7.8 3.38 Granitic (Owen ei al.. 2011) Blasingame, CA. USA This study 36.954 -119.631 23 3 0.26 38.7 2.23 Tonalite (Dixon el al., 2009) 3 Dritl Creek. OR. USA Tbic sindy 44.517 -123.844 167i37 2.55 223 1.95 Tyee Formation (Eocene graywacke sandstone) (Bierman el al., 200 1) 3 This study 35.622 -83.204 1 1.38 154 0.44 Quartzite (3aifan et al., 2003) 3 Ridgetop Laplacian and erosion rates 2 riduene feret 4 This 34.051 -116.934 21 0.59 72.9 2.07 and 3 Ridgelep Leplcin and erosion rates 2 Chapparral and oak identifed in satellite 3 Pimntly grnitic nd metcmerpbi reks (DiBise el al., 2010) 3 Great Smokey Mountains, NC. USA San Bernardino Mountains, CA. USA 19+ Ridgetop ned Laplacian erenienres 2 Primarily granitic rocks study 175 (quartz monzonite gneiss) (Binnie et al., 2007) San Gabriel Mountains, CA. USA This study 34.364 -117.992 71 12 R66 77.1 2.33 .l images. Chaparral, decidious and visible in sanellite and aerial images . Intensely sheared coniferous Dense ferrest conifers Ridgetop Laplacian and erosion raetes thrust of greenstone. greywacke sandstone and chert (Franciscan sheets Tennesse Valley. CA. USA This study 37.850 -122.550 174 21 0.89 84.4 2.45 assemblage)(Heimsath el al.. 1997) 165 Ridgetop Laplacian and erosion rates 2 Coastal grassland and scrub 4 WasatchMontain. UT, USA location Thin study 40.892 -111.865 used location literature and 83 3 0.45 1.03 31.0 Gneiss (Stock 2009) el a.. a region, a Ridgetop Laplacian and erosion rates lat/lon 2 for the study. I report the mean that best matched the site description. If multiple measurements were made for the was not able to be identified, I exact the standard error D D were included. I If raw standard standard deviation, may reflect the reported in the the report that instead. 'Rock category: 1= unconsolidated, 2 =sedimentary, 3 = Igneous/metamorphic. Technique category: I = Scarp modeling. 2 =Laplacian and erosion rates. 3 = Reliefand erosion rate, 4 = LEM. 5 = Colluvial flux and slope. forested, 4=frested. I = Arid/desert, 2 = grasslands/scrublands, 3 = Xgetation "Ifihe 'Uncertainties ae category: uncertainties range, or error. savannaMightly 166 estimates of Patchy vegetation with mixture trees sage, and grasses visible in photographs of site, satellite, and aerial image" calculated of and of 3