Open Collections

UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Forest and hydrogeomorphic processes in shallow landslide initiation zones Sakals, Matthew Egan 2010

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Item Metadata

Download

Media
24-ubc_2010_fall_sakals_matthew.pdf [ 33.75MB ]
Metadata
JSON: 24-1.0070976.json
JSON-LD: 24-1.0070976-ld.json
RDF/XML (Pretty): 24-1.0070976-rdf.xml
RDF/JSON: 24-1.0070976-rdf.json
Turtle: 24-1.0070976-turtle.txt
N-Triples: 24-1.0070976-rdf-ntriples.txt
Original Record: 24-1.0070976-source.json
Full Text
24-1.0070976-fulltext.txt
Citation
24-1.0070976.ris

Full Text

FOREST AND HYDROGEOMORPHIC PROCESSES IN SHALLOW LANDSLIDE INITIATION ZONES  by Matthew Egan Sakals  A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF  DOCTOR OF PHILOSOPHY in  THE FACULTY OF GRADUATE STUDIES (Forestry)  THE UNIVERSITY OF BRITISH COLUMBIA (Vancouver)  May, 2010  Matthew Egan Sakals, 2010  ABSTRACT Following a shallow landslide, hydrogeomorphic processes in the initiation zone respond to the discontinuities of soil depth, topographic expression, and hydrologic and forest conditions. The shallow landslides occurring at the study sites are episodic erosion processes punctuating periods marked by the deposition of hillslope material. Recovery of soil depth and topographic expression occurs through smallscale processes that infill the failure area. Rates of small-scale redistribution of hillslope material decline with time and are attributable to both vegetation re-establishment and diminished surface topographic variability. Understory forest vegetation in failure areas resemble the vegetation on adjacent undisturbed slopes within approximately 100 years of failure; re-establishment of the forest canopy may require several centuries longer. Stochastic elements of the surrounding forests strongly affect soil accumulation through influences on material transport and deposition. The central tendency of soil accumulation approximates a sigmoid curve with the majority of accumulation occurring within 100 years of failure. Soil depth on adjacent hillslopes positively influences soil accumulation in failure areas, but repeated shallow landslides can deplete hillslope materials from the contributing area. The sediment balance of weathering, storage, and evacuation strongly influences future cycles of failure and recharge. The frequency of preferential flow pathways in shallow landslide initiation zones was found to be spatially variable with fewer preferential flow pathways in the infilled soils of failure areas; thus landslide occurrence and subsequent infilling may negatively influence future slope stability. As a result of this study, recommendations for future research are made regarding the effects of forests on the regime of small-scale material transport processes occurring after shallow landslides, on the spatial and temporal development of preferential flow pathways in shallow landslide initiation zones, and on the long-term stability of sites of shallow landslides. ii  TABLE OF CONTENTS  ABSTRACT .................................................................................................................................. ii TABLE OF CONTENTS ............................................................................................................ iii LIST OF TABLES .....................................................................................................................viii LIST OF FIGURES .................................................................................................................... xi ACKNOWLEDGEMENTS ...................................................................................................... xvi CO-AUTHORSHIP STATEMENT........................................................................................ xvii 1. INTRODUCTION ...................................................................................................................... 1 References ................................................................................................................................ 12 2. PROCESSES OF HILLSLOPE ADJUSTMENT FOLLOWING SHALLOW LANDSLIDES ........................................................................................................................................................ 22 Introduction ............................................................................................................................. 22 Methods .................................................................................................................................... 24 Study areas ..................................................................................................................................... 24 Site selection .................................................................................................................................. 30 Site characterization ....................................................................................................................... 31 Results ...................................................................................................................................... 35 iii  Recent material transport volumes ................................................................................................. 35 Microtopography ............................................................................................................................ 41 Surface cover.................................................................................................................................. 50 Surface cover-microtopography interaction ................................................................................... 54 Processes ........................................................................................................................................ 57 Sediment size analysis.................................................................................................................... 59 Effects of timber harvesting ........................................................................................................... 60  Discussion................................................................................................................................. 62 Conclusions .............................................................................................................................. 68 References ................................................................................................................................ 70 3. SOIL ACCRETION FOLLOWING SHALLOW LANDSLIDES ...................................... 78 Introduction ............................................................................................................................. 78 Methods .................................................................................................................................... 80 Study areas ..................................................................................................................................... 80 Selection of candidate landslides ................................................................................................... 85 Site description ............................................................................................................................... 88 Data analysis .................................................................................................................................. 96 iv  Results ...................................................................................................................................... 99 Non-linear modeling ...................................................................................................................... 99 Conceptual model of stochastic infilling ...................................................................................... 113 Discussion............................................................................................................................... 118 Conclusions ............................................................................................................................ 133 References .............................................................................................................................. 135 4. SUBSURFACE SOIL VOIDS IN THE VICINITY OF SHALLOW LANDSLIDES ...... 144 Introduction ........................................................................................................................... 144 Methods .................................................................................................................................. 147 Study areas ................................................................................................................................... 147 Selection of sites .......................................................................................................................... 148 Site description ............................................................................................................................. 153  Results .................................................................................................................................... 158 Microtopography and preferential flow pathways ....................................................................... 158 Relative spatial distribution of hydrologic features ..................................................................... 159 Position of hydrologic features in the soil profile ........................................................................ 162 Temporal variability of preferential flow pathways ..................................................................... 165 v  Discussion............................................................................................................................... 169 Conclusions ............................................................................................................................ 174 References .............................................................................................................................. 176 5. THE EXTENDED TEMPORAL EFFECT OF SHALLOW LANDSLIDES ................... 183 Introduction ........................................................................................................................... 183 Increased convergence following a shallow landslide ........................................................ 184 Soil surface alteration ................................................................................................................... 185 Basal surface degradation ............................................................................................................ 185  Discontinuities of the subsurface hydrologic network ....................................................... 187 Geomorphic adjustments ..................................................................................................... 191 Land use and the geomorphic cycle of shallow landslides ................................................. 193 Conclusions ............................................................................................................................ 194 References .............................................................................................................................. 196 6. CONCLUSION....................................................................................................................... 204 Small-scale processes ............................................................................................................ 204 Material accumulation.......................................................................................................... 207 Preferential flow pathways ................................................................................................... 209  vi  Contributing areas ................................................................................................................ 211 Amalgamation ....................................................................................................................... 214 Land use ................................................................................................................................. 215 Study design ........................................................................................................................... 216 Final comments ..................................................................................................................... 218 References .............................................................................................................................. 220 APPENDIX A: MAPS OF THE SURVEYED STUDY SITES .............................................. 225 APPENDIX B: PHOTOGRAPHS ............................................................................................ 256 Characteristic photographs .................................................................................................. 257 Photographs of study sites .................................................................................................... 279 APPENDIX C: COMPARISON OF FOUR STUDY AREAS ............................................... 320 APPENDIX D: DATA TABLES ............................................................................................... 325  vii  LIST OF TABLES Table 2.1. Selected information of the study areas (Sutherland-Brown, 1968; Holland, 1976; Imaizumi and Sidle, 2005; Imaizumi et al., 2008; Environment Canada, 2008; Tutiempo.net ). ............................... 27 Table 2.2. Surface cover classes. ........................................................................................................................... 34 Table 2.3. Distribution of surveyed points and number of points with active material transport processes by topographic curvature............................................................................................................................. 44 Table 2.4. Number of points with active material transport processes by topographic curvature. Expected frequencies are in parentheses. .................................................................................................................... 45 Table 2.5. Volume and percent of recent material transport by topographic form. ............................................... 46 Table 2.6. Comparison of active and inactive points for four different surface cover classes. ............................. 51 Table 2.7. Observed and expected (in parentheses) frequencies of survey points for four surface covers by topographic quadrant. .................................................................................................................................. 55 Table 2.8. Observed and expected (in parentheses) frequencies of survey points with recent material transport for four surface covers by topographic quadrant. ........................................................................ 56 Table 2.9. Distribution of material transport volume and frequency by process for the three study regions including all ages of landslides. .................................................................................................................. 59 Table 2.10. Sediment size analysis for infilled soils of 18 failure areas ............................................................... 59  viii  Table 2.11. Percent of surveyed points with recent material transport at unlogged and logged landslide sites.............................................................................................................................................................. 62 Table 3.1. Selected information of the study areas (Sutherland-Brown, 1968; Holland, 1976; Muller 1977; Imaizumi and Sidle, 2005; Imaizumi et al., 2008; Environment Canada, 2008; Tutiempo.net). ................ 87 Table 3.2. Geomorphic rock mass classification system from Selby (1980) with both original and modified joint orientation scores. ................................................................................................................ 93 Table 3.3. Hillslope variables included in models................................................................................................. 94 Table 3.4. Candidate basic models and formulae. ............................................................................................... 101 Table 3.5. Details for candidate models, n=41 for all models, k is the number of parameters in the model. ..... 109 Table 3.6. Parameter estimates for all models. .................................................................................................... 110 Table 3.7. Model descriptions, R2, and AICc for each model including a geology variable. .............................. 111 Table 3.8. Model testing results, ranked by PMSE (squared mean residual + error variance), AICc Rank and Evidence ratio from the calibration have been included for ease of comparison. .............................. 112 Table 4.1. Selected information of the study areas (Sutherland-Brown, 1968; Holland, 1976; Muller 1977; Imaizumi and Sidle, 2005; Imaizumi et al., 2008; Environment Canada, 2008, Tutiempo.net). .............. 152 Table 4.2. Relative morphologic positions in landslide initiation zones. ............................................................ 158 Table 4.3. Statistical test results of differences between topographic variables for points with and without Type I pathways. Significant results are in bold-face type. ...................................................................... 160  ix  Table 4.4. Statistical test results of differences between topographic variables for points with and without Type II pathways. Significant results are in bold-face type. ..................................................................... 161 Table 4.5. Summary table of logistic regressions of preferential flow pathways between relative morphologic positions around shallow landslide head scarps. Positions are compared to hillslope (M – Margin of failure area, F – Failure area). Significance codes in parentheses: 0 < *** < 0.001 < ** < 0.01 < * < 0.05 < . < 0.1 < NS ....................................................................................................... 162  x  LIST OF FIGURES Figure 2.1. Study areas in British Columbia, Canada and on the Kii peninsula of central Japan. ........................ 26 Figure 2.2.Mean daily temperatures by month for study areas in British Columbia (North Coast and Haida Gwaii) and Japan (Owase). Dashed line represents an estimate from Owase station data to the elevation of the study sites based on 500 m of adiabatic cooling. .............................................................. 28 Figure 2.3.Mean monthly precipitation for study areas in British Columbia (North Coast and Haida Gwaii) and Japan (Owase). ......................................................................................................................... 29 Figure 2.4. Relation between observed volumes of recent material transport and time since failure. Volumes have been normalized by the perimeter length of the failure areas. ............................................ 37 Figure 2.5. Histogram of measured volumes of recent material transport. ........................................................... 38 Figure 2.6. Relation between the percent of surveyed points with evidence of recent material transport and the elapsed time since failure. Dashed line is a hand-fit envelop curve. ..................................................... 39 Figure 2.7. Relation between the volume of recent material transport and the percent of surveyed points with evidence of recent material transport. ................................................................................................. 40 Figure 2.8. Relation between the percent of surveyed points with evidence of recent material transport and the ratio of mean failure area soil depth to mean hillslope soil depth. ........................................................ 42 Figure 2.9. Relation between four classes of material transport volume and microtopography for surveyed points (1.5 m analysis scope). ..................................................................................................................... 47  xi  Figure 2.10. Relation between percent of doubly concave points in failure areas with time since failure. Dashed line is a hand-fit envelope curve to accentuate increasing value and decreasing variability with time. .................................................................................................................................................... 48 Figure 2.11. Relation between percent of doubly concave points and the percent of points with recent material transport. ....................................................................................................................................... 49 Figure 2.12. Relation between percent of vegetated points in failure areas and time since failure. Dashed line is a hand-fit envelop curve. .................................................................................................................. 52 Figure 2.13. Relation between percent of survey points with recent material transport and the percent of vegetated points in failure areas. ................................................................................................................. 53 Figure 2.14. Temporal trends of material transport processes for 33 shallow landslide initiation zones in British Columbia and Japan. ....................................................................................................................... 58 Figure 2.15. Surface cover distributions for unlogged and logged failure areas and adjacent hillslopes for four time periods following failure. ............................................................................................................ 61 Figure 2.16. Coarse woody debris contributes structure in the failure area that leads to both direct and indirect volume accretion. ........................................................................................................................... 68 Figure 3.1. Study areas in British Columbia, Canada and in central Japan. .......................................................... 82 Figure 3.2.Mean daily temperatures by month for study areas in British Columbia (North Coast, Haida Gwaii, and South Coast) and Japan (Owase). Dashed line represents an estimate from Owase station data to the elevation of the study sites based on adiabatic cooling.................................................. 83  xii  Figure 3.3.Mean monthly precipitation for study areas in British Columbia (North Coast, Haida Gwaii, and South Coast) and Japan. ....................................................................................................................... 84 Figure 3.4. Graphical representation of the method of determining convergence variable. Red and blue lines demarcate failure areas. See Appendix D for an explanation of letter notation. ................................ 91 Figure 3.5. Relation between soil depth and modified index of rock mass strength with predicted values for 8 study sites with missing hillslope soil depth data. Dashed lines indicate the 95% prediction intervals. ...................................................................................................................................................... 95 Figure 3.6. Relation between average infilled soil depth in shallow landslide failure areas and time since landslide for all 58 study landslides from three areas of coastal BC and south-central Japan. Models accepted at the conceptual review stage were fit to the calibration data. .................................................. 102 Figure 3.7. Recent (within 2 years) material transport volumes normalized by failure area and converted to an annual value for 32 failure areas and surrounding slopes. Heavy black bars represent an average of time and volume; thin bars indicate the temporal scope of these average values. The dashed line represents annual accumulation as predicted by the sigmoid curve of Figure 3.5 for an average-size failure area. ........................................................................................................................... 103 Figure 3.8. Residual plot for the sigmoid model. ............................................................................................... 104 Figure 3.9. Histogram of convergence variable for failure areas; 180º represents a planar slope, lower values are concave slopes, higher values are convex. ............................................................................... 106 Figure 3.10. Histograms of failure area lengths (scarLength) and widths (scarWidth)....................................... 107 Figure 3.11. Observed versus predicted soil depths for the primary sigmoid with geology variable. Grey line represents a 1:1 relation. .................................................................................................................... 113 xiii  Figure 3.12 Conceptual model component 1. Relation of probability of small-scale stochastic deposition on plot of ratio of failure area soil depth compared to hillslope soil depths through time. ....................... 115 Figure 3.13. Conceptual model component 2. Relation of probability of small-scale stochastic erosion on plot of ratio of failure area soil depth compared to hillslope soil depths through time. ............................ 116 Figure 3.14. Conceptual model component 3. Relation of probability of catastrophic erosion on plot of ratio of failure area soil depth compared to hillslope soil depths through time. ....................................... 117 Figure 3.15. Relation between mean hillslope soil depth and the joint orientation score of the geology variable. ..................................................................................................................................................... 124 Figure 3.16. Windthrown root system that has lifted a large block of competent gneissic bedrock with moderately spaced joints (0.3–1.0 m). ...................................................................................................... 130 Figure 3.17. Colluvial material accumulation upslope of coarse woody debris in a shallow landslide failure area in south-central Japan. Dry ravel of coarse material as depicted here was a minor process at study sites. ................................................................................................................................ 131 Figure 3.18. Relation between gradient of failure axis (axisSlope) and mean hillslope gradient. Solid line is a 1:1 relation. ......................................................................................................................................... 132 Figure 4.1. Dynamic cone penetrometer. ............................................................................................................ 150 Figure 4.2. Study areas in British Columbia, Canada and central Japan. ............................................................ 151 Figure 4.2. Examples of Type II preferential flow pathways. Frame A: A Type II preferential flow pathway in a head scarp. Note that the base of the flow pathway is on bedrock. Frame B: Two Type  xiv  II preferential flow pathways that were originally identified with the knocking cone penetrometer. Pipes are on either side of the pencil, note outflow from left feature. ...................................................... 155 Figure 4.3. Soil resistance profile measured with a dynamic cone penetrometer with a 2.5 kg hammer weight and a cone diameter of 30 mm. A Type I pathway is located between 0 and 10 cm below the surface and a Type II pathway is located at 35–45 cm below the surface. ............................................... 156 Figure 4.4. Relative frequencies of Type I pathway occurrence compared to depth within the soil profiles. .... 163 Figure 4.5. Relative frequencies of Type II pathways compared to depth within the soil profiles. .................... 164 Figure 4.6. Relation between percent of points in failure areas with Type I pathways and time of material accumulation since last landslide. Dashed line is an envelope curve for all study areas except for Japan. Values on extreme right are average percent of hillslope points with Type I pathways. ............... 166 Figure 4.7. Relation between percent of points in failure areas with Type II pathways and time of material accumulation since last landslide. Solid black line is a local regression trend line with a span of 0.75. Values on extreme right are average percent of hillslope points with Type II pathways................. 168 Figure 5.1. Landslide failure area with shallow soils compared to adjacent upslope hillslope locations. Approximate boundary of failure area is indicated by yellow annotation line. ........................................ 189 Figure 5.2. Orientation of coarse woody debris in 26 shallow landslide failure areas (n=115). ......................... 192 Figure 6.1. A hillslope in Japan with slope-perpendicular post-harvesting litter arrangement and subsequent small failures. ......................................................................................................................... 207  xv  ACKNOWLEDGEMENTS I am grateful to John Innes for supervising me during this project; I appreciate the freedom that he granted me in all aspects of this work. Many times during this study his words came back to me in times of need. Roy Sidle provided enthusiasm and a journey to Japan that deepened my understanding of hillslope processes. The remainder of the supervisory committee (Dan Moore, Oldrich Hungr, and Matthias Jakob) provided the right balance of doubt and encouragement through their comments and suggestions. Funding for this project was from Dave Wilford, Jim Schwab, and Tom Millard of the BC Ministry of Forests Research Section; Dave was also a willing and able reviewer. This project would not have been possible without the help of all of my field assistants; I am particularly indebted to my dad, Chris Davy, and Rav Bal. This work would not have been possible without the continued support of my wife and family.  xvi  CO-AUTHORSHIP STATEMENT The broad direction of this research was a result of discussion between Professor John Innes, Professor Roy Sidle and me. Further direction was gained from preliminary fieldwork with Prof. John Innes, Jim Schwab (BC Ministry of Forests), and supervisory committee member Matthias Jakob. The identification and design of the research questions, the study areas, and the study design was under my own direction with suggestions from the supervisory committee. I conducted the research including the selection of field sites, the development of the methods, and the collection of the data. The direction and scope of data analysis was defined by me, though I did receive some assistance regarding statistical techniques for Chapter 4 from Peter Ott (BC Ministry of Forests). Chapter 2 will be published with me as the first author and John Innes and Roy Sidle as co-authors. Chapter 3 will be published with me as first author and all members of the committee as co-authors. The contribution of the supervisory committee to Chapter 3 has resulted in the broadening of the analysis and the inclusion of the qualitative model as well as suggestions regarding the statistical modeling. Chapter 4 will be published with me as the first author and John Innes and Roy Sidle also listed. For Chapters 2 through 4, the contributions of the co-authors have not yet been realized but are expected as we progress towards publication. Chapter 5 was derived of my own notion, but related discussions with John Innes strengthened the manuscript and he will be a co-author. All coauthors will assist in the preparation and revision of these manuscripts but I will conduct and lead revisions and submissions for publication.  xvii  1. INTRODUCTION The objective of this research was to develop a better understanding of the forest and hydrogeomorphic processes acting in areas of shallow landslide initiation in the period following failure. Shallow landslides are long relative to their depth (Turner and Schuster, 1996) and are the most important mechanism of sediment transport in areas of steep relief and high precipitation (Rapp, 1960; Hack and Goodlett, 1960). Shallow landslides often trigger debris flows that transport materials down channels and can result in damage to infrastructure and natural resources as well as loss of human life (Innes, 1983; Jakob et al., 2000; Wilford et al., 2003; Sidle and Chigira, 2004). In an attempt to better manage natural resources in and below landslide prone terrain, past research has mainly focused on the initiation of shallow landslides; hydrologic, geomorphic, and forest processes have been studied with the intent of understanding landslide triggers and controls on debris flow initiation and runout (Jakob and Weatherly, 2003; Johnson and Wilcock, 2002; Jakob et al., 2005; Brayshaw and Hassan, 2009). Some hillslope areas have been observed to produce repeat landslides (e.g., Shimokawa, 1984). To experience subsequent failure, the failure area must be infilled with sufficient material. In this study the initiation zone refers to the broader area generating the failure, including the failure area with attendant head and side scarps, and also the adjacent slopes that contribute hillslope materials and water to the failure area through smallscale forest and hydrogeomorphic processes. Although limited information is available describing the rates of material recharge in these features (Shimokawa, 1984; Shimokawa et al., 1989; Dietrich et al., 1982; Smale et al., 1997), only cursory information exists regarding controls on the rates and processes of this material transfer (Dietrich et al., 1982; Sidle, 1987; Yamada, 1999). Hydrogeomorphology is an interdisciplinary science that includes the linkages among various hydrologic and geomorphic processes. Hydrogeomorphology was first defined as the study of landforms caused by 1  the action of water (Scheidegger, 1973). The concept of combined hydrologic and geomorphic processes may stem from earlier work by Horton (1945) describing the development of drainage basin topography. More recent additions by Okunishi (1994), Dunne (1997), and Sidle and Onda (2004) refined the concept of hydrogeomorphology. Sidle and Onda (2004) define hydrogeomorphology as: ‘an interdisciplinary science that focuses on the interaction and linkage of hydrologic processes with landforms or earth materials and the interaction of geomorphic processes with surface and subsurface water in temporal and spatial dimensions’. The field of study of this thesis is consistent with this quotation and I have therefore referred to hydrogeomorphic processes throughout this work. Hydrogeomorphology is a science of the surface of the earth (Scheidegger, 1973; Okunishi, 1994; Dunne, 1997; Sidle and Onda, 2004) and much of the earth's surface (~30%) is covered with forests (FAO, 2009). Due to the shallow soils and strong interactions with near surface processes, forests play an important role in the failure and recovery of shallow landslide initiation zones (Smith et al., 1986; Guariguata, 1990; Larsen et al., 1999). On forested hillslopes surrounding shallow landslide failure areas, forests can influence the water balance of the hillslope (Hewlett, 1982), fine and coarse organic matter inputs vary depending upon various forest conditions (DeLong et al., 2008), flora and fauna of the forest can promote the development of preferential flow pathways (Noguchi et al., 1999), and the binding action of roots has been found to resist slope failures (Abe and Ziemer, 1991; Roering et al., 2003; Hales et al., 2009). In this study, forests were found to affect processes of soil accumulation and material availability for geomorphic adjustments. Due to the complexity of the forests at the study sites, the spatial and temporal scales of the investigation, and the presence of the topographic, hydrologic, and forest discontinuity following shallow landslides (hillslope versus failure area), the linkage of the hydrogeomorphic processes with their controlling environment must be considered. The result of the forest environment is that the stochastic nature of the hydrogeomorphic processes is enhanced.  2  The effects of forests on various hillslope processes have been studied (Hack and Goodlett, 1960; Wondzell and King, 2003; Dorren et al., 2004); however knowledge gaps still remain. The occurrence of the relatively unstudied field of hillslope material transfer in shallow landslide initiation zones may be attributed to the interdisciplinary nature of the research: pure geomorphologists would preferably avoid forest ecology along with its stochastic variability (e.g., Band et al., 1993; Moorcroft et al., 2001); foresters interested in forest processes would likewise preferably avoid unstable terrain and the episodic disturbances that characterize initiation zones. Researchers interested in hazards to humans and infrastructure are generally better served studying the integration of hillslope processes by focusing on resultant downslope processes and conditions (Wilford et al., 2003; Wilford et al., 2004; Jakob et al., 2000). Thus, much remains to be studied regarding shallow landslides initiating in forested terrain and the associated post-failure adjustments. Humans influence forests through land use activities, and thus forest management can influence the hydrogeomorphic regimes of hillslopes. Knowledge of these regimes and their associated hazards becomes desirable when considering landslide risks. The practice of protection forestry blends the hydrogeomorphic and forest realm. Forests can provide protection to people and resources from hydrogeomorphic hazards including floods, debris floods, debris flows, snow avalanches and rockfalls (Sidle et al., 1985; Brang, 2001; Cheng et al., 2002; Sakals et al., 2006; Sidle and Ochiai, 2006). There are two general roles through which forests act to reduce hydrogeomorphic hazards: 1) forests promote the retention of organic and inorganic material in situ; and 2) the physical structure of forests resists, confines, and contains the transport of mobilized material thereby limiting the extent of destruction along the transport pathway as well as in the deposition zone. Management of the forest influence to enhance these functions is progressing (Sakals et al., 2006). This study uses the concept of zero-order basins. The terms „zero-order basin‟ and „hollow‟ are often used interchangeably, but definitions in the literature commonly distinguish between the two. Zero-order 3  basins refer to an unchannelized feature that may include noses, side slopes, and hollows, and encompasses the hillslope from the ridgeline down to the initiation of a first-order stream; a hollow is the axis of the hillslope depression (Hack and Goodlett, 1960; Thorne et al., 1987; Montgomery and Dietrich, 1989; Yamada, 1999). Tsukamoto et al. (1982), define the features from a more hydrological standpoint as hillslope units where flow lines converge on a hollow. Tsukamoto and Minematsu (1987) stated that a zero-order basin is an area in which saturated overland flow may appear during periods of storm runoff. Although this study includes zero-order basins, several of the study sites are in areas of divergent topography where prior to failure there existed no incipient basin (see Appendix A). At these sites, a convergent surface form was only achieved through failure of the soil mass. Further, many sites would only be considered convergent sites at the scale of 100 up to 101 m, as opposed to the typical 101 to 102 m scale. To eliminate confusion over expected surface configurations for the study sites, the more generic term of 'shallow landslide initiation zones' has been used throughout this study to refer to the specific failure area and the surrounding and contributing slopes. Topographic depressions on steep hillslopes have been observed to be more susceptible to shallow landslides than other landforms in both glaciated and non-glaciated regions (Dietrich and Dunne, 1978; Tsukamoto et al., 1982; Sidle 1984; Rollerson, 1992; Montgomery et al., 2000; Borga et al., 2002). In some areas, shallow landslides have been found to occur repeatedly from these features (Okunishi and Iida, 1981; Dietrich et al., 1982; Tsukamoto et al., 1982; Onda et al., 1992; Casadei et al., 2003). Theories have been proposed regarding the rates and processes of material infilling (Dietrich et al., 1982; Sidle, 1987; Yamada, 1999), and a few have attempted to measure soil accumulation following landslides (Shimokawa, 1984; Shimokawa et al., 1989; Dietrich et al., 1982; Smale et al., 1997). However, previous studies have only included a limited number of sites and factors that affect the rate of soil accretion have had only limited study (Shimokawa, 1984; Yamada, 1999).  4  While the terminology varies (swale, hollow, hopper, dale, dell, depression, zero-order basin, headwall, cove, rampa, wedge and colluvial-filled ravine [Dietrich et al., 1987]), the concept of hydrologic and geomorphic processes intermediate and overlapping between those of hillslopes and those of channels remains (Tsukamoto and Minematsu, 1987). The coupling of upslope areas with channels has been the focus of considerable research (e.g., Benda and Dunne, 1997; Gomi et al., 2001) but within hillslopes, the coupling of slopes with convergent features that fail preferentially has received less attention, particularly in glaciated terrain. Studies related to the accumulation of soil material in shallow landslides, or more generally to soil thicknesses on hillslopes, have generally been conducted in areas that did not experience Late Pleistocene glaciation (Shimokawa, 1984; Dietrich et al., 1995; Heimsath et al., 2001). This study uses a synoptic study design applied in three areas in British Columbia: the North Coast, in the vicinity of Prince Rupert; Haida Gwaii (formerly the Queen Charlotte Islands), with sampling centred in Rennell Sound; and south-west Vancouver Island, with study sites located in the Klanawa drainage. Due to the poor access to many of the study sites in the South Coast study area, some data collection was omitted and thus the South Coast study area is not included in Chapter 2. Several sites were also sampled in Japan in the south-central portion of Honshu Island on the Kii peninsula. The challenge of including sites from both British Columbia and Japan in this analysis concerns three factors, previous glaciation, differences of climate, and differences in forest conditions. All of the studied hillslopes in British Columbia were glaciated during the Late Pleistocene, and the glacial legacy results in differences of inorganic material availability that may affect infilling rates and the attendant small-scale hydrogeomorphic processes. Some hillslopes have abundant sediment in high topographic positions; other hillslopes were scoured and were left almost completely devoid of sediment (O‟Loughlin, 1972). Study sites in British Columbia were restricted to those of dominantly colluvial material to avoid the variability associated with morainal materials; however variation in colluvium and its provenance also exist. Some of the colluvium at the study sites may have been derived from morainal deposits rather than the 5  underlying bedrock, though the till itself may also have been derived from the local bedrock. In the Japanese study area on the Kii Peninsula of south-central Honshu, the area has not experienced glaciation and hillslope materials represent a balance of erosion and downslope material transport (Reneau and Dietrich, 1991; Stock and Dietrich, 2003). However, this balance may not be in equilibrium with the current climate (Heimsath et al., 2000) and the balance will exhibit disequilibrium at the temporal scale of the study of the study and the spatial scale of individual hillslope segments (Heimsath et al., 2001). The difference between the study sites from British Columbia and Japan at the spatial and temporal scale of this investigation may be less dramatic than anticipated as in each case the failures are bound by a shallow, predominantly colluvial soil. Differences in several other factors (e.g., geological material, adjacent soil depth, hillslope gradient) are incorporated into the general study design. The climates of coastal British Columbia and the Kii Peninsula of Japan are different; however they are not incomparable. Annual and monthly and even daily climate variables are not drastically different once temperatures are corrected for the elevations of the study sites. For instance, the extreme daily precipitation for the weather station in Owase, Japan (<60 km from all the study sites in Japan) for the period of 1973–2000 was 351 mm/day (Tutiempo.net); for the same period in the South Coast study area, the Port Renfrew climate station (<50 km from all South Coast study sites) had an extreme daily precipitation of 293 mm (www.climate.weatheroffice.gc.ca). Temperatures adjusted for adiabatic cooling allow the possibility of freezing during the winter months in all study areas (for more climate comparison, please see the Methods section of Chapter 3). The forests of the study areas in British Columbia are all relatively similar, but the forests of the study sites in Japan are dramatically different. The forests in Japan have been intensively managed for several rotations. Such plantation forests have less diversity and less stand structure and in general are less complex than the old-growth and relatively unmanaged second-growth forests of British Columbia (Puettman et al., 2009) (see Appendix B for photographs). The complexity of the forest environment was 6  found to result in differences of small-scale processes and this effect also influences the accumulation rates in failure areas and processes within the contributing areas to shallow landslides. In this thesis, I sought to answer the general question of: What attributes of shallow landslide initiation zones most strongly influence the rate and processes of material infilling failure areas? My study of the topic was exploratory in nature as previous studies have provided preliminary results on infilling rates (e.g., Dietrich et al., 1982; Shimokawa, 1984; Shimokawa et al., 1989; Smale et al., 1997) and have implied infilling processes (Dietrich et al., 1982; Shimokawa, 1984), but I am not aware of another analysis devoted to shallow landslide failure areas and the associated small-scale hydrogeomorphic processes. I employed two basic methods to address the question: 1) an inventory of recent material transport at study landslides combined with topographic and surface cover data; and 2) the development of an empirical model of soil depth accretion. The inventory of material transport was intended to provide process backing for the empirical soil accretion model by providing information regarding material transport and hydrogeomorphic linkages within the shallow landslide initiation zone (Chapter 2). The objective of the empirical model of soil depth accretion was to provide reasonable estimates of soil depths in failure areas and to capture the range of variability of infilling rates (Chapter 3). The final two chapters forming the body of this thesis were conceived during the investigation. The discovery of frequent soil voids (inferred to be pathways of preferential flow) led to questions regarding the spatial and temporal frequencies of the features (Chapter 4). In the final chapter of the body of the thesis (Chapter 5), a new hypothesis regarding shallow landslides is developed and discussed with reference to the preceding chapters. As a result of this exploratory study, recommendations for future research regarding shallow landslide initiation zones are included in the concluding chapter (Chapter 6). Following the Introduction (Chapter 1), the second chapter of this thesis investigates the small-scale processes of material infilling and controls on material transport. Shallow landslide failure areas infill by small-scale forest and hydrogeomorphic processes and frequency-magnitude relations indicate that small7  scale landslide processes can occur frequently (Brardinoni et al., 2003). The processes of material transport in failure areas following failure have been identified (Shimokawa, 1984; Dietrich et al., 1982) though they have been the subject of only limited data collection (Larsen et al., 1999). The more general study of sediment transport on hillslopes provides a body of literature to draw upon (e.g., Carson and Kirkby, 1972; Slaymaker, 1967; Croke et al., 1999; Roering et al., 1999; Merritt et al., 2003). The purpose of this component of the study was to provide process understanding to the empirical model and further the understanding of small-scale processes of infilling and two of their controlling factors: microtopography and ground surface cover/re-establishment of vegetation. In this study, I wanted to test whether certain topographic forms and/or surface cover types were more associated with material transport events and which of the two factors was more influential with respect to occurrences and volumes of material transport. The influence of these factors has been studied together in general (Dunkerly and Brown, 1999; Luce and Black, 1999; Gabet, 2003), but not in the hydrogeomorphic environment of shallow landslide initiation zones. These two features are also of a discontinuous nature at failure sites, with the discontinuity declining with time (i.e., topographic break becomes subdued and changes in vegetation become less abrupt). Topographic surveys of initiation zones and a spatially explicit inventory of both recent material transport and dominant ground surface cover were used for the analysis. In the third chapter, the development of empirical models of soil accretion is described. The specific hypothesis is that the rate of soil accretion can be predicted by a suite of variables characterizing the conditions of the initiation zone. Soil depth is one of the most important factors influencing the stability of a soil mass (Selby, 1993). Following the shallow landslides of this study, the remaining soils were of insufficient depth to fail, but over time soil accretion was expected to infill the failure area to the point where a subsequent failure would be possible (Shimokawa, 1984). Material accumulation will be dependent on both the production or availability of material and also the transport and deposition of 8  material into the failure area. Seven quantitative factors are included in an attempt to predict infilling rates: geology, climate, topography (including slope gradient and a measure of slope convergence), vegetation (including forest productivity and the occurrence of previous timber harvesting), the dimensions of the failure, the mean depth of soils on slopes adjacent to the failure, and the time elapsed since the previous failure. These factors are similar to the soil-forming factors of Jenny (1941); however it is important to note that this work does not focus on pedological soil formation, rather on the accumulation of unconsolidated soil material in failure areas. Debris accumulation rates have been found for small samples of landslides (e.g., Shimokawa, 1984; Shimokawa et al., 1989; Dietrich et al., 1982; Smale et al., 1997) and for a larger sample of gully systems (Jakob et al., 2005), though gullies may be expected to respond somewhat differently than initiation zones due to the presence of the active stream. To test the specific hypothesis, data was collected regarding the seven factors and accumulated soil depths at 58 failure areas. An empirical model to predict soil depth at these sites was the goal and thus the data were split into a calibration data set and a testing data set. Non-linear models were then assessed for conceptual and theoretical fit to empirical data, and for agreement with the volume of material transport events observed in Chapter 2. Despite the stochastic nature of soil infilling that is largely attributed to the forest influence and other factors, support is found for the central tendency to approximate a sigmoid curve. In response to the stochastic nature of infilling processes, a probabilistic model is introduced with components to describe small-scale deposition, small-scale erosion, and catastrophic erosion. In the fourth chapter, an opportunistic study regarding preferential flow pathways is presented. During the course of collecting data regarding material depths, the occurrence of a single strike driving the knocking cone penetrometer down into the soil >10 cm was at first unexpected but eventually came to be common. The presence of subsurface soil voids was not outside my conceptual constructions of hillslope hydrology, but the size and frequency of the features warranted further investigation. As the study sites have periodically high water tables (Kirkby and Chorley, 1967; Sidle, 1984), it was assumed that any 9  available conduit within the soil profile would be used for water transmission. Excavation of several of these features confirmed them as preferential flow pathways. In steep forested terrain, subsurface stormflow is often a major mechanism of storm runoff (Hursh and Brater, 1941; Anderson and Burt, 1990). Subsurface flow can occur by matrix flow, where water flows through the voids between solid soil particles, or by the more rapid preferential flow pathways, where water moves through larger and less constricted openings (Atkinson, 1978). Preferential flow pathways can consist of a number of subsurface soil features such as enlarged or decayed root pathways, animal and insect burrows, desiccation or freeze/thaw cracks, and large inter-aggregate pore spaces (Sidle, 1980; Noguchi et al., 1999). The length of individual preferential flow pathways have been reported as generally being <0.5 m (excluding soil pipes in peatlands [Holden, 2005]); however, subsurface flow on hillslopes is transmitted by a network of self-organizing features that respond to progressive soil wetting (Sidle et al., 2001). The frequency and spatial distribution of preferential flow pathways is not only important to stormflow generation (Brown et al., 1999), but is also thought to be important to the stability of hillslopes. Positive influence over slope stability is related to the rapid drainage provided by the network of preferential flow features in the hillslope soil (McDonnell, 1990; Onda et al., 2004). The negative influence on slope stability relates to the obstruction of flow within an established network. When a flow pathway is blocked and alternative network pathways are insufficient to convey the flow, there is a rapid elevation of hydrostatic pressures (Pierson, 1983; McDonnell, 1990; Onda et al., 2004). Such increases in hydrostatic pressures are associated with shallow landslide initiation (McDonnell, 1990; Sidle and Tsuboyama, 1992; Montgomery et al., 1997; Fannin and Jaakkola, 1999). Due to the infilling of landslide failure areas, the hydrologic pathways within the accumulated soils may be expected to vary from the soils outside the immediate failure area. The different transmissivities, defined as the (effective) saturated hydraulic conductivity integrated over depth (Band et al., 1993), of these adjacent soils could influence future slope stability. Little research has addressed this hydrologic discontinuity resulting from shallow landslides and subsequent soil depth recharge. Onda et al., (1992) is a notable exception; that investigation found a 10  greater ratio of large pores in the infilled soil of one failure area compared to the soils above the head scarp of the failure. The soil composition of the infilled soil would likely have a large influence over the relative hydrologic transmissivity, and Onda et al., (1992) state that the occurrence of previous landslides should be considered when analyzing future landslide initiation. With the spatial position of the knocking cone penetrometer profiles known both qualitatively and quantitatively, I was able to hypothesize and test queries of the spatial position of the preferential flow pathways. Specifically, I was interested in whether the surface and subsurface topographic configurations for profiles with preferential flow pathways could be distinguished from profiles without (including both the surface and subsurface). Further, I questioned whether the frequency of preferential flow pathways was different between the failure area and the surrounding hillslope as a significant difference could have implications to the subsurface hydrology of the sites and possibly induce subsequent failures. All the observations and analysis of the study culminate in a new hypothesis presented in the fifth chapter. The preceding chapters illustrate the complexities and sensitivities of initiation zones to perturbation, together the second and third chapters show that shallow landslide failure areas refill over time, and the fourth chapter provides some evidence that the hydrologic recovery may lag behind the soil depth recovery. As a result, a self-regulating cycle of shallow landsliding is proposed. When combined with land use activities, the initiation of a single shallow landslide on a previously stable slope may trigger a complex hillslope response and the initiation of a cycle of shallow landsliding. A sixth chapter presents concluding remarks regarding this study. Contributions made to the science of forest and hydrogeomorphic processes in shallow landslide initiation zones are discussed. Suggestions and considerations for further work are also included.  11  REFERENCES Abe K, Ziemer RR. 1991. Effect of tree roots on shallow seated landslides. General Technical Report PSW-GTR 130, United States Forest Service, Washington, D.C.  Anderson MG, Burt TP. 1978. The role of topography in controlling throughflow generation. Earth Surface Processes 3: 331–344.  Atkinson TC. 1978. Techniques for measuring subsurface flow on hillslopes. In: Hillslope hydrology. Kirkby MJ (ed.) John Wiley & Sons, Toronto.  Band LE, Patterson P, Nemani R, Running SW. 1993. Forest ecosystem processes at the watershed scale: incorporating hillslope hydrology. Agricultural and Forest Meteorology 63: 93–126.  Benda L, Dunne T. 1997. Stochastic forcing of sediment supply to channel networks from landsliding and debris flow. Water Resources Research 33: 2849–2863.  Borga M, Dalla Fontana G, Gregoretti C, Marchi L. 2002. Assessment of shallow landsliding by using a physically based model of hillslope stability. Hydrological Processes 16: 2833–2851.  Brang P. 2001. Resistance and elasticity: promising concepts for the management of protection forests in the European Alps. Forest Ecology and Management 145: 107–119.  Brardinoni F, Slaymaker HO, Hassan MA. 2003. Landslide inventory in a rugged forested watershed: a comparison between air-photo and field survey data. Geomorphology 54: 179–196.  12  Brayshaw D, Hassan MA. 2009. Debris flow initiation and sediment recharge in gullies. Geomorphology 109: 122–131.  Brown VA, McDonnell JJ, Burns DA, Kendall C. 1999. The role of event water, rapid shallow flowpaths and catchment size in summer stormflow. Journal of Hydrology 217L: 171–190.  Carson MA, Kirkby MJ. 1972. Hillslope Form and Process. Cambridge University Press: New York.  Casadei M, Dietrich WE, Miller NL. 2003. Testing a Model for Predicting the Timing and Location of Shallow Landslide Initiation in Soil-Mantled Landscapes. Earth Surface Processes and Landforms 28: 925-950.  Cheng JD, Lin LL, Lu HS. 2002: Influences of forests on water flows from headwater watersheds in Taiwan. Forest Ecology and Management 165: 11–28.  Croke, J., Hairsine, P., and Fogarty, P. 1999. Sediment transport, redistribution and storage on logged forest hillslopes in southeastern Australia. Hydrological Processes 13: 2705–2720.  DeLong SC, Sutherland GD, Daniels LD, Heemskerk BH, Storaunet KO. 2008. Temporal dynamics of snags and development of snag habitats in wet spruce–fir stands in east-central British Columbia. Forest Ecology and Management 255: 3613–3620.  Dietrich WE, Dunne T, 1978. Sediment budget for a small catchment in mountainous terrain. Z. Geomorph. N. F. Suppl. Bd. 29: 191–206.  Dietrich WE, Dunne T, Humphrey NF, Reid LM. 1982. Construction of sediment budgets for drainage basins. In Proceedings of a workshop on sediment routing in forested catchments, Swanson F, Janda R, Dunne T (eds). US Forest Service General Technical Report PNW-141. 13  Dietrich EW, Reiss R, Hsu M-L, Montgomery DR. 1995. A process-based model for colluvial soil depth and shallow landsliding using digital elevation data. Hydrological Process 9: 383–400. Dietrich WE, Reneau SL, Wilson CJ. 1987. Overview: “zero-order basins” and problems of drainage density, sediment transport and hillslope morphology. In Erosion and sedimentation in the Pacific Rim, Beschta RL, Blinn T, Grant GE, Swanson FJ, Ice GG (eds). IAHS Pub. 165: 39–48. Dorren LKA, Maier B, Putters US, Seijmonsbergen AC. 2004. Combining field and modelling techniques to assess rockfall dynamics on a protection forest hillslope in the European Alps. Geomorphology 57: 151–167. Dunkerley DL, Brown KJ. 1999. Banded vegetation near Broken Hill, Australia: significance of surface roughness and soil physical properties. Catena 37: 75–88. Dunne T. 1994. Hydrogeomorphology: an introduction. Transactions of the Japanese Geomorphological Union 15A: 1–4.  Fannin RJ, Jaakkola J. 1999. Hydrological response of hillslope soils above a debris-slide headscarp. Canadian Geotechnical Journal 36: 1111–1122.  FAO, 2009. http://www.fao.org/forestry/28808/en/, 2 December 2009.  Gabet EJ. 2003. Sediment transport by dry ravel. Journal of Geophysical Research 108(B1), 2049, doi:10.1029/2001JB001686.  Gomi T, Sidle RC, Bryant MD, Woodsmith RD. 2001. Characteristics of woody debris and sediment distribution in headwater streams, southeast Alaska. Canadian Journal of Forest Research 31: 1386– 1399. 14  Guariguata, M. 1990. Landslide disturbance and forest regeneration in the Upper Luquillo Mountains of Puerto Rico. Journal of Ecology 78: 814–832. Hack JT, Goodlett JC. 1960. Geomorphology and forest ecology of a mountain region in the central Appalachians, US Geological Survey Professional Paper, 347, 66 pp.  Hales TC, Ford CR, Hwang T, Vose JM, Band LE. 2009. Topographic and ecologic controls on root reinforcement, Journal of Geophysical Research 114: F03013. Heimsath AM, Chappell J, Dietrich WE, Nishiizumi K, Finkel RC. 2000. Soil production on a retreating escarpment in southeastern Australia. Geology 28: 787–790.  Heimsath AM, Dietrich WE, Nishiizumi K, Finkel RC. 2001. Stochastic processes of soil production and transport: Erosion rates, topographic variation, and cosmogenic nuclides in the Oregon Coast Range, Earth Surface Processes and Landforms 26: 531–552.  Hewlett JD. 1982. Principles of forest hydrology. University of Georgia Press, Athens, Georgia. 183 p.  Holden J. 2005. Peatland hydrology and carbon release: Why small-scale process matters. Philosophical Transactions of the Royal Society of London, Series A 363:2891–2913.  Horton RE. 1945. Erosional development of streams and their drainage basins: Hydrophysical approach to quantitative geomorphology. Geological Soceity of America Bulletin 56: 275–370.  Hursh CR, Brater EF. 1941. Separating storm-hydrographs from small drainage-areas into surface- and subsurfaceflow. Transactions of the American Geophysical Union 3: 863–871.  Innes JL. 1983. Debris flows. Progress in Physical Geography 7: 469–501.  15  Jakob M, Anderson D, Fuller T, Hungr O, Ayotte D. 2000. An unusally large debris flow at Hummingbird Creek, Mara Lake, British Columbia. Canadian Geotechnical Journal 37: 1109–1125.  Jakob M, Bovis M, Oden M. 2005. The significance of channel recharge rates for estimating debris-flow magnitude and frequency. Earth Surface Processes and Landforms 30: 755–766. doi: 10.1002/esp.1188  Jakob M, Weatherly H. 2003. A hydroclimatic threshold for shallow landslide initiation at the North hore Mountains of Vancouver, BC. Geomorphology 57: 137–156.  Jenny H. 1941. The factors of soil formation. McGraw-Hill: New York.  Johnson AC, Wilcock P. 2002. Association between cedar decline and hillslope stability in mountainous regions of southeast Alaska. Geomorphology 46: 129–142.  Kirkby M, Chorley R. 1967. Throughflow, overland flow and erosion. Bulletin of the International Association of Science and Hydrology 12: 73–80.  Larsen MC, Torres-Sanchez AJ, Concepcion IM, 1999. Slopewash, surface runoff and fine-litter transport in forest and landslide scars in humid-tropical steeplands, Luquillo Experimental Forest, Puerto Rico. Earth Surface Processes and Landforms 24: 481-502.  Luce CH, Black TA. 1999. Sediment production from forest roads in western Oregon. Water Resources Research 35: 2561–2570.  McDonnell JJ. 1990. The influence of macropores on debris flow initiation. Quarterly Journal of Engineering Geology 23: 325-331.  16  Merritt WS, Letcher RA, Jakeman AJ. 2003. A review of erosion and sediment transport models. Environmental Modelling and Software 18: 761–99.  Montgomery DR, Dietrich WE. 1989. Source areas, drainage density and channel initiation. Water Resources Research 25: 1907–1918.  Montgomery DR, Dietrich WE, Torres R, Anderson SP, Jeffner JT, Loague K, 1997. Hydrologic response of a steep, unchanneled valley to natural and applied rainfall. WRR 33: 91-109.  Montgomery DR, Schmidt KM, Greenberg HM, Dietrich WE. 2000. Forest clearing and regional landsliding. Geology 28: 311–314.  Moorcroft PR, Hurtt GC, Pacala SW. 2001. A method for scaling vegetation dynamics: the ecosystem demography model (ED). Ecological Monographs 71:557–586.  Noguchi S, Tsuboyama Y, Sidle RC, Hosoda I. 1999. Morphological characteristics of macropores and distribution of preferential flow pathways in a forested slope segment. Soil Science Society of America Journal 63: 1413–1423.  Okunishi K. 1994. Concepts and method of hydrogeomorphology. Transactions of the Japanese Geomorphological Union 15A: 5–18.  Okunishi K, Iida T. 1981. Evolution of hillslopes including landslides. Transactions of the Japanese Geomorphological Union 2: 291300. O’Loughlin CL. 1972. An investigation of the slope stability of the steepland forest soils in the Coast Mountains, southwest British Columbia. PhD dissertation, The University of British Columbia, Vancouver. 147 pp. 17  Onda Y, Mori A, Shindo S. 1992. The effects of topographic convergence and location of past landslides on subsurface water movement on granitic hillslope. Journal of Natural Distaster Science 14: 4558.  Onda Y, Tsujimura M, Tabuchi H. 2004. The role of subsurface water flow paths on hillslope hydrological processes, landslides and landform development in steep mountains of Japan. Hydrological Processes 18: 637–650.  Pierson, TC. 1983. Soil pipes and slope stability. Quarterly Journal of Engineering Geology 16: 1–11.  Puettmann KJ, Coates KD, Messier C. 2009. A critique of siliviculture: Managing for complexity. Island Press, Washington DC.  Rapp A. 1960. Recent development of mountain slopes in Karkevagge and surroundings, northern Scandinavia. Geografiska Annaler Series A 42: 65–200.  Reneau SL, Dietrich WE. 1991. Erosion rates in the southern Oregon Coast Range: Evidence for an equilibrium between hillslope erosion and sediment yield. Earth Surface Processes and Landforms 16: 307–322.  Roering JJ, Kirchner JW, Dietrich WE. 1999, Evidence for nonlinear, diffusive sediment transport on hillslopes and implications for landscape morphology. Water Resources Research 35: 853–870.  Roering JJ, Schmidt KM, Stock JD, Dietrich WE, Montgomery DR. 2003. Shallow landsliding, root reinforcement, and the spatial distribution of trees in the Oregon Coast Range. Canadian Geotechnical Journal 40: 237–253.  Rollerson TP. 1992. Relationships between landscape attributes and landslide frequencies after logging: Skidegate Plateau, Queen Charlotte Islands. BC Ministry of Forests Land Management Report 076. 18  Sakals ME, Innes JL, Wilford DJ, Sidle RC, Grant GE. 2006. The role of forests in reducing hydrogeomorphic hazards. Forest Snow and Landscape Research 80: 11–22.  Scheidegger AE. 1973. Hydrogeomorphology. Journal of Hydrology 20: 193–215.  Selby MJ. 1993. Hillslope materials and processes, 451 pp., Oxford University Press, Oxford.  Shimokawa E. 1984. A natural recovery process of vegetation on landslide scars and landslide periodicity in forested drainage basins. In Symposium on Effects of Forest Land Use on Erosion and Slope Stability O’Loughlin CL, Pierce AJ (eds). East-West Center, University of Hawaii, Honolulu; 99107.  Shimokawa E, Jitouzono T, Takano S. 1989. Periodicity ofshallow landslide on Shirasu (Ito pyroclastic flow deposits) steep slopes and prediction of potential landslide sites. Transactions of the Japanese Geomorphological Union 10: 267–284.  Sidle RC. 1980. Slope stability on forest land Pacific NW Extension Publication PNW 209. Oregon State Universtiy, Corvallis.  Sidle RC. 1984. Shallow groundwater fluctuations in unstable hillslopes in coastal Alaska. Zeitschrift f¨ur Gletscherkunde und Glazialgeoleologie 20: 79–95.  Sidle RC. 1987. A Dynamic Model of Slope Stability in Zero-Order Basins. International Association of Hydrological Sciences, Publication 165: 101–110.  Sidle RC, Chigira M. 2004. Landslides and debris flows strike Kyushu, Japan. EOS Transactions American Geophysical Union 85: 145–156.  Sidle RC, Noguchi S, Tsuboyama Y, Laursen K. 2001. A conceptual model of preferential flow systems in forested hillslopes: evidence of self-organization. Hydrological Processes 15: 1675–1692. 19  Sidle RC, Ochiai H. 2006. Landslides: Processes, Prediction, and Land Use. American Geophysical Union: Washington DC.  Sidle RC, Onda Y. 2004. Hydrogeomorphology: an overview of an emerging science. Hydrological Processes 18: 597–602. Sidle RC, Pearce AJ, O’Loughlin CL. 1985. Hillslope Stability and Land Use. American Geophyscial Union: Water Resources Monograph 11.  Sidle RC, Tsuboyama Y. 1992. A comparison of piezometric response in unchanneled hillslope hollows: coastal Alaska and Japan. Journal of the Japanese Society of Hydrology & Water Resources 5: 3–11.  Slaymaker O. 1967. Patterns of erosion in small catchments in mid-Wales. Unpublished Ph. D. thesis, University of Cambridge.  Smale MC, McLeod M, Smale PN. 1997. Vegetation and soil recovery on shallow landslide scars in Tertiary Hill Country, East Cape Region, New Zealand. New Zealand Journal of Ecology 21: 3141.  Smith RB, Commandeur PR, Ryan MW. 1986. Soils, vegetation, and forest growth on landslides and surrounding logged and old-growth areas on the Queen Charlotte Islands. BC Ministry of Forests: Victoria, Land Management Report 41.  Stock JD, Dietrich WE. 2003. Valley incision by debris flows: evidence of a topographic signature. Water Resources Research 39. DOI:10.1029/2001WR001057.  Thorne CR, Zevenbergen LW, Burt TP, Butcher DP. 1987. Terrain analysis for quantitative description of zero-order basins. In Erosion and sedimentation in the Pacific Rim (Proceedings of the Corvallis  20  Symposium, August, 1987) International Association of Hydrological Sciences Publication No. 165: 121–130.  Tsukamoto Y, Ohta T, Noguchi H. 1982. Hydrological and geomorphological study of debris slides on forested hillslope in Japan. IAHS Publication 137: 89–98. Tsukamoto Y, Minematsu H. 1987. Hydrogeomorphological Characteristics of a Zero-Order Basin In: Erosion and Sedimentation in the Pacific Rim. IAHS Publication No. 165. International Association of Hydrological Sciences, Washington, DC. p 61–70.  Turner AK, Schuster RL (eds.). 1996. Landslides: investigation and mitigation. Transportation Research Board, National Research Council. National Academy Press: Washington DC; Special Report 247. Wilford DJ, Sakals ME, Innes JL. 2003: Forestry on fans: a problem analysis. Forestry Chronicle 79: 291–296.  Wilford DJ, Sakals, ME, Innes JL, Sidle RC, Bergerud WA. 2004. Recognition of debris flow, debris flood and flood hazard through watershed morphometrics. Landslides 1: 61–66. Wondzell SM, King J. 2003. Postfire erosional processes in the Pacific Northwest and Rocky Mountain regions. Forest Ecology and Management 178: 75–87.  Yamada S. 1999. The role of soil creep and slope failure in the landscape evolution of a head water basin: field measurements in a zero order basin of northern Japan. Geomorphology 28: 329–344.  21  2. PROCESSES OF HILLSLOPE ADJUSTMENT FOLLOWING SHALLOW LANDSLIDES1 INTRODUCTION The initiation of shallow landslides from sites of repeat failure has been modeled as a combination of stochastic and deterministic components with the degree of material infilling as the deterministic component (Sidle, 1987). With greater infilling, the initial volume of future failures will be larger, which results in a greater likelihood of debris flow (Millard, 1999; Takahashi, 2007). If debris flow is not triggered at the time of the initial shallow landslide, failed materials can accumulate in other hillslope or channel locations such as headwater stream channels (Benda and Dunne, 1997). Accumulated material can eventually be mobilized by a subsequent mass movement (Sidle and Ochiai, 2006; Brayshaw and Hassan, 2009). Without quantifying and understanding the small-scale processes that infill failure areas, we lack information regarding one fundamental control on shallow landslide (and debris flow) frequency and the sediment cascade operating on steep hillslopes. It is frequently assumed that following a shallow landslide, the failure area will be immune to failure for several hundred to thousands of years, but much of evidence to support this contention comes from the Oregon Coast Range where a selection of arguably stable features was used to make assumptions about erosion rates (Reneau et al., 1989; Reneau and Dietrich, 1991). More generally, the period of immunity lasts only as long as material are of insufficient depth to fail (Shimokawa, 1984), and then only requires a sufficient hydroclimatic event (Jakob and Weatherly, 2003). In other regions, the period of immunity may  1  A version of this chapter will be submitted for publication. Sakals ME, Innes JL, Sidle RC. 2010. Processes of hillslope adjustment following shallow landslides  22  be as short as decades (Shimokawa, 1984; Shimokawa et al., 1989). Following failure, small-scale processes begin to infill the failure area; the efficacy of soil accumulation is variable and dependent on site and climatic conditions (Adams and Sidle, 1987). The oversteepened slopes of the head and side scarps retrogress and the eroded material is transported downslope into the failure area by small landslides and surface erosion (Dietrich et al., 1982). Where sufficient material exists on the contributing hillslope, the failure area will infill but the rate of infilling is expected to slow as local topographic variation diminishes. Thus, landslide occurrence and subsequent hillslope adjustment will result in a more convergent surface morphology (cross-slope concavity). The associated enhancement of the topographic and hydrologic concentration to the failure area will increase the likelihood of subsequent failure (Dietrich and Dunne 1978; Tsukamoto et al., 1982; Shimokawa, 1984; Sidle 1987; Montgomery et al., 2000). Previous observations suggest that microtopography (scale of <3 m) may also be important to areas of soil erosion and deposition (Adams and Sidle, 1987; Kirkby et al., 1998; Bryan, 2000). Ground surface cover (i.e., extent of vegetative cover or lack thereof) is another strong factor in various types of sediment transport (Hack and Goodlett, 1960; Slaymaker, 1967; Carson and Kirkby, 1972; Larsen et al., 1999; Gabet and Dunne, 2002). In vegetated areas, disturbance of the vegetative cover commonly results in a strong increase in material movement (Slaymaker, 1967; Norman et al., 1995; Gabet et al., 2003; Ludwig et al., 2005). On steep forested hillslopes, shallow landslides strip the vegetation cover and thus post-failure geomorphic adjustments occur with less restriction during the unvegetated period. However, in humid forests, vegetation generally recovers quickly (Smith et al., 1986; Bovis and Pellerin, 1998; Larsen et al., 1999). A better understanding of the affects of ground surface cover and microtopography on material transport in shallow landslide initiation zones will provide a better basis for the conceptual and physical modeling of material dynamics in failure areas in the period between landslides; quantification of material transport  23  volumes will provide support for empirical models of soil depth accretion. Coincident with soil depth accretion beyond the immunity depth is a return of landslide hazard (Shimokawa, 1984; Sidle, 1987). In this study, we used surveys of shallow landslide failure areas and geomorphically contributing slopes that included the occurrence and volumes of recent material transport events and dominant surface cover types for surveyed points. Surveys were then used to develop local digital elevation models and microtopographic data were determined at each survey point. These data were then used to test the hypotheses that both concave and unvegetated features would be correlated with greater material transport. The data were further explored in an attempt to assess which factor – microtopographic form or dominant surface cover – was more influential on the presence of material transport events. Regional differences on material transport regimes and the effects of timber harvesting are also discussed.  METHODS Study areas Study areas were located in two areas of British Columbia: the North Coast, in the vicinity of Prince Rupert; and Haida Gwaii (formerly the Queen Charlotte Islands), with sampling centred in Rennell Sound. A third study area was located on the Kii peninsula of south-central Japan (Figure 2.1, Table 2.1). A comparison of climate between the three study areas was done to provide the context under which infilling processes act. The climate of the two areas in British Columbia was predicted using ClimateBC (Hamann and Wang, 2005; Wang et al., 2006) and was found to be similar (Figures 2.2 and 2.3). Climate data for Owase, Japan during the period of 1973 to 2000 was attained online (Tutiempo.net). Although the station was in reasonable proximity to the studied landslides (<60 km from furthest study site), the station was located just above sea level (27 m) while the lowest study site was at 150 m elevation, the other sites were at 440, 490, 500, 1 060 and 1 100 m elevation. Adiabatic cooling was estimated for the sites of approximately 500 m elevation by applying the environmental lapse rate (Oke, 1987). No 24  information is available regarding the increase in precipitation with altitude, but Imaizumi and Sidle (2005) reported a range of 1 600–4 500 mm (average of 3 300 mm) for the Miyagawa Dam area where three of the study sites were located. It is important to note that although the climate of the Kii Peninsula may be warmer during the summer months, that the average daily temperature during the winter months allows the possibility of freezing, particularly for the higher elevation sites; therefore needle ice and freeze-thaw processes remain possible during the winter months for all three study areas. Precipitation at the Owase station has different annual patterns than in British Columbia, but the wettest and driest months in each study area are comparable. Differences in precipitation intensities are more dramatic with daily values three times greater at the Owase station compared to the higher of the two values in coastal British Columbia (Table 2.1). A qualitative comparison of other measures of the study sites from the three study areas is included in Appendix C. Both study areas in British Columbia were glaciated during the Late Pleistocene. The landslide sites from Japan have not experienced glaciation and thus lack the paraglacial legacy of the study areas in British Columbia. However, steep slopes with colluvium of <2 m depth are common in all three study areas.  25  Figure 2.1. Study areas in British Columbia, Canada and on the Kii peninsula of central Japan.  26  Table 2.1. Selected information of the study areas (Sutherland-Brown, 1968; Holland, 1976; Imaizumi and Sidle, 2005; Imaizumi et al., 2008; Environment Canada, 2008; Tutiempo.net ). North Coast, BC  Haida Gwaii, BC  Kii Peninsula, Japan  Landslides sampled  10  16  7  Late Pleistocene glaciation  yes  yes  no  pyroclastic rocks, shale, siltstone,  schist, sandstone, slate, shale,  sandstone, conglomerate, argillite  mudstone  Geology  granite, gneiss, shale, slate steep slopes rising from sea level to  steep slopes rising from sea level to  500 m  600 m  1 600–3 350 mm  2 400–3 700 mm  ~4 000 mm  118.2 mm /day (Prince Rupert, BC)  79.5 mm /day (Sandspit, BC)*  351 mm /day (Owase, Japan)**  colluvial soils of <1 m in depth  colluvial soils of <1 m in depth  colluvial soils of 0.5-1.5 m in depth  second and old-growth coniferous  second and old-growth coniferous  coniferous plantations  Predominant natural disturbance type  shallow landslide, debris flow  shallow landslide, debris flow  shallow landslide, debris flow  Predominant human disturbance type  forest harvesting  forest harvesting  forest harvesting  yes  yes  yes  Typical morphology Annual precipitation at study sites Extreme rainfall at nearest climate station Typical soils Forests  Landslide inventory available  steep slopes with local relief of 300 m  *The distribution of precipitation on Haida Gwaii is strongly influenced by orographic effects. The climate station (Sandspit) is on the leeward, eastern side of the major mountain range of the island archipelago; the study area was located on the windward, western side of the range. Precipitation is accentuated on the western side of the islands where incoming Pacific frontal systems first encounter the land mass. Therefore, precipitation at the Sandspit station underestimates precipitation at the study sites (Karanka, 1986). **Highest noted value during data compilation of period from 1973–2000.  27  Figure 2.2.Mean daily temperatures by month for study areas in British Columbia (North Coast and Haida Gwaii) and Japan (Owase). Dashed line represents an estimate from Owase station data to the elevation of the study sites based on 500 m of adiabatic cooling.  28  Figure 2.3.Mean monthly precipitation for study areas in British Columbia (North Coast and Haida Gwaii) and Japan (Owase).  29  Site selection Previously compiled landslide inventories were used to select candidate shallow landslide initiation sites and were available for both study areas in British Columbia (Schwab, unpub.) as well as in Japan (Imaizumi et al., 2008). A series of aerial photographs of different ages were used to bracket the date of failure, or the failure was associated with a substantial storm that occurred during the period between two successive photo sets. The error of this method is assumed to be <5 years. Data accompanying the inventories were used to select candidate landslides based on the feasibility of access. Some failures were visible on aerial photographs, in fresh condition or indicated by distinctive strips of forest cover. Some sites included in this study were not included in the aerial-photograph based inventories and were not visible on 1:20 000 scale aerial photographs because of their small size and the crowns of adjacent trees; these sites were found during field traverses of failure-prone hillslopes. An effective lower boundary to failure area width was 5 m. A range of elapsed time since failure was sought so that the progression from recent failure could be analyzed. The distribution of time since previous failure was designed to have a skewed distribution with greater sampling of the most recent failures and similar distributions in each study area. Young sites were expected to have higher rates of material accumulation and possibly different processes of material infilling due to the recent disturbance; therefore, a skewed sample was warranted. Several criteria were used for site selection. Assuming that the initiation zone was not too steep or dangerous to work on, the first requirement was that the failed material was colluvium. Colluvium was assumed to be more directly related to the current conditions, slopes, and underlying geology than materials such as till or fluvioglacial sediments. Till was avoided because of the variability in textures and differences in consolidation that could result in different root penetrations and resistances to surface erosion processes that could introduce extra variability into the study. All failure areas had failed down to 30  bedrock, with material remaining after failure <20 cm depth; only a few sites were omitted from the study because of deeper remaining soils. No specific size criteria were applied, so long as the initial failure was of the full soil depth. Landslides included in precompiled inventories were already dated; for the remaining sites and those where the date was suspected of being incorrect, dendrochronological techniques were used to determine the date of failure (e.g., Wilford et al., 2005). Tree scars were the preferred morphologic feature for dating followed by abrupt growth changes of trees on the margins of the failure that developed due to changes in nutrient, light, and water availability. Dating trees established on the recharged soil material produced a minimum age only, as the period of accumulation prior to establishment was unknown. Other methods, such as the in situ accumulation of litter material, were used to determine the age of failures <5 years. The temporal error of time since failure at each study site was assessed independently and was dependent on the number, quality, and consistency in the dendrochronological samples and other evidence; estimated errors are presented in Appendix D. Finally, landslides caused by excessive anthropogenic disturbance, such as major drainage alterations by roads or trails, were excluded. Areas of previous timber harvesting were acceptable as long as there were no obvious upslope diversions of drainage pathways or areas of extensive soil or other disturbance.  Site characterization Site surveys of topography and infilling processes were conducted at 33 landslide sites. The entire failure area and the immediately adjacent slopes of the initiation zone (typically 5-10 m beyond the side scarps and 10–15 m above the head scarp) were surveyed. The extent of the surveys was limited to 10–15 m upslope for several reasons. Firstly, in many cases, the upslope topographic expression did not indicate geomorphic or hydrologic concentration; in some cases, the upslope form was divergent. Second, the 31  distance to the slope divide was typically several hundreds of metres with terrain and vegetation making the completion of a detailed survey difficult. Thirdly, no small-scale processes of infilling found during the preliminary fieldwork or during the course of the study progressed downslope more than 10 m. Thus the geomorphic contributing area for recent processes did not extend far upslope. For these reasons, the detailed survey was limited to the immediately adjacent slopes. Unfortunately, this also meant an incomplete characterization of the hydrological contributing area; however, with the relatively planar configuration of most of the studied hillslopes, the determined watershed area would have been of questionable use considering the effects of preferential flow pathways not entirely following surficial relief (Hutchinson and Moore, 2000). An irregular network approximating a 3 m grid with additional points to capture slope transitions was used (see Appendix A). At each station, the relative x, y, and z coordinates were determined with a theodolite and a laser range finder, the dominant surface cover was determined (Table 4.2), and recent (within 2 years) processes of organic and inorganic material transport in the vicinity of the point were noted. Recent material transport was identified by the lack of substantial non-transported litter or vegetative cover and limited surface alteration by rain drop impact or surface washing; some fresh deposits were still under-consolidated. The temporal error associated with seasonality of processes and misclassification is estimated at 25%, not significant to affect the results. A variety of materials were found to be transported including fine and coarse woody debris, litter, and fine and coarse sediment. Evidence of material transport was classified as one of the following: dry ravel, landslide, litterfall, needle ice, slope wash (including soil pedestals), or windthrow (see Appendix B). The volume of material transported was estimated with measurements of the length, width, and depth of the deposition or erosion if the deposit was >1 cm in depth. Lengths and widths were measured to the nearest 0.05 m, depths were measured to the nearest 0.01 m. A threshold on the magnitude of small-scale processes is that they must not have progressed directly through the failure area, in such case they would be considered a separate landslide event. In the case of uneven deposit geometry of small-scale processes, 32  an average value was estimated. Care was taken not to count both the erosion and deposition of the same material. The inverse volume of soil pedestals (the volume of eroded soil) was combined with the volume of slope wash. For windthrow, the volume of the wood (and sediment) delivered to the scar area was estimated by assuming a cylindrical shape of an average diameter. Windthrow was measured only if it was deposited into the failure area. Where overlapping deposits were suspected, the stratigraphy was inspected to search for separating layers of material or organic veneers indicating distinct events. The degree of „over-printing‟ at the study sites is not known as the initial transport of material could have occurred by one process and then a subsequent process may have transported the material prior to the site investigation (Jakob et al., 2005). Thus, the importance of dry ravel or freeze-thaw processes in preparing material for subsequent transport may be under-represented. The temporal error of observations of recent material transport is expected to be similar at all study sites. The volume of material transport was normalized by the length of the perimeter of the failure according to the following:  where scarLength and scarWidth are the length and width of the failure area measured to the nearest 0.1 m. The location of the lower boundary of the failure area was subjective, but commonly determined by a slight narrowing of the disturbed width where debris begins to be focused into the flow pathway; for debris avalanches, an increase in the width of the disturbed area was used to determine the lower bound of the failure area. The final term (2 years) was included to produce an annual value as material transport events were recorded up to 2 years of age. Normalizing transport volumes by the area of the failure area would produce an average depth accumulation from recent processes that is more easily interpreted; however, since material recharge was expected to be derived from the unsupported material along the 33  failure margins, observed volumes were normalized by the perimeter length of the failure area (length of head and side scarps). Fifty-one soil samples were collected from 18 different failure areas. Samples were dried at 105ºC for 24 hours then sieved through 32, 16, 8, 4, and 2 mm screens; organic matter was included in weighed mass. Clasts >64 mm were not included in the samples. Hillslope topography was described using measures derived from 0.5 m digital elevation models (DEMs) developed from the site survey data. Inverse distance weighting was used to interpolate elevation (z) values between survey points. Slope, aspect, and cross-slope and longitudinal curvatures were extracted from the DEMs at each survey point using GRASS (Neteler and Mitasova, 2004) and an analysis scale of 1.5 m (including the 0.5 m raster cells containing the point and those adjacent). Longitudinal and crossslope curvatures were calculated for both the surface and subsurface with the r.param.scale command which fits a quadratic equation to the analyzed surface by least-squares. For the analysis of material transport volumes, the number of instances favoured grouping the data by slope form as positive (convex) or negative (concave) values of both cross-slope and longitudinal curvatures. Table 2.2. Surface cover classes. Surface cover  Description  Soil  Bare soil; rotting leaves; decomposed organic material  Rocks  Bedrock; loose, angular colluvium  Vegetation  Shrubs; grasses; ferns; mosses  Woody debris  coarse and fine woody debris  34  Several initiation zones contained failure areas representing separate failure events, where the most recent failure did not completely cover a previous failure. For these sites, the date of failure of both surfaces was estimated and the two were included as separate failures. Statistical analysis was completed using the statistical package R (Ihaka and Gentleman, 1996). Chisquare tests of homogeneity were used to assess the difference between observed and expected values for microtopographic and surface cover descriptors. A one-tailed t-test with α=0.05 was used to test for a difference in activity between logged and unlogged sites.  RESULTS Recent material transport volumes The surveys of the 33 landslide initiation zones included 1412 points. Active (within the past 2 years) geomorphic processes were found at 860 points. Volumes of material transport were calculated at 186 points for the identified processes; the total surveyed volume of recent material transport was 20.48 m3. The normalized material transport volumes observed in or contributing to the failure areas decreased with time since failure (Figure 2.4). A non-linear (negative exponential) regression was fitted to the normalized annual accretion data (m3/m/yr). The points of the North Coast and Haida Gwaii study areas seem reasonably scattered. However, the sites from Japan all plot below the non-linear regression line. Residuals between the observed and predicted normalized volumes did not have statistically significant relations with site-level variables, including descriptors of the geology (Selby, 1980), slope of the slide axis, hillslope convergence, or average fall and winter precipitation values (for British Columbia sites only).  35  Nine of the initiation areas surveyed had no evidence of recent material transport; others were extremely active, with up to nearly 6 m3 of material transported in the last two years (0.06 m3/m/yr of normalized material transport). Most material transport events were <0.1 m3; the largest individual material transport event observed was 2.8 m3 (see Appendix D for a sample of observation data). The histogram of volumes of recent material transport shows a strongly skewed distribution (Figure 2.5). As a result of the strongly decreasing volumes of recent material transport with increasing time since failure, the analysis of associated factors is strongly influenced by the occurrence of several relatively high magnitude material transport events. To compensate for this, we have instead used the percent of surveyed points with evidence of recent material transport at each landslide initiation zone (Figure 2.6). The plot still shows a temporal effect with decreasing variability over time, as shown by the declining hand-fit envelop curve. The effect of the large magnitude occurrences are removed, and activity such as the dispersed slope wash deposits observed at the study sites in Japan are better represented. Though three sites from Japan have some of the highest values of recent activity, there does not appear to be a strong regional pattern. Several sites of <50 years since failure have a small percent of recently active points, specific site conditions such as shallow surrounding soils, downslope transport of material through the failure area (and thus no accumulated deposits), and the limited temporal extent of the processes captured by the <2 year time frame are considered responsible. No pattern is evident between the volume of recent material transport and the percent of points with active processes (Figure 2.7). Processes of relatively high magnitude and low frequency were often present at just a single observation point, a small percent of the points in the entire initiation zone, and thus the two measures are decoupled. Over the temporal scale of 2 years and the spatial scale of the surveys (~3 m), the small-scale processes are assumed to be independent. Therefore, a higher proportion of active points will generally be associated with greater material transport. 36  Figure 2.4. Relation between observed volumes of recent material transport and time since failure. Volumes have been normalized by the perimeter length of the failure areas.  37  Figure 2.5. Histogram of measured volumes of recent material transport.  38  Figure 2.6. Relation between the percent of surveyed points with evidence of recent material transport and the elapsed time since failure. Dashed line is a hand-fit envelop curve.  39  Figure 2.7. Relation between the volume of recent material transport and the percent of surveyed points with evidence of recent material transport.  40  Microtopography Figure 2.8 presents the relation between the percent of points with evidence of recent material transport and the ratio of failure area soil depths to soil depths on the surrounding hillslope. Although the data show significant scatter, there is an overall trend of decreased activity as the ratio of soil depths approaches unity. Some explanation of the points plotting with low soil depth ratios and small percents of recent transport is warranted. One site (RenFace2 – Haida Gwaii) likely had more active processes occurring, but the morphology of the site would result in the immediate downslope ravel of the material thus removing evidence from many of the surveyed points. A second site (Gregory3B – Haida Gwaii) with a small percent of active sites and small soil depth ratio has a negatively influenced soil depth ratio because it is partially bounded by a more recent failure (Gregory3A – Haida Gwaii) and this has not been captured in the calculation of hillslope soil depths. A more correct soil depth ratio for this site would be larger and would move this site closer to the predicted norm. The site from Japan with forty percent activity and a soil depth ratio of near zero (Waka12-2A) was one site where material was not accumulating in the failure area but was instead passing directly through down into the transport zone and channel domain. Due to the lack of material trapping, the percent of active points was negatively influenced. Above the trend line, one of the North Coast sites (Silver1) was a very recent landslide, visited within weeks of failure. The other North Coast site plotting well above the line (PwrLineB) may have a scale effect as it was a very small landslide with few surveyed points.  41  Figure 2.8. Relation between the percent of surveyed points with evidence of recent material transport and the ratio of mean failure area soil depth to mean hillslope soil depth.  42  The majority (56%) of all surveyed points were concave in both the cross-slope and longitudinal directions. When drawn on a plot of cross-slope curvature against longitudinal curvature, the majority of points occur in the topographically defined third quadrant (Q3) where both curvatures are negative (Figure 2.9 and Table 2.3). A chi-square test of homogeneity indicated that a significant difference exists among the four quadrants and that the third quadrant is the only quadrant where the total was larger than the expected value. Failure areas are generally concave features, so the large percent of doubly concave sites is statistically, but not physically, unexpected. Figure 2.9 is a graphical representation of curvature values; the greatest concentration of points with recent material transport is close to the origin, but more frequently in Q3. Testing also indicated a significant difference for the number of points with recent material transport (active), and Q3 is the only quadrant with observed values greater than the expected values (Table 2.3). Again, this is to be expected as the survey was primarily of deposits that would preferentially occur at sites in the topographically defined third quadrant. Separating the points with active material transport by only longitudinal or cross-slope morphology revealed a significant difference in both cases between the concave and convex morphologies (Table 2.4). The greater activity of the concave points could be explained by the predominantly depositional nature of activity occurring preferentially at concave up points. Any erosion by concentrated runoff would also occur at Q3 points. Further, processes at convex sites may transport material to other sites, namely proximal third-quadrant sites. The distribution of points with larger material transport volumes (0.1 m3 < Volume < 1 m3 and Volume > 1 m3) could not be tested statistically as the sample size was insufficient. However, the total material transport in each quadrant is summarized in Table 2.5, where it is readily apparent that the majority of material transport occurs in the third quadrant.  43  Plotting the percent of points in the failure area at each site that are in Q3 (or doubly concave) against the time since failure shows a trend of decreasing variability and increasing mean of percent of points in Q3 with time (Figure 2.10). The two points plotting below the hand-fit envelop curve both have many points with concave curvatures, but no doubly concave points. Decreasing topographic variability and local relief, as described by an increasing percent of points in Q3 and an increasing soil depth ratio, may suggest a relation with the decline in material transport volumes and the percent of points with recent activity with time since failure. However, plotting percent of points in the failure area at each site that are in Q3 against the percent of points with recent material transport does not produce any discernible pattern (Figure 2.11). Thus, it could be inferred that the decreasing topographic variability is a resultant feature rather than a forcing feature as the local relief appears to be. In both plots, study areas appear to be well mixed, with no conspicuous differences between the regions. Table 2.3. Distribution of surveyed points and number of points with active material transport processes by topographic curvature.  Quadrant  Cross-slope curvature  Longitudinal Observed points Inactive Active curvature (Expected) (Expected) (Expected)  1  convex  convex  174 (353)  169 (152)  5 (22)  2  concave  convex  227 (353)  210 (198)  17 (29)  3  concave  concave  863 (353)  718 (751)  145 (111)  4  convex  concave  148 (353)  133 (129)  15 (19)  Chi-square test statistic p-value  44  992  34.2  <0.001  <0.001  Table 2.4. Number of points with active material transport processes by topographic curvature. Expected frequencies are in parentheses. Cross-slope Slope Morphology  Inactive  Active  Longitudinal Inactive  Active  concave  928 (950) 162 (140) 851 (881) 160 (130)  convex  302 (280)  Chi-square test statistic p-value  45  20 (42)  379 (349)  22 (52)  14.4  23.8  <0.001  <0.001  Table 2.5. Volume and percent of recent material transport by topographic form.  Longitudinal curvature  Percent of all points  Volume of recent material transport (m3)  Percent of recent material transport volume  Percent of all points with recent material transport  convex  convex  8.9  2.0  9.9  7.4  2  concave  convex  15.2  0.7  3.5  12.6  3  concave  concave  55.5  15.4  75.4  63.2  4  convex  concave  8.6  1.6  8.1  8.8  Quadrant  Crossslope curvature  1  46  Figure 2.9. Relation between four classes of material transport volume and microtopography for surveyed points (1.5 m analysis scope).  47  Figure 2.10. Relation between percent of doubly concave points in failure areas with time since failure. Dashed line is a hand-fit envelope curve to accentuate increasing value and decreasing variability with time.  48  Figure 2.11. Relation between percent of doubly concave points and the percent of points with recent material transport.  49  Surface cover Vegetation was the dominant surface cover (Table 2.6) when all study sites were analyzed. It was significantly more common than the other surface covers as it was the only one of the four surface cover types to have more points observed than expected. Considering active processes of material transport, the chi-square test statistic indicated that a significant difference existed within the four surface cover types. Two groups were revealed: soil and rocks both had more active points than expected and vegetation and woody debris both had fewer active points than expected. Plotting the percent of points in the failure area at each site that have surface covers of vegetation or woody debris against the time since failure shows a increasing trend with time (Figure 2.12). The one site lying below the hand-fit envelop curve was a recently logged site in Japan where low-lying vegetation was likely destroyed during felling and yarding activities. Limbing was not conducted prior to yarding, so excess slash (fine woody debris) was not present at the site. The plot of the percent of active points against the percent of failure area points with vegetation or woody debris has a relatively strong pattern with an R2 = 0.33 (Figure 2.13). One outlier is from the North Coast, this site may be affected by scale as it consists of just six points, four of which had recent processes. Three study sites from Japan have both high levels of activity and relatively high levels of vegetation and woody debris coverage compared to the trend line. This could be a reflection of the high intensity precipitation in Japan resulting in sediment transport despite vegetation presence, but the other Japanese sites do not display such a relation plotting close to the trend line or below it. More likely, this is the presence of stochastic processes that have not been entirely designed around. Removal of the single outlier from the North Coast resulted in an R2 = 0.42. The additional removal of all the Japanese sites resulted in an R2 = 0.53.  50  Table 2.6. Comparison of active and inactive points for four different surface cover classes.  Surface cover  Observed points (Expected)  Inactive Active (Expected) (Expected)  RX  276 (386)  226 (244)  50 (32)  S  224 (386)  154 (198)  70 (26)  VEG  809 (386)  764 (715)  45 (94)  WD  233 (386)  219 (206)  14 (27)  Chi-square test statistic  624.3  131.6  p-value  <0.001  <0.001  51  Figure 2.12. Relation between percent of vegetated points in failure areas and time since failure. Dashed line is a hand-fit envelop curve.  52  Figure 2.13. Relation between percent of survey points with recent material transport and the percent of vegetated points in failure areas.  53  Surface cover-microtopography interaction The relation between slope morphology and surface cover as found by Chi-square tests of homogeneity found surface covers of all types in frequencies reasonably close to the predicted value for each topographic quadrant (Table 2.7). One exception is the larger frequency of woody debris in the first quadrant and the smaller frequency in the third quadrant. An explanation for this could be the progressive burial of woody debris at the Q3 points and the erosive protection offered by the woody debris acting to preserve the topography of Q1 points. This is the reciprocal of the pattern of soil surface covers where a smaller frequency than expected at Q1 points and larger frequency than expected at Q3 points was found. A comparison of Q3 points with surface covers of vegetation and woody debris with the percent of recently active points did not result in an increase of explained variance relative to Figure 2.13; the R2 value fell to 0.17. When considering just points with recent material transport (Table 2.8), rocks and soil were less frequent on longitudinally convex sites and more common on longitudinally convex sites. This is consistent with longitudinally concave sites being sites of preferential deposition and that longitudinally convex sites are relatively more stable as indicated by the frequency of vegetation as the dominant surface cover. Combined with the results of Table 2.4, it would appear that the longitudinal slope form is more influential than surface cover as a forcing agent of slope adjustment in the post-failure period.  54  Table 2.7. Observed and expected (in parentheses) frequencies of survey points for four surface covers by topographic quadrant.  Quadrant  Cross-slope curvature  Longitudinal curvature  RX  S  VEG  WD  1  convex  convex  19 (26.9)  13 (23.2)  80 (80.9)  43 (24.0)  2  concave  convex  25 (37.7)  26 (32.5)  127 (113.3)  39 (33.6)  3  concave  concave  169 (143.9)  133 (124.0)  420 (432.8)  107 (128.3)  4  convex  concave  19 (23.6)  28 (20.3)  71 (71.0)  18 (21.1)  Chi-square statistic  43.12  p-value  <0.001  55  Table 2.8. Observed and expected (in parentheses) frequencies of survey points with recent material transport for four surface covers by topographic quadrant.  Quadrant  Cross-slope curvature  Longitudinal curvature  RX  S  VEG  WD  1  convex  convex  3 (10.4)  6 (16.2)  100 (76.7)  6 (11.7)  2  concave  convex  9 (15.7)  22 (24.4)  120 (115.3)  22 (17.6)  3  concave  concave  22 (21.1)  42 (32.7)  143 (154.7)  25 (23.6)  4  convex  concave  24 (10.8)  20 (16.8)  63 (79.3)  12 (12.1)  Chi-square statistic  50  p-value  <0.001  56  Processes Processes were not evenly distributed with time since failure (Figure 2.14). Small landslides comprised 84% of recent material transport in the first 20 years following the site-evacuating failure. Older study sites (20–49 yrs, 50–99 yrs, and >100 yrs) had a more balanced distribution of volumes transported by each process. Slope wash remained relatively consistent over each time class. Needle ice, dry ravel and windthrow processes contributed less than landslides and slope wash except for the second age category where the combined processes transported 36% of the total volume. The volumes of material transported by the three process types were different across the regions (Table 2.9). Japan had the highest relative rate of transport by slope wash and both the Japanese sites and the sites on Haida Gwaii had high percentages of transport by the „Other processes‟ category. In Japan, dry ravel was a common process; in Haida Gwaii, freeze-thaw soil heaving by needle ice was common, but this could be affected by the season (late winter) of fieldwork. In addition to the volume of material transported, Table 2.8 includes the percent by process of all surveyed points with recent material transport. The similarity of the study sites from British Columbia can be seen with the high percent of landslide volume, but much lower percent of occurrence. In Japan, the dominance of slope wash is evident in both volume and percent of occurrence.  57  Figure 2.14. Temporal trends of material transport processes for 33 shallow landslide initiation zones in British Columbia and Japan.  58  Table 2.9. Distribution of material transport volume and frequency by process for the three study regions including all ages of landslides. Percent of transport volume (%) Region  Percent of process occurrence (%)  Landslide  Slope wash  Other processes  Landslide  Slope wash  Other processes  Japan  15.7  60.7  23.6  1.5  90  8.5  North Coast  64.3  35.4  0.2  20.6  35.2  44.2  Haida Gwaii  50.2  19.3  30.5  17.5  22.2  60.3  Sediment size analysis The sediment size analysis shows that the fine (<2 mm) component of the sampled soils is approximately equivalent to the sum of the larger sizes (Table 2.10). The 32, 16, and <2 mm sizes all strongly dominated some samples. All sizes, with the exception of the <2 mm size, were absent in some samples. Table 2.10. Sediment size analysis for infilled soils of 18 failure areas Sieve size (mm)  Mean ± standard deviation (%)  Maximum component of sample (%)  Minimum component of sample (%)  32  4.9±14.3  71.1  0.0  16  9.2±14.7  70.2  0.0  8  9.0±8.5  42.6  0.0  4  9.8±6.8  29.5  0.0  2  11.9±6.9  41.7  0.0  <2  54.5±25.5  98.1  0.4  59  Effects of timber harvesting The effect of timber harvesting on the surface covers of both failure areas and surrounding hillslopes through time is presented in Figure 2.15. The distribution of surface covers is relatively similar between the logged and unlogged failure areas. The greater percent of woody debris in the logged failure areas is a result of just two failure areas; both were from Japan and dominated by fine woody debris from the Hinoki trees at the site. The hillslopes were less similar between the logged and unlogged sites. The difference may be explained by the site disturbance due to falling and yarding at the logged sites. Logged sites of >100 years since failure were not found and so are absent from this data. Logged and unlogged sites within 50 years of failure were not statistically different in the volume of recently transported material. However, when the stochastic effects of material volumes are removed and only the percent of survey points with recent material transport are considered, logged sites had a significantly higher rate of active points (Table 2.11).  60  Figure 2.15. Surface cover distributions for unlogged and logged failure areas and adjacent hillslopes for four time periods following failure.  61  Table 2.11. Percent of surveyed points with recent material transport at unlogged and logged landslide sites. Percent of surveyed points with recent processes Unlogged  6.1  Logged  19.2  p-value  0.014  DISCUSSION Hillslope adjustment following shallow landslides decreases with increasing time since failure. Regardless of initial condition, the incision into the hillslope by slope failure results in a renewed cycle of material transport focused into the failure area. The decline is approximated with a negative exponential curve occurring coincidently with the dampening of topographic variability and the establishment and expansion of vegetative cover. Through soil transport processes (Roering et al., 1999) and preferential deposition and erosion in failure areas and surrounding slopes, hillslope forms are moderated resulting in a recovery of the ratio of failure area soil depths to hillslope soil depths. Concave hillslope features are believed to be the most common failure sites because they channel convergent surface and subsurface flow as well as surficial materials (Dietrich and Dunne, 1978). Following failure, the relaxation of oversteepened scarp slopes propagates surface gradient adjustments upslope in the form of knick-point erosion (Kirkby, 1971). This is consistent with the findings of this study that the incoming material will be preferentially deposited in micro-topographic depressions and, through time, results in a decreased level of hillslope topographic complexity (Figure 2.8) (see also Kirkby et al., 1998). Just as the convergence of the hillslope acts to extend the area capable of contributing material to landslide failure 62  areas beyond the head scarp and side scarps, there is a microtopographic contributing area that accompanies each surveyed point. The higher frequency and greater volume of material transport at Q3 points identifies the active nature of concave hillslope features at the scale of <5 m. Combining this finding with the decrease in local relief over time (Figure 2.8) suggests that material accumulates at Q3 points. Eventually the general elevation of the failure area rises and engulfs convexities, if they exist, resulting in a characteristically evenly concave form to the failure area and surrounding slopes. Despite infilling, the site will continue to be at least a minor concavity and thus have a much higher likelihood of subsequent failure (Tsukamoto et al., 1982; Dietrich et al., 1982; Sidle, 1987; Crozier et al., 1990). The morphology of underlying geologic structures contributes variability but it appears to be largely overridden in older failures. The dominantly gently-concave morphology of the failure areas approximates the more mature topographic form of nose-and-hollow (or gully-and-interfluve) topography. Such topographic form in areas of shallow soil is rarely found in British Columbia as the landscape is generally still in a period of adjustment from Late Pleistocene glaciation (Martin et al., 2002; Campbell and Church, 2003). Instead, it is characteristic of landscapes that have not been glaciated, such as the study area in Japan (Tsukada, 1982), the Oregon Coast Range (Montgomery et al., 1997), or the central Appalachians (Hack and Goodlett, 1960). Several of the Japanese failures had surfaces formed almost entirely of Q3 points, even immediately after failure. However, not all landslides in mature topographies occur in the axis of failures; those that occur on more planar slopes, such as side slopes, may have hillslope material transport regimes similar to that of the study landslides in British Columbia. Processes of transport are dominated by small landslides and slope wash. However, each study region had a different regime of material transport when both the volumes and frequencies of occurrence were considered. Landslides transport the majority of material soon after failure and the decay of the local 63  relief between failure areas and the surrounding hillslopes is largely attributable to their work. Considering all sites, landslides were the dominant process with respect to the volume of material transported at the North Coast and Haida Gwaii sites, but this is as result of several occurrences of relatively high magnitude, low frequency events. The potential for individual infilling processes with volumes >1 m3 is restricted to landslides and the windthrow of large stems into the failure area. Regarding the frequency of occurrence, in Table 2.9, landslides were the least common for all study areas. The „Other processes‟ category was dominantly dry ravel at the Japanese sites and dominantly freezethaw processes and needle-ice in Haida Gwaii. Reasonable similarity is shown between the North Coast and Haida Gwaii sites excepting that the majority of fieldwork in Haida Gwaii was done under a high pressure system during late winter. Consequently, conditions for needle ice formation on bare soil areas were good and the process was commonly identified early in the day prior to its melting. In the case of frequent precipitation, needle ice would not have been so frequently observed. Weathering by physical and chemical processes can also accumulate material from beneath the basal plane of the failure that could be mobilized in a subsequent landslide. Field observations in this study found this to be an uncommon process except in cases of weak, highly fractured bedrock that could be lifted by the root systems of windthrown trees following tree establishment and windthrow in the failure area. Given the ubiquitous presence of vegetation on the slopes surrounding the study sites, litterfall is presumably a contributing process at all sites; nearly all surveyed points of all landslides showed some evidence of litterfall, but the occurrence of measurable volumes was rare. In addition to the volume accretion of deposited material, litterfall provided structure within the scar that augments depositional processes (Shimokawa, 1984; Wilford, 1984) (Figure 2.16). Fine woody debris, litter, and patches of 64  vegetation were also observed to form small debris jams that enhance deposition (e.g., Rey 2004; Ziegler et al., 2006). The predominance of slope wash, and its preferential transport of small particles, is evidenced in the fine texture of most of the infilled soils (Carson and Kirkby, 1972). Distributed sheet erosion was the dominant process in Japan, probably because of high intensity precipitation during the annual Baiu and typhoon seasons and the greater importance of precipitation intensity over precipitation amount (Jungerius and ten Harkel, 1994). Further, the difference in the forest conditions through the reduced influence of coarse woody debris, litter, and complex vegetation communities relative to the sites in British Columbia maybe responsible for the different material transport regimes. A major land use for steep hillslopes in both British Columbia and Japan is timber harvesting. Understory vegetation communities are affected by the removal of the canopy influence, which also results in changes to the micro-climate (Moore et al., 2005). Decreased water availability for plant uptake in failure areas may be a particularly strong limiting control on vegetation type before the accumulation of a forest soil, particularly when much of the failure areas have surface covers of rock and overconsolidated soil. This could lead to both drier periods and rapid stormflow response where overland flow could disturb vegetation and discourage establishment and reduce vigour thus leading to reduced direct accumulations of organic material. Reduced soil moisture conditions could also reduce the frequency of small landslides and slope wash. Water retention may be less limiting in the environment under the forest canopy that does not experience such elevated evaporation rates. Examination of the regression residuals of the normalized material transport from Figure 2.4 found the logged and unlogged sites to be statistically inseparable. However, the frequency of survey points with recent transport showed a clear difference with more activity at logged sites, most likely attributable to the higher percent of soil and rock surface covers on logged hillslopes relative to unlogged sites. Exposed mineral soils and general soil disturbance are 65  commonly associated with timber harvesting (e.g., Rab, 1996; Merino et al., 1998; Laffan et al., 2001). Changes to the micro-climate and soil conditions, induced by logging and landslides, may hamper colonization of the failure areas and result in more surface covers of soil and rock. This effect may be enhanced if the hillslope vegetation communities adapt poorly to the altered growing conditions following disturbance. Soil is commonly exposed as a result of material transport events and its link with material transport is evident in both this study and others (Carson and Kirkby, 1972, Prosser and Williams, 1998; Ludwig et al., 2005). Timber felling and log yarding can also result in soil disturbance (Bockheim et al., 1975). Larger material transport events will likely be followed by a period of elevated material transport in the freshly disturbed area and immediate surroundings, leading to a further expansion of geomorphic activity. But within failure areas, the surface covers of logged and unlogged sites are similar for the first 50 years following failure. The stochastic nature of material transport processes created challenges for the characterization of the volumes of material transport and led to the analysis of the percent of surveyed points with recent activity. In 27% of the sites, including sites of all ages, no active material transport was measured. At the younger sites without measured activity, it was assumed that the period of observation (2 year) was too short, given the episodic nature of the infilling processes. Although inactive initiation zones were common, the rate of material recharge at these sites was not necessarily low if the infilling processes were episodic. Causal field observations at one well-vegetated site, where no other depositional processes were occurring (excepting minor litterfall), revealed a single landslide deposit of ~8 m3. Such large-volume events will result in substantial infilling and serve to explain much of the variation in transport and deposition rates. A significant feature present at a few sites was a wildlife trail that traversed the head scarp area. In each case, the dominant surface cover was still vegetation but the presence of the trail resulted in material 66  transport. Trails may also cause alterations to drainage at a micro scale that could further alter the site (Sidle et al., 2006). The geomorphic effect of trails was not large but alludes to the sensitivity of smallscale processes acting in initiation zones. Once established, vegetation appears to be effective at diminishing the geomorphic activity regardless of topographic form. Thus, there is a similarity between scales. At the scale of the initiation zone, the occurrence of the concave failure area into the hillslope initiates geomorphic adjustments that decline as the topographic variability decreases and the vegetation establishes. At the scale of an individual surveyed point and its neighbours (~5 m scale), small-scale processes of material transport are more likely while the concave and convex topographies are both present and prior to the establishment of vegetation.  67  Figure 2.16. Coarse woody debris contributes structure in the failure area that leads to both direct and indirect volume accretion.  CONCLUSIONS Following a shallow landslide, hillslopes undergo a period of adjustment. Eventually, the failure site can appear similar in form, ground surface cover, vegetation community, and soil depth to adjacent hillslopes (Smith et al., 1986). Field data reveal that recent material transport volumes decline strongly in the first few decades following failure, roughly coinciding with the diminishment of the differences in soil depths between surrounding 68  hillslope and failure areas and also coinciding with vegetation establishment. This finding agrees with the results of Jakob et al., (2005) who observed exponential decline in recharge rate of debris-flow gullies on Haida Gwaii and south-western BC. Over time, the distribution of surface cover in the failure area tends towards that of the adjacent hillslope. Similar to the rate of infilling and the reduction of microtopographic complexity, geomorphic adjustment is largely completed within the first several decades after failure unless a subsequent failure renews the cycle. Concave surface microtopography in both cross-slope and longitudinal directions had significantly greater percentages of survey points with active processes compared to other topographic configurations. Over time, the dominant processes of small landslides and slope wash appear to infill microtopographic depressions and cause a decrease in the variability of the soil surface. Ground surface covers of soil and rocks were significantly more likely to have active processes than vegetation cover or woody debris. Vegetation cover responded differently at logged and unlogged sites on the hillslopes adjacent to the failure areas taking longer to recover from the landslide and land-use change disturbance. Although hillslope adjustments occur episodically despite vegetation cover, the rate of adjustment is highest immediately following failure and declines exponentially with increasing time since failure.  69  REFERENCES Adams PW, Sidle RC. 1987. Soil conditions in three recent landslides in southeast Alaska. Forest Ecology and Management 18: 93–102.  Benda L, Dunne T. 1997. Stochastic forcing of sediment supply to channel networks from landsliding and debris flow. Water Resources Research 33: 2849–2863.  Bockheim JG, Ballard TM, Wellington RP. 1975. Soil disturbance associated with timber harvesting in southwestern British Columbia. Canadian Journal of Forest Research 5: 285–290. doi:10.1139/x75039  Bovis MJ, Pellerin D. 1998. Rates of revegetation of gullies in coastal British Columbia: implications for fine-sediment production. Unpublished Annual Report to Science Council of British Columbia, Victoria, Canada.  Brayshaw D, Hassan MA. 2009. Debris flow initiation and sediment recharge in gullies. Geomorphology 109: 122–131.  Bryan RB. 2000. Soil erodibility and processes of water erosion on hillslope. Geomorphology 32: 385– 415. doi:10.1016/S0169-555X(99)00105-1  Campbell D, Church M. 2003. Reconnaissance sediment budgets for Lynn Valley, British Columbia: Holocene and contemporary time scales. Canadian Journal of Earth Science. 40: 701-713. doi: 10.1139/E03-012  70  Carson, M.A., Kirkby, M.J., 1972. Hillslope form and process. Cambridge University Press, Cambridge, 475 pp.  Crozier MJ, Vaughan EE, Tippett JM. 1990. Relative instability of colluvium-filled bedrock depressions. Earth Surface Processes and Landforms 15: 329–339.  Dietrich WE, Dunne T. 1978. Sediment budget for a small catchment in mountainous terrain. Zeitschrift fur Geomorphologie N.F. Suppl. 29: 191–206.  Dietrich WE, Dunne T, Humphrey NF, Reid LM, 1982. Construction of sediment budgets for drainage basins. In Sediment Budgets and Routing in Forested Drainage Basins, Swanson FJ, Janda RJ, Dunne T, Swanston DN (eds). USFS Gen. Tech. Rept. PNW 141: 5–23.  Environment Canada, 2009. http://www.climate.weatheroffice.ec.gc.ca/climate_normals/index_e.html  Gabet EJ, Dunne T. 2002. Landslides on coastal sage-scrub and grassland hillslopes in a severe El Niño winter: The effects of vegetation conversion on sediment delivery. Geological Society of America Bulletin 114: 983–990. doi: 10.1130/0016-7606(2002)114<0983:LOCSSA>2.0.CO;2  Gabet EJ, Reichmann OJ, Seabloom EW. 2003. The effects of bioturbation on soil processes and sediment transport. Annual Review of Earth and Planetary Sciences 31: 249–273.  Hack JT, Goodlett JC. 1960. Geomorphology and forest ecology of a mountain region in the central Appalachains. US Geological Survey Professional Paper 347. 66 pp.  Hamann A, Wang T. 2005. Models of climatic normals for genecology and climate change studies in British Columbia. Agricultural and Forest Meteorology 128: 211–221.  71  Holland SS, 1976. Landforms of British Columbia a physiographic outline; BC Department of Mines and Petroleum Resources, Bulletin 48.  Hutchinson DG, Moore RD. 2000. Throughflow variability on a forested hillslope underlain by compacted glacial till. Hydrological Processes 14: 1751–1766.  Ihaka R, Gentleman R. 1996. R: A language for data analysis and graphics. Journal of Computational and Graphical Statistics 5: 299–314.  Imaizumi F, Sidle RC. 2005. Relationship between sediment supply and transport processes in Miyagawa dam catchment. Annals of Disaster Prevention Research Institute, Kyoto University: Kyoto No. 48C.  Imaizumi F, Sidle RC, Kamei R. 2008. Effects of forest harvesting on the occurrence of landslides and debris flows in steep terrain of central Japan, Earth Surface Processes and Landforms 33: 827-840.  Jakob M, Bovis M, Oden M. 2005. The significance of channel recharge rates for estimating debris-flow magnitude and frequency. Earth Surface Processes and Landforms 30: 755–766. doi: 10.1002/esp.1188  Jakob M, Weatherly H. 2003. A hydroclimatic threshold for shallow landslide initiation at the North hore Mountains of Vancouver, BC. Geomorphology 57: 137–156.  Jungerius PD, ten Harkel MJ. 1994. The effect of rainfall intensity on surface runoff and sediment yield in the grey dunes along the Dutch coast under conditions of limited rainfall acceptance. Catena 23: 269-279.  72  Karanka EJ. 1986. Trends and fluctuatlons in precipitation and stream runoff in the Queen Charlotte Islands. British Columbia Ministry of Forests, Research Program: Victoria British Columbia; Land Management Report 040.  Kirkby MJ. 1971. Hillslope process-response models based on the continuity equation, Institute of the British Geographical Society, Special Publication 3: 15–30.  Kirkby MJ, Abrahart R, McMahon MD, Shao J, Thornes JB. 1998. MEDALUS soil erosion models for global change. Geomorphology 24: 35–49.  Laffan M, Jordan G, Duhig N. 2001. Impacts on soils from cable-logging steep slopes in northeastern Tasmania, Australia. Forest Ecology and Management 144: 91-99.  Larsen MC, Torres-Sanchez AJ, Concepcion IM. 1999. Slopewash, surface runoff and fine-litter transport in forest and landslide scars in humid-tropical steeplands, Luquillo Experimental Forest, Puerto Rico. Earth Surface Processes and Landforms 24: 481–502.  Ludwig JA, Wilcox BP, Breshears DD, Tongway DJ, Imeson AC. 2005. Vegetation patches and runoff– erosion as interacting ecohydrological processes in semiarid landscapes. Ecology 86: 288–297. doi: 10.1890/03-0569  Martin Y, Rood K, Schwab JW, Church M. 2002. Sediment transfer by shallow landsliding in the Queen Charlotte Islands, British Columbia. Canadian Journal of Earth Science 39: 189–205.  Merino A, Edeso JM, González MJ, Marauri P. 1998. Soil properties in a hilly area following different harvesting management practices. Forest Ecology and Management 103: 235–246.  73  Millard T. 1999. Debris flow initiation in coastal British Columbia gullies. Forest Research, Technical Report, BC Ministry of Forests, Nanaimo, BC.  Montgomery DR, Dietrich WE, Torres R, Anderson SP, Jeffner JT, Loague K, 1997. Hydrologic response of a steep, unchanneled valley to natural and applied rainfall. WRR 33: 91-109.  Montgomery DR, Schmidt KM, Greenberg HM, Dietrich WE. 2000. Forest clearing and regional landsliding. Geology 28: 311–314.  Moore RD, Spittlehouse DL, Story A. 2005. Riparian microclimate and stream temperature response to forest harvesting. Journal of the American Water Resources Association 41: 813–834.  Neteler M, Mitasova H. 2002. Open Source GIS: A GRASS GIS Approach, Kluwer Academic Press, Dordrecht: Boston, 464 pp.  Oke TR. 1978. Boundary layer climates. London: Methuen and Sons 372 pp.  Prosser IP, Williams L. 1998. The effect of runoff and erosion in native Eucalyptus forest. Hydrological Processes 12: 251–265.  Rab MA. 1996. Soil physical and hydrological properties following logging and slash and burning in the Eucalyptus regnans forest of southeastern Australia. Forest Ecology and Management 84: 159–176.  Reneau SL, Dietrich WE. 1991. Erosion rates in the southern Oregon Coast Range: Evidence for an equilibrium between hillslope erosion and sediment yield. Earth Surface Processes and Landforms 16: 307–322.  74  Reneau SL, Dietrich WE, Rubin M, Donahue DJ, Jull AJT. 1989. Analysis of Hillslope Erosion Rates Using Dated Colluvial Deposits. The Journal of Geology 97: 45–63.  Rey F. 2004. Effectiveness of vegetative barriers for marly sediment trapping. Earth Surface Processes and Landforms, 29:1161–1169.  Roering JJ, Kirchner JW, Dietrich WE. 1999, Evidence for nonlinear, diffusive sediment transport on hillslopes and implications for landscape morphology. Water Resources Research 35: 853–870.  Schwab JW. Unpublished Report. Research Project EP 782.07. Historical documentation of mass wasting in North Coastal British Columbia. Forest Sciences Section, BC Forest Service, Smithers, BC.  Selby MJ, 1980. A rock mass strength classification for geomorphic purposes: with tests from Antarctica and New Zealand. Zeitschrift für Geomorphologie 24: 31–51.  Shimokawa E. 1984. A natural recovery process of vegetation on landslide scars and landslide periodicity in forested drainage basins. In Proceedings of Symposium on Effects of Forest Land Use on Erosion and Slope Stability, East-West Centre, Honolulu 99–107.  Shimokawa E, Jitouzono T, Takano S. 1989. Periodicity ofshallow landslide on Shirasu (Ito pyroclastic flow deposits) steep slopes and prediction of potential landslide sites. Transactions of the Japanese Geomorphological Union 10: 267–284.  Sidle RC. 1987. A dynamic model of slope stability in zero-order basins. In Erosion and sedimentation in the Pacific Rim (Proceedings of the Corvallis Symposium, August, 1987). IAHS Publ. 165: 101–110.  Sidle RC, Ochiai H. 2006. Landslide processes, prediction, land use. American Geophysical Union, Washington, DC, pp 1–312. 75  Sidle RC, Ziegler AD, Negishi JN, Abdul Rahim N, Siew R, Turkelboom F. 2006. Erosion processes in steep terrain – truths, myths, and uncertainties related to forest management in Southeast Asia. Forest Ecology and Management 224: 199–225.  Slaymaker O. 1967. Patterns of erosion in small catchments in mid-Wales. Unpublished Ph. D. thesis, University of Cambridge.  Smith RB, Commandeur PR, Ryan MW. 1986. Soils, vegetation, and forest growth on landslides and surrounding logged and old-growth areas on the Queen Charlotte Islands. BC Ministry of Forests Land Management Report 41, Victoria, British Columbia.  Sutherland-Brown A. 1968. Geology of the Queen Charlotte Islands. Bulletin 54. British Columbia Department of Mines and Petroleum Resources.  Takahashi T. 2007. Debris flows: mechanics, prediction and countermeasures. Taylor and Francis / Balkema, 448pp.  Tsukada, M. (1982) Cryptomeria japonica: glacial refugia and late glacial and post glacial migration. Ecology 63: 1091-1105.  Tsukamoto Y, Ohta T, Noguchi H. 1982. Hydrological and geomorphological studies of debris slides in Japan. International Association of Hydrologic Sciences. Publication 137: 89–98.  Tutiempo.net, http://www.tutiempo.net, 20 April 2010.  Wang T, Hamann A, Spittlehouse D, Aitken SN. 2006. Development of scale-free climate data for western Canada for use in resource management. International Journal of Climatology 26: 383–397.  76  Wilford DJ. 1984. The sediment-storage function of large organic debris at the base of unstable slopes. In: Meehan, W.R.; Merrell, T.R.; Hanley, T.A., ed. Fish and wildlife relationships in old-growth forests: Proceedings of a symposium. American Institute of Fishery Research Biologists, 115–119.  Wilford DJ, Cherubini P, Sakals ME. 2005. Dendroecology: a guide for using tree to date geomorphic and hydrologic events. BC Ministry of Forests Research Branch: Victoria BC, Land Management Handbook 58, 25 pp.  Ziegler AD, Negishi J, Sidle RC, Preechapanya P, Sutherland RA, Giambelluca TW, Jaiaree S. 2006. Reduction of Stream Sediment Concentration by a Riparian Buffer: Filtering of Road Runoff in Disturbed Headwater Basins of Montane Mainland Southeast Asia. Journal of Environmental Quality 35: 151–162.  77  3. SOIL ACCRETION FOLLOWING SHALLOW LANDSLIDES1 INTRODUCTION In many areas of high rainfall and steep topography, shallow landslides are a significant natural hazard as well as an important natural disturbance agent creating diversity in the landscape and soil environment (Adams and Sidle, 1987; Geertsema, et al., 2009). Shallow landslides are slope failures that are long relative to their depth (Turner and Schuster, 1996) and are a common form of slope failure contributing significantly to sediment transport (Rapp, 1960; Selby, 1993; Sidle and Ochiai, 2006). Shallow landslides are known to occur repeatedly from the same, or nearly the same, hillslope positions (Shimokawa 1984; Dietrich and Dunne 1978; Schwab, unpub.). Following failure, the soil mantle is commonly of insufficient thickness to overcome the frictional cohesion at its base, and therefore generally considered unlikely to fail again for a considerable period. This idea has support in some areas (Reneau and Dietrich, 1991); however data from other regions suggest more rapid soil depth accretion and a shorter period of immunity from slope failure (Shimokawa, 1984; Shimokawa et al., 1989; Smale et al., 1997). Greater accumulations of material will also characteristically result in larger subsequent failures (Innes, 1983) and the greater the magnitude of the failure, the more likely it is to generate a channelized debris flow (Millard, 1999; Imaizumi et al., 2006).  1  A version of this chapter will be submitted for publication. Sakals ME, Innes JL, Sidle RC, Moore RD, Hungr O, Jakob M. 2010. Soil accretion following shallow landslides  78  There is an impetus for greater knowledge concerning these slope hazards as human development encroaches upon steep and unstable terrain and into the runout zones of shallow landslides and associated hazards. Each year in steepland areas throughout the world, shallow landslides and debris flows result in the loss of many lives (Sidle and Chigira, 2004) and costly damage to human developments (Jakob and Hungr, 2005). Debris flows also damage natural resources, including aquatic habitat, water quality and forest land (Gomi et al., 2001; Wilford et al., 2003; Kobayashi et al., 2010). Deposits from smaller failures also contribute to debris flow activity through accumulation and subsequent mobilization in loworder channels (Millard, 1999; Gomi et al., 2001; Hungr et al., 2005, Brayshaw and Hassan, 2009). Many sites of repeat failures are hollows (Hack and Goodlett, 1960; Reneau et al., 1990), the axes of zero-order basins (Dietrich et al., 1982) or colluvium-filled bedrock depressions (Crozier et al., 1990). Studies of zero-order basins have primarily occurred in unglaciated terrain (e.g., Dietrich and Dunne, 1978; Dietrich et al., 1982; Alger and Ellen, 1987; Reneau et al., 1990; Montgomery and Dietrich, 1994), although some work has also been completed in glaciated terrain (Crozier et al., 1990; Yamada, 1999). In both glaciated and unglaciated terrain, these features are strongly linked to the generation of debris flows from steep forested slopes (Dietrich and Dunne, 1978; Adams and Sidle, 1987; Alger and Ellen, 1987; Montgomery and Dietrich, 1994; Schwab unpublished). Although research has been conducted for more than 35 years on the hydrologic and geomorphic processes in zero-order basins (e.g., Tsukamoto et al., 1982; Adams and Sidle, 1987; Benda and Cundy, 1990; Iida, 1999; Ohnuki et al., 1997; Iida, 2004), only a limited body of literature relates to rates of soil accumulation and each study includes only a small sample of shallow landslide initiation zones (Dietrich and Dunne, 1978; Shimokawa et al., 1989; Tsukamoto and Minematsu, 1987; Reneau and Dietrich, 1991; Smale et al., 1997). Even less work has addressed the relations between rates of accumulation and the environment of the shallow landslide initiation zone (Shimokawa, 1984; Yamada, 1999); some work has been done in debris-flow gullies (Jakob et al., 2005). Theoretical work has also been completed; Sidle (1987) proposed a sigmoid curve to 79  model the recovery of soil depth following landslide and suggested that it would apply over a range of infilling rates. Investigating recharge by directly measuring soil depths will assist in quantifying recharge rates at shallow landslide failure areas, which will provide information regarding controls on shallow landslide return periods and rates of sediment flux across the landscape. In this study, 58 shallow landslide initiation sites on colluvium-mantled hillslopes were surveyed with the intention of creating a model of soil material accumulation. A synoptic approach, briefly visiting many landslide initiation sites, was preferred because of the variability in soil accretion rates, geological materials, and hillslope forms. The factors included in the models are loosely based on the soil-forming factors of Jenny (1941); however, this work focuses not on pedological soil formation, but on the prediction of unconsolidated soil material in the failure area that could subsequently fail as a shallow landslide. Both a statistical and a conceptual model are presented in an attempt to characterize material accretion. First, attributes of the initiation zones were included in non-linear regression models of soil accretion with the objective of quantifying material recharge rates in coastal British Columbia and south-central Japan and determining the important rate-controlling variables. Second, a conceptual model of stochastic material infilling is presented that may partially address the unexplained variance of the statistical method.  METHODS Study areas Study areas were located in three areas of British Columbia: the North Coast, in the vicinity of Prince Rupert; Haida Gwaii (formerly the Queen Charlotte Islands), with sampling centred in Rennell Sound; and the South Coast, with sampling in the Klanawa drainage of south-central Vancouver Island (Figure 3.1, Table 3.1). The climate of the three areas in British Columbia was predicted using ClimateBC 80  (Hamann and Wang, 2005; Wang et al., 2006) with daily precipitation extremes recorded at nearby weather stations (Environment Canada, 2008) and resulted in being very similar (Figures 3.2 and 3.3). The South Coast sites might have been warmer, but much of the sampling was at somewhat higher elevations and the ClimateBC model has adjusted the temperature values. A fourth study area was located on the Kii peninsula of south-central Japan. Climate data for Owase, Japan during the period of 1973 to 2000 was attained online (Tutiempo.net). Although the station was in reasonable proximity to the studied landslides (<60 km from furthest study site), the station was located just above sea level (27 m) while the lowest study site was at 150 m elevation, the other sites were at 440, 490, 500, 1 060 and 1 100 m elevation. Adiabatic cooling has been estimated for the sites of approximately 500 m elevation by applying the environmental lapse rate (Oke, 1987) to the Owase station data. No information is available regarding the increase in precipitation with altitude, but Imaizumi and Sidle (2005) reported a range of 1 600–4 500 mm (average of 3 300 mm) for the Miyagawa Dam area where three of the study sites were located. It is important to note that although the climate of the Kii Peninsula may be warmer during the summer months, the average daily temperature during the winter months allows the possibility of freezing, particularly for the higher elevation sites. Therefore needle ice and other freeze-thaw processes remain possible during the winter months. The daily precipitation extreme is higher at the Owase station compared to the South Coast study area (Port Renfrew station) and three and four times higher than the North Coast (Prince Rupert) and Haida Gwaii (Sandspit) stations respectively. However, the Sandspit station is known to be drier than the west coast of Haida Gwaii due to local orographic effects (Karanka, 1986). Precipitation intensities over shorter periods are not known but could be expected that the sites on the Kii Peninsula may experience higher intensity rainfall during the Baiu (early summer monsoon period) and typhoon seasons. Monthly  81  precipitation as predicted by ClimateBC for the areas in British Columbia and from the Owase station record for the Japanese study area appear to be similar in the magnitude for the wettest and driest months. All three study areas in British Columbia were glaciated during the Late Pleistocene. The landslide sites from Japan have not experienced glaciation and thus the landscape lacks paraglacial legacy of the study areas in British Columbia. However, steep slopes with soil covers of <2 m are common in all regions.  Figure 3.1. Study areas in British Columbia, Canada and in central Japan.  82  Figure 3.2.Mean daily temperatures by month for study areas in British Columbia (North Coast, Haida Gwaii, and South Coast) and Japan (Owase). Dashed line represents an estimate from Owase station data to the elevation of the study sites based on adiabatic cooling.  83  Figure 3.3.Mean monthly precipitation for study areas in British Columbia (North Coast, Haida Gwaii, and South Coast) and Japan.  84  Selection of candidate landslides Landslide inventories were used to select candidate landslide initiation sites and were available for all three study areas in British Columbia (Schwab, unpub., Maynard et al., 2004) as well as in Japan (Imaizumi et al., 2008). A time-series of aerial photographs of different ages were used to bracket the date of failure, or the failure was associated with a substantial storm that occurred during the period between two successive photo sets. The error of this method is assumed to be <5 years. Data accompanying the inventories were used to select candidate landslides based on material type, geology, and the feasibility of access. Some failures were visible on aerial photographs, in fresh condition or indicated by distinctive strips of forest cover. Some sites included in the study were not included in the inventories and were not visible on 1:20 000 scale aerial photographs because of their small size and the crowns of adjacent trees; these sites were found by reviewing aerial photographs for candidate hillslopes and completing field traverses to identify possible study landslides. The distribution of time since previous failure was designed to have a skewed distribution with greater sampling of the most recent failures and similar distributions in each study area. Young sites were expected to have higher rates of material accumulation due to the recent disturbance, and thus a skewed sample was warranted. Several criteria were used for site selection. Beyond safety concerns, the first requirement at the landslide site was that the failed material was colluvium. Colluvium is assumed to be more directly related to the current conditions, slopes and underlying geology than other hillslope materials such as till. Till was avoided because of the variability in textures and differences in consolidation that could result in different root penetrations and resistances to surface erosion processes that could introduce extra variability into the study. Great variation exists in colluvium but at most of the study sites the material appeared to be ultimately derived from the local bedrock. A failure that comprised the full depth of colluvium (to 85  competent rock) was required to easily measure the depth of soil accumulation. A soil pit was excavated to bedrock or until a failure plane surface was reached within the soil. In the latter case, the landslide would not be included if the depth of soil above bedrock remaining after the previous failure was >20 cm as this would necessitate too much addition effort to determine the lag soil depth; only a few sites were excluded based on this criterion. No specific size criteria were applied, so long as the initial failure was of the full soil depth.Soil depths were measured with a knocking cone penetrometer or a steel rod driven by a mallet; the change in penetration resistance was used to distinguish accumulated material from the basal plane. The date of failure was required in order to assess the period of material accumulation. Some dates were previously reported in landslide inventories (Schwab, unpub.; Maynard et al., 2004; Imaizumi et al., 2008); dendrochronological techniques were used where required to determine the date of failure for the remaining sites or where the reported date was suspected of being incorrect (e.g., Wilford et al., 2005). Tree scars were the preferred feature for dating followed by abrupt growth releases of trees on the margins of the failure that developed due to changes in water, nutrient, and light availability. Dating trees established on the recharged soil material produced a minimum age only, as the period of accumulation prior to establishment was unknown. Other methods, such as the accumulation of litter material, were used to determine the age of failures <5 years. The temporal error of time since failure at each study site was assessed independently and was dependent on the number, quality, and consistency in the dendrochronological samples and other evidence; 11 sites had errors >5 years with only 2 of the 11 being estimated at <20 years since failure. Estimated errors are presented in Appendix D. Finally, landslides caused by excessive anthropogenic disturbance such as major alterations by roads or trails were excluded. Areas of timber harvest were acceptable as long as there were no obvious upslope diversions of drainage pathways or areas of excessive soil or other disturbance.  86  Table 3.1. Selected information of the study areas (Sutherland-Brown, 1968; Holland, 1976; Muller 1977; Imaizumi and Sidle, 2005; Imaizumi et al., 2008; Environment Canada, 2008; Tutiempo.net). North Coast, BC 17 yes  Haida Gwaii, BC South Coast, BC Kii Peninsula, Japan Landslides sampled 16 18 7 Late Pleistocene glaciation yes yes no pyroclastic rocks, shale, granodiorite, quartz diorite, schist, sandstone, slate, shale, Geology granite, gneiss, shale, slate siltstone, sandstone, granite intrusions, gneiss, mudstone conglomerate, argillite weak volcanic rocks, steep slopes rising from sea steep slopes rising from sea steep slopes rising from sea steep slopes with local relief Typical morphology level to 500 m level to 600 m level to 1 000 m of 300 m Annual precipitation at study sites 1 600–3 350 mm 2 400–3700 mm 3 450–4 000 mm ~4 000 mm Extreme rainfall at nearest climate 118.2 mm /day (Prince 79.5 mm /day (Sandspit, 293 mm /day (Port Renfrew, 351 mm /day (Owase, station Rupert, BC) BC)* BC) Japan)** colluvial soils of <1 m in colluvial soils of <1 m in colluvial soils of <1 m in colluvial soils of 0.5-1.5 m in Typical soils depth depth depth depth second and old-growth second and old-growth second and old-growth Forests coniferous plantations coniferous coniferous coniferous Predominant natural disturbance type shallow landslide, debris flow shallow landslide, debris flow shallow landslide, windthrow shallow landslide, debris flow Predominant human disturbance type forest harvesting forest harvesting forest harvesting forest harvesting Landslide inventory available yes yes yes yes *The distribution of precipitation on Haida Gwaii is strongly influenced by orographic effects. The climate station (Sandspit) is on the leeward, eastern side of the major mountain range of the island archipelago; the study area was located on the windward, western side of the range. Precipitation is accentuated on the western side of the islands where incoming Pacific frontal systems first encounter the land mass. Therefore, precipitation at the study sites is under represented by the Sandspit station (Karanka, 1986). **Highest noted value during data compilation of period from 1973–2000.  87  Site description Soil depths were collected over the entire initiation area and the immediately adjacent upslope contributing area (typically 5–10 m beyond side scarps and 10–15 m above the head scarp) in an irregular grid pattern. The extent of the surveys was limited to 10–15 m upslope for several reasons. Firstly, in many cases, the upslope topographic expression did not indicate geomorphic or hydrologic concentration; in some cases, the upslope form was even divergent. Second, the distance to the slope divide was typically several hundreds of metres with difficult terrain and vegetation for the completion of a detailed survey. Thirdly, no small-scale processes of infilling found during the preliminary fieldwork or during the course of the study progressed downslope more than 10 m. For these reasons, the detailed survey was limited to the immediately adjacent slopes. Unfortunately, this also meant an incomplete characterization of the hydrological contributing area; however, with the relatively planar configuration of most of the studied hillslopes, the determined watershed area would have been of questionable use considering the effects of preferential flow pathways not entirely following surficial relief (Hutchinson and Moore, 2000). The surveys used a ~3 m inter-point spacing, but this was reduced if required to capture important slope transitions. At each point, the following information was recorded: 1. The depth of the soil to bedrock, as determined by a knocking cone penetrometer or steel rod. Penetration resistance values of the underlying substrate varied between sites, but were generally ~10 (i.e., 10 strikes of a 5 kg weight falling 50 cm to drive the 3 cm diameter penetrometer cone 10 cm into the soil) (e.g., Ohnuki et al., 1997). At each site, measurements with the steel rod and mallet were roughly calibrated with the penetrometer. For the Japanese sites a standard 5 kg weight was used; for the British Columbia sites a 2.5 kg weight was used to facilitate fieldwork in remote locations. 88  2. The morphologic position relative to the initiation zone. At 33 sites in the North Coast, Haida Gwaii, and Japanese study areas, a theodolite and laser distance measuring device were used to derive topographic information (x, y, and z coordinates) for each survey point (see maps of study areas in Appendix A). During the survey, the volume of recent (within 2 years) sediment or debris transport, >1 cm in thickness, in the vicinity of the survey point was recorded. See Chapter 2 for detailed methods. Topographic and material transport data was not collected at the study sites in the South Coast study area. The geology of the failure area was described with both the original and a modified version of Selby‟s (1980) Geomorphic Rock Mass Classification (geology, geologyMod) (Tables 3.2 and 3.3). Rock mass scores were calculated by estimating and summing values for each category. High scores for all components of the geomorphic rock mass classification, except joint orientation, were expected to be associated with lower recharge rates. Therefore, a modified geological variable (geologyMod) was created by reversing the scores of the joint orientation component and retaining the remainder of the classification. Thus, the expected negative correlation pattern between the components of the classification and material recharge was preserved throughout. Bedrock dipping steeply out of the slope was expected to result in less basal roughness in the failure area and thus more material passing through the failure area rather than being held and contributing to soil accretion. Bedrock dipping steeply into the slope was expected to result in greater basal friction that would enhance material accumulation in the failure area. The remaining components of the classification described the condition of the rock mass: intact rock strength; degree of rock mass weathering; and the spacing, width, and continuity of joints. The component relating to the outflow of groundwater was omitted from both classifications as the typically wet soil conditions made it difficult to determine the source of the water. Component scores were summed to produce the geology and geologyMod (with modified joint orientation score) variables. In 89  cases where accumulated soil material had completely covered the bedrock in the failure area, an inference was made from rock exposed at the base of the soil pit(s), as well as any rock exposures in the immediate vicinity. The average length and width of the initiation area was measured to the nearest 0.1 m (scarLength, scarWidth). The transition between the failure area and the transport zone was determined by a narrowing of the disturbed width or, in other cases, by a lack of widening of the disturbed area. As the study sites were all mid-slope failures, discernable difference in slope gradient between the hillslope above the failure and the failure axis were noted at only a few sites; therefore, the slope of the failure axis through the initiation area was measured to the nearest degree with a handheld clinometer (axisSlope). Crossslope morphology (convergence) was measured by first taking the bearing parallel to the left and right hillslopes in the ~10 m adjacent to the failure, then determining the difference (Figure 3.4). Values <180º indicate concave morphology; values >180º indicate convex morphology.  90  Figure 3.4. Graphical representation of the method of determining convergence variable. Red and blue lines demarcate failure areas. See Appendix D for an explanation of letter notation. Climate normals for British Columbia sites were estimated using the ClimateBC application, which predicts temperature and precipitation normals across British Columbia with a spatial resolution of 400 m (Hamann and Wang, 2005; Wang et al., 2006). The predicted fall precipitation (average September to November precipitation total in mm) was used (PPT911). This value is the best available information to describe the intense precipitation events that frequently occur in coastal BC during the fall (Lewis and Moran, 1985). 91  The forest productivity at the British Columbian sites was described quantitatively by the site index (siteIndex). Site index is the average height of the 100 tallest trees per hectare at an index age of 50 year at breast height (1.3 m) (Mitchell, 1988). A dummy variable (logged) was used to describe whether the site had been timber harvested. Several initiation areas contained surfaces representing separate failure events; these surfaces were distinguished in the field and the two failure ages were included as separate landslides. Fifty-three initiation zones were included in this study; five of these had surfaces representing two failure areas and thus the total for analysis is 58. Eight initiation zones required a linear approximation for the average depth of soil on adjacent hillslopes. Using the full data set, a significant (p<0.001, R2adj = 0.21) relation was found between the mean hillslope soil depth and the geologyMod score for the failure. The relation was then used to predict the missing values for the eight failure sites (Figure 3.5).  92  Table 3.2. Geomorphic rock mass classification system from Selby (1980) with both original and modified joint orientation scores. Component  1 Very strong  2 Strong  3 Moderate  4 Weak  5 Very weak  Intact rock strength  r: 20 e.g., quartzite, dolerite, gabbro  r: 5 e.g., chalk, rocksalt, lignite  Unweathered r: 10  Spacing of joints  >3 m r: 30 Very favourable, Steep dips into slope, cross joints interlock r: 20 Very unfavourable, Steep dips out of slope r: 20  r: 14 e.g., slate, shale, sandstone, mudstone, ignimbrite Moderately weathered r: 7 0.3-1 m r: 21 Fair, Horizontal dips, or nearly vertical (hard rocks only) r: 14  r: 10 e.g., coal, siltstone, schist  Weathering  r: 18 e.g., marble, limestone, dolomite, andesite, granite, gneiss Slightly weathered r: 9 1-3 m r: 28 Favourable, Moderate dips into slope r: 18  Highly weathered r: 5 0.05-0.3 m r: 15 Unfavourable, Moderate dips out of slope r: 9  Completely weathered r: 3 <0.05 m r: 8 Very unfavourable, Steep dips out of slope r: 5  Unfavourable, Moderate dips out of slope r: 18  Fair, Horizontal dips, or nearly vertical (hard rocks only) r: 14  Favourable, Moderate dips into slope r: 9  <0.1 mm r: 7  0.1-1 mm r: 6  1-5 mm r: 5  5-20 mm r: 4  Very favourable, Steep dips into slope, cross joints interlock r: 5 >20 mm r: 2  Continuity of joints  none continuous r: 7  few continuous r: 6  continuous, no infill r: 5  continuous, thin infill r: 4  continuous, thick infill r: 1  Outflow of ground-water  none r: 6  trace r: 5  slight <25 litres/min/10 m2 r: 4  moderate 25-125 litres/min/10 m2 r: 3  great >125 litres/min/ 10 m2 r: 1  Joint orientations (Original)  Joint orientations (Modified)  Width of joints  93  Table 3.3. Hillslope variables included in models. Variable  Units  axisSlope  º  convergence  º  Description Slope of failure axis measured with a clinometer Difference between the tangents of the left and right hillslopes adjacent to the head scarp  geology  Geomorphic Rock Mass Classification (Selby, 1980)  geologyMod  Modified Geomorphic Rock Mass Classification (Selby, 1980) Dummy variable: 0 for natural surrounding forests; 1 for  logged  PPT911  previous timber harvest  mm  scarLength  m  scarWidth  m  siteIndex  m  Average precipitation from September to the end of November for the landslide site as predicted by ClmateBC (*) Slope distance from head scarp to lower limit of failure area Measured perpendicular to scarLength at the widest point of the failure area Predicted average height of the 100 tallest trees per hectare at an index age of 50 year at breast height (Mitchell, 1988)  94  Figure 3.5. Relation between soil depth and modified index of rock mass strength with predicted values for 8 study sites with missing hillslope soil depth data. Dashed lines indicate the 95% prediction intervals. 95  Data analysis Non-linear modeling Prior to analysis, study sites from all four study areas were compared to identify the regional differences. The plots are presented in Appendix C. Data were randomly split into a model calibration set and a testing set. Some adjustments were then made: 1) landslide initiation zones with two failure ages present were kept together (i.e., both in model calibration, or both in testing) to prevent contamination of the testing data set; and 2) all seven Japanese sites were reserved for testing so that differences between the study sites from British Columbia and Japan would be noted, rather than incorporated into the models. The result is a model calibration set of 41 study sites and a model testing set of 17 sites including 10 sites from British Columbia and 7 sites from Japan. For all soil-depth prediction models, the independent variable was the average depth of soil material in the failure area. Various non-linear model forms were evaluated for conceptual fit and those accepted were fit to the data using the Likelihood package for R (Goffe et al., 1994; Ihaka and Gentleman, 1996). The Likelihood package uses simulated annealing to optimize parameter estimates. Simulated annealing refers to the process of searching, initially randomly and increasingly focused on sites of greatest likelihood, for the combination of parameter estimates that results in the best estimate. In comparison to non-linear regression by least squares, the simulated annealing method is less susceptible to being „trapped‟ in local maxima because it searches throughout the n-dimensional space determined by the high and low parameter bounds specified by prior to the model run. Model controls are available to balance thoroughness against the focus on high likelihood estimates, as well as the number of model iterations. Basic models were first tested without extra variables, only with the time since failure and the parameters necessary to generate the specific curve (Table 3.4). 96  The basic sigmoid equation as specified by Sidle (1987) indicated that the parameter b represented the asymptotic soil depth; here we will assume it to be equivalent to the adjacent hillslope soil depth (HsdMean). This model was tested with and without a parameter controlling the scale of the HsdMean variable. The sigmoid model including the substitution of HsdMean for b was considered the primary model:  where a and c are constants, HsdMean is the mean soil depth of adjacent hillslope soils, and time is the time since landslide in years. Each of the descriptive variables was plotted against the residuals of basic model form to assess the potential relation. Each of the variables was then added as a linear term to the primary model; linear terms were used because given the variability in the data, only basic relations were sought. A constant was also added to each model. In order to assess potentially multivariate relations, model residuals were recompared with the other descriptive variables; further models were then specified. All models that included geology were analyzed twice, once using the original geology variable and once using the modified geologyMod variable to determine whether the modification was justified. Akaike‟s Information Criterion (Akaike, 1973), with a correction to account for the moderately small sample size (AICc), was used to assess the support of each of the models (Burnham and Anderson, 2002): AICc = –2·log(L) + 2k + 2k · (k+1)/(n–k–1)  97  where k is the number of parameters in the model, L is the maximized value of the likelihood function for the estimated model, and n is the sample size. AICc scores were calculated, ranked, and their differences from the best model calculated. The model with the lowest AICc score is assumed to be the best model, but the score alone does not describe the support relative to other models. Evidence ratios are calculated by comparing and calculating ratios for the Akaike scores to create values that reflect the relative strength of the models. Evidence ratios >10 are taken as very strong support for one model over another (Burnham and Anderson, 2002). Likelihood, Akaike scores, and evidence ratios were compared for candidate models applied to the calibration data; residual plots were also inspected for fit. Models were evaluated against the test data by descriptive statistics including the coefficient of determination for observed versus predicted plots and the pseudo mean square error (PMSE) which is a measure of model bias and precision (Zhang, 1997): PMSE = ē2 + v where ē is the mean error and v is the variance. The form of observed versus predicted plots were also assessed for model suitability to the testing data. Conceptual model of stochastic infilling For the conceptual model of stochastic processes infilling, we described sites according to the ratio of the infilled soil depth divided by the depth of soils on the adjacent slopes. This provided a basic platform on which to expand our theory.  98  RESULTS Non-linear modeling Basic model selection Six models were assessed for conceptual agreement with field observations and data; four models were accepted at the conceptual level (Table 3.4; Figure 3.6). Observed volumes of recent (<2 year) material transport decreased with time and so the form of the selected model(s) should also indicate reduced accumulation with time (Figure 3.7). The exponential and linear models predicted nearly constant (and constant) soil accumulations from the initial period of rapid soil accumulation to beyond 200 years and thus were rejected from further analysis. These models better fit the sites with >0.8 m of infilled soil but the model form disagreed with observations of decreased material transport through time as vegetation establishes and head and side scarps become less defined and less steep with time. The remainder of the models were generally consistent with both the form of the recent material transport curve and the soil depth data. The four accepted models were fitted to the calibration data and all four fit the field observations reasonably equally, given the variation in the data (Figure 3.6). The sigmoid model had the lowest AICc score (sigmoid model -13.43; Verhulst model -11.24, logarithmic model -7.43, asymptotic model -5.89). The residual plots of the sigmoid, Verhulst, and asymptotic models are very similar and all were accepted despite a degree of heteroskedasticity. The heteroskedasticity is mostly due to two observations with strongly positive residuals (Figure 3.8). The effect of these two points was exacerbated by the limited observations of failures >50 years old. The logarithmic model appears to have the worst fit of the  99  four models, and displayed a systematic trend with negative values for younger sites and dominantly positive residuals beyond 20 years; the logarithmic model was thus rejected from further analysis. The sigmoid model fit well with qualitative field observations of recent failure sites where the initial period of high material transport was not matched with corresponding soil depth accretion because the material was being transferred directly through the failure area and down into the transport zone of the event. Although the conditions of a given failure site could conceivably be better suited to one of the other models, field observations indicate more recently failed sites would be better approximated by the form of the sigmoid model rather than the immediate accumulation portrayed by the Verhulst and asymptotic models. Elevated sediment delivery to downslope areas following failure was noted at many sites of <20 years since failure. The sigmoid model had a simpler model form allowing easier interpretation of parameter estimates; it has also been used previously to describe soil accumulation in failure areas (Sidle, 1987).  100  Table 3.4. Candidate basic models and formulae.  Name  Conceptual  Model  Review  Residual Plot Review  Sigmoid  y = a + (b-a)·exp(-2·c/time)  Accepted  Accepted  Logarithmic  y = a+b·log(time)  Accepted  Rejected  Verhulst  y = a·b/[a+(b-a)·exp(-c·time)]  Accepted  Accepted  Asymptotic  y = a+(b-a)·[1-exp(-c·time)]  Accepted  Accepted  Exponential  y = a+b·timec  Rejected  NA  Linear  y = a+b·time  Rejected  NA  101  Figure 3.6. Relation between average infilled soil depth in shallow landslide failure areas and time since landslide for all 58 study landslides from three areas of coastal BC and south-central Japan. Models accepted at the conceptual review stage were fit to the calibration data. 102  Figure 3.7. Recent (within 2 years) material transport volumes normalized by failure area and converted to an annual value for 32 failure areas and surrounding slopes. Heavy black bars represent an average of time and volume; thin bars indicate the temporal scope of these average values. The dashed line represents annual accumulation as predicted by the sigmoid curve of Figure 3.5 for an average-size failure area. 103  Figure 3.8. Residual plot for the sigmoid model.  104  Models with additional variables For models with a single additional variable added to the primary model, the AICc value was decreased for three descriptive variables (geology, axisSlope, and geologyMod) (Table 3.5); the remaining six variables (siteIndex, scarWidth, logged, PPT911, scarLength, and convergence) increased the AICc score. When compared to the residuals of the basic model, additional parameters did not show strong relations. The highest ranking model included the geologyMod variable combined with axisSlope. Models ranked second through fifth included the unmodified geology variable. The parameter estimates for all models are presented in Table 3.6. Most of the variables that had a negative impact on the AICc score were minimized by the modeling. That is, the best parameter estimate was near zero and led to a negligible contribution to the estimate of infilled soil depth. The evidence ratio indicated that the top four models (evidence ratios <4) all had a high likelihood of contributing relatively useful information (Burnham and Anderson, 2002). The fifth ranked model still had an evidence ratio of <10 and may be a valuable model. Table 3.7 contains the comparison between the original and modified geology variables. Although geologyMod was included in the top model, based on AICc, the average AICc score for models including the geology variable was somewhat lower (-22.97) than for those models including geologyMod (-18.71). Not all variables were collected over wide ranges. Convergence, scarLength, and scarWidth all had distributions that may have limited their use in characterizing the range of responses across their values (Figures 3.9 and 3.10).  105  Figure 3.9. Histogram of convergence variable for failure areas; 180º represents a planar slope, lower values are concave slopes, higher values are convex.  106  Figure 3.10. Histograms of failure area lengths (scarLength) and widths (scarWidth).  107  Model testing Table 3.8 presents summary information regarding the fit of the candidate models to the reserved testing data set. The PMSE applies to the entire testing data set, while the PMSE (BC) and PMSE (Japan) refer to only the British Columbian and Japanese test sites, respectively. Of the top five models ranked by PMSE, four of them were ranked within the top five AICc ranked models. The top AICc ranked model, including the geologyMod variable, ranks 17th. When the AICc ranking and the PMSE Ranking are combined, three models share a combined rank score of six: the primary sigmoid with the geology and convergence variables, the primary sigmoid with the geology variable alone, and the primary sigmoid with the geology and axisSlope variables.  108  Table 3.5. Details for candidate models, n=41 for all models, k is the number of parameters in the model.  Rank  k  R2  AICc  Δi  Likelihood  Akaike weight  Evidence ratio  Primary sigmoid + geologyMod + axisSlope + constant  1  5  0.47  -24.98  0.000  1.000  0.331  1.0  Primary sigmoid + geology + axisSlope + constant  2  5  0.40  -24.36  0.623  0.732  0.244  1.4  Primary sigmoid + geology + scarLength + constant  3  5  0.30  -24.04  0.943  0.624  0.208  1.6  Primary sigmoid + geology + constant  4  4  0.35  -22.82  2.164  0.339  0.113  2.9  Primary sigmoid + geology + convergence + constant  5  5  0.37  -20.67  4.314  0.116  0.039  8.6  Primary sigmoid + axisSlope + constant  6  4  0.47  -20.19  4.795  0.091  0.030  10.9  Primary sigmoid + geologyMod + constant  7  4  0.45  -18.20  6.778  0.034  0.011  29.4  Primary sigmoid  8  2  0.45  -17.40  7.577  0.023  0.008  43.9  Primary sigmoid + siteIndex + constant  9  3  0.46  -16.04  8.937  0.011  0.004  86.7  Primary sigmoid + geologyMod + scarLength + constant  10  5  0.47  -16.01  8.969  0.011  0.004  88.1  Primary sigmoid + geologyMod + convergence + constant  11  5  0.44  -15.63  9.347  0.009  0.003  106.4  Primary sigmoid + scarWidth + constant  12  4  0.43  -15.18  9.802  0.007  0.002  133.6  Primary sigmoid with internal scalar  13  3  0.45  -15.07  9.909  0.007  0.002  140.9  Basic sigmoid  14  3  0.38  -13.43  11.555  0.003  0.001  320.8  Primary sigmoid + logged + constant  15  4  0.45  -13.34  11.640  0.003  0.001  334.8  Primary sigmoid + PPT911 + constant  16  4  0.46  -12.69  12.287  0.002  0.001  462.8  Primary sigmoid + scarLength + constant  17  4  0.44  -12.63  12.350  0.002  0.001  477.5  Primary sigmoid + convergence + constant  18  4  0.45  -12.61  12.368  0.002  0.001  481.7  Model description  109  Table 3.6. Parameter estimates for all models.  Model description  c  Parameter 1  Parameter 2  Constant  Primary sigmoid + geologyMod + axisSlope + constant  0.30  a  14.60  -0.53  0.01  -0.14  Primary sigmoid + geology + axisSlope + constant  0.23  12.51  -0.52  0.01  0.02  Primary sigmoid + geology + scarLength + constant  0.16  13.63  -0.86  -0.01  0.69  Primary sigmoid + geology + constant  0.10  18.33  -0.63  Primary sigmoid + geology + convergence + constant  0.14  17.51  -0.64  Primary sigmoid + axisSlope + constant  0.30  11.86  0.01  -0.44  Primary sigmoid + geologyMod + constant  0.21  23.96  -0.44  0.32  Primary sigmoid  0.25  18.09  Primary sigmoid + siteIndex + constant  0.19  19.27  0.01  -0.32  Primary sigmoid + geologyMod + scarLength + constant  0.24  21.76  -0.47  0.00  0.34  Primary sigmoid + geologyMod + convergence + constant  0.22  23.36  -0.45  0.00  0.30  Primary sigmoid + scarWidth + constant  0.22  19.52  0.00  -0.03  Primary sigmoid with internal scalar  0.25  1.01  18.33  Basic sigmoid  0.25  0.80  22.83  Primary sigmoid + logged + constant  0.24  18.23  0.01  0.00  Primary sigmoid + PPT911 + constant  0.25  17.83  0.00  0.02  Primary sigmoid + scarLength + constant  0.27  17.61  0.00  -0.03  Primary sigmoid + convergence + constant  0.26  17.43  0.00  -0.02  110  b  0.49 0.00  0.51  Table 3.7. Model descriptions, R2, and AICc for each model including a geology variable. R2  AICc  Primary sigmoid + geology + axisSlope + constant  0.40  -24.36  Primary sigmoid + geology + scarLength + constant  0.30  -24.04  Primary sigmoid + geology + constant  0.35  -22.82  Primary sigmoid + geology + convergence + constant  0.37  -20.67  Average  0.36  -22.97  Primary sigmoid + geologyMod + axisSlope + constant  0.47  -24.98  Primary sigmoid + geologyMod + constant  0.45  -18.20  Primary sigmoid + geologyMod + scarLength + constant  0.47  -16.01  Primary sigmoid + geologyMod + convergence + constant  0.44  -15.63  Average  0.46  -18.71  Model description Geology  GeologyMod  The accuracy of the models to predict the soil depths of the test data set was visualized by observed versus predicted plots. The coefficient of determination was also calculated for each model against the test data set. The observed versus predicted plots for the primary sigmoid model combined with the geology and axisSlope variables is presented in Figure 3.11.  111  Table 3.8. Model testing results, ranked by PMSE (squared mean residual + error variance), AICc Rank and Evidence ratio from the calibration have been included for ease of comparison. AICc rank  Evidence ratio  PMSE rank  PMSE  PMSE (BC) rank  PMSE (BC)  PMSE (Japan) rank  PMSE (Japan)  Primary sigmoid + geology + convergence + constant  5  8.6  1  0.028  2  0.022  9  0.040  Primary sigmoid + geology + constant  4  2.9  2  0.028  4  0.027  3  0.033  Primary sigmoid + siteIndex + constant  9  86.7  3  0.030  5  0.030  NA  NA  2  1.4  4  0.032  3  0.025  13  0.047  3  1.6  5  0.032  1  0.020  15  0.056  Primary sigmoid  8  43.9  6  0.035  9  0.039  4  0.034  Primary sigmoid + convergence + constant  18  481.7  7  0.036  10  0.039  5  0.034  Primary sigmoid + logged + constant  15  334.8  8  0.036  11  0.039  1  0.026  Primary sigmoid + scarLength + constant  17  477.5  9  0.036  12  0.040  6  0.035  Primary sigmoid + scarWidth + constant  12  133.6  10  0.036  15  0.041  2  0.033  Primary sigmoid + axisSlope + constant  6  10.9  11  0.037  6  0.037  10  0.041  Primary sigmoid + PPT911 + constant  16  462.8  12  0.038  8  0.038  NA  NA  Primary sigmoid + geologyMod + convergence + constant  11  106.4  13  0.040  16  0.043  8  0.040  Primary sigmoid + geologyMod + constant  7  29.4  14  0.040  13  0.041  11  0.043  Primary sigmoid + geologyMod + scarLength + constant  10  88.1  15  0.041  7  0.038  14  0.049  Primary sigmoid with internal scalar  13  140.9  16  0.044  18  0.051  7  0.037  Primary sigmoid + geologyMod + axisSlope + constant  1  1.0  17  0.045  14  0.041  16  0.056  Basic sigmoid  14  320.8  18  0.046  17  0.050  12  0.046  Model description  Primary sigmoid + geology + axisSlope + constant Primary sigmoid + geology + scarLength + constant  112  Figure 3.11. Observed versus predicted soil depths for the primary sigmoid with geology variable. Grey line represents a 1:1 relation.  Conceptual model of stochastic infilling The conceptual model of stochastic infilling is presented in Figures 3.12 through 3.14. The underlying plot in each frame is the relation of the ratio of mean soil depth in the failure area divided by the mean 113  soil depth on the adjacent hillslope plotted against time. The dashed lines indicate the envelope inside which most study sites exist. The lower envelope represents the approximate rate of accumulation of small scale processes following shallow landslide. The upper envelope represents the maximum infilling at the failure area, the actual ratio of which is determined by the slope morphology in the vicinity of each failure and sites may be unlikely to reach such a maximum state except in cases of long periods of stability (Reneau and Dietrich, 1991). Here the upper envelop has been set at just greater than unity to capture the effect of slight hillslope convergence common at the study sites (Figure 3.9). For each figure, the stronger colour signifies higher probability of stochastic processes and higher volumes of material affected. In Figure 2.12, the probability of small-scale deposition is greatest for failure areas <50 years old due to incomplete vegetation establishment (Chapter 2) and also greater for failure areas with a smaller ratio of failure area soil depths compared to hillslope soil depths. Small-scale processes of material deposition will continue to be active and contribute to soil depth accretions even in areas of the figure that are not shaded, but the probability of infilling is relatively smaller. In the second figure of the conceptual model (Figure 2.13), the probability of small-scale erosive events (i.e., transport of material out of the failure area) is portrayed. The erosive potential is restricted to failure areas of <30 years since landsliding. The thinnest soils have slightly higher probabilities of small-scale erosion because much of the subsurface flow from above the contributing area will issue from the scarps as surface flow and increase the likelihood of rill erosion (Bryan, 2000). Figure 2.14 portrays the probability of subsequent failure. It is a time-independent relation that is controlled entirely by the depth of soil. It has been presented on the same basic plot but it should be noted that the probability of failure will be governed by other site level variables such as the hillslope soil depth (even 100% recovery may not provide enough soil depth for subsequent failure), the degree of cohesion at the failure plane, and other factors. Of the sites >50 years since failure, 50% have failure area to hillslope soil depth ratios of >0.75 and only a highly divergent site (Return) has a failure area to hillslope soil depth ratio of <0.5. 114  Figure 3.12 Conceptual model component 1. Relation of probability of small-scale stochastic deposition on plot of ratio of failure area soil depth compared to hillslope soil depths through time.  115  Figure 3.13. Conceptual model component 2. Relation of probability of small-scale stochastic erosion on plot of ratio of failure area soil depth compared to hillslope soil depths through time.  116  Figure 3.14. Conceptual model component 3. Relation of probability of catastrophic erosion on plot of ratio of failure area soil depth compared to hillslope soil depths through time.  117  DISCUSSION One statistical and one conceptual model have been proposed to explain soil depth accretion in the failure areas of shallow landslide initiation zones. As a result of field observations, the stochastic nature of forest and hydrogeomorphic processes acting in the vicinity of failure areas are thought to be responsible for the inability of the statistical model to characterize the variability in measured soil depths. The statistical model has provided information regarding the central tendency of accumulation rates and has provided some indication of controlling factors; the conceptual model addresses the probability of stochastic processes to affect infilling rates based on observations from >60 shallow landslide failure areas. The depth of material accumulations in failure areas has a temporal trend. The non-linear modeling was an attempt to portray the typical progression following full-depth evacuation. The sigmoid model was selected because it has both conceptually and statistically good fit. This model has been used previously in the literature (Sidle, 1987) but other model forms have also been used; Dietrich and Dunne (1978) used a power-law relation, Shimokawa (1984) used an undescribed non-linear curve, and Shimokawa et al. (1989) used a linear relation. Field data collected in this study support a non-linear relation such as represented by the selected sigmoid model but also could support one of the other non-linear models evaluated. Two general discrepancies exist between the sigmoid model and the measured volumes of recent material transport (Figure 3.7). The early discrepancy (0-6 year) may be explained by varying material retention in this first period. If a high proportion of material is not deposited but instead is transferred through the failure area, an occurrence verified anecdotally in the field, there would be little soil depth accretion in the failure area. Landslide sites are known to contribute sediment beyond the catastrophic pulse at failure (Imaizumi et al., 2008) and the sigmoid model supports such a period of material export from the failure area. Alternatively, if material inputs are trapped with high efficiency in the failure area, then soil depths would begin increasing immediately, particularly in the first period 118  following failure when material transport rates are high. This would then be best modelled with an immediately rising curve, such as produced by the Verhulst or asymptotic models. Predictions for all models are relatively close in this first period, and both model types fit the pattern reasonably considering the scatter in the soil depths at recently failed sites. The other discrepancy, in the range of 20 to 50 years following failure, predicts accumulation rates well above observed material transport volumes. The small number of sites in that age range may have been insufficient to capture the nature of the recharge processes. Beyond the second period (>50 years), the observed processes and predicted accumulation for an average-sized failure area are quite close. At that age, the failure areas are mostly vegetated and the head and side scarps attain a shallower and more stable slope angle. At this time, the value of the modeled sigmoid curve also declines, and the observed volumes of recent material transport decrease to agree well with the first derivative of the predicted soil accumulation curve. Considering the hillslope sediment budget (Slaymaker, 2003), there is a balance between inputs, outputs, and change in storage. On hillslopes in British Columbia, the major input of sediment occurred during Late Pleistocene glaciation. Although the study sites were restricted to colluvial materials, there remains an abundance of glacial sediment on many slopes throughout the province (Ashmore, 1993). This hillslope storage provides a medium for production forestry but also provides ample material for shallow landslides. In Japan, long-term weathering replaces glaciation as the method of producing hillslope materials. Depending upon the rate of organic soil material production, the rate of bedrock weathering, the rate of material transfer from upslope sources, and the rate of erosion (via shallow landsliding), contributing areas of the shallow landslides could become depleted of soil material over repeated failure cycles. This would negatively affect the rate of material transfer into the failure area as material accumulation is positively influenced by hillslope soil depths. Eventually, material accretions in the failure area could be solely the result of organic matter deposition and the products of bedrock weathering with little material being transferred to the failure area from surrounding hillslopes. The depletion of 119  hillslope surficial material would occur relatively more rapidly in cases of competent, weatheringresistant bedrock and on slopes that do not support recharge from upslope. Failure sites with weak, easily weathered bedrock may continue to produce both mineral and organic material for infilling failure sites. Organic matter accumulation is controlled by the productivity of the site and the rate of organic matter breakdown. Qualitative estimates of organic matter at the study sites found the recharged material in several of the landslides with the lowest recharge rates being predominantly organic matter supporting the concept of lesser contributions from sites with more competent geological parent materials. At several study sites, more than one failure area was found indicating that successive failures had not failed in exactly the same location or to the same extent, but were depleting the hillslope material storage. As hillslope soils in the vicinity of failure margins are depleted over repeated failure cycles, upslope contributing areas will become more important for material infilling. Either hillslope material will move down towards the established failure site, or more likely, the location of the future failures will change. Subsequent failures could be expected be located further upslope initially, evacuating source areas. Once the source area is depleted of excess storage, future failures could be expected to be located further downslope, thereby increasing the contributing area. Unfortunately, in the study area, such long-term processes are both dependent upon forests for the production of organic material and simultaneously the forest obscures the evidence of such past events. Therefore, the study of such long-term processes at such a small spatial scale is severely challenged. Non-linear model performances were relatively consistent between the calibration and testing results. However, seven models performed better than the primary sigmoid model during the modeling phase but only three of these out-performed the primary model in the testing phase (Tables 2.5 and 2.8). An obvious advantage for some models was the addition of extra variables, particularly as they were assessed with the PMSE measure, which does not include a penalty for additional variables as does the AICc. The influence 120  of the Japanese sites on the testing data set (7 of 17 testing sites) may have altered the results; however a balance of calibration and testing performance was considered when selecting the best models. Most models were similar in their predictions as a consequence of using the primary model and the inherent weakness of most of the descriptive variables. Considering the relative strength of the geology variable and its inclusion in most of the top models, the primary sigmoid with the geology variable is quite likely the best model to predict soil accretion. However, the soil depths at some sites were not well predicted. Substantial under-prediction occurred at two sites used in the calibration set from the South Coast study area. One of these sites (Stihl1) had highly convergent morphology and steep slopes, effectively funnelling materials down into the failure area; the other site (Max) had a rather low score of the geology parameter, but no other outstanding features to explain its deep accumulation. A South Coast study area site from the model testing set (Home2) was also under-predicted. It had relatively average hillslope soil depths (0.75 m, mean = 0.69 m), but had a mean head scarp height of 1.7 m. The large disturbed face was delivering a great deal more sediment than would be expected given the hillslope soil depths. One South Coast site (Return) was strongly overpredicted (observed = 0.26 m, predicted = 0.68 m). The site had relatively deep soils above the head scarp, but it was a strongly divergent site and material was being dispersed over the topographic nose. From Japan, the sites of <25 years since failure were all overpredicted. Although material is being transferred into the failure areas, the high intensity precipitation events during the typhoon and Baiu seasons could be scouring the failure areas. Especially in the first decades after failure, this scouring could be detrimental to vegetation establishment and the subsequently trapping further material inputs (Rey, 2004); this is confirmed by field observations and encompassed within the proposed conceptual model. Although recent material transport volumes recorded during fieldwork were small, this could be due to the distribution of small-scale processes. In Japan, slope wash deposits were noted at many survey points but not included in measured material transport volumes as the deposits did not exceed the 1 cm lower threshold for deposit thickness. The forests adjacent to all the 121  Japanese sites are intensively managed with little coarse woody debris or understory vegetation to capture hillslope material into distinct deposits of >1 cm depth. The combined effect of forest conditions with more intense precipitation could explain the over-prediction of the sites <25 years old in Japan. Once material begins to deposit in the failure area and vegetation is established, the accumulation rate at the sites in Japan may be higher than at the study sites in British Columbia creating a potential for underprediction. As an illustration, one of the older failure areas in Japan was estimated nearly perfectly, the other was under-estimated. The RedPine site (observed = 0.86 m, predicted = 0.54 m) had one of the highest scores of the geology variable. The expected and realized negative parameter estimate for the geology parameter was intended to reflect the decreased material availability from highly scoring bedrock types. However, this relation did not extend to this particular site as it had relatively deep soils on the contributing hillslopes and therefore material infilling was not limited by material availability. The other older Japanese site (Waka17) also had deep soils compared to many of the sites in the other study areas, but it also had a moderate score of the geology parameter that resulted in a more moderate prediction. Both of the older Japanese sites were dominated by litterfall as an infilling process, but both had some occurrences of slope wash of mineral material. Within the non-linear modeling, two site variables performed well, the average depth of hillslope soils (HsdMean) and the variable describing the geomorphic rock mass (geology). The use of the HsdMean variable in the primary sigmoid model provided better results than using a constant value (b). This is intuitive, as, stronger microtopographic relief at the failure site will provide more energy for soil diffusive processes to redistribute hillslope materials (Roering et al., 2001). The geology variable includes intact rock strength, weathering, joint spacing, joint orientation, joint width, and joint continuity. Selby‟s (1980) classification system reduced the bedrock source material to a single numerical value, and in most cases, it appeared to describe the geological contributions to infilling well. For the application of this classification to modeling soil accumulation, including a sub-parameter regarding the texture of 122  weathering products may have strengthened its performance. Comparison of the original (geology) and modified (geologyMod) geological variables revealed that the original geological variable (geology) was better suited to describing material recharge. The better result of geology suggests that bedrock with dip out of slope may result in greater material transport from the geomorphically contributing area and that it has a greater effect on the accumulation of material than the increased roughness, and thus material retention, of bedrock with dip into the slope in the failure area. Bedrock with dip steeply out of the slope is associated with increased landslide activity (Sidle et al., 1985). Over time, through a balance of bedrock weathering and soil erosion, thinner soils could result on contributing hillslopes with out-of-slope dip, and thicker soils on contributing slopes with into-slope dips. Field data support this relation with a significant (p < 0.05) regression, though the R2 is <0.1 (Figure 3.15).  123  Figure 3.15. Relation between mean hillslope soil depth and the joint orientation score of the geology variable. The contributions of other site-level variables, beyond HsdMean and geology, to the prediction of infilled soil depths are less substantial. Both convergence and axisSlope were poor performers in both the model 124  calibration and testing phases, yet when combined with geology produced two of the top models. Slope angle is the basic driver of hillslope processes and the outcome may reflect this. Yet in cases of little soil material, the effect of slope gradient will not be great. Similarly, the effect of convergence may be limited by available hillslope material. The siteIndex variable performed well in the testing phase even though it was in the 9th ranked model after the model calibration. Vegetation can facilitate bedrock weathering through both physical and chemical processes. Windthrow of Sitka alder (Alnus viridis (Chaix.) D.C.) was observed to displace ~0.03 m3 of bedrock in a single exposed root system. Even blocks of competent bedrock types with high geomorphic rock mass scores were observed in windthrown root systems (Figure 3.16). Both fine and coarse woody debris contribute to organic accumulations but coarse woody debris also contributes structure that results in deeper accumulations of colluvial material (Wilford, 1984) (Figure 3.17; also see photographs in Appendix B). The site index for the failure sites in Japan could not be determined as forest productivity in Japan is protected information and thus models including the siteIndex variable were only tested on the sites from British Columbia. The poor performance of the precipitation variable is likely due to the use of the average fall precipitation; a better measure may have been the frequency of precipitation above a certain intensity (Jungerius and ten Harkel, 1994) but no such data were available. Typically the highest precipitation storms arrive on the coast of British Columbia during the fall (Lewis and Moran, 1985), and therefore the spatially-distributed average fall precipitation from ClimateBC was thought to be the best available information to characterize the climate influence over the entire period of material accumulation. No distributed precipitation data was available for the sites of the Kii Peninsula in Japan. In the tested linear relation, the convergence variable was a surprisingly poor predictor of soil accumulation. Topographic depressions on steep hillslopes have been observed to be more susceptible to shallow landsliding than other slope forms (Dietrich and Dunne 1978; Tsukamoto et al., 1982; Rollerson, 1992; Montgomery et al., 2000). However, hillslope convergence is commonly measured qualitatively rather than quantitatively as in this study. In this opportunistic study 125  design, where the majority of landslides encountered in the field were sampled, the preferential distribution of landslides on hillslopes with slight convergence is evident (Figure 3.9). The limited range and distribution of convergence in the calibration data prevents a thorough examination of its effects on rates of infilling. Unfortunately, details of subsurface topography were only collected at a subsample of the studied landslides and thus prevented the inclusion of a subsurface convergence variable in the models. A limited sample range and distribution of both the scarLength and scarWidth variables likely contributed to their poor performance as well (Figure 3.10). The logged variable was not strong in either modeling or testing. In British Columbia, the oldest sites with timber harvesting were only 27 years old; although two sites in Japan were >90 years old. Changes to soil hydrology can occur as a result of log yarding (Purser and Cundy, 1992) and root decay results in decreased root reinforcement at logged sites within the first decade after harvest (Sidle, 1991). Therefore, these changes would have been realized prior to sampling many of the logged sites. Investigation of timber harvesting and landslides in British Columbia is difficult because much of the logging prior to ~30 years ago was in valley bottoms and other sites less prone to failure. Further, changes to harvesting methods over time may confound the data. Stringent field site requirements were employed to limit the variability of other factors such as topography and material types. A criterion of site suitability was a mid-slope position; this, and the shallow soils, was expected to result in minimal differences between the slope of the failure axis (axisSlope) and the slope of upslope contributing areas. Figure 3.18 displays the relation between the slope of the failure axis and the slope of above head scarp areas as determined by digital elevation models constructed from survey data at 33 sites. The study sites from Japan seem to plot relatively high, this is likely due to the more developed nature of the sites displaying a more graded form with lower slopes in the axis of the failure relative to the surrounding slopes. Such form reinforces the sites as sites of hillslope deposition. Contributions of slope material moving from outside of the margins of the failures were not common. Typically, material transported on the contributing hillslopes was trapped in microtopographic 126  depressions or by vegetation or woody debris before reaching the failure area except in Japan where these features were less common. Despite the similarity of slope values in Figure 3.18, and the limited contributions of material from upslope positions noted during fieldwork, the gradient of areas above head scarps may have proven useful in this analysis. This measure would be best limited to the local slope of the area immediately above the head scarp, as this is the dominant source region. At many of the study sites, the upslope hydrogeomorphic contributing area had only weak topographic expression and preferential flow pathways have been found to have some independence from topography (Hutchinson and Moore, 2000). As a result, defining attributes of the upslope areas was not pursued in the field, and information available from regional digital elevation models was also not thought to be valuable. The initial depth of material at the failure site may be a source of error and create problems for the predictive models. Relations were sought and a significant (p<0.01) correlation was found between remaining soil depth and the geology variable for failure areas of the modeling data that were <10 years old. Extending the relation to the older sites to correct for initial conditions was unsatisfactory as many of the older failures had corrected material depths of <0 cm indicating that the relation was likely not accurate or transferrable. The controls on the depth of soil immediately following landslide initiation remains uncertain but very few potential sites were excluded due to evidence of a residual soil depth >20 cm. Following the inability of the analytical methods to reduce the unexplained variance in the data of soil depth accretion, a conceptual model was developed. The conceptual model emphasizes the stochastic elements of forest and geomorphic processes that are not well captured with the collected site-level variables. Small-scale processes, such as inventoried occurrences of recent material transport (Chapter 2), result in material accumulation, but these are not entirely sufficient to explain the measured increases in soil depth in failure areas. Stochastic forest and hydrogeomorphic processes may explain this discrepancy 127  and the variability in the non-linear modeling results. Stochastic forest processes would include the occurrence of windthrow, including both the contribution of large woody debris volumes and the contribution of physical structure to the failure area, as well as the displacement of soil material and bedrock in close proximity to the failure area. Windthrown root systems can initiate sediment transport; previous studies have shown that much of the displaced sediment returns to the site where the root system previously occupied except on steep slopes (Norman et al., 1995). However, these studies have not been conducted at failure sites where tree stability may be compromised as a result of the failure and the presence of head and side scarps. Stochastic forest hydrogeomorphic processes also include litterfall or windfall of material that facilitates material accumulation, and the root reinforcement, or release, of hillslope soils (Hales et al., 2009). Observations at landslide initiation zones within the study areas but not included in the study found individual material recharge events up to ~8 m3. The occurrence of these low-frequency, and relatively high-magnitude events could strongly alter the measured material accumulation at specific failed sites. This strongly stochastic nature of accumulation events challenges the model of gradual accretion as supported by the curve in Figure 3.6. However, the model should be viewed as an approximation of the average condition, not the specific trajectory of individual failure areas at a particular time. The probability of stochastic processes is expected to remain constant through time, but the influence on material accumulation will be much greater when combined with other factors associated with material transport such as unvegetated slopes and strong surface topographic gradients. Thus, the probability of stochastic processes that influence material accumulation is greater for sites of < 50 years since failure when large portions of the failure area remain unvegetated and for sites where the difference in soil depths between failure area and hillslope is large.  128  The probability of small-scale erosive events is largely controlled by the period that the failure area is unvegetated. The position of soil-water outlets from the head scarp may also be important as the relatively shallow soils in the failure area have an increased likelihood of saturated overland flow and associated erosion. The final component of the conceptual model relates to the potential for catastrophic erosion. Although this is more a factor of absolute, rather than relative, soil depth, the same base plot was kept for consistency. Typically, the hillslope soil depth will not be dramatically affected by a single failure and so an increasing soil depth ratio represents a progression towards a condition similar to the previous failure and an associated rise in the probability of subsequent failure. Additional sample sites would make both the models more robust. This research would have particularly benefited from more failures >100 years in age. These older failures are difficult to locate in the field, particularly without introducing a bias towards larger failures with increasing age. The failures of <10 m width quickly become obscured and determining the age of the failure is a particular problem as forest ecological processes obscure the evidence. After the passing of time the failure becomes almost indiscernible and dendrochronological methods would no longer provide a useful dating proxy. The use of other dating methods such as optical dating methods could provide more opportunities for dating older landslides (Prescott and Robinson, 1997).  129  Figure 3.16. Windthrown root system that has lifted a large block of competent gneissic bedrock with moderately spaced joints (0.3–1.0 m).  130  Figure 3.17. Colluvial material accumulation upslope of coarse woody debris in a shallow landslide failure area in south-central Japan. Dry ravel of coarse material as depicted here was a minor process at study sites.  131  Figure 3.18. Relation between gradient of failure axis (axisSlope) and mean hillslope gradient. Solid line is a 1:1 relation.  132  CONCLUSIONS Statistical models based on field data from three study areas in British Columbia and one study area in Japan were developed to predict the depth of accumulated soil at landslide failure areas. Six basic model forms were evaluated against field observations of soil accumulations and the volume of recent material transport in the initiation zone. The sigmoid model was selected as the strongest model as it most closely matched the field data and observations. Although soil accumulation in the failure areas of shallow landslides depends on various factors, including some not parameterized in this study, the accumulation rate of soil material is expected to decline beyond ~200 years after failure. Field observations of material transport and deposition in the failure area indicate that the majority of soil accumulation occurs within 50 years of failure. The addition of attributes describing the failure area and immediately adjacent slopes resulted in an increase in model accuracy when models were compared against a near-random selection of the data from British Columbia and Japan. Of the variables assessed, two appeared most effective at increasing model accuracy: 1) HsdMean, the average depth of soils on the adjacent slopes was included as an integral component of the primary sigmoid model; and 2) geology, a quantitative measure of the rock mass (Selby, 1980). The variable describing the slope of the failure axis (axisSlope) performed well when combined with the geology variable. The remaining variables did not appear to contribute strongly to the modelling: average fall precipitation, the dimensions of the failure area, a measure of the hillslope convergence, and the presence or absence of timber harvesting.  133  A conceptual model was developed in response to the stochastic nature of forest and hydrogeomorphic processes in the vicinity of shallow landslide failure areas. Based on field observations of >60 failure areas, the qualitative probability of stochastic processes resulting in substantial influence to material transport is presented. Small-scale deposition and erosion, as well as catastrophic erosion are related to the ratio of failure area soil depths to hillslope soil depths and the time since failure. The time since failure is a surrogate for the degree of revegetation and the stability of head and side scarps.  134  REFERENCES Adams PW, Sidle RC. 1987. Soil conditions in three recent landslides in southeast Alaska. Forest Ecology and Management 18: 93–102.  Akaike H. 1973. Information theory and an extension of the maximum likelihood principle. In Second International Symposium on Information Theory, B. N. Petrov and F. Csake (eds). Akadamiai Kiado: Budapest; 267–281.  Alger CS, Ellen SD. 1987. Zero-order basins shaped by debris flows, Sunol, California, USA. In Erosion and sedimentation in the Pacific Rim (Proceedings of the Corvallis Symposium, August, 1987) International Association of Hydrological Sciences Publication No. 165: 111–119.  Ashmore P. 1993. Contemporary erosion of the Canadian landscape. Progress in Physical Geography 17(2): 190–204. DOI: 10.1177/030913339301700206  Benda LE, Cundy TW. 1990. Predicting deposition of debris flows in mountain channels. Canadian Geotechnical Journal 27: 409–417.  Brayshaw D, Hassan MA. 2009. Debris flow initiation and sediment recharge in gullies. Geomorphology 109: 122–131. DOI:10.1016/j.geomorph.2009.02.021  Bryan RB. 2000. Soil erodibility and processes of water erosion on hillslope. Geomorphology 32: 385– 415.  135  Burnham KP, Anderson DR. 1998. Model selection and inference: A practical information-theoretic approach. Springer-Verlag: NewYork.  Crozier MJ, Vaughan EE, Tippett MJ. 1990. Relative instability of colluvium-filled bedrock depressions, Earth Surface Processes Landforms, 15: 329–339.  Dietrich WE, Dunne T, 1978. Sediment budget for a small catchment in mountainous terrain. Z. Geomorph. N. F. Suppl. Bd. 29: 191–206.  Dietrich WE, Dunne T, Humphrey NF, Reid LM, 1982. Construction of sediment budgets for drainage basins. In Sediment Budgets and Routing in Forested Drainage Basins, Swanson FJ, Janda RJ, Dunne T, Swanston DN (eds). USFS Gen. Tech. Rept. PNW 141: 5–23.  Environment Canada. 2008. http://www.climate.weatheroffice.ec.gc.ca/climate_normals/index_e.html, June 2008. Geertsema M, Highland LM, Vauegois L. 2009. Environmental impact of landslides. In Landslides – Disaster Risk Reduction, Sassa K, Canuti P (eds.). Springer-Verlag: Berlin Heidelberg 585–604.  Goffe WL, Ferrier GD, Rogers J. 1994. Global optimization of statistical functions with simulated annealing. Journal of Econometrics 60: 65–100.  Gomi T, Sidle RC, Bryant MD, Woodsmith RD. 2001. Characteristics of woody debris and sediment distribution in headwater streams, southeast Alaska. Canadian Journal of Forest Research 31: 1386– 1399.  Hack JT, Goodlett JC. 1960. Geomorphology and forest ecology of a mountain region in the central Appalachians, US Geolological Survey Professional Paper 347, 66 pp. 136  Hales TC, Ford CR, Hwang T, Vose JM, Band LE. 2009. Topographic and ecologic controls on root reinforcement, Journal of Geophysical Research 114: F03013, doi:10.1029/2008JF001168.  Hamann A, Wang T. 2005. Models of climatic normals for genecology and climate change studies in British Columbia. Agricultural and Forest Meteorology 128: 211–221.  Holland SS. 1976. Landforms of British Columbia: A physiographic outline. British Columbia Department of Mines and Petroleum Resources: Victoria, British Columbia; Bulletin 48.  Hungr O, McDougall S, Bovis M. 2005. Entrainment of material by debris flows. In Debris-flow hazard and related phenomena, Jakob M, Hungr O (eds). Springer-Praxis: Berlin, Germany; 135–158.  Hutchinson DG, Moore RD. 2000. Throughflow variability on a forested hillslope underlain by compacted glacial till. Hydrological Processes 14: 1751–1766.  Ihaka R, Gentleman R. 1996. R: A language for data analysis and graphics. Journal of Computational and Graphical Statistics 5: 299–314.  Iida T. 1999. A stochastic hydro-geomorphological model for shallow landsliding due to rainstorm. Catena 34: 293–313.  Iida T. 2004. Theoretical research on the relationship between return period of rainfall and shallow landslides. Hydrological Processes 18: 739–756.  Imaizumi F, Sidle RC. 2005. Relationship between sediment supply and transport processes in Miyagawa dam catchment. Annals of Disaster Prevention Research Institute, Kyoto University: Kyoto No. 48C.  137  Imaizumi F, Sidle RC, Tsuchiya S, Ohsaka O. 2006. Hydrogeomorphic processes in a steep debris flow initiation channel. Geophysical Research Letters 33, L10404, doi:10.1029/2006GL026250  Imaizumi F, Sidle RC, Kamei R. 2008. Effects of forest harvesting on the occurrence of landslides and debris flows in steep terrain of central Japan. Earth Surface Processes and Landforms 33: 827–840. DOI:10.1002/esp.1574.  Innes JL. 1983. Debris flows. Progress in Physical Geography 7: 469–501.  Jakob M, Bovis M, Oden M. 2005. The significance of channel recharge rates for estimating debris-flow magnitude and frequency. Earth Surface Processes and Landforms 30: 755–766. doi: 10.1002/esp.1188  Jakob M, Hungr O. 2005 Debris-flow hazards and related phenomena. Springer: Berlin.  Jenny H. 1941. The factors of soil formation. McGraw-Hill: New York.  Jungerius PD, ten Harkel MJ. 1994. The effect of rainfall intensity on surface runoff and sediment yield in the grey dunes along the Dutch coast under conditions of limited rainfall acceptance. Catena 23: 269-279.  Karanka EJ. 1986. Trends and fluctuatlons in precipitation and stream runoff in the Queen Charlotte Islands. British Columbia Ministry of Forests, Research Program: Victoria British Columbia; Land Management Report 040.  Kobayashi S, Gomi T, Sidle RC, Takemon Y. 2010. Variations of headwater stream ecosystems related to stand age of managed forest catchments: responses of macroinvertebrate community to forest  138  clearcutting and debris flow occurrence. Canadian Journal of Fisheries and Aquatic Sciences 67 (in press). doi:10.1139/F09-186 Lewis PJ, Moran MD. 1985. Severe storms off Canada’s West Coast: A catalogue summary of the period 1957 to 1983. Atmospheric Environment Service, Environment Canada.  Maynard (Denny) and Associates Ltd. and Golder Associates, 2004. Klanawa terrain and landslide hazards mapping: Pilot study report. Prepared for Weyerhaeuser Company Limited BC Coastal Group – Nanaimo Woodlands.  Millard T. 1999. Debris flow initiation in Coastal British Columbia Gullies. British Columbia Ministry of Forests: Nanaimo British Columbia; Technical Report.  Mitchell KJ. 1988. Site index curves and tables for British Columbia, coastal species. Forest Resource Development Agreement Report 037, Victoria BC.  Montgomery DR, Dietrich WE, 1994. A physically based model for the topographic control on shallow landsliding. Water Resources Research. 30: 1153–1171.  Montgomery DR, Schmidt KM, Dietrich WE, Greenberg H. 2000. Forest clearing and regional landsliding. Geology 28: 311–314.  Muller JE. 1977. Geology of Vancouver Island. Geological Survey of Canada Open File 463.  Norman SA, Schaetzl RJ, Small TW. 1995. Effects of slope angle on mass movement by tree uprooting. Geomorphology 14: 19–27.  139  Ohnuki Y, Terazono R, Ikuzawa H, Hirata I, Kanna K, Utagawa H. 1997. Distribution of colluvia and saprolites and their physical properties in a zero-order basin in Okinawa, southwestern Japan. Geoderma 80: 75–93.  Prescott JR, Robertson GB, 1997. Sediment dating by luminescence: A review. Radition Measurements 27: 893–922.  Purser MD, Cundy TW. 1992. Changes in soil physical properties due to cable yarding and their hydrologic implications. Western Journal of Applied Forestry 7(2): 36–39.  Rapp A. 1960. Recent development of mountain slopes in Karkevagge and surroundings, northern Scandinavia. Geografiska Annaler 42: 65–200.  Reneau SL, Dietrich WE. 1991. Erosion rates in the southern Oregon Coast Range: Evidence for an equilibrium between hillslope erosion and sediment yield. Earth Surface Processes and Landforms 16: 307–322.  Reneau SL, Dietrich WE, Donahue DJ, Hull AJT, Rubin M. 1990. Late Quaternary history of colluvial deposition and erosion in hollows, central California Coast Ranges. Geological Society of America Bulletin 102: 969–982.  Rey F. 2004. Effectiveness of vegetative barriers for marly sediment trapping. Earth Surface Processes and Landforms, 29:1161–1169.  Roering JJ, Kirchner JW, Sklar LS, Dietrich WE. 2001. Hillslope evolution by nonlinear creep and landsliding: An experimental study. Geology 29 (2):143–146.  140  Rollerson TP. 1992. Relationships between landscape attributes and landslide frequencies after logging: Skidegate Plateau, Queen Charlotte Islands. BC Ministry of Forests Land Management Report 076.  Schwab JW. Unpublished Report. Research Project EP 782.07. Historical documentation of mass wasting in North Coastal British Columbia. Forest Sciences Section, BC Forest Service, Smithers, BC.  Selby MJ, 1980. A rock mass strength classification for geomorphic purposes: with tests from Antarctica and New Zealand. Zeitschrift für Geomorphologie 24: 31–51.  Selby MJ. 1993. Hillslope materials and processes. 2nd ed., Oxford University Press: Toronto.  Shimokawa E. 1984. Natural recovery process of vegetation on landslide scars and landslide periodicity in forested drainage basins. Proceedings of the Symposium on effect on forest land use on erosion and slope stablility East-West Center, Honolulu Hawaii, 99-108.  Shimokawa E, Jitouzono T, Takano S. 1989. Periodicity ofshallow landslide on Shirasu (Ito pyroclastic flow deposits) steep slopes and prediction of potential landslide sites. Transactions of the Japanese Geomorphological Union 10: 267–284.  Sidle RC. 1987. A dynamic model of slope stability in zero-order basins. In Erosion and sedimentation in the Pacific Rim (Proceedings of the Corvallis Symposium, August, 1987) International Association of Hydrological Sciences Publication No. 165: 101–110.  Sidle RC. 1991. A conceptual model of changes in root cohesion in response to vegetation management. Journal of Environmental Quality 20: 43–52.  Sidle RC, Chigira M. 2004. Landslides and debris flows strike Kyushu, Japan. EOS Transactions American Geophysical Union 85: 145–156. 141  Sidle RC, Ochiai H. 2006. Landslides: Processes, Prediction, and Land Use. American Geophysical Union: Washington DC. Sidle RC, Pearce AJ, O’Loughlin CL. 1985. Hillslope Stability and Land Use. American Geophyscial Union: Water Resources Monograph 11.  Sidle RC, Swanston DN. 1982. Analysis of a small debris slide in coastal Alaska. Canadian Geotechnical Journal, 19: 167–174.  Slaymaker O. 2003. The sediment budget as conceptual framework and management tool. Hydrobiologia 494:71–82.  Smale MC, McLeod M, Smale PN, 1997. Vegetation and soil recovery on shallow landslide scars in Tertiary Hill Country, East Cape Region, New Zealand. New Zealand Journal of Ecology 21: 31–41.  Sutherland-Brown A. 1968. Geology of the Queen Charlotte Islands. British Columbia Department of Mines and Petroleum Resources. Bulletin 54.  Tsukamoto Y, Minematsu H. 1987. Hydrological characteristics of a zero-order basin. Erosion and Sedimentation in the Pacific Rim (Proceedings of the Corvallis Symposium, August, 1987) IAHS Publication 165. 61– 70.  Tsukamoto Y, Ohta T, Noguchi H. 1982. Hydrological and geomorphological study of debris slides on forested hillslope in Japan. In Recent Developments in the Explanation and Prediction of Erosion and Sediment Yield. Walling DE (ed). International Association of Hydrological Sciences Publication No. 137: 89–98.  142  Turner AK, Schuster RL. 1996. Landslides: investigation and mitigation. Transportation Research Board, National Research Council. National Academy Press: Washington DC; Special Report 247.  Tutiempo.net, http://www.tutiempo.net, 20 April 2010.  Wang T, Hamann A, Spittlehouse D, Aitken SN. 2006. Development of scale-free climate data for western Canada for use in resource management. International Journal of Climatology 26: 383–397.  Wilford DJ. 1984. The sediment-storage function of large organic debris at the base of unstable slopes. In: Meehan, W.R.; Merrell, T.R.; Hanley, T.A., ed. Fish and wildlife relationships in oldgrowthforests: Proceedings of a symposium. American Institute of Fishery Research Biologists; 115– 119.  Wilford DJ, Sakals ME, Innes JL. 2003. Forestry on fans: a problem analysis. Forestry Chronicle 79: 291–296.  Wilford DJ, Cherubini P, Sakals ME. 2005. Dendroecology: a guide for using trees to date geomorphic and hydrologic events. British Columbia Ministry of Forests, Research Branch: Victoria British Columbia; Land Management Handbook No. 58. .http://www.for.gov.bc.ca/hfd/pubs/Docs/Lmh/Lmh58.htm  Yamada S. 1999. The role of soil creep and slope failure in the landscape evolution of a head water basin: field measurements in a zero order basin of northern Japan. Geomorphology 28: 329-344.  Zhang L. 1997. Cross-validation of non-linear growth functions for modelling tree height-diameter relationships. Annals of Botany 79: 251–257.  143  4. SUBSURFACE SOIL VOIDS IN THE VICINITY OF SHALLOW LANDSLIDES1 INTRODUCTION Subsurface preferential flow pathways are significant features for the drainage of hillslopes (Noguchi et al., 1999, Jones, 1978). Networks of connected flow pathways can rapidly convey stormflow down and through the soil profile (Sidle et al., 2001; Uchida et al., 2001). Preferential flow pathways can have both positive and negative influences on hillslope stability (Pierson, 1983; McDonnell, 1990; Uchida et al., 2001). Positive influences relate to the routing of stormflow so that excess water pressure does not develop on the hillslope. Negative influences occur when the capacity of the network is exceeded by rapid water delivery from upslope or reduced capacity due to pipe collapse or blockage (McDonnell, 1990; Nieber and Sidle, 2010). If the downslope transmission of water is impeded, the result will be elevated hydrostatic pressures in the soil matrix and possibly a landslide. Preferential flow paths are frequently found in landslide head scarps; however, these may be a result of post-landslide erosion processes (Tsukamoto et al., 1982). The rapid response of stormflow through preferential flow pathways influences hillslope and watershed hydrologic response (Fannin and Jaakkola, 1999; Uchida et al., 2001). In many studies, a high to very high proportion of subsurface flow occurs through preferential flow pathways (variously termed soil pipes, macropores, or more generally preferential flow pathways) (Tsukamoto et al., 1982; Ziermer and Albright, 1987; Sidle et al., 1995). These features are thought to be responsible for much of the error in  1  A version of this chapter will be submitted for publication. Sakals ME, Innes JL, Sidle RC. 2010. Subsurface soil voids in the vicinity of shallow landslide initiation zones  144  hydrologic and slope stability models and are generally thought to be responsible for model inaccuracies and inconsistencies (Borga et al., 1997; Casadei et al., 2003; Dhakal and Sidle 2004). The genesis of these hydrologically important soil features has been explored and they are believed to result from animal burrowing, root development and decay, and subsurface erosion (Wilson and Smart, 1984; Sidle et al., 1985; McDonnell, 1990; Noguchi et al., 1999; Chigira, 2002). Subsurface soil erosion in weathered granitic soils can be strong enough to result in topographic changes of the order of metres (Terajima and Sakura, 1993). Hillslope excavations reveal that soil pipes frequently occur immediately above a confining layer (McDonnell, 1990; Freer et al., 2002). This implies that the water has established a path at the lowest possible location on top of the impermeable boundary. General subsurface flow has been found to be controlled by subsurface topography during low flow but better approximated by surface topography during periods of high flow (Hutchinson and Moore, 2000). However, at scales of <10 m, the importance of existing preferential flow pathways (such as soil pipes and macropores) can be greater than either surface or subsurface topographic control (Jones, 1987; Tsuboyama et al., 1994; Hutchinson and Moore, 2000). Although the distribution of soil pipes in the soil profile has been observed in several investigations using excavated pit faces (Noguchi et al., 1999; Uchida et al., 2001), a major drawback of these observations is that the excavation may alter the active hydrologic pathways (Onda et al., 1992; Freer et al., 2002). Pit faces are also less well suited for the study of vertically oriented preferential flow paths. Okunishi and Iida (1981) identified the characteristic profile of higher resistance soils over lower resistance soils and associated it with shallow landslide initiation zones, but were unable to attribute the phenomenon to a particular geomorphic location. In this investigation, many soil profiles were assessed for the presence of preferential flow pathways using a dynamic cone penetrometer (Figure 4.1). Dynamic (or knocking) cone penetrometers have been used to conduct investigations of subsurface soil properties on hillslopes, particularly in Japan (Okunishi 145  and Iida, 1981; Onda et al., 1992; Ohnuki et al., 1997; Uchida et al., 1999; Onda et al., 2004). They are robust, relatively easily transported instruments that provide information regarding subsurface conditions without the effort or hillslope disturbance required for in situ observations. Dynamic cone penetrometers measure the vertical resistance profile of the soil by driving a hardened-steel cone down through the soil by dropping a weight of known mass a set distance onto a striking plate (Herrick and Jones, 2002). A key limitation of the instrument is its sensitivity to differences in soil moisture; this also makes the comparison between sites difficult. However, in this work, the difference in field moisture has a negligible effect as the analysis is of soil voids indicated by little or no resistance to penetration and limited comparison between sites is made. In this paper, data from soil resistance profiles gathered with a dynamic cone penetrometer are analyzed to determine the spatial distribution of preferential flow pathways nearby shallow landslide failure areas. Information regarding the spatial distribution of these features may provide insight into the location of past and future landslide failure areas as well as deepening our understanding of stormflow responses in headwater areas subject to shallow landslides. The spatial distribution of preferential flow pathways is compared to both surface and subsurface topographic curvatures, to the morphologic position relative to the shallow landslide failure area, and to the vertical position within the soil profile to determine the typical soil hydrology conditions and attempt to relate these to the occurrence of past shallow landslides. Finally, a limited analysis of the temporal aspects of observed preferential flow pathways is made and a supposition is made regarding future landslides.  146  METHODS Study areas Study areas were located in three areas of British Columbia: the North Coast, in the vicinity of Prince Rupert; Haida Gwaii (formerly the Queen Charlotte Islands), with sampling centred in Rennell Sound; and the South Coast, with sampling in the Klanawa drainage of south-central Vancouver Island. A fourth study area was located in Japan on the Kii peninsula of south-central Honshu Island (Figure 4.1, Table 4.1). A qualitative comparison of study sites from the four study areas is included in Appendix C. The climate of the three areas in British Columbia was predicted using ClimateBC (Hamann and Wang, 2005; Wang et al., 2006) and resulted in being very similar (see Figures 3.2 and 3.3). The sites on the South Coast might have been warmer, but much of the sampling was at somewhat higher elevations and the ClimateBC model adjusts predicted temperatures based on elevation. At the Japanese study area, the weather station (Owase) was located in reasonable proximity to the studied landslides (<60 km from furthest study site) but just above sea level (27 m) (Tutiempo.net). The lowest study site was at 150 m elevation, the other sites were at 440, 490, 500, 1 060 and 1 100 m elevation and were likely cooler and may have had greater precipitation than recorded at the station. No specific information is available regarding the increase in precipitation with altitude, but Imaizumi and Sidle (2005) reported a range of 1 600 –4 500 mm (average of 3 300 mm) for the Miyagawa Dam area where three of the study sites were located. The daily precipitation extreme is higher at the Owase station compared to the South Coast study area (Port Renfrew station) and three and four times higher than the North Coast (Prince Rupert) and Haida Gwaii (Sandspit) stations respectively. However, the Sandspit station is known to be drier than the west coast of Haida Gwaii due to local orographic effects (Karanka, 1986). Precipitation intensities over shorter periods are not known but could be expected that the sites on the Kii Peninsula may experience  147  higher intensity rainfall during the Baiu (early summer monsoon period) and typhoon seasons. A more complete comparison of study area climates can be found in the Methods section of Chapter 3. All three study areas in British Columbia were glaciated during the Late Pleistocene. The landslide sites from Japan have not experienced glaciation and thus the landscape lacks paraglacial legacy of the study areas in British Columbia. However, steep slopes with soil covers of <2 m and subsurface stormflow routing are common in all regions.  Selection of sites Landslide inventories were used to select candidate landslide initiation sites and were available for all three study areas in British Columbia (Schwab, unpub., Maynard et al., 2004) as well as in Japan (Imaizumi et al., 2008). A time-series of aerial photographs of different ages were used to bracket the date of failure, or the failure was associated with a substantial storm that occurred during the period between two successive photo sets. The error of this method is assumed to be <5 years. Data accompanying the inventories were used to select candidate landslides based on material type, geology, and the feasibility of access. Some failures were visible on aerial photographs, in fresh condition or indicated by distinctive strips of forest cover. Some sites included in the study were not included in the inventories and were not visible on 1:20 000 scale aerial photographs because of their small size and the crowns of adjacent trees; these sites were found by reviewing aerial photographs for candidate hillslopes and completing field traverses to identify possible study landslides. Several criteria were used for site selection. Assuming that the initiation area was not too steep or dangerous to work on, the first requirement was that the failed material consisted of colluvium. Colluvium is assumed to be more directly related to the current conditions, slopes, and underlying geology than materials such as till or fluvioglacial sediments. Till was avoided because of the variability 148  in textures and differences in consolidation that could result in different root penetrations and resistances to surface erosion processes that could introduce extra variability into the study. Great variation exists in colluvium but in most of the study sites the parent material of the colluvium appeared to be the local bedrock. A full depth (to competent rock) failure was required. A soil pit was dug down to bedrock or until a failure plane surface was found within the soil. In the case of the latter, the landslide would not be included in the study if the depth of soil remaining after the previous failure was >20 cm but this excluded very few potential study sites. No specific size criteria were applied, so long as the initial failure was of the full soil depth. The date of failure was required in order to correctly place the failure with respect to all facets of the study. For some landslides the year, or approximate year, of failure had been determined in the landslide inventory (Schwab Unpub.; Maynard et al., 2004; Imaizumi et al., 2008). In other cases, dendrochronological techniques were used to determine the date of failure (Wilford et al., 2005). Tree scars were the preferred feature for dating; abrupt growth releases of trees on the margins of the failure were next most preferred. Dating trees established on the recharged soil material produced only a minimum age as the period prior to establishment was unknown. Other methods, such as the accumulation of litter material, were used to determine the age of failures <5 years. The temporal error of time since failure at each study site was assessed independently and was dependent on the number, quality, and consistency in the dendrochronological samples and other evidence; estimated errors are presented in Appendix D. Finally, the failure could not be the result of excessive anthropogenic disturbance. Areas that had been previously timber harvested were acceptable as long as there were no obvious upslope diversions of drainage pathways or areas of extensive soil or other disturbance.  149  Figure 4.1. Dynamic cone penetrometer.  150  Figure 4.2. Study areas in British Columbia, Canada and central Japan.  151  Table 4.1. Selected information of the study areas (Sutherland-Brown, 1968; Holland, 1976; Muller 1977; Imaizumi and Sidle, 2005; Imaizumi et al., 2008; Environment Canada, 2008, Tutiempo.net). North Coast, BC 8 yes  Haida Gwaii, BC South Coast, BC Kii Peninsula, Japan Landslides sampled 14 12 7 Late Pleistocene glaciation yes yes no pyroclastic rocks, shale, granodiorite, quartz diorite, schist, sandstone, slate, shale, Geology granite, gneiss, shale, slate siltstone, sandstone, granite intrusions, gneiss, mudstone conglomerate, argillite weak volcanic rocks, steep slopes rising from sea steep slopes rising from sea steep slopes rising from sea steep slopes with local relief Typical morphology level to 500 m level to 600 m level to 1 000 m of 300 m Annual precipitation at study sites 1 600–3 350 mm 2 400–3 700 mm 3 450–4 000 mm ~4 000 mm Extreme rainfall at nearest climate 118.2 mm /day (Prince 79.5 mm /day (Sandspit, 293 mm /day (Port Renfrew, 351 mm /day (Owase, station Rupert, BC) BC)* BC) Japan)** colluvial soils of <1 m in colluvial soils of <1 m in colluvial soils of <1 m in colluvial soils of 0.5-1.5 m in Typical soils depth depth depth depth second and old-growth second and old-growth second and old-growth Forests coniferous plantations coniferous coniferous coniferous Predominant natural disturbance type shallow landslide, debris flow shallow landslide, debris flow shallow landslide, windthrow shallow landslide, debris flow Predominant human disturbance type forest harvesting forest harvesting forest harvesting forest harvesting Landslide inventory available yes yes yes yes *The precipitation on Haida Gwaii is strongly influenced by orographic effects. The Sandspit climate station is on the leeward, eastern side of the major mountain range of the island archipelago; the study area was located on the windward, western side of the range. Precipitation is accentuated on the western side of the islands where incoming Pacific frontal systems first encounter the land mass. Therefore, precipitation at the study sites is under represented (Karanka, 1986). **Highest noted value during data compilation of period from 1973–2000.  152  Site description Thirty-seven failure sites were included in this study; five of these had surfaces representing two separate failures and thus the total available for analysis was 42. Site surveys were conducted over the entire failure area and the immediately contributing slopes (typically 5–10 m laterally beyond side scarps and 10–15 m above the head scarp). Although surveys did not cover the entire contributing area, this distance was considered sufficient given the typical length of preferential flow path features is < 1 m (Noguchi et al., 1999). An irregular survey approximating a grid with a ~3 m inter-point spacing was used. At each point, the relative x, y, and z coordinates were derived from data collected by a theodolite and laser range finder. Digital elevation models were constructed from site survey data. The resolution of the digital elevation models was 0.5 m; this ensured the representation of small scale features (such a head and side scarps) that were captured in the site survey. Inverse distance weighting was used to interpolate surface and subsurface elevation values between survey points. Two different analysis windows were used: 1.5 m and 4.5 m. The 1.5 m analysis window included the grid cell including the survey point and the two adjacent grid cells; the 4.5 m analysis window included the grid cell of the survey point and the cells up to 2 m away in each direction of the analysis. The 1.5 m extent was used as it was the smallest available with the field data. The 4.5 m analysis extent was as large as practical given the small size of many of the landslide failure areas; a larger analysis extent would result in many cells without values as the analysis window would exceed the failure area. Slope and cross-slope and longitudinal curvatures, for both the surface and subsurface, were determined from the digital elevation models based on fitting a quadratic function by the least squares method using GRASS (Neteler and Mitasova, 2004). At approximately half of the points, the dynamic cone penetrometer was used to determine subsurface soil resistance. Where these profiles encountered stones, a second profile was completed within 30 cm of the surveyed point. The knocking cone penetrometer has a cone diameter of 3 cm and a falling height of 50 cm. For the Japanese sites a standard 5 kg weight was used; for the British Columbia sites a 2.5 kg 153  weight was used to facilitate fieldwork in rugged and remote locations. Strike counts for all the measurements using the 5 kg hammer were doubled to be consistent with the majority of sites using the 2.5 kg hammer. Strikes were counted for each decimeter (10 cm) of penetration producing decimeter sections used in the analysis (see Appendix D for observation data). Preferential flow pathways (soil pipes and macropores) have been defined in various ways (Jones, 1978; Onda et al., 1992; Noguchi et al., 1997; Cey and Rudolph, 2009). Two types of preferential flow pathways have been identified in this study: Type I and Type II. It is inferred that Type I pathways would be sites of preferential flow but not necessarily of concentrated flow as in Type II pathways. Preferential flow pathway Type I is defined as a region of low-density soils where one or fewer strikes was required to drive the penetrometer through an entire decimeter section (10 cm depth increment). That is, with <2 strikes, the penetrometer would pass by two consecutive decimeter depth indicators. The size of Type I pathways is defined as the number of consecutive decimetre sections meeting the criterion. The first criterion for Type II pathways is soil that no strikes were required to drive the penetrometer through a decimeter section. The second criterion was that in the decimetre layer above the potential feature, the number of strikes was >0; this excluded vertically oriented soil voids and general areas of loose soils. Thus, Type II pathways were always located >10 cm in depth and were always 10 cm in vertical dimension. This technique is relatively coarse, selecting only rather large features. However, Tsukamoto et al., (1982) noted the average size of soil pipes in landslide head scarps to be 14 cm, and Chigira (2002) identified soil pipes in landslide head scarps up to 1 m in diameter, in both cases the sizes of the features are within the sensitivity of this method. The cross-sectional shape of the feature is also not known, though Terajima et al. (2000) found that most pipes were roughly circular in cross-section. Figure 4.2 displays a few instances of Type II preferential flow pathways; Figure 4.3 illustrates the resistance profile for a single point with both Type I and Type II pathways present. Several instances of Type II pathways  154  were excavated to confirm that the subsurface soil void showed evidence of water transport. Smooth walls and the lack of animal activity were taken as sufficient evidence of water conveyance.  Figure 4.2. Examples of Type II preferential flow pathways. Frame A: A Type II preferential flow pathway in a head scarp. Note that the base of the flow pathway is on bedrock. Frame B: Two Type 155  II preferential flow pathways that were originally identified with the knocking cone penetrometer. Pipes are on either side of the pencil, note outflow from left feature.  Figure 4.3. Soil resistance profile measured with a dynamic cone penetrometer with a 2.5 kg hammer weight and a cone diameter of 30 mm. A Type I pathway is located between 0 and 10 cm below the surface and a Type II pathway is located at 35–45 cm below the surface. All statistical analysis was completed using R (Ihaka and Gentleman, 1996; R Development Core Team, 2008). Surveyed points with Type I or II pathways were assessed against surveyed points without pathways to determine if there is a difference in the mean of the assessed topographic variable. The null hypothesis is that there is no difference in the topographic variable between the points with and without pathways. A Shapiro-Wilk normality test was used to test non-normality and was followed with a nonparametric Wilcoxon ranked sum test (=0.05). A Kolmogorov-Smirnov two sample test was used to determine whether a difference in the distributions could be detected (=0.05). 156  To assess potential differences in the frequency and size of preferential flow pathways between hillslope soils and the soils accumulated in failure areas, logistic regression was used. Each survey point was assigned with the presence/absence of each of pathway type. Relative morphologic positions from field assessments were classified into 3 classes (Table 4.2). The margins of the slides include the head and side scarps where the surface is disturbed but the soil material is that of the hillslope, not of infilled material. Hydrologic conditions in the slide margins may be altered by the presence of the open face of the fresh landslide, similar to the effects of observation trenches. For these reasons, the inclusion of the slide margin category was warranted. The percent fraction of Type I and II pathways at each relative morphological position was calculated as the number of point occurrences divided by the total number of points with penetrometer profiles. To eliminate the effect of varying soil depths between the three morphologic positions, the total number of decimetre analysis units included in the analysis was incorporated into the models. Values were transformed using the arcsine transformation: t = arcsin(sqrt(f)); where t is the transformed fraction and f is the untransformed fraction. The average size of Type I pathways was calculated as the number of decimetre sections classified as Type I pathways divided by the number of points with Type I pathways.  157  Table 4.2. Relative morphologic positions in landslide initiation zones. Position  Description  Hillslope  Undisturbed hillslopes outside of the failure site.  Margin of failure area  Disturbed hillslope positions including the top, face and bottom of the head scarp and side scarps of the failure area.  Failure area  Includes the entire failure area except for the head scarp and side scarps.  RESULTS Penetrometer measurements were completed at 1037 of 1994 surveyed points. Of the 1037 surveyed points with penetrometer profiles, 463 points (45%) contained Type I pathways; 94 (9.1%) were classified as containing Type II pathways.  Microtopography and preferential flow pathways Of the 1037 points with penetrometer profiles, 756 also have microtopographic information of the surface and subsurface form. The mean value of the slope gradient for both the surface and the subsurface was steeper for points with Type I pathways than for those without (Table 4.3). The difference in subsurface gradient was >3º, and for the surface was 2.7º. All the surface topographic variables were statistically different for points with and without Type I pathways, except for cross-slope curvature of the surface in the 4.5 m analysis scale that resulted in a p-value of 0.052. In each case, the means of the points with Type I pathways were more divergent or more convex than those without. For the subsurface, only the longitudinal curvature at the 4.5 m analysis scale was significant. Again, the points with Type I pathways had less concavity on average.  158  Table 4.4 summarizes the statistical results regarding Type II pathways. Neither surface nor subsurface slope was significant in separating points with Type II pathways from those without. Cross-slope curvature of the ground surface was significantly different between the points with and without Type II pathway for the 4.5 m analysis scale. Longitudinal curvature of the ground surface was significant for separating points with and without Type II pathways for both the 1.5 m and 4.5 m analysis scales. In each case of significant difference between the two groups, the points with Type II pathways were found on less concave sites (for the 1.5 m analysis scale the mean longitudinal surface curvature was essentially zero indicating a planar surface). For the analysis of the subsurface, only the longitudinal curvature of the 4.5 m analysis scale was significant in separating the two groups; the mean curvature of points with Type II pathways was slightly convex and the mean curvature of the points without Type II pathways was concave.  Relative spatial distribution of hydrologic features The transformed percent of all soil profiles with Type I pathways was larger on the adjacent hillslopes than in the failure areas. Accounting for differences in soil depth was significant for Type I pathways; the frequency in the margins of the failures was less than the frequency in hillslope soils but not as low as the frequency in failure areas. Failure areas not only had fewer Type I pathways, but those that were present were also smaller; Type I pathways in failure areas (mean = 15.9 cm) were significantly smaller (familywise p<0.05) than in both the hillslope (mean = 23.6 cm) and slide margin areas (mean = 22.4 cm). The transformed fraction of points with Type II pathways in failure areas was significantly less than the transformed fraction on hillslopes (Table 4.5). The margins of slides were not found to differ in the frequency of Type II pathways from either hillslopes or failure areas. Accounting for the difference in soil depth between the positions was not significant and did not alter the results for Type II pathways.  159  Table 4.3. Statistical test results of differences between topographic variables for points with and without Type I pathways. Significant results are in bold-face type.  Surface of analysis  Curvature direction  Analysis scale (m)  Points without Type II  Points with Type II  Wilcoxon signed rank test (p-value)  Mean value (no Type II)  Mean value (Type II)  Kolmogorov -Smirnov test (pvalue)  Surface  Slope  NA  442  81  0.59  41.0º  40.6 º  0.66  Subsurface  Slope  NA  436  78  0.93  40.1º  40.4 º  0.99  Surface  Cross-slope  1.5  476  86  0.20  -0.035 m-1  -0.028 m-1  0.06  Surface  Longitudinal  1.5  476  86  <0.01  -0.029 m-1  0.0001 m-1  0.02  Surface  Cross-slope  4.5  741  94  0.13  -0.037 m-1  -0.030 m-1  0.04  Surface  Longitudinal  4.5  741  94  <0.01  -0.023 m-1  -0.003 m-1  0.03  Subsurface  Cross-slope  1.5  469  83  0.28  -0.023 m-1  -0.054 m-1  0.55  Subsurface  Longitudinal  1.5  469  83  0.15  -0.024 m-1  0.006 m-1  0.13  Subsurface  Cross-slope  4.5  450  78  0.31  -0.030 m-1  -0.026 m-1  0.64  Subsurface  Longitudinal  4.5  450  78  <0.01  -0.018 m-1  0.009 m-1  <0.01  160  Table 4.4. Statistical test results of differences between topographic variables for points with and without Type II pathways. Significant results are in bold-face type.  Curvature direction  Analysis scale (m)  Points without Type I  Points with Type I  Wilcoxon signed rank test (p-value)  Mean value (no Type I)  Mean value (Type I)  Kolmogoro v-Smirnov test (pvalue)  Surface  Slope  NA  170  353  <0.01  39.1 º  41.8 º  0.001  Subsurface  Slope  NA  169  345  <0.01  38.0 º  41.2 º  0.001  Surface  Cross-slope  1.5  187  375  <0.01  -0.042 m-1  -0.030 m-1  0.005  Surface  Longitudinal  1.5  187  375  <0.01  -0.035 m-1  -0.019 m-1  <0.001  Surface  Cross-slope  4.5  390  445  0.02  -0.042 m-1  -0.033 m-1  0.05  Surface  Longitudinal  4.5  390  445  <0.001  -0.032 m-1  -0.014 m-1  <0.001  Subsurface  Cross-slope  1.5  186  366  0.18  -0.031 m-1  -0.025 m-1  0.28  Subsurface  Longitudinal  1.5  186  366  0.85  -0.025 m-1  -0.017 m-1  0.48  Subsurface  Cross-slope  4.5  185  343  0.06  -0.038 m-1  -0.025 m-1  0.11  Subsurface  Longitudinal  4.5  185  343  0.01  -0.029 m-1  -0.005 m-1  <0.01  Surface of analysis  161  Table 4.5. Summary table of logistic regressions of preferential flow pathways between relative morphologic positions around shallow landslide head scarps. Positions are compared to hillslope (M – Margin of failure area, F – Failure area). Significance codes in parentheses: 0 < *** < 0.001 < ** < 0.01 < * < 0.05 < . < 0.1 < NS  Intercept  Position (M)  Position (F)  Analyzed soil depth (dm)  Probability of points with Type I pathways  0.800 (***)  -0.207 (NS)  -0.367 (*)  NA  Probability of points with Type I pathways considering soil depth  1.085 (***)  -0.413 (.)  -0.572 (**)  -0.003 (.)  Probability of points with Type II pathways  -1.601 (***)  -0.205 (NS)  -1.122 (***)  NA  Probability of points with Type II pathways considering soil depth  -1.525 (***)  -0.260 (NS)  -1.177 (***)  -0.001 (NS)  Model  Position of hydrologic features in the soil profile The vertical distribution of the frequency of Type I pathways was consistent over the three morphologic positions (Figure 4.4). A steady decrease in the frequency of Type I pathways was found for each position with failure areas generally having a slightly lower percent of Type I pathways. The vertical distribution of the frequency of Type II pathways in the profile is shown in Figure 4.5. The pattern was relatively similar for all three morphologic positions with a strong rise in the near surface and then a decrease in the frequency with depth. Slide margins had the highest frequency of Type II pathways at 9.1%. The frequency for the hillslope areas was relatively consistent, dropping from the highest value of 6.1% at about one quarter of the way into the profile. The frequency of Type II pathways in the failure areas decreased to zero approximately 2/3 of the way down the profile. 162  Figure 4.4. Relative frequencies of Type I pathway occurrence compared to depth within the soil profiles.  163  Figure 4.5. Relative frequencies of Type II pathways compared to depth within the soil profiles.  164  Temporal variability of preferential flow pathways When comparisons are made across the all the study sites, the data is not inconsistent with an increase in both Type I and II pathways with increasing elapsed time since failure (Figures 4.6 and 4.7). In the first 20 years following failure, the percent of failure area points with Type I pathways is nearly evenly distributed. Following the first 20 years after landslide, at least half of the points in failure areas had Type I pathways except for a single site from the Japanese study area (RedPine). At the other older (>50 years since landslide) Japanese site 86% of failure area points had Type I pathways. The points at the right end of the plot represent the average percent of points with Type I pathways on the hillslopes adjacent to the failure areas. For each study area, it appears as though the older sites come closer to the average hillslope value with time. The difference between the average hillslope values for the three study areas in British Columbia and the sites in Japan in considerable. Small-scale hydrogeomorphic processes of material transport may be responsible for this strong difference.  165  Figure 4.6. Relation between percent of points in failure areas with Type I pathways and time of material accumulation since last landslide. Dashed line is an envelope curve for all study areas except for Japan. Values on extreme right are average percent of hillslope points with Type I pathways.  166  The occurrence of Type II pathways is generally less frequent and so the data to support an increase in Type II pathways with time is limited. Three points of <30 years since failure have percents of between 10% and 20%. In each case, the points with the Type II pathways had deeper soils than the average of the failure area; therefore, these features may have been present in the soil that was not evacuated during the previous failure. Beyond 80 years after landslide, only 2 of the 7 failure areas had no instances of Type II pathways. The failure area from the South Coast at 89 years since failure (Stihl1) had an average failure area soil depth of 1.3 m and a highly convergent contributing area; these two features may have acted collaboratively to result in the high percent of points with Type II pathways. At the far right of Figure 4.7 is plotted the average percent of points from the adjacent hillslopes with Type II pathways. While the soil properties of the failure area and the hillslopes may remain different as a result of different hydrological conditions, the young failure areas appear to have a less developed hydrologic network compared to the hillslopes; older failure areas display a range of conditions.  167  Figure 4.7. Relation between percent of points in failure areas with Type II pathways and time of material accumulation since last landslide. Solid black line is a local regression trend line with a span of 0.75. Values on extreme right are average percent of hillslope points with Type II pathways.  168  DISCUSSION These data support some commonly held perceptions of the position of preferential flow pathways, contest other perceptions, and introduce some new perspectives on the potential pathways of hillslope stormflow. The concentration by surface and subsurface topography was expected to result in a higher frequency of preferential flow pathways: 1) with concave topographic variables (Noguchi et al., 1999); and 2) at the bedrock interface (Uchida et al., 2001). Instead, these data show that for every statistically significant difference between points with and without Type II flow pathways, the points with the features were found on more convex topography. The occurrence of less concave mean surface morphologies for points with Type II pathways suggests that there is not a simple relation between the lowest topographic positions and the location of preferential subsurface pathways detected in this investigation. The relationship between topographic curvature and the presence of water is well supported (Anderson and Burt, 1978; O‟Loughlin, 1981; Sinai et al., 1981), but this does not provide information on the presence of preferential pathways for water transmission. However, Sasaki et al. (2000) found that soil pipes formed in response to heavy subsurface flow and hydraulic gradients. Zero-order basins, such as landslide initiation zones, could have these conditions and Sidle (1984) found artesian water pressure in such situations in coastal Alaska. Landslide initiation zones are also known to have episodically high water tables (Sidle and Swanston, 1982). One explanation for the lack of preferential flow paths on the more concave/convergent sites is that in hillslope depositional areas, such as landslide failure areas, the deposition of material will occur preferentially in areas of concave topography (Chapter 2). The accumulation of material from small landslides and slope wash processes could create a soil that does not initially include preferential flow pathways, but they may develop over time. Thompson and Moore (1996) found that the scale of analysis was important in predicting the response of the groundwater table and that it could be affected by the presence of preferential flow pathways. For 169  scales of <10 m, Hutchinson and Moore (2000) found that soil pipes and macropores could ‟overwhelm topographic controls on throughflow‟. Freer et al., (2002) found that including bedrock topographic indices improved the spatial estimation of flow at a trench face. In this study, both types of flow pathways were more strongly associated with the topography of the surface than with the subsurface confining layer. This is likely because many of the preferential flow pathways were in the upper part of the soil profile and the control of the substrate was undetected or irrelevant. Preferential flow pathways have been found to be more common in the upper soil horizons (developed from root pathways) and at horizon boundaries (developed through subsurface erosion) (Tsukamoto et al., 1982; Noguchi et al., 1997). The results of this study are consistent with these findings. The highest frequencies of both Type I and II pathways are near the top of profiles (Figures 4.4 and 4.5). Subsurface flow is known to flow through joints and open structures of near-surface bedrock and move between shallow bedrock and the soil (Wilson and Dietrich, 1987; Anderson et al., 1997; Noguchi et al., 1999). Noguchi et al. (1999) found that bedrock topography influences hydrologic pathways at a hillslope scale. In this study, the lack of soil pipes near to the bedrock surface could be related to the thin (< 10 cm) high resistance soils (overconsolidated soil or weathered bedrock) that were commonly present at the base of soil profiles. The presence of preferential flow pathways in areas of greater hillslope convexity may also reflect the association of the features with root pathways (Uchida et al., 2001). Topographic concavities and landslide sites may have episodically high water tables (Sidle and Swanston, 1982) that will discourage the establishment of roots and differences in root reinforcement have been linked to hillslope morphology (Hales et al., 2009). In this study, roots were found in conjunction with most of the excavated Type II pathways. The formation of flow pathways by tree roots is also supported by the temporal data presented in Figure 4.6. Conifer forests had established on each of the landslide failure area areas with >1 Type II 170  flow pathways except for one site (Kaien57A) that had a dense cover of red alder. This site had recently experienced windthrow of ~30% of the stems that would have considerably loosened some of the sampled soils; such loose soils maybe sites of preferential flow (Noguchi et al., 1999). A systematic relationship was found between the frequency of occurrence of preferential flow pathways and the relative morphology with respect to the landslide initiation zone. The failure areas had experienced a range of soil infilling but in all cases the soils of the failure areas were much younger than the soils of the adjacent hillslopes. The failure areas had fewer of both Types I and II flow pathways than the surrounding hillslopes. Uchida et al. (2001) reported that 50-90% of landslide head scarps had macropores and Tsukamoto et al. (1982) found 60 of 64 landslide head scarps had soil pipes; these frequencies are much higher than found in this study, but this could be entirely a result of the size and definition of flow pathways used. Further, these frequencies were presumably collected by an inventory of head scarps where the likelihood of missing a preferential flow pathway would have been much lower than in this study and where the trenching effect may have resulted in subsurface erosion (Onda et al., 1992). Okunishi and Iida (1981) identified four characteristic resistance profiles, and spatially assigned two to hillslopes and one to failure areas. The failure area profile had loose soils down to the subsurface confining layer. Contrasting this, Onda et al. (1992) found lower porosity but increased large pores in landslide failure areas, but the study only included a single landslide failure area. Our findings of less developed hydrologic pathways in the failure areas could have implications for the repetition of shallow landsliding at the same, or nearly the same, hillslope position. During runoff events, hillslopes above and adjacent to the head scarps could be draining with relative efficiency, aided by the drainage capabilities of the relatively abundant preferential flow pathways. As the stormflow encounters the much thinner soils of the landslide failure area, the total hydrologic transmissivity is strongly reduced due to the thinner soil profile having less cross-sectional area for matrix flow and also having a lesser hydrologic network. Not only is the thin soil of the failure area less effective at conducting the stormflow, but the gently 171  converging topography common above hillslope failures (Talebi et al., 2008) concentrates discharge from the contributing slopes. The result is a hillslope section with a highly effective drainage network funnelling into a smaller hillslope area with less effective drainage. This discontinuity could be responsible for the elevation of pore water pressures in the vicinity of the head scarp and a subsequent landslide (Sidle 1984; McDonnell, 1990; Onda et al., 1992; Uchida et al., 2001). Despite the limited data regarding the recovery of preferential flow pathways in landslide failure areas, the trend of increasing flow Type I flow pathways with time seems physically and conceptually likely for the sites in British Columbia where infilled soils include a high degree of organic material (Figure 4.6). The soils of the failure areas could be expected to eventually reflect or exceed the hillslope value and such a trend in the data appears reasonable for the study areas in British Columbia, though only limited in nature. This relation does not seem to be supported by the data from the Japanese site, but with only two failure areas with much soil depth recovery, any pattern would remain hidden. However, a physical explanation for the lower level of hillslope Type I pathways could relate to the dominance of slope wash deposits noted in the study of small-scale processes (Chapter 2). The slope wash deposits may deposit hillslope material at microtopographic lows and deter the formation of flow pathways at these sites. Over time, the development of a thickening soil as well as the continued hydrologic pressure from the interaction of the convergent topography and the stormflow coming from the surrounding more mature soils may be expected to result in a temporal sequence of soil-hydrologic development. Regarding the temporal pattern of Type II flow pathways (Figure 4.7), many of the numerous points without Type II flow paths can be explained by the soil depths not exceeding the minimum of 20 cm to allow a Type II pathway. At the three points of <30 years since failure with Type II pathways, the pathways were likely found in soil not evacuated at the time of the landslide. The five older failure areas with Type II pathways may represent a hint of an underlying trend that warrants further investigation. These sites are known to experience strong subsurface flow and the collection of organic debris that upon decay may create such 172  subsurface features. Thus, the recovery of hillslope hydrology may lag behind the accumulation of material depth, taking ≥150 yr to fully develop (Shimokawa, 1984). In failure areas, the period following the majority of soil accumulation but before the recovery of preferential flow pathways may represent a period during which the site is of increased risk of failure. A limitation of the methods used here is that smaller soil pipes located on the confining layer would have been missed. As reported by Noguchi et al., (1999), the 2x2 m grid size of McDonnell et al. (1996) was too coarse to capture the small-scale variability in the bedrock surface topography and the related flow paths. Although a ~3 m grid was used in this study, replication at many sites was intended to capture the effect of slightly larger bedrock topographic patterns on the preferential flow pathways of >10 cm size. Further investigations of subsurface preferential flow pathways would benefit by increasing the sampling intensity for both the spatial network and the vertical analysis units of the penetrometer (i.e., from 10 to 5 cm or less) enabling the detection of smaller features of hydrologic significance. Definitions are important in the scientific discussion of any features. Here, we have defined two preferential flow pathway types. We cannot be certain that all the instances of the features identified are used in the hillslope hydrologic network as flow pathways, but given that these are generally convergent features with seasonally high water tables (Sidle and Swanston, 1982) it seems likely that zones of loose soils (Type I pathways) and soil voids (Type II pathways) would be sites of water conduction. However, some Type I pathways on strongly divergent sites may remain dry despite being in the vicinity of areas of saturation. Besides being zones of loose soils, Type I pathways did not necessarily provide other evidence of use as flow pathways. All of the Type II pathways excavated were clearly features of preferential flow. The detection of both Types I and II pathways was relatively coarse, but highlighted the ability of a knocking cone penetrometer to detect features of potential hydrologic significance. A study combining a slope trench and intensive sampling with a dynamic cone penetrometer could provide a better estimate of 173  the capabilities and limitations of the instrument for soil-hydrology investigations. Further work could include the detailed mapping of subsurface preferential flow pathways with the use of a dynamic cone penetrometer. Examining tenuously stable and marginally convergent slopes that have not failed for >200 yr could also provide insight into the development of preferential flow networks in shallow landslide failure areas.  CONCLUSIONS A dynamic cone penetrometer was used to identify low resistance features in profiles of soils in the initiation zones of shallow landslides. All low resistance features were defined as Type I preferential flow pathways but Type II pathways were discrete decimeter thick voids bound within a stronger matrix. With the methods used the smallest features identified were 10 cm in the vertical dimension, but with refinements in future studies, smaller features of hydrologic significance could be identified. Both preferential flow pathways were preferentially located in the upper portions of soil profiles. Type I pathways were found in the top 10% of soil depth at between 60 and 70% of profiles investigated and declined relatively consistently with depth reaching values of <5% at the bottom of the profile. Type II pathways on hillslopes adjacent to failure areas reached their maximum frequency (8% of profiles investigated) at 25% of the soil profile depth. Type II pathways in failure areas were most frequent between 30 and 60% of the soil profile depth (~3% occurrence) and none were found in the lowest third of the soil profiles. Several topographic variables were significantly different between points with and without the preferential flow pathways investigated here; in all cases of significant difference between points with and without preferential flow pathways, the points with the features were found at less concave points. Since processes of material movement in initiation zones are preferentially active at concave locations 174  (Chapter 2), it is hypothesized that this accretion activity may restrict the development of preferential flow pathways. Penetrometer profiles on the hillslopes adjacent to failure areas had significantly higher frequencies of both Type I and II pathways than the infilled soils of the failure areas. The limited nature of the subsurface hydrologic network in shallow landslide failure areas may lead to a discontinuity of hillslope stormflow transmissivity that results in more frequent cases of elevated pore water pressures in the previously failed area. This occurrence indicates an increased likelihood of subsequent failures at the same hillslope position. A temporal analysis of the limited data is not inconsistent with Type II pathways becoming more abundant with increasing time since failure. The period of hydrologic adjustment may extend beyond the period encompassing the majority of soil accumulation following failure (Chapter 3). In such a case, a critical period may exist in shallow landslide failure areas after the majority of soil accumulation yet prior to hydrologic maturity.  175  REFERENCES Anderson MG, Burt TP. 1978. The role of topography in controlling throughflow generation. Earth Surface Processes 3: 331–344.  Anderson SP, Dietrich WE, Montgomery DR, Torres R, Conrad ME, Loague K. 1997. Subsurface flow paths in a steep, unchanneled catchment. Water Resources Research 33(12): 2637–2653.  Borga M, Dalla Fontana G, Da Ros D, Marchi L. 1997. Shallow landslide hazard assessment using a physically based model and digital elevation data. Environmental Geology 35: 81–88  Casadei M, Dietrich WE, Miller NL. 2003. Testing a model for predicting the timing and location of shallow landslide initiation in soil-mantled landscapes. Earth Surface Processes and Landforms 28: 925–950.  Cey E, Rudolph D. 2008. Field study of macropore flow processes using tension infiltration of a dye tracer in partially saturated soils. Hydrological Processes. 23: 1768–1779.  Chigira M. 2002. Geologic factors contributing to landslide generation in a pyroclastic area: August 1998 Nishigo Village, Japan. Geomorphology 46:117–128.  Dhakal AS Sidle RC. 2004. Distributed simulations of landslides for different rainfall conditions. Hydrological Processes 18: 757–776.  Environment Canada, 2009. http://www.climate.weatheroffice.ec.gc.ca/climate_normals/index_e.html  176  Faeh AO, Scherrer S, Naef F. 1997. A combined field and numerical approach to investigate flow processes in natural macroporous soils under extreme precipitation, Hydrology and Earth System Sciences 1: 787–800.  Fannin RJ, Jaakkola J. 1999. Hydrological response of hillslope soils above a debris-slide headscarp. Canadian Geotechnical Journal. 36: 1111–1122.  Freer J, McDonnell JJ, Beven K, Peters NE, Burns DA, Hooper RP, Aulenbach B, Kendall C. 2002. The role of bedrock topography on subsurface storm flow. Water Resources Research 38: 1269. doi: 10.1029/2001WR000872.  Hales TC, Ford CR, Hwang T, Vose JM, Band LE. 2009. Topographic and ecologic controls on root reinforcement, Journal of Geophysical Research 114: F03013, doi:10.1029/2008JF001168.  Hamann A, Wang T. 2005. Models of climatic normals for genecology and climate change studies in British Columbia. Agricultural and Forest Meteorology 128: 211–221.  Herrick JE, Jones TL, 2002. A dynamic cone penetrometer for measuring soil penetration resistance. Soil Science Society of America Journal 66: 1320–1324.  Holland SS. 1976. Landforms of British Columbia a physiographic outline. BC Department of Mines and Petroleum Resources, Bulletin 48.  Hutchinson DG, Moore RD. 2000. Throughflow variability on a forested hillslope underlain by compacted glacial till. Hydrological Processes 14: 1751–1766.  Ihaka R, Gentleman R. 1996. R: A language for data analysis and graphics. Journal of Computational and Graphical Statistics 5: 299–314. 177  Imaizumi F, Sidle RC. 2005. Relationship between sediment supply and transport processes in Miyagawa dam catchment. Annals of Disaster Prevention Research Institute: Kyoto University, No. 48C.  Imaizumi F, Sidle RC, Kamei R. 2008. Effects of forest harvesting on the occurrence of landslides and debris flows in steep terrain of central Japan, Earth Surface Processes and landforms, 33: 827–840.  Jones JAA. 1978. Soil pipe networks: distribution and discharge. Cambria 5:1–21.  Jones JAA. 1987. The effects of soil piping on contributing area and erosion pattern. Earth Surface Processes and Landforms 12: 229–248.  Karanka EJ. 1986. Trends and fluctuatlons in precipitation and stream runoff in the Queen Charlotte Islands. BC Ministry of Forests Research Program: Victoria BC, Land Management Report 040.  Maynard (Denny) and Associates Ltd. and Golder Associates, 2004. Klanawa terrain and landslide hazards mapping: Pilot study report. Prepared for Weyerhaeuser Company Limited BC Coastal Group – Nanaimo Woodlands.  McDonnell JJ, 1990. The influence of macropores on debris flow initiation. Quarterly Journal of Engineering Geology 23: 325–331.  McDonnell JJ, Freer J, Hopper R, Kendall C, Burns D, Beven K, Peters J. 1996. New method developed for studying flow on hillslopes. Transactions of the American Geophysical Union 77: 465.  Muller JE. 1977. Geology of Vancouver Island. Geological Survey of Canada Open File 463.  Neteler M, Mitasova H. 2002. Open Source GIS: A GRASS GIS Approach, Kluwer Academic Press Dordrecht: Boston, 464 pp. 178  Nieber JL, Sidle RC. 2010. How do disconnected macropores in hillslopes facilitate preferential flow? Hydrological Processes (accepted).  Noguchi S, Tsuboyama Y, Sidle RC, Hosoda I. 1997. Spatially distributed morphological characteristics of macropores in forest soil of Hitachi Ohta Experimental Watershed, Japan. Journal of Forest Research 2: 207–215.  Noguchi S, Tsuboyama Y, Sidle RC, Hosoda I. 1999. Morphological characteristics of macropores and the distribution of preferential flow pathways in a forested slope segment. Soil Science Society of America Journal. 63: 1413–1423.  Ohnuki Y, Terazono R, Ikuzawa H, Hirata I, Kanna K, Utagawa, H. 1997. Distribution of colluvia and saprolites and their physical properties in a zero-order basin in Okinawa, southwestern Japan. Geoderma 80: 75-93.  Oke TR. 1978. Boundary layer climates. London: Methuen and Sons 372 pp.  Okunishi K, Iida T. 1981. Evolution of hillslopes including landslides. Transactions of the Japanese Geomorphological Union 2: 291–300. O’Loughlin EM. 1981. Saturation regions in catchments and their relations to soil and topographic properties. Journal of Hydrology 53: 229–246.  Onda Y, Mori A, Shindo S. 1992. The effects of topographic convergence and location of past landslides on subsurface water movement on granitic hillslope. Journal of Natural Distaster Science 14: 45–58.  179  Onda Y, Tsujimura M, Tabuchi H. 2004. The role of subsurface water flow paths on hillslope hydrological processes, landslides and landform development in steep mountains of Japan. Hydrological Processes 18: 637–650.  R Development Core Team. 2008. R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria. ISBN 3-900051-07-0, URL http://www.Rproject.org.  Sasaki Y, Fuhii A, Asai K. 2000. Soil creep precess and its role in debris slide generation - field measurements on the north side of Tsukuba Mountain in Japan. Engineering Geology 56: 163–183.  Schwab JW. Unpublished report. Research Project EP 782.07. Historical documentation of mass wasting in North Coastal British Columbia. Forest Sciences Section, BC Forest Service, Smithers, BC.  Shimokawa, E. 1984. A natural recovery process of vegetation on landslide scars and landslide periodicity in forested drainage basins. In Symposium on effects of forest land use on erosion and slope stability, O’Loughlin CL, Pierce AJ (eds). East-West Center, University of Hawaii, Honolulu, 99–107  Sidle RC. 1984. Shallow groundwater fluctuations in unstable hillslopes of coastal Alaska. Zeitschrift für Gletscherkunde und Glazialgeologie 20: 79–95.  Sidle RC, Noguchi S, Tsuboyama Y, Laursen K. 2001. A conceptual model of preferential flow systems in forested hillslopes: evidence of self-organization. Hydrological Processes 15: 1675-1692. Sidle RC, Pearce AJ, O’Loughlin CL. 1985. Hillslope stability and land use. Water Res. Monograph 11. American Geophysical Union: Washington DC, 140 pp. 180  Sidle RC, Swanston DN. 1982. Analysis of a small debris slide in coastal Alaska. Canadian Geotechnical Journal 19:167–174.  Sidle RC, Tsuboyama Y, Noguchi S, Hosoda I, Fujieda M, Shimizu T. 1995. Seasonal hydrologic response at various spatial scales in a small forested catchment, Hitachi Ohta, Japan. Journal of Hydrology 168: 227–250. Sinai G, Zaslavsky D, Golany P. 1981. The effect of soil surface curvature on moisture and yield – Beer Sheba observation. Soil Science 132: 367–375.  Sutherland-Brown A. 1968. Geology of the Queen Charlotte Islands. Bulletin 54. British Columbia Department of Mines and Petroleum Resources.  Talebi A, Troch PA, Uijlenhoet R. 2008. A steady-state analytical hillslope stability model for complex hillslopes. Hydrological Processes 22: 546–553. DOI: 10.1002/hyp.6881  Terajima T, Sakamoto T, Shirai T. 2000. Morphology, structure and flow phases in soil pipes developing in forested hillslopes underlain by a Quaternary sand-gravel formation, Hokkaido, northern main island in Japan. Hydrological Processes 14: 713–726.  Thompson JC, Moore RD. 1996. Relations between topography and water table depth in a shallow forest soil. Hydrological Processes 10: 1513-1525.  Tsuboyama Y, Sidle RC, Noguchi S, Hosoda I. 1994, Flow and solute transport through the soil matrix and macropores of a hillslope segment, Water Resources Research 30: 879–890.  Tsukamoto Y, Ohta T, Noguchi H. 1982. Hydrological and geomorphological study of debris slides on forested hillslope in Japan. IAHS Publ. 137: 89–98. 181  Tutiempo.net, http://www.tutiempo.net, 20 April 2010.  Uchida T, Kosugi K, Mizuyama T. 2001. Effects of pipeflow on hydrological process and its relation to landslide: a review of pipeflow studies in forested headwater catchments. Hydrological Processes 15: 2151–2174.  Wang T, Hamann A, Spittlehouse D, Aitken SN. 2006. Development of scale-free climate data for western Canada for use in resource management. International Journal of Climatology 26: 383–397.  Wilford DJ, Cherubini P, Sakals ME. 2005. Dendroecology: A guide for using trees to date geomorphic and hydrologic events. BC Ministry of Forests Research Branch: Victoria BC, Land Management Handbook 58. <http://www.for.gov.bc.ca/hfd/pubs/Docs/Lmh/Lmh58.htm>  Wilson CJ, Dietrich WE. 1987. The contribution of bedrock groundwater flow to storm runoff and high pore pressure development in hollows. Erosion and Sedimentation in the Pacific Rim (Proceedings of the Corvallis Symposium, August, 1987) IAHS Publication 165: 49–59.  Wilson CM, Smart PL. 1984. Pipes and pipe flow processes in an upland catchment, Wales. Catena 11: 145–158.  Zeimer RR, Albright JS. 1987. Subsurface pipeflow dynamics of north-coastal California swale systems. Erosion and Sedimentation in the Pacific Rim (Proceedings of the Corvallis Symposium, August, 1987) IAHS Publication 165: 71–80.  182  5. THE EXTENDED TEMPORAL EFFECT OF SHALLOW LANDSLIDES1 INTRODUCTION Landslides are a natural process in many environments (Swanston, 1991). They occur as a result of increased shear stress or decreased shear strength (Turner and Schuster, 1996). Increases in landslides and soil erosion lead to many detrimental effects to the environment. Drinking water supplies can be contaminated through the introduction of fine sediment. Elevated levels of fine particulate matter cause increased problems for water treatment facilities and infrastructure and sediment-laden drinking water can cause gastrointestinal problems for humans (Strauss et al., 2001). Streams inundated with both fine and coarse sediment can directly affect fish populations through the destruction of habitat or the clogging of spawning gravels (Lisle, 1989; Wu, 2000). Fish can also experience indirect effects through the reduced productivity of tributary streams (Vannote et al., 1980; Wipfli et al., 2007; Kobayashi et al., 2010). Forest growing sites can be lost as a result of soil removal from initiation and transport zones (Smith et al., 1986). Although they can be extremely destructive, landslides are also known for producing variability and diversity of the landscape through the reconfiguration of soil, topographic, and hydrologic features (Geertsema et al., 2009). Fish can benefit from increased biotic production in the years following a landslide (Wipfli and Musslewhite, 2004). Some forest sites will become more productive following landslide (Smith et al., 1986) and the regeneration of some forest sites are assisted by landslide disturbance (Veblen, 1982; Banner et al., 2005).  1  A version of this chapter will be submitted for publication. Sakals ME, Innes JL. 2010. The extended temporal effect of shallow landslides  183  An increase in landslide activity as a result of land use has been found by many investigators in mountainous regions (e.g., Schwab, 1988; Jakob, 2000; Montgomery et al., 2000; Sidle and Ochiai, 2006). Land use and land conversion has been implicated in geomorphic activity over widespread areas, such as the extensive landsliding following deforestation in Scotland (Innes, 1997), and New Zealand (Smale et al., 1997; Glade, 2003), or the anthropogenic terracing affecting slope stability in Peru (Alexander, 1992). Landslides triggered by land use can be divided into two broad categories: initial failures and repeated failures. Initial failures are those that, prior to the change in land use, do not appear to have previously occurred for an extended duration. Over the same period, repeated failures have produced at least one landslide. In the case of repeated failure, the land-use induced failure may not have occurred as soon, or to the same extent, under natural conditions. The remainder of this paper discusses three major hydrogeomorphic features that develop following a shallow landslide and that can affect the long-term stability of a slope: increased convergence, creation of discontinuities in the subsurface hydrologic network, and continued geomorphic adjustments. A final section will comment on the differences between natural and land-use induced landslides.  INCREASED CONVERGENCE FOLLOWING A SHALLOW LANDSLIDE Landslides are known to occur preferentially in convergent topography (Tsukamoto et al., 1982; Montgomery et al., 2000; Chigira, 2002; Casadei et al., 2003; Talebi et al., 2008; Chapter 3). The convergent, or concave in the cross-slope direction (Tsukamoto and Minematsu, 1987), morphology of the initiation zone is enhanced due to the removal of material from the axis of the failure and the retention of material above and beyond the head and side scarps. Two factors can contribute to the increased convergence. The first is the topography of the underlying bedrock surface, while the second is the 184  difference in soil surface elevations between failure area and head scarp areas. Together, these affect the routing of both water and hillslope materials (Okunishi and Iida, 1982; Tsukamoto et al., 1982; Reneau and Dietrich, 1987; Sidle, 1987; Onda et al., 1992).  Soil surface alteration Characteristically, the outcome of a shallow landslide is that the failure area is lower than the surrounding surface. Thus, concurrent with the landslide there is the creation, or reinforcement, of a zero-order basin. Strong surface gradients are immediately created and rapid re-alignment of slope vectors into the incipient basin occurs. Through soil diffusion processes (Roering et al., 1999), the scarp slopes are moderated and material is transported toward the axis of the zero-order basin. There is also an associated increase in the extent of the zero-order basin as the scarp slopes are moderated by post-event geomorphic and biological processes. Surface expression has been found to control subsurface storm flow at high flows (Hutchinson and Moore, 2000), and any occurrence of overland flow will also be focused into the convergence of the failure area. Sediment transport will be associated with the movement of the water on the hillslope. In Chapter 2, we showed that sites of cross-slope convergence had higher levels of sediment and debris transport than other morphologic configurations. Both the surface and subsurface water flows are influenced by the presence of the zero-order basin and the resulting increased probability of saturation (Sidle and Swanston, 1982; Onda et al., 1992) and delivery of water (Thompson and Moore, 1996; Hutchinson and Moore, 2000) will increase the probability of a subsequent failure.  Basal surface degradation Shallow landslides commonly encompass the full depth of the soil profile (Chapter 3), leading to the exposure of the underlying material. In shallow landslides this is often bedrock, though it can be compact 185  till or another material (e.g., saprolite). Following failure and the resultant thinning of the soil depth, the geologic weathering of the basal material is accelerated (Dietrich et al., 1995). The basal surface has increased exposure to various agents such as fire, frost action, acid rain, wetting-drying cycles, and biological weathering (Birkeland, 1999). Bedrock can be cracked and comminuted by the action of roots. Windthrow of trees with roots anchored in bedrock removes surprising amounts of material from the basal surface depending upon the structure and competence of the rock. Field measurements of windthrown root systems of ~20 yr old Sitka alder (Alnus viridis (Chaix.) D.C.) have found ~0.03 m3 of previously intact but strongly jointed schistose bedrock in a single exposed root plate, and larger trees may lift proportionally greater amounts of bedrock. Such rapid degradation of the underlying surface will accentuate positive feedback mechanisms for convergence. In contrast, lightly jointed, competent rock or compact till will exclude roots and weather at slower rates. Another set of factors contributing to the positive feedback of convergence on the basal layer is the spatially varying properties of the bedrock. The orientation of joints in the underlying bedrock has been found to control the location of surface gullies (Beavis, 2000), and landslides are often located above zones of geologic discontinuities such as shear zones or areas of hydrothermal alteration (Hasegawa et al., 2009). These geologic features increase the effectiveness of the elements of rock weathering (Birkeland, 1999). Much of the material removed from the basal layer will be incorporated into the soil and be available for mobilization in subsequent failures. On steep slopes, greater soil depth is associated with decreased stability. Although erosion of the basal surface occurs at a much lower rate than the alterations of surface material, it determines the baseline for future accumulations of soil and debris and for surface expressions during the early phases of the accumulating soil. The surface expression of the basal layer may act to enhance or retard soil accumulation, in addition to other factors such as failure geometry. Surface expressions of low roughness may not provide enough 186  friction for the capture of material moved by slope wash or small landslides. Large organic debris, which could contribute structure to the failure area and act as a platform for further accumulation of both organic and inorganic material, would also be less likely to be arrested in the failure area if the surface expression has low roughness. Bedrock with dip steeply out of the slope tends to have lower surface roughness than other orientations. In Chapter 3, we found that bedrock with steep dip out of the slope was associated with thinner soils surrounding the failure areas and that the depth of surrounding soil was positively correlated with the rate of accumulation. Interactions exist between the hydrology of hillslopes and the bedrock mass. Zones of soil saturation are governed by the topography of the underlying confining layer (Hutchinson and Moore, 2000), and subsurface preferential flow paths are also influenced by impermeable bedrock topography (Noguchi et al., 1999; Freer et al., 2002). The bedrock mass is not hydrologically passive, as evidenced by shallow bedrock pathways of preferential flow and artesian exfiltration of water from bedrock fractures (Anderson et al., 2001). Bedrock degradation will cause an increase in the concentration of hillslope water and sediment. Landsliding removes the soil layer and may therefore reduce flow resistance for water flowing out of near-surface bedrock fractures.  DISCONTINUITIES OF THE SUBSURFACE HYDROLOGIC NETWORK Prior to an initial landslide, the soil at stable and unstable hillslope positions is likely to be at a similar stage of development. Minor differences in topographic expression can create slight differences in soil properties (Anderson and Burt, 1978), but the change is continuous through morphologic hillslope positions. After a shallow landslide, a hydrologic discontinuity is created that may contribute to subsequent landslides at the site. The frequencies of subsurface soil voids, which may conduct water, are smaller in the soils of failure areas than in the soils of the adjacent hillslopes (Chapter 4). Further, the soil of the failure area lacks depth where matrix flow could contribute to the downslope transmission of 187  stormflow and this could result in more frequent saturation. The occurrence of hillslope „seepage‟ sites at previous landslide locations exemplifies this occurrence (Figure 5.1). Saturation is one method to form high levels of water pressure that are often associated with hillslope failure (Van Asch et al., 1999). As material in the failure area accumulates, the newly deposited material is formed within a similar hydrologic environment to the previous soil, but with increased surface and potentially subsurface convergence. „Closed-end pipes‟ could possibly be formed as material is deposited over the outlets of preferential flow pathways (Kosugi et al., 2004). „Closed-end pipes‟ are pipes whose ends are blocked and when no alternative hydrologic pathway of similar capacity is available, hydrostatic pressures can rise rapidly. This flow restriction may elevate pore water pressures in the soils immediately above the head scarp and result in a failure and an extension of the failure area to include the area previously above the head scarp (McDonnell, 1990). The period during which a strong variation in soil depths exists is also a period when the „open-end pipes‟ originating above the head scarp can readily erode their lower segments and increase their flow capacity. Thus the discrepancy between the hydrologic efficiency of either side of the head scarp is accentuated. The occurrence of „open-end pipes‟ is expected to be much greater than „closed-end pipes‟, as most pipes are located in the upper portion of the profile (Chapter 4) and for most sites it takes several decades to result in enough material accumulation that the lowest of these pipes would begin to be blocked (Chapter 3). Eventually, it could be speculated that the accumulated soil in the scar will adjust as it becomes conditioned to accepting and transferring water from upslope contributing slope segments.  188  Figure 5.1. Landslide failure area with shallow soils compared to adjacent upslope hillslope locations. Approximate boundary of failure area is indicated by yellow annotation line. 189  One of the methods by which soil pipes are thought to be formed is through the establishment and decay of tree roots (Noguchi et al., 1999). In the situation of an accumulated soil that is in a convergent hillslope position, is of limited thickness, and lacks the hydrologic network of preferential flow pathways, substantial restrictions to the establishment of trees and their roots exist. The establishment of trees in the period following landslide is limited due to the thin soils over bedrock and the mobile nature of the sediment remaining in the failure (Smith et al., 1986; Chapter 2). Further, small failures in mature forests will receive little light at the soil surface and thus further discourage seedling establishment (Coates, 2000). For trees originating on the margin of the slide, it is the frequent saturation of the soil profile given the thin soil profile, increased convergence and lack of a developed subsurface drainage network that will discourage root growth into the failure area. Most trees also require access to soil air which would be in short supply during periods of saturation (Brady and Weil, 1996); thus, recent landslide failure areas are not favourable sites for the establishment of tree roots, or for the creation of associated preferential flow pathways. The incorporation of wood into the accumulated soil in the landslide failure area could create subsurface pathways; particularly since the orientation of coarse woody debris in failure areas is preferentially slope parallel or diagonal with a smaller portion being cross-slope (Figure 5.2). The preferential flow pathways formed in association with the wood may be of a continuous nature, more so than with root-derived pathways. The conditions for root development have further implications for root reinforcement in the accumulated soil. Root reinforcement has long been included in distributed models of slope stability (Wu and Sidle, 1995; Pack et al., 1998; Montgomery et al., 2000) but the spatial distribution of trees and their root systems, and therefore of root reinforcement, is seldom addressed (Sakals and Sidle, 2004; Danjon et al., 2008; Stokes et al., 2009). Favourable geomorphic, hydrologic and dendrologic conditions are all required for significant root development for either root reinforcement (Hales et al., 2009) or for the  190  development of preferential flow paths; these conditions often do not exist at the location of a recent landslide.  GEOMORPHIC ADJUSTMENTS Geomorphic adjustment occurs over two time scales with respect to the oversteepening of hillslopes surrounding the failure. The first response is the immediate regrading of hillslope form in the vicinity of the head and side scarps. This process begins soon after failure and declines within several decades of failure to a background level. In the first 20 yr following failure, rates are ~5 times higher than beyond 50 yr after failure (Chapter 2). The failure areas of recent shallow landslides also have more soil and rocks as surface cover (Chapter 2). Surfaces covered with these materials have increased frequency of material movement events relative to surface covers of vegetation and woody debris, which are more typical of undisturbed hillslopes (Chapter 2). The occurrence of geomorphic adjustments in temperate climates is highly episodic and may be the result of the influences of vegetation on material transport (Slaymaker, 1967). But the influences of vegetation are temporarily removed in the period immediately following a landslide, exposing soil and rock and resulting in higher rates of sediment movement.  191  Figure 5.2. Orientation of coarse woody debris in 26 shallow landslide failure areas (n=115). The second time scale of adjustment is much longer and spans multiple landslide events within the zeroorder basin. The development and extent of the zero-order basin is largely controlled by the underlying substrate competence and initial topographic expression as well as the magnitude of available material stored within the basin at the time of initial failure. In cases of readily weathered and erodible bedrock or other basal material, a gully may evolve. As development continues, the strong concentration in the axis of the gully will tend towards fluvial processes with the potential for episodic side-wall and gully confined debris flows depending upon the geologic properties (Innes, 1983). Hillslope water will issue freely into the cleft and the geomorphic development will be governed by the factors of basal material weathering. In cases of competent material, limited basal erosion will limit the contributing area. As material is removed from the basin through repeat landslides the positive feedbacks identified here (convergence, soil-hydrology discontinuities and geomorphic adjustments) are dampened by the lack of available material. Increased convergence will be controlled by the rate of bedrock degradation as the soil will have been thinned by repeat landsliding and subsequent adjustments. Discontinuities with respect to 192  the frequency of subsurface preferential flow pathways will be less dramatic as the increased period prior to exceeding the immunity depth will allow more hydrologic maturity of the infilled soils. Geomorphic adjustments will slow as the available material is reduced. Eventually, an extended duration will be required to accumulate enough materials in the axis of the basin to exceed the immunity depth. As the rate of material accumulation is positively related to the depth of soils on the adjacent slopes (Chapter 3), this may be realized rapidly if soils were originally thin or if previous failures had begun the process of hillslope evacuation. Over time, the accumulation of organic material will increase in importance. Eventually, the entire zero-order basin could host an organic soil, especially on competent, weatheringresistant bedrock. While failures continue to be possible, the duration between periods when the soil depth exceeds the immunity depth may be extended and the slope may be in a stable configuration for long periods. Regardless of whether the morphologic development tends towards the erosion of basal material and gullying or the lack of basal erosion and material exhaustion, eventually, much of the material stored on the hillslope at the initiation of the shallow landsliding sequence may be transported downslope.  LAND USE AND THE GEOMORPHIC CYCLE OF SHALLOW LANDSLIDES When a failure is associated with land use change, the main difference between primary and failed slopes is that the geomorphic adjustments associated with primary failures are initiated by the land-use change, whereas with failed slopes, the land use change affects the pre-established cycle of hillslope instability. As a result, the impact on failed slopes will be less likely to have a dramatic effect on the long-term processes of the hillslope. Land-use change can induce primary failures as well as triggering failures in sites of repeated failure. Vegetation conversion has been demonstrated to have measurable affects on slope stability in many locations (O‟Loughlin and Pearce, 1976; Innes, 1989; Prosser and Soufi, 1998). In more stable 193  landscapes, landslides are often the result of human activities such as drainage diversions as a result of road construction (Montgomery, 1994) or vegetation conversion. These slides are generally small and individually their effect is considered negligible, but exceptions occur where the magnitude of intercepted drainage is large, where the slope water is delivered to steep terrain, or where particular types of sediment deposits exist (Schwab, 2002). For failed slopes, land-use related activities such as road construction, may add road sidecast material to zero-order basins (VanBuskirk et al., 2005), or may result in extra hillslope water arriving at the site from drainage concentration in ditches (Montgomery, 1994). During road construction, the excess weight and vibrations of machinery may exceed the bearing strength of the hillslope materials, particularly when the soils are of high organic content representing a late stage of the shallow landsliding cycle on competent basal material. The hillslope response may be viewed from a sediment budget perspective at the hillslope scale (Slaymaker, 2003). Inputs of material come from upslope and from the weathering of bedrock; outputs are episodic shallow landslides from the axis of the zero-order basin. The change in storage refers to the gradual depletion of hillslope soil material. Hillslopes with greater amounts of hillslope storage will require more time or larger or more frequent landslides to reach extended periods of stability again once the process has been initiated. Thus, hillslopes with extensive sediment deposits, such as many slopes with a glacial legacy, should be managed with utmost care. The depletion of this storage represents a catastrophic loss of hillslope soil material that could support hillslope forests for timber harvesting or protection forestry.  CONCLUSIONS In this manuscript the results of recent work have been combined to develop a model of the cycle of shallow landslides. The central tenet is that landslide sites will develop three features that may propagate further hillslope disturbance: increased convergence into the failure area; discontinuous hydrologic 194  pathways between the hillslope and the failure area; and the geomorphic adjustment of contributing slopes that results in repeated material accumulation in shallow landslide failure areas until the supply of available hillslope material is depleted. The first two hillslope responses will not assuredly result in further landslides, but both increase the driving forces for slope movement or decrease the resisting factors. The geomorphic adjustment of contributing slopes will respond relative to the depths of soil on those slopes, as material available for transport into the failure area declines, the geomorphic adjustment will correspondingly slow. Hillslopes with greater material storage will require greater adjustment before extended periods of relative hillslope stability are experienced. Land managers must continue to keep slope stability as a very high priority in determining actions on landscapes that are prone to landsliding or are liable to become prone to a cycle of shallow landsliding.  195  REFERENCES Alexander D. 1992. On the causes of landslides – human activities, perception, and natural processes. Environmental Geology 20:165–179.  Anderson MG, Burt TP. 1978. The role of topography in controlling throughflow generation. Earth Surface Processes 3: 331–344.  Anderson SP, Dietrich WE, Montgomery DR, Torres R, Conrad ME, Loague K. 1997. Subsurface flow paths in a steep, unchanneled catchment. Water Resources Research 33(12): 2637–2653.  Banner A, LePage P, Moran J, de Groot A (eds). 2005. The HyP3 Project: pattern, process, and productivity in hypermaritime forests of coastal British Columbia – a synthesis of 7-year results. BC Ministry of Forests Research Branch: Victoria BC, Special Report 10. <http://www.for.gov.bc.ca/hfd/pubs/Docs/Srs/Srs10.htm>  Beavis SG. 2000. Structural controls on the orientation of erosion gullies in mid-western New South Wales, Australia. Geomorphology 33: 59–72.  Birkeland PW. 1999. Soils and geomorphology. Oxford University Press: New York. 430 pp.  Brady NC, Weil RR. 1996. The nature and properties of soils. 11th ed. Prentice Hall: New Jersey, 740 p.  Casadei M, Dietrich WE, Miller NL. 2003. Testing a model for predicting the timing and location of shallow landslide initiation in soil-mantled landscapes. Earth Surface Processes and Landforms 28: 925–950.  196  Chigira M. 2002. Geologic factors contributing to landslide generation in a pyroclastic area: Augsut 1998 Nishigo Village, Japan. Geomorphology 46: 117–128.  Coates KD. 2000. Conifer seedling response to northern temperate forest gaps. Forest Ecology and Management 127: 249–269.  Danjon F, Barker DH, Drexhage M, Stokes A. 2008. Using Three-dimensional Plant Root Architecture in Models of Shallow-slope Stability. Annals of Botany, 101:1281–1293. doi:10.1093/aob/mcm199  Dietrich WE, Reiss R, Hsu ML, Montgomery DR. 1995. A process-based model for colluvial soil depth and shallow landsliding using digital elevation data. Hydrological Processes 9: 383–400.  Freer J, McDonnell JJ, Beven K, Peters NE, Burns DA, Hooper RP, Aulenbach B, Kendall C. 2002. The role of bedrock topography on subsurface storm flow. Water Resources Research 38: 1269. DOI: 10.1029/2001WR000872.  Geertsema M, Highland LM, Vauegois L. 2009. Environmental impact of landslides. In Sassa K, Canuti P (eds). Landslides – Disaster Risk Reduction, Springer-Verlag: Berlin Heidelberg; 585–604.  Glade T. 2003. Landslide occurrence as a response to land use change: A review of evidence from New Zealand. Catena 51: 297–314.  Hales TC, Ford CR, Hwang T, Vose JM, Band LE. 2009. Topographic and ecologic controls on root reinforcement, Journal of Geophysical Research 114: F03013. DOI:10.1029/2008JF001168.  Hasegawa S, Dahal RK, Yamanaka M, Bhandary NP, Yatabe R, Inagaki H. 2009. Causes of large-scale landslides in the Lesser Himalaya of central Nepal Environmental Geology 57: 1423–1434. DOI: 10.1007/s00254-008-1420-z 197  Hutchinson DG, Moore RD. 2000. Throughflow variability on a forested hillslope underlain by compacted glacial till. Hydrological Processes 14: 1751–1766.  Innes JL. 1983. Debris flows. Progress in Physical Geography 7: 469–501.  Innes JL. 1989. Rapid mass movements in Upland Britian: A review with particular reference to debris flows. Studia Geomorphologica Carpatho-Balcanica: Krakow, 23: 53–66.  Jakob MJ. 2000. The impacts of logging on landslide activity at Clayoquot Sound, British Columbia. Catena 38: 279–300.  Kobayashi S, Gomi T, Sidle RC, Takemon Y. 2010. Variations of headwater stream ecosystems related to stand age of managed forest catchments: responses of macroinvertebrate community to forest clearcutting and debris flow occurrence. Canadian Journal of Fisheries and Aquatic Sciences 67 (in press). doi:10.1139/F09-186  Kosugi K, Uchida T, Mizuyama T. 2004 Numerical calculation of soil pipe flow and its effect on water dynamics in a slope. Hydrological Processes 18: 777–789. DOI: 10.1002/hyp.1367  Lisle TE. 1989. Sediment transport and resulting deposition in spawning gravels, North Coastal California. Water Resources Research 25: 1303–1319.  McDonnell JJ. 1990. The influence of macropores on debris flow initiation. Quarterly Journal of Engineering Geology 23: 325–331.  Montgomery D. 1994. Road surface drainage, channel initiation, and slope instability. Water Resources Research 30: 192–193.  198  Montgomery DR, Schmidt KM, Greenberg HM, Dietrich WE. 2000. Forest clearing and regional landsliding. Geology 28: 311–314.  Noguchi S, Tsuboyama Y, Sidle R, Hosoda I. 1999. Morphological characteristics of macropores and the distribution of preferential flow pathways in a forested slope segment. Soil Science Society of America Journal 63: 1413–1423.  Okunishi K, Iida T. 1981. Evolution of hillslopes including landslides. Transactions of the Japanese Geomorphological Union 2: 291–300. O’Loughlin CL, Pearce AJ. 1976. Influence of Cenozoic geology on mass movement and sediment yield response to forest removal, North Westland, New Zealand. Bulletin of the International Association of Engineering Geologists 14: 41–46.  Onda Y, Mori A, Shindo S. 1992. The effects of topographic convergence and location of past landslides on subsurface water movement on granitic hillslope. Journal of Natural Distaster Science 14: 45–58.  Pack RT, Tarboton DG, Goodwin CN. 1998. The SINMAP approach to terrain stability mapping. In Proceedings, International Congress of the International Association for Engineering Geology and the Environment 8. Moore DP Hungr O (eds). AA Balkema: Rotterdam, Netherlands, 8: 1157–1165.  Prosser IP, Soufi M. 1998. Controls on gully formation following forest clearing in a humid temperate environment. Water Resources Research 34: 3661–3671.  Reneau SL, Dietrich WE. 1987. The importance of hollows in debris-flow studies; examples from Marin County, California. In Debris Flows/Avalanches. Costa JE, Wieczorek GF (eds). Geological Society of America, Review of Engineering Geology 7: 165–180. 199  Roering JJ, Kirchner JW, Dietrich WE. 1999. Evidence for nonlinear, diffusive sediment transport on hillslopes and implications for landscape morphology. Water Resources Research 35: 853–870.  Sakals ME, Sidle R. 2004. A spatial and temporal model of root cohesion in forest soils. Canadian Journal of Forest Research 34: 950–958.  Schwab JW. 1988. Mass wasting impacts to forest land: Forest management implications, Queen Charlotte Timber Supply Area. In Degradation of forested land: forest soils at risk. Proceedings of the 10th British Columbia Soil Science workshop, Lousier JD, Still GW (eds). British Columbia Ministry of Forests, Land Management Report 56: 104–115. Schwab JW. 2002. Donna Creek – What did we learn? In Terrain stability and forest management in the interior of British Columbia. Jordan P, Orban J (eds). BC Ministry of Forests Research Program: Victoria BC; Tech. Rep. 003: 1–13. www.for.gov.bc.ca/hfd/pubs/Docs/Tr/Tr003.htm  Sidle RC. 1987. A dynamic model of slope stability in zero-order basins. In Erosion and sedimentation in the Pacific Rim (Proceedings of the Corvallis Symposium, August, 1987) International Association of Hydrological Sciences Publication 165: 101–110.  Sidle RC, Ochiai H. 2006. Landslide processes, prediction, land use. American Geophysical Union, Washington, DC, pp 1–312.  Sidle RC, Swanston DN. 1982. Analysis of a small debris slide in coastal Alaska Canadian Geotechnical Journal 19: 167–174. DOI:10.1139/t82-018  Slaymaker O. 1967. Patterns of erosion in small catchments in mid-Wales. Unpublished Ph. D. thesis, University of Cambridge. 200  Slaymaker O. 2003. The sediment budget as conceptual framework and management tool. Hydrobiologia 494:71–82.  Smale MC, McLeod M, Smale PN. 1997. Vegeatation and soil recovery on shallow landslide scars in tertiary hill country, East Cape region, New Zealand. New Zealand Journal of Ecology 21: 31–41.  Smith RB, Commandeur PR, Ryan MW. 1986. Soils, vegetation, and forest growth on landslides and surrounding logged and old-growth areas on the Queen Charlotte Islands. British Columbia Ministry of Forests Land Management Report 41: Victoria British Columbia; Land Management Report 41.  Stokes A, Atger C, Bengough AG, Fourcaud T, Sidle RC. 2009. Desirable plant root traits for protecting natural and engineered slopes against landslides. Plant and Soil 324: 1–30.  Strauss B, King W, Ley A, Hoey JR. 2001. A prospective study of rural drinking water quality and acute gastrointestinal illness. BMC Public Health 1:8. http://www.biomedcentral.com/1471-2458/1/8  Swanston DN. 1991. Natural processes. American Fisheries Society Special Publication 19: 139–179.  Talebi A, Uijlenhoet R, Troch PA. 2008. A low-dimensional physically based model of hydrologic control of shallow landsliding on complex hillslopes. Earth Surface Processes and Landforms 33: 1964–1976.  Thompson JC, Moore RD. 1996. Relations between topography and water table depth in a shallow forest soil. Hydrological Processes 10: 1513–1525.  Tsukamoto Y, Minematsu H. 1987. Hydrological characteristics of a zero-order basin. Erosion and Sedimentation in the Pacific Rim (Proceedings of the Corvallis Symposium, August, 1987) IAHS Publication 165: 61–70.Tsukamoto Y, Ohta T, Noguchi H. 1982. Hydrological and 201  geomorphological study of debris slides on forested hillslope in Japan. In Recent Developments in the Explanation and Prediction of Erosion and Sediment Yield. Walling DE (ed). International Association of Hydrological Sciences Publication 137: 89–98.  Turner AK, Schuster RL (eds.). 1996. Landslides: investigation and mitigation. Transportation Research Board, National Research Council. National Academy Press: Washington DC; Special Report 247.  Vannote RL, Minshall GW, Cummins KW, Sedell JR, Cushing CE. 1980. The river continuum concept. Canadian Journal of Fisheries and Aquatic Sciences 37: 130–137. DOI:10.1139/f80-017  Van Asch TWJ, Buma J, Van Beek LH. 1999. A view on some hydrological triggering systems in landslides. Geomorphology 30: 25–32.  VanBuskirk CD, Neden RJ, Schwab JW, Smith FR. 2005. Road and terrain attributes of road fill landslides in the Kalum Forest District. BC Ministry of Forests and Range, Research Branch: Victoria BC; Technical Report 24. <http://www.for.gov.bc.ca/hfd/pubs/Docs/Tr/Tr024.htm>  Veblen T. 1982. Natural hazards and forest resources in the Andes of South-Central Chile. GeoJournal 6:141–150.  Wipfli MS, Musslewhite J. 2004. Density of Red Alder (Alnus rubra) in headwaters influences invertebrate and detritus subsidies to downstream fish habitats in Alaska. Hydrobiologia 520: 153– 163.  Wipfli MS, Richardson JS, Naiman RJ. 2007. Ecological linkages between headwaters and downstream ecosystems: transport of organic matter, invertebrates, and wood down headwater channels. Journal of the American Water Resources Association, 43: 72–85. DOI: 10.1111 ⁄ j.1752-1688.2007.00007.x 202  Wu FC. 2000. Modeling embryo survival affected by sediment deposition into salmonid spawning gravels: Application to flushing flow prescriptions. Water Resources Research 36: 1595–1603.  Wu W, Sidle RC. 1995. A distributed slope stability model for steep forested basins. Water Resources Research 31: 2097–2110.  203  6. CONCLUSION This thesis has increased our understanding of the rates and processes of soil redistribution and the spatial occurrence of subsurface hydrologic features in the initiation zones of shallow landslides that originate from forested hillslopes. Slope stability concerns are widespread in mountainous regions and with increasing land use in steepland areas there is an impetus for greater knowledge concerning slope hazards. This knowledge will lead to an improved understanding of geomorphic hazards and land use effects within the context of natural background geomorphic activity. The preceding chapters illustrate the complexities and sensitivities of initiation zones to perturbation. Together, the second and third chapters show that shallow landslide failure areas are sites of material accumulation as a result of small-scale forest and hydrogeomorphic processes; the fourth chapter provides evidence that the hydrologic properties of infilled soils is different than the surrounding hillslopes. The fifth chapter outlines a hypothesis regarding the cycle of shallow landsliding in light of the findings of the previous chapters.  SMALL-SCALE PROCESSES Following a shallow landslide, localized processes deposit material in the failure area resulting in soil accretion (Dietrich et al., 1987). Recent processes of material transport in shallow landslide initiation zones were found to decline in volume with time since the landslide (Chapter 2). Although the volume of material transport was found to have a temporal pattern, relations with both microtopography and surface cover were highly variable. Relations, typically in the form of envelop curves, were found between occurrences of material transport events and both surface cover and local topographic relief. These relations provide information regarding the adjustment of initiation zones following shallow landslide and can thus be used to refine models of material transfer on hillslopes. The reduction of local relief and the establishment of vegetation appear sufficient to consider failure areas mature with respect to small-scale 204  processes of hillslope readjustment. At the small spatial (on the order of 10 m) and temporal scale (2 years prior to field visit) of the process study, stochastic forest (and hydrogeomorphic) processes were thought to be responsible for differences of several orders of magnitude in material transport volumes. Strong forest effects were observed on processes of soil surface erosion and deposition, contribution of organic matter, and contribution to scarp stability through root reinforcement. Differences in the forest influence between the study sites in British Columbia and those in Japan was likely greatest at the point-scale, the smallest scale in this investigation. As a result of intensive forest management and high utilization, coarse woody debris is scarce at the Japanese sites. Further, the plantation forests are less complex, typically comprised of just a discontinuous herbaceous layer and then the canopy of conifers (Puettman et al., 2009). In British Columbia and at the scale of a failure, the sites were essentially unaltered by human activities; sites that had been timber harvested were influenced but not to the same degree as the sites in Japan. Differing forest harvesting methods also create much more fine woody debris at the previously harvested sites in British Columbia. Whole-tree harvesting, or at least whole-tree yarding, was witnessed during the investigation of the Wakayama sites in Japan and its effects were noted in the paucity of woody debris at the sites in Wakayama and the other sites in Japan. In British Columbia, most limbing and some bucking (cutting to length) occurs prior to yarding and thus is distributed on the hillslope. Further, the harvesting of old-growth stands with a spectrum of tree types results in much greater waste than in an even-aged stand such as those in Japan. Commercial thinning, or selective harvesting of immature second growth was also practiced in Japan; such practices are uncommon in British Columbia. Occurring along with these different forest conditions and forest management and harvesting regimes, there is a difference in the distribution of recent processes of material transport. Slope wash was the dominant method of material transfer in Japan (Table 2.9), but most of the occurrences were not large enough to be captured as a volume of material transfer. Slope materials are gradually moved downslope into hillslope sites of preferential deposition by many relatively 205  small movements. The more consistent (in both space and time) material transfer appeared to be a result of the more consistent forest conditions on the studied hillslopes in Japan. Unfortunately, this is somewhat a factor of developing sampling methods in one location and then applying them to another location that functions slightly differently. In British Columbia, the occurrence of slope wash occurred only in areas of recent vegetation disturbance as elsewhere the soil was protected by a complex of vegetation and organic materials. In British Columbia, material transport may be characterized as more episodic, more stochastic, and of greater individual volumes compared to Japan, again this could be a reflection of the more variable forest conditions. Future study relating to the complexities of forest conditions in shallow landslide initiation zones and the related effects on the associated material transport regime would be a welcome scientific contribution especially for applied forest land management. As evidence of such inclination of land managers in Japan is the practice of gathering forest debris into slope perpendicular wattles following harvesting. Presumably, the intent is to minimize the magnitude of slope wash, but the effects of such practices may not have been studied scientifically and while it may reduce slope wash, it may also exacerbate other processes of hillslope material transfer (Figure 6.1) An alternative perspective could be that the differences in material transport regimes are a sole result of the difference in precipitation intensities. Although the difference between extreme daily totals for the South Coast study area and the Japanese study area are considerable but not incomparable, the differences between the North Coast and Haida Gwaii study areas are larger. A major detriment is the poor representation of the Haida Gwaii study area by the Sandpit weather station (Karanka, 1986).  206  Figure 6.1. A hillslope in Japan with slope-perpendicular post-harvesting litter arrangement and subsequent small failures.  MATERIAL ACCUMULATION Soil failure has a critical thickness (Crozier et al., 1990); thus a degree of soil depth accumulation must occur prior to a further shallow landslide being possible from the failure area. The occurrence of previous landslides at a site has an influence over the material available for site recharge (Onda et al., 1992). This 207  study was restricted to shallow-soiled slopes and thus if a hillslope segment has recently experienced greater erosion than deposition or weathering, it may have temporarily reduced material storage in the geomorphically contributing area such that hillslope soil depths are rather thin, comprised of a greater proportion of organic materials, and less prone to moving downslope into the failure area. In the failed area of a shallow landslide initiation zone, soil depths are chiefly dependent upon the time since failure (Shimokawa, 1984; Shimokawa et al., 1989; Dietrich et al., 1982; Smale et al., 1997, Chapter 3). With more time elapsed since failure, there is an increasing probability that stochastic forest and hydrogeomorphic processes will have produced soil material (Heimsath et al., 2001; DeLong et al., 2008) and/or subsequently transported it into to the failure area. In this study, the central tendency (i.e., within the variability provided by stochastic effects) of accumulations was found to approximate a non-linear (sigmoid) relation with time (Chapter 3). As a result of the stochastic nature of forest processes and the influence of the forest conditions on the small-scale hydrogeomorphic processes, the accumulation of soil material in failure areas is likewise marked by a substantial degree of variance; this variance was not expected to decline with further sampling. At the scale of a failure area, the forests in coastal British Columbia are thought to be strong determinants of the rate of material accumulation. As a negative influence, the reinforcement of scarp slopes by living root systems could maintain strong surface gradients for much longer than in the case of a much smaller root network. A positive influence could be the windthrow of a stem across a failure area. Not only is the deposition of the stem a large contribution of material to the failure area, but the structure of the stem could then act to retain materials in the failure area that may otherwise have passed through. A similar, but smaller-scale effect occurred with the deposition of fine woody debris. Each of these processes was found to affect the accumulated soil depth at sites in British Columbia; in Japan, only limited occurrences of fine woody debris trapping were noted at the study sites. Despite the stochastic nature of the forest influence all the study areas, the volumes of material transfer (Chapter 2) and accumulation (Chapter 3) 208  reasonably agree. Considering the occurrence of deposition events of several cubic metres of material that were observed several times during field traverses, this may be considered a measured success. Forest processes result in individual forest elements growing, dying, falling, and decaying, but the geomorphic and hydrologic conditions are also varying with time (Chapters 2, 3, and 4). The degree to which material deposition and soil accretion variability can be assigned to geomorphic, hydrologic, or forest effects would be worthy of further research. In response to such forest influence, an observation-based qualitative probabilistic model is proposed to describe the likelihood of three components regarding the soil depth in previous failure areas. Together, the probabilities for small-scale deposition, small-scale erosion, and catastrophic erosion describe the regime of material movement for typical conditions of local relief and time since failure (revegetation). This model could be used in the field to qualitatively assess the potential for future material contributions and in combination with the accumulated soil depth to predict the susceptibility of the failure area to future landsliding. This does not preclude failures from the remainder of the initiation zone; however, in all of the initiation zones visited, at least a portion of the failure areas overlapped and thus the volume of the accumulated material has some bearing on the volume of subsequent failure and the likelihood of inducing a debris flow (Millard, 1999).  PREFERENTIAL FLOW PATHWAYS On hillslopes dominated by subsurface flow, disruption of subsurface flow pathways occurs coincidently with failure of the soil material by landslide. The hydrology of initiation zones is typically dominated by the presence of preferential flow pathways (Uchida et al., 2001); however, the analysis of the frequency of preferential flow pathways in landslide initiation zones suggests that the infilled soils of failure areas are less capable of transferring hillslope throughflow than adjacent hillslope soils (Chapter 4). Although subsurface features of low resistance have been previously identified (Okunishi and Iida, 1981), little 209  spatially explicit analysis has been conducted prior to this investigation, especially in shallow landslide initiation zones. Here, analysis regarding the spatial occurrence of the features was conducted of both quantitative topographic position and relative morphologic position. Both qualitative and quantitative descriptions of preferential flow pathway frequencies/locations can be compared to regions of saturation with respect to topography (i.e., O‟Loughlin, 1981) to further the understanding of hillslope hydrologic networks in the vicinity of shallow landslides. Many occurrences of forest road failures are a result of a disruption to an unexpectedly high magnitude of subsurface flow. Similar discontinuities of the subsurface hydrologic network are thought to be important to shallow landslide initiation (McDonnell, 1990; Kosugi et al., 2004; Nieber and Sidle, 2010), yet these discontinuities may not be important where unrestricted downslope flow is occurring such as in recently evacuated landslide failure areas. As the as these sites are infilled, the occurrence of a discontinuity could have strong implications to future landsliding. This study has shown that the infilled soils of shallow landslide failure areas are hydrologically distinct from their contributing areas, previous studies had identified a change but information was only found reported from a single site (Onda et al., 1992). Despite the limited data, the introduction of a temporally varying frequency of preferential flow pathways in failure areas could prove important for understanding sites of repeated failure, further research is warranted. The frequency of Type II pathways was not dramatic between study areas. However, there appeared to be a lack of Type I pathways in Japan. This could be due to the predominance of slope wash deposits that will occupy the lowest topographic positions. In contrast, the lesser contributions of slope wash in British Columbia leaves such sites to recharge by a lesser degree of slope wash, and greater contributions from the deposition of organic debris and the episodic occurrences of small-scale landslides from the scarp slopes of the failure. These landslides are often comprised of a great deal of organic matter and would be expected to result in more frequent occurrences of Type I pathways.  210  The effectiveness of the knocking cone penetrometer to determine the presence and location of subsurface soil voids could be strengthened simply by recording more information from each profile. Combining penetrometer profiles with an observation trench would allow the assessment of the capabilities and limitations of the instrument. In forest engineering, subsurface conditions are rarely well known and thus the increased use of knocking cone penetrometers in both research and forest operational assessments could provide valuable subsurface information.  CONTRIBUTING AREAS The use of the term “zero-order basin” extends the fluvial classification of stream order (Strahler, 1957; Shreve, 1966) to the hillslope, although the “zero” designation connotes an endpoint or at least a threshold. I viewed the dimensions of the zero-order basin to be defined topographically, and at some shallow landslide sites (such as those on divergent topography) the zero-order basin was not formed prior to failure and the extent of the zero-order basin was no larger than the failure itself. After failure, the site attains a convergent form due to the difference in surface elevation between the failure area and the adjacent hillslope. The lack of a contributing surface topographic basin does not preclude these sites from being areas of subsurface flow concentration, as preferential flow pathways can overwhelm topographic controls and redistribute hillslope throughflow to certain hillslope locations (Tsuboyama et al., 1994; Hutchinson and Moore, 2000) and preferential flow paths have been noted in many head scarps in this study and others (Tsukamoto et al., 1982; Uchida et al., 2001). Zero-order basins are implied to represent the boundary between hillslope and fluvial processes, though this boundary is temporally and spatially variable and the active processes are subject to episodic interruptions. The nature of material evacuation and the limited role of fluvial processes indicate that the zero-order basins of this study belong primarily in the hillslope domain, though somewhat less so in the case of the Japanese study sites. The hillslopes features in British Columbia generally do not share the 211  same level of connection to fluvial processes as many of the study sites in Japan or those reported in the literature from other unglaciated terrains (e.g., Dietrich et al., 1987). The zero-order basins in Japan typically displayed more mature morphologic form in terms of larger, more obviously convergent, contributing areas where landslides were most certainly a repeat process of material transfer. Further, the more continuous nature of material transport is more similar to a fluvial system than the more episodic movements characterizing the material transport in British Columbia. Not all the failures in the study area in Japan occurred in the axis of the zero-order basin, some occurred on the side slopes or in secondary axes. From the perspective of competitive valley incision by preferential erosion, landslides occur not only at the point of maximum erosion but also at other competing, though less competitive, hillslope positions. The difference between the probability of the focal location of local deposition to receive hillslope materials from the contributing area and secondary, or tertiary candidates, should produce a divergence in infilling rates. In British Columbia, few sites were comparable to the well-developed zeroorder basins typical of the Japanese study area. However, not only was mature topographic form developed at some sites in British Columbia, such as those with strongly jointed or rapidly weathering bedrock (Hasegawa et al., 2009); some sites in Japan were of low maturity and the landslide appeared to be the seminal, or near-seminal, failure at the site. Individual hillslope characteristics determine the effectiveness and direction of hillslope evolution. Where the basal material is erodible, failures may incrementally lower the failure plane and result in either increased slope gradients into the failure area or an expanded contributing area. A gully feature may thus be created and material accumulating in the cleft will be predominantly derived from the underlying material (bedrock) as the surficial material will have been removed over progressive failures. On the other extreme, a weathering-resistant failure plane will have negligible change in elevation even over repeated failure cycles; the size of the contributing area will thus be static and controlled by the topography of the substrate. To some degree this difference was captured in the soil accumulation modeling of Chapter 3 212  with the inclusion of the average soil depth of the adjacent hillslope and a measure of the competence of the geomorphic rock mass. Soil depths have been found to vary by topographic position (Heimsath et al., 1999) and the presence of strong differences in soil depths between the failure area and the remainder of the initiation zone was found here to be associated with greater occurrences of material transport (Chapter 2). Therefore, characterizing the sediment delivery regime to failure areas benefits from the inclusion of the mean depth of hillslope soils to estimate the amount of available material (Chapter 3). The characterization of the rock mass provides an indication of the degree of material contribution though topographic gradients are necessary to move the material into the failure area. Some indication of such association was also found in Chapter 3 with the performance of the model combining the geological variable with the descriptor of slope gradient. Downslope from the study sites in Japan, the continued accumulation of hillslope water leads to a fluvially dominated system or a subsurface flow system buried by channel recharge (Jakob et al., 2005). In contrast, many failure paths at the study sites in British Columbia displayed a lower degree of organization and the initial debris slide failure may not have progressed to the valley bottom. Instead, the transported material would redeposit some distance downslope as an aggregation of organic and inorganic material. Deposition would sometimes result from the interaction of the moderate size of the events (~50 m3) with logs or trees that may or may not have been included in the transported debris. Observation of such debris piles was found repeatedly on slopes >35º. The site of the deposition is then loaded with tenuously stable material and primed for a further event of this sediment cascade; thus, the processes of hillslopes in British Columbia area again more stochastic and less typical, than in Japan. Slope failures of importance to the safety of forest workers and of relevance to regenerating forests in British Columbia may operate at a smaller spatial scale and as a result require more intensive slope stability assessments prior to forest harvesting and the associated silvicultural obligation to promote regeneration.  213  The implications and magnitude of landscape maturity, at the scale of a zero-order basin or contributing area, and its effect over repeated cycles of hillslope adjustment is one of the central problems encountered in this research. Landscape maturity at the scale of the contributing area here defined as the proximity of the site to a stochastic steady state. Presumably, the Japanese study sites should display greater maturity due to the lack of Late Pleistocene glaciation and thus represent a landscape closer to steady state and a nearly consistent term of hillslope storage in the sediment budget. With the lack of glacial disturbance, slopes would also tend towards shapes characteristic of their formative processes (Kirkby, 1971), though spatial and temporal variation from characteristic slopes will occur. Not all study sites in British Columbia exhibited storage of excess hillslope surficial material. Some sites may have been glacially scoured and thus had little initial hillslope material; other sites appear to have evacuated their hillslope stores through repeated shallow landslides. The future direction of these initiation zones is dependent upon geologic and forest characteristics. The infilled material in some sites in the North Coast study region was almost completely organic matter. Thus, over a sub-geologic time scale (~104 years), the site may have already attained a steady state with episodic evacuation of predominantly organic material; a further analysis of return periods in these organic-matter-dominated systems would be useful as a scientific landmark and for operational guidance of timber harvesting.  AMALGAMATION The combination of the findings in Chapters 2 through 4 led to a model of hillslope evolution in shallow landslide initiation zones (Chapter 5). Examining initiation zones and characterizing them from the standpoint not only of accumulated soil depth but also recent material transport, topography, ground surface covers, and preferential flow pathways provided an understanding of relevant process controls and the variability of individual hillslope segments. Precipitation events of a magnitude sufficient to cause a landslide occur with higher frequency compared to the rate of material accumulation (Jakob et al., 214  2006; Chapter 3); therefore this discontinuity of the subsurface network of preferential flow pathways is thought to have the potential to induce further slope instability, particularly after the site has been largely infilled. The inclusion of several initiation zones with evidence of more than one failure indicates that repeat failures are possible, though not necessarily restricted to infilled material. Some of the slopes in British Columbia have excess storage of surficial material as a result of Late Pleistocene glaciation. Such material beyond the head and side scarps has an increased likelihood of subsequent failure due to the loss of toe slope support. The probability of these failures resulting in debris flows would be partially dependent on their magnitude and the extent to which the recharged, and often fully saturated, volume in the axis of the failure could be entrained. Small failures from the hillslopes would be incorporated as recharge; large hillslope failures may travel through the previous failure area regardless of the amount of recharged material. Various mechanisms of shallow landslide and debris flow initiation are possible depending upon the characteristics and conditions of the initiation area, but an increase in the volume of material in the axis of the previous failure will act to increase the likelihood of debris flow initiation (Millard, 1999).  LAND USE Further complication in the study of landslide initiation zones arises due to temporally varying conditions. Timber harvesting and road building may remove much of the forest influence but not all. Following harvesting at hillslopes in the study areas in British Columbia, a great deal of woody debris and other features of the complex forest remain. The decline of root cohesion following forest harvesting is known to have an effect on the initiation of shallow landslides (Marden and Rowan, 1993; Ekanayake and Philips, 1999), but it may also affect the infilling rates. Although, in this study, volumes of material transport were not statistically differentiated between timber harvested and natural slopes, the number of surveyed points with recent material transport was higher for harvested areas. Considering that the results 215  of Chapter 3 indicate a period of ~150 years for the near complete infilling of failure areas, and timber harvesting rotations in British Columbia may be as short as ~60 years in areas of high productivity, timber harvesting could occur repeatedly during a single infilling cycle. Soil disturbance from felling and yarding operations could rejuvenate geomorphic adjustments resulting in more sediment and material deposition in the failure area and a shortened „immunity period‟ (Shimokawa, 1984) during which the soil depth is insufficient for failure. The effects of land use on the forest and hydrogeomorphic processes of initiation zones were hypothesized, with the conclusion that the avoidance of landslide activity be a high priority. Caution is advised for the management of shallow landslide prone terrain when considering the forest influence over hydrogeomorphic regimes, the rate, controls, and processes of soil accumulation, and the properties of preferential flow pathways in initiation zones. Considering the proposed cycle of shallow landsliding, this is particularly important at sites that have not failed since Late Pleistocene deglaciation as these sites are not naturally active and these sites will generally have the most sediment available for the repeated recharge of the failure area. Although still a hypothesis, this amalgamation of findings may indicate that the repercussions of land management in the last century may still be occurring. Careful observation of timber harvesting of both old-growth and second-growth stands in steep terrain could address hydrogeomorphic (Chapters 2, 3, and 4) and forest factors (i.e., Sidle, 1991) that span across multiple rotations of industrial forestry.  STUDY DESIGN While the data from British Columbia form the majority of the study observations, the inclusion of the Japanese sites provides contrast. While both study regions have steep, forested, thinly mantled slopes and both produce shallow landslides, the landscape in British Columbia experienced Late Pleistocene glaciation and the area of study in Japan did not. Deeper weathering and more mature slope forms 216  characterizes the study area in Japan; in British Columbia, the landscape is still largely paraglacial (Church and Ryder, 1972). As discussed in the Introduction, the climate variables for the sites from coastal British Columbia and the study sites in Japan are not drastically different, though the Japanese sites may receive higher intensity and greater total amounts of precipitation. In both study regions, orographic effects may increase the intensity of precipitation in certain locations and the information regarding precipitation over shorter time scales (and spatial scales in Japan) remains unknown. Possibly the greatest contrast between the two regions is the difference of the forest influence. All of the Japanese study sites were located in plantation forests while the forests at the study sites in British Columbia were either old-growth or young second-growth (<40 years old). The intensively managed forests at the Japanese study sites have much smaller trees and thus are expected to have much smaller root systems (Schmidt et al., 2001; Sakals and Sidle, 2004), less coarse woody debris, and less complex structure (Puettman et al., 2009). Unfortunately, no failures in natural forests were sampled in Japan, nor were any failures sampled in the most intensively managed forests in British Columbia. Despite the paraglacial conditions in British Columbia, and likely as a result of the strong influence of the forests, the variability within study areas appears to mask the differences between study areas. Two requirements of the field sampling were that the failure plane was on bedrock and the hillslope material was dominantly colluvium. The requirement of the failure plane to be on bedrock was to ease the measurement of infilled soil depths as the full depth of the soil column above bedrock could be measured remotely with relative confidence. Colluvium is expected to be more closely associated with local bedrock than other materials such as till. Remnants of till were sometimes found remaining at the bottom of the soil profile, these deposits were of a discontinuous nature and not thought to influence the processes of the site in a substantial manner. As a dominant hillslope material, till was excluded to minimize the variability of the study sites associated with differences in density and the inclusion of materials different than the local bedrock. However, morainal deposits blanket many slopes in British 217  Columbia and many shallow landslides occur within them. Inclusion of these slopes would have allowed field sampling to occur with much less restriction, allowing less onerous field sampling and access to stronger gradients across precipitation, forest type and productivity, and topography. The properties of failure planes of various materials could be assessed quantitatively using the knocking cone penetrometer and thus the gradient of basal material entrainment could be extended from competent bedrock through to morainal deposits of various conditions and penetration resistances. The relevance of parent material to soil accretion and changes in failure area form could then be made more directly.  FINAL COMMENTS While an elegant solution of soil depth with time since failure continues to be elusive, all of the previous chapters culminate into a hypothesis that is presented in Chapter 6 – the effects of a single landslide predispose the site to future instability. While in many land-use associated failures, the cause may be removed following some remediation, the conditions evoked as a result of the failure may persist. As a conclusion from this study, it may be expected that this could lead to a cycle of hillslope instability. This would not only have implications for the growing site of the failure area, but also for the transport and deposition zones. As settlements proceed into mountainous areas, this process could result in a second wave of elevated hazard despite that trees may have regrown from the previous timber harvesting. The hydrogeomorphic effects of past landslides will persist and given the rotation lengths of industrial forestry in British Columbia and the approximate rates of material infilling as found in the third chapter of this thesis, the landscape may be becoming progressively more unstable and at risk of subsequent failure given past and future forest activities. While this can be considered the result of a poor understanding of forest, geomorphic, hydrologic processes, the evaluation of current forest practices is ongoing. Throughout this study there has been on emphasis on the forest control over geomorphic and hydrologic processes. The overlap of the two realms receives little attention, particularly in North America where the 218  value of forests in reducing hydrogeomorphic hazards is not as well recognized as in other parts of the world. Typically, where protection is required, designs call for the complete removal of timber, even in cases where it appears as though the forest strongly mitigated the hazard. The management of a forest stand to maintain its protection role may be an economic, silvicultural and ecological challenge, but recognizing, researching, and advocating the benefit of the forest influence on hydrogeomorphic processes is a task for all hydrogeomorphologists working in forested areas.  219  REFERENCES Banner A, MacKenzie W, Haeussler S, Thomson S, Pojar J, Trowbridge R. 1993. A field guide to site identification and interpretation for the Prince Rupert Forest Region. BC Ministry of Forests: Victoria BC. Land Management Handbook 26.  Brayshaw D, Hassan MA. 2009. Debris flow initiation and sediment recharge in gullies. Geomorphology 109: 122–131. DOI:10.1016/j.geomorph.2009.02.021  Church M, Ryder JM. 1972. Paraglacial sedimentation: A consideration of fluvial processes conditioned by glaciation. Bulletin of the Geological Society of America 83: 3059–3072.  Crozier MJ, Vaughan EE, Tippett MJ. 1990. Relative instability of colluvium-filled bedrock depressions. Earth Surface Processes and Landforms 15: 329–339.  DeLong SC, Sutherland GD, Daniels LD, Heemskerk BH, Storaunet KO. 2008. Temporal dynamics of snags and development of snag habitats in wet spruce–fir stands in east-central British Columbia. Forest Ecology and Management 255: 3613–3620.  Dietrich WE, Dunne T, Humphrey NF, Reid LM. 1982. Construction of sediment budgets for drainage basins. In Proceedings of a workshop on sediment routing in forested catchments, Swanson F, Janda R, Dunne T (eds). US Forest Service General Technical Report PNW-141. Dietrich WE, Reneau SL, Wilson CJ. 1987. Overview: “zero-order basins” and problems of drainage density, sediment transport and hillslope morphology. In Erosion and sedimentation in the Pacific Rim, Beschta RL, Blinn T, Grant GE, Swanson FJ, Ice GG (eds). IAHS Pub. 165: 39–48.  220  Ekanayake, J.C., and Phillips, C.J. 1999. A method for stability analysis of vegetated hillslopes: an energy approach. Canadian Geotechnical Journal 36: 1172–1184.  Hasegawa S, Dahal RK, Yamanaka M, Bhandary NP, Yatabe R, Inagaki H. 2009. Causes of large-scale landslides in the Lesser Himalaya of central Nepal Environmental Geology 57: 1423–1434. DOI: 10.1007/s00254-008-1420-z  Heimsath AM, Dietrich WE, Nishiizumi K, Finkel RC. 2001. Stochastic processes of soil production and transport: Erosion rates, topographic variation, and cosmogenic nuclides in the Oregon Coast Range, Earth Surface Processes and Landforms 26: 531–552.  Hutchinson DG, Moore RD. 2000. Throughflow variability on a forested hillslope underlain by compacted glacial till. Hydrological Processes 14: 1751–1766.  Jakob M, Bovis M, Oden M. 2005. The significance of channel recharge rates for estimating debris-flow magnitude and frequency. Earth Surface Processes and Landforms 30: 755–766.  Jakob M, Holm K, Lange O, Schwab JW. 2006. Hydrometeorological thresholds for landslide initiation and forest operation shutdowns on the north coast of British Columbia. Landslides 3: 228–238.  Karanka EJ. 1986. Trends and fluctuatlons in precipitation and stream runoff in the Queen Charlotte Islands. British Columbia Ministry of Forests, Research Program: Victoria British Columbia; Land Management Report 040.  Kirkby MJ. 1971. Hillslope process-response models based on the continuity equation, Institute of the British Geographical Society, Special Publication 3: 15–30.  221  Kosugi K, Uchida T, Mizuyama T. 2004 Numerical calculation of soil pipe flow and its effect on water dynamics in a slope. Hydrological Processes 18: 777–789. DOI: 10.1002/hyp.1367  Marden, M., and Rowan, D. 1993. Protective value of vegetation on Tertiary terrain before and during Cyclone Bola, East Coast, North Island, New Zealand. New Zealand Journal of Forest Science 23: 255–263.  McDonnell JJ. 1990. The influence of macropores on debris flow initiation. Quarterly Journal of Engineering Geology 23: 325–331.  Millard T. 1999. Debris flow initiation in Coastal British Columbia Gullies. British Columbia Ministry of Forests: Nanaimo British Columbia; Technical Report.  Nieber JL, Sidle RC. 2010. How do disconnected macropores in hillslopes facilitate preferential flow? Hydrological Processes (accepted).  Okunishi K, Iida T. 1981. Evolution of hillslopes including landslides. Transactions of the Japanese Geomorphological Union 2: 291300. O’Loughlin EM. 1981. Saturation regions in catchments and their relations to soil and topographic properties. Journal of Hydrology 53: 229–246.  Onda Y, Mori A, Shindo S. 1992. The effects of topographic convergence and location of past landslides on subsurface water movement on granitic hillslope. Journal of Natural Distaster Science 14: 4558.  Puettmann KJ, Coates KD, Messier C. 2009. A critique of siliviculture: Managing for complexity. Island Press, Washington DC.  222  Sakals ME, Sidle RC. 2004. A spatial and temporal model of root cohesion in forest soils. Canadian Journal of Forest Research 34: 950–958. doi: 10.1139/X03-268  Schmidt KM, Roering JJ, Stock JD, Dietrich WE, Montgomery DR, Schaub T. 2001. The variability of root cohesion as an influence on shallow landslide susceptibility in the Oregon Coast Range. Canadian Geotechnical Journal 38: 995–1024. DOI: 10.1139/cgj-38-5-995  Shimokawa E. 1984. Natural recovery process of vegetation on landslide scars and landslide periodicity in forested drainage basins. Proceedings of the Symposium on effect on forest land use on erosion and slope stablility East-West Center, Honolulu Hawaii; 99–108.  Shimokawa E, Jitouzono T, Takano S. 1989. Periodicity ofshallow landslide on Shirasu (Ito pyroclastic flow deposits) steep slopes and prediction of potential landslide sites. Transactions of the Japanese Geomorphological Union 10: 267–284.  Sidle RC. 1991. A conceptual model of changes in root cohesion in response to vegetation management. Journal of Environmental Quality 20: 43–52.  Sidle RC, Ochiai H. 2006. Landslides: processes, prediction, and land use. American Geophysical Union, Water Resoures Monograph 18: 350 pp.  Smale MC, McLeod M, Smale PN. 1997. Vegetation and soil recovery on shallow landslide scars in Tertiary Hill Country, East Cape Region, New Zealand. New Zealand Journal of Ecology 21: 3141.  Strahler, A.N. 1957. Quantitative analysis of watershed geomorphology. Transactions of the American Geophysical Union 38(6): 913-920.  223  Tsuboyama Y, Sidle RC, Noguchi S, Hosoda I. 1994. Flow and solute transport through the soil matrix and macropores of a hillslope segment, Water Resources Research 30: 879–890.  Tsukamoto Y, Ohta T, Noguchi H. 1982. Hydrological and geomorphological study of debris slides on forested hillslope in Japan. IAHS Publication 137: 89–98.  Uchida T, Kosugi K, Mizuyama T. 2001. Effects of pipeflow on hydrological process and its relation to landslide: A review of pipeflow studies in forested headwater catchments. Hydrological Processes 15: 2151–2174.  224  APPENDIX A: MAPS OF THE SURVEYED STUDY SITES This appendix includes topographic maps of all the study sites with site surveys (Figures 1–29). All maps are oriented with North to the top of the page. Maps have 1 m contour lines plotted on them for the survey area. All maps have the approximate bounds of the most recent failure indicated in red. Where two separate failure surfaces were identified, the bounds of the older failure are indicated in blue. Positions surveyed are indicated with the assigned morphologic position a key is located in Table A.1.  225  Table A.1. Survey positions and description. Position code  Description  A  Axis of the failure  AAHS  In the axis of the failure, above the head scarp  AHS  Above the head scarp  BHS  Bottom of head scarp  BLS  Bottom of left side of failure  BRS  Bottom of right side of failure  HS  Head scarp  LH  Left hillslope  LS  Left side of failure  RH  Right hillslope  RS  Right side of failure  S  Failure area (scar)  THS  Top of head scarp  TLS  Top of left side of failure  TRS  Top of right side of failure  226  Figure A.1. BigHill1 study site.  227  Figure A.2. BigHill2 study site.  228  Figure A.3. CP144-6 study site.  229  Figure A.4. Crest1 study site.  230  Figure A.5. Firewood study site.  231  Figure A.6. Gregory1 study site.  232  Figure A.7. Gregory2 study site.  233  Figure A.8. Gregory3 study site.  234  Figure A.9. Kaien57 study site.  235  Figure A.10. Lachmach1 study site.  236  Figure A.11. Lachmach2 study site.  237  Figure A.12. PwrLine study site.  238  Figure A.13. RedPine study site.  239  Figure A.14. RenFace1 study site.  240  Figure A.15. RenFace2 study site.  241  Figure A.16. Rennell1 study site.  242  Figure A.17. Shelley1 study site.  243  Figure A.18. Shelley2 study site.  244  Figure A.19. Shelley3 study site.  245  Figure A.20. Silver1 study site.  246  Figure A.21. skeenaView study site.  247  Figure A.22. Skidegate1 study site.  248  Figure A.23. Skidegate2 study site.  249  Figure A.24.SwimPool study site.  250  Figure A.25. Waka12-2 study site.  251  Figure A.26. Waka17 study site.  252  Figure A.27. Whitebtm1 study site.  253  Figure A.28. Whitebtm2 study site.  254  Figure A.29. Yakoun1study site.  255  APPENDIX B: PHOTOGRAPHS In this appendix, select photographs of the study sites and other relevant photos are presented. The first section contains photos of processes and characteristic soil profiles (Figures 1–24); the second section contains selected photographs of some of the study sites (Figures 25-65).  256  CHARACTERISTIC PHOTOGRAPHS  Figure B.1. A photograph of a candidate study site showing typical size and degree of hillslope convergence. This site was not selected for study because the confining layer was till.  257  Figure B.2. A hillslope in the North Coast study area that included Lachmach1 and Lachmach2 study sites.  258  Figure B.3.Debris from the PwrLineA failure deposited in a headwater stream. 259  Figure B.4. Two adjacent soil profiles at the margin of a candidate landslide showing differences in soil development.  260  Figure B.5.An example of a small landslide transporting hillslope material.  261  Figure B.6.Recent material transport processes during wet weather on the North Coast of BC. 262  Figure B.7.A material transport event that moved ~8 m3 of material down into the failure area of a shallow landslide.  263  Figure B.8. Active landsliding and slope wash occurring on this 10 year old side scarp. 264  Figure B.9. Dry ravel occurring at this 10 year old side scarp. 265  Figure B.10. A wildlife trail across this headscarp caused soil disturbance. 266  Figure B.11. Incorporation of woody material into the infilled soils at Shelley2.  267  Figure B.12. Coarse woody debris capturing material at the Kaien57A study site.  268  Figure B.13. Large woody debris on hillslopes contain material and, in this case, water. 269  Figure B.14. A windthrown root system that has lifted bedrock. Note the structure of the bedrock has remained intact. 270  Figure B.15. Soil pedestals, an indication of surface wash.  271  Figure B.16. A slope wash deposition area (downslope from Figure B.15).  272  Figure B.16. Needle ice lifted soil particles.  273  Figure B.17. Infilled soil over bedrock at the Kaien57A study site.  274  Figure B.18. A soil profile from the Home2 study site showing buried organic material.  275  Figure B.19. Soil profile with thin colluvium over weak rock. Note geological structure continuing into brown soil near shovel.  276  Figure B.20. An infilled soil of predominantly organic material at the Lachmach 2 study site.  277  Figure B.21. A soil comprised of coarse clasts and organic matter.  278  PHOTOGRAPHS OF STUDY SITES  Figure B.22. The Lachmach1 study site from below.  279  Figure B.23. The Lachmach1 study site.  280  Figure B.24. The Silver1 study site.  281  Figure B.25. The Lachmach2 study site.  282  Figure B.26. The Silver1 study site showing uneven bedrock topography.  283  Figure B.27. The PwrLine study site.  284  Figure B.28. The Kaien57A study site.  285  Figure B.29. The Kaien57A study site with the Kaien57B study site in the background.  286  Figure B.30. The RenFace2 study site showing the accumulation of dry ravel material.  287  Figure B.31. Gregory1 study site showing lack of regeneration in the failure area.  288  Figure B.32. The Rennell1 study site.  289  Figure B.33. The Yakoun1 study site. This site failed during or prior to logging, hence the abundant fine and coarse woody debris.  290  Figure B.34. The Skidegate1 study site.  291  Figure B.35. The BigHill1 study site.  292  Figure B.36. The BigHill1 study site.  293  Figure B.37 The lower portion of the BigHill2 study site.  294  Figure B.38. The Gregory2 study site.  295  Figure B.39 Gregory3 study site showing little recharge in the upper portion of the failure but some material accumulation in the foreground. 296  Figure B.40. RenFace1 study site.  297  Figure B.41. CP144-6 study site. At 180 years since failure, the evidence of this failure is being obscured by forest and material transport processes.  298  Figure B.42. CP144-6 study site as seen from above the head scarp.  299  Figure B.43. Ridley3 study site showing deposition of fine woody debris into the failure area.  300  Figure B.44. 209 study site.  301  Figure B.45. The Hidden study site.  302  Figure B.46. The Max study site. The landslide occurred 100 years ago and the margins of the failure area have stabilized.  303  Figure B.47. Pork’n’cheese study site. 304  Figure B.48. The Home1 study site that had been grass seeded. Grass seeding is used to decrease surface erosion of exposed soils.  305  Figure B.49. Home2 study site showing active material transport processes and the incorporation of coarse woody debris. 306  Figure B.50. The Home4 study site.  307  Figure B.51. The skeenaView study site.  308  Figure B.52. The Shelley1 study site.  309  Figure B.53. The Shelley2 study site with a small recent failure and the incorporation of coarse woody debris.  310  Figure B.54. The Crest1 study site.  311  Figure B.55. The Minerva2 study site showing scoured bedrock source area and deposition on a forest road below. 312  Figure B.56. The head scarp of the Minerva1 study site.  313  Figure B.57. The Lachmach80 study site showing strong regeneration of red alder.  314  Figure B.58. The Lachmach79 study site. 315  Figure B.59. The Waka12-2 study site showing smooth bedrock with a thin veneer of soil.  316  Figure B.60. The SwimPool study site.  317  Figure B.61. The RedPine study site.  318  Figure B.62. The Firewood study site.  319  APPENDIX C: COMPARISON OF FOUR STUDY AREAS This appendix presents a comparison of the four study areas; comparisons made of regional climates are included in each chapter. Here, several basic descriptors of the study sites are compared. No regional patterns are identified in the four plots.  320  Figure C.1. Comparison of geomorphic rock mass classification (geology) and time since failure.  321  Figure C.2. Comparison of lengths and widths of failure areas.  322  Figure C.3. Comparison of slope gradient of failure areas and values of convergence for study sites.  323  Figure C.4. Comparison of soil depths adjacent to failure areas and the bearing of failure axes.  324  APPENDIX D: DATA TABLES In this section the observation and analysis data for site level observations are presented. Point level data have been abridged to present data for two study sites. Explanations for column names are provided in Tables D.1 and D.2. Data are presented in Tables D.3 and D.4. Table D.1. Column names, units of measure, and description for site level data.  Name  Units  name  Description Site name  time  yrs  Time (years) since previous landslide  timeError  yrs  Approximate error of time since previous landslide  region  Study region. One of: North Coast, South Coast, Haida Gwaii (Queen Charlotte Islands), or Japan.  XsdMean  m  Mean soil depth in the failure area.  HsdMean  m  Mean soil depth on the hillslope surrounding the failure area.  HsdMeanPred  "X" indicates that the HsdMean value was predicted based on a linear regression.  Calibrate/Test  Indicates whether site was included in model calibration or model testing. All Japanese sites were used for model testing; BC sites were randomly selected for model calibration or testing.  Lat  decimal deg.  Latitude of study site.  Long  decimal deg.  Longitude of study site.  alt  m  Elevation of study site.  IRS  Intact Rock Strength: a component of both rock mass descriptive variables (Selby, 1980).  W  Weathering: a component of both rock mass descriptive variables (Selby, 1980).  JS  Joint Spacing: a component of both rock mass descriptive variables (Selby, 1980).  JOMod  Joint Orientation for the rock mass classification system of Selby (1980)  JO  Joint Orientation of the modified rock mass classification (Selby, 1980).  JW  Joint Width: a component of both rock mass descriptive variables (Selby, 1980).  JC  Joint Continuity: a component of both rock mass variables (Selby, 1980).  325  Table D.1.(Continued)  Name  Units  Description  geology  Geomorphic Rock Mass Classification system of Selby (1980).  geologyMod  Modified geomorphic rock mass classification system.  scarLength  m  Average length of failure area  scarWidth  m  Average width of failure area  axisBear  Bearing of the axis of the failure area.  convergence  º  RWT - LWT with corrections for discontinuity at 0/360º  axisSlope  º  Slope of the axis of the failure area  hsheight  m  Average height of head scarp where a defined head scarp existed  siteIndex  m  Site Index. Basic site index for the biogeoclimatic ecosystem variant, corrected for the percent of each major tree species  logged  -  Dummy variable indicating logged (1) or unlogged (0) condition.  failureType  -  Type of mass movement initiated by failure  PPT911  mm  Average precipitation at the site from September through November as predicted by ClimateBC (Hamman and Wang, 2005)  FWPPT  mm  Average precipitation at the site during fall and winter (September through February) as predicted by ClimateBC  allProcVol LF LS NI DR SP  m3  Sum of all recent material transport events at the site  3  Sum of all recent litterfall measured at the site.  3  Sum of all recent landslide deposits measured at the site.  3  Sum of all recent needle ice lifted material measured at the site.  3  Sum of all recent dry ravel measured at the site.  3  Sum of all recent soil pedestal volumes (volume of removed soil material)  m m m m m  measured at the site. SW W.1 Xs9croscSD  m3  Sum of all recent slope wash volumes measured at the site.  3  Sum of all recent windthrow volumes measured at the site.  -1  Standard deviation of surface cross-slope convexity within failure areas analyzed  m m  in a 4.5 m analysis window Xs9longcSD  -1  m  Standard deviation of surface longitudinal convexity within failure areas analyzed in a 4.5 m analysis window  scarPts  Number of survey points in the failure area  326  Table D.1.(Continued)  Name  Units  Description  count_cccc  Number of survey points concave in both cross-slope and longitudinal directions  active_cccc  Number of survey points concave in both cross-slope and longitudinal directions that had recent material transport  vol_cccc  m3  Volume of recent material transport at survey points concave in both cross-slope and longitudinal directions  count_cccv  Number of survey points concave in the cross-slope direction and convex in the longitudinal direction  active_cccv  Number of survey points concave in the cross-slope direction and convex in the longitudinal direction that had recent material transport  vol_cccv  m3  Volume of recent material transport at survey points concave in the cross-slope direction and convex in the longitudinal direction  count_cvcc  Number of survey points convex in the cross-slope direction and concave in the longitudinal direction  active_cvcc  Number of survey points convex in the cross-slope direction and concave in the longitudinal direction that had recent material transport  vol_cvcc  3  m  Volume of recent material transport at survey points convex in the cross-slope direction and concave in the longitudinal direction  count_cvcv  Number of survey points convex in both cross-slope and longitudinal directions  active_cvcv  Number of survey points convex in both cross-slope and longitudinal directions that had recent material transport  vol_cvcv  3  m  Volume of recent material transport at survey points convex in both cross-slope and longitudinal directions  327  Table D.2. Column names, units of measure, and description for point level data.  Name  Units  Description  name  Name of site  position  Description of the position of the point relative to the failure area (see Table A.1.1)  scar  “X” denotes that the point was within the failed area  x  m  The horizontal distance from the initial theodolite position in the East-West direction. Initial position of theodolite was set at 100 m, points East are >100 m, points West are <100 m.  y  m  The horizontal distance from the initial theodolite position in the North-South direction to the survey point. Initial position of theodolite was set at 100 m, points North are >100 m, points South are <100 m.  z  m  The vertical distance from the initial theodolite position to the survey point. Initial position of theodolite was set at 100 m elevation, points higher are >100 m, points lower are <100 m.  ssz  m  Elevation of the subsurface confining layer. Determined by subtracting the soil depth at the survey point from the determined z value.  sd  m  sitePoint  Soil depth at the survey point Survey point identifier  expRod1  cm  Length of steel rod exposed once driven down to the confining layer  expRod2  cm  Where multiple attempts were made to determine the depth to the confining layer, the length of steel rod exposed once driven down to the confining layer  KCP10  Number of strikes to drive penetrometer 10 cm into the soil  KCP20  Number of strikes to drive penetrometer 20 cm into the soil  KCP30  Number of strikes to drive penetrometer 30 cm into the soil  KCP40  Number of strikes to drive penetrometer 40 cm into the soil  KCP50  Number of strikes to drive penetrometer 50 cm into the soil  KCP60  Number of strikes to drive penetrometer 60 cm into the soil  KCP70  Number of strikes to drive penetrometer 70 cm into the soil  KCP80  Number of strikes to drive penetrometer 80 cm into the soil  KCP90  Number of strikes to drive penetrometer 90 cm into the soil  KCP100  Number of strikes to drive penetrometer 100 cm into the soil  KCP110  Number of strikes to drive penetrometer 110 cm into the soil  KCP120  Number of strikes to drive penetrometer 120 cm into the soil  328  Table D.2. (Continued)  Name  Units  Description  KCP130  Number of strikes to drive penetrometer 130 cm into the soil  KCP140  Number of strikes to drive penetrometer 140 cm into the soil  KCP150  Number of strikes to drive penetrometer 150 cm into the soil  KCPstrikes  Total number of strikes  KCPmax  cm  Final penetration  KCPnotes  Any notes regarding the penetrometer profile  procType  The most dominant process of recent material transport  procL  mm  Length of deposit of recent material transport measured to the nearest 50 mm  procW  mm  Width of deposit of recent material transport measured to the nearest 50 mm  procD  mm  Depth of deposit of recent material transport measured to the nearest 50 mm  3  procVol  m  Volume of recent material transport (procL x procW x procD)  LWDL  m  Length of coarse woody debris (>10 cm in diameter) present in the failure area  LWDD  cm  Diameter of coarse woody debris present in the failure area  3  LWDVol  m  Volume of coarse woody debris present in the failure area  LWDOrient  º  Orientation of coarse woody debris, 90º is cross-slope, 45º is diagonal, 0º is up/down slope Dominant surface cover in the vicinity of the survey point, R – rocks, S – soil, VEG -  surfCover  vegetation, WD – woody debris aspect  º  Aspect of the cell containing the survey point. Raster cells are 0.5 m on each side and were created from inverse distance weighting of the elevations of the survey points.  slope s3longc  º  Slope of the raster cell containing the survey point -1  m  Longitudinal surface convexity of the raster cell when compared to its neighbours (1.5 m analysis window) in the direction of maximum slope  s3crosc  -1  m  Cross-slope surface convexity of the raster cell when compared to its neighbours (1.5 m analysis window) in the direction of maximum slope  s9longc  m-1  s9crosc  -1  Longitudinal surface convexity of the raster cell when compared to its neighbours (4.5 m analysis window) in the direction of maximum slope  m  Cross-slope surface convexity of the raster cell when compared to its neighbours (4.5 m analysis window) in the direction of maximum slope  ssaspect  º  Aspect of the confining layer in the cell containing the survey point. Cells are 0.5 m on each side and were created from inverse distance weighting of the elevations of the survey points.  329  Table D.2. (Continued)  Name  Units  Description  ssslope  º  Slope of the confining layer in the raster cell containing the survey point  ss3crosc  -1  m  Cross-slope subsurface convexity of the raster cell when compared to its neighbours (1.5 m analysis window) in the direction of maximum slope  ss9longc  m-1  ss9crosc  -1  Longitudinal subsurface convexity of the raster cell when compared to its neighbours (4.5 m analysis window) in the direction of maximum slope  m  Cross-slope subsurface convexity of the raster cell when compared to its neighbours (4.5 m analysis window) in the direction of maximum slope  330  Table D.3. Site level data. name 209 BigHill1 BigHill2 CP144-6 Crest1 Firewood Gregory1 Gregory2 Gregory3A Gregory3B Hidden Home1 Home2 Home4 JDay K800A K800B Kaien57A Kaien57B Lachmach1 Lachmach2 Lachmach79 Lachmach8 LogPull Max Minerva1 Minerva2 Minerva3 PorkNCheese PwrLineA PwrLineB RedBowl RedPine RenFace1 RenFace2 Rennell1 Return Ridley1 Ridley3 Shelley1 Shelley2 Shelley3 Silver1 skeenaView Skidegate1 Skidegate2  time 24 27 27 180 15 15 10 114 1 11 7 1 9 30 14 9 35 31 48 2 130 5 4 10 100 61 1 67 1 1 17 49 92 12 27 24 69 27 27 35 70 114 1 2 16 16  timeError 3 3 2 40 5 2 2 2 1 3 1 1 7 10 10 5 5 0.5 0.5 1 40 2 2 5 20 5 0 5 0.5 0.5 2 10 15 2 3 2 14 5 5 10 3 5 1 2 2 2  region SC HG HG HG JAPAN JAPAN HG HG HG HG SC SC SC SC SC SC SC NC NC NC NC NC NC SC SC NC NC NC SC NC NC SC JAPAN HG HG HG SC NC NC HG HG HG NC NC HG HG  XsdMean 0.36 0.50 0.51 0.62 0.16 0.19 0.21 0.47 0.22 0.23 0.15 0.31 0.66 0.54 0.32 0.28 0.31 0.42 0.48 0.12 0.45 0.02 0.02 0.11 1.00 0.24 0.00 0.31 0.06 0.39 0.26 0.49 0.86 0.19 0.14 0.40 0.26 0.12 0.20 0.63 0.51 0.41 0.33 0.18 0.36 0.36 331  HsdMean 0.64 0.60 0.82 0.80 0.88 0.46 0.75 0.51 0.72 0.72 0.90 0.77 0.75 0.77 0.67 0.73 0.73 0.75 0.75 0.36 0.65 0.57 0.57 0.19 0.90 0.40 0.37 0.42 0.53 0.57 0.57 0.67 0.82 0.33 0.48 0.73 1.06 0.51 0.66 0.71 0.65 0.75 0.55 0.71 0.93 0.77  HsdMeanPred  X X X X  X X  X  X  Calibrate/Test Calibrate Calibrate Test Calibrate Test Test Calibrate Test Calibrate Calibrate Calibrate Calibrate Test Calibrate Test Calibrate Calibrate Calibrate Calibrate Calibrate Calibrate Calibrate Calibrate Calibrate Calibrate Calibrate Test Calibrate Test Test Test Calibrate Test Calibrate Calibrate Calibrate Calibrate Calibrate Calibrate Test Calibrate Calibrate Calibrate Calibrate Calibrate Calibrate  Table D.3. (Continued) name Somerset1 Somerset2 Stihl1 Stihl2 SwimPool Waka12-2A Waka12-2B Waka17 Whitebtm1 Whitebtm2 Yakoun1 YFin  time 7 13 89 84 23 1 16 91 29 95 19 1  timeError 3 5 25 25 5 0.5 5 5 0.5 1 2 1  region SC SC SC SC JAPAN JAPAN JAPAN JAPAN NC NC HG SC  XsdMean 0.21 0.34 1.31 0.52 0.12 0.02 0.07 0.76 0.62 0.54 0.28 0.36  332  HsdMean 1.00 0.85 0.80 0.57 0.35 0.90 0.90 1.09 0.74 0.73 1.14 0.69  HsdMeanPred  Calibrate/Test Calibrate Calibrate Calibrate Test Test Test Test Test Calibrate Calibrate Calibrate Calibrate  Table D.3. (Continued) name 209 BigHill1 BigHill2 CP144-6 Crest1 Firewood Gregory1 Gregory2 Gregory3A Gregory3B Hidden Home1 Home2 Home4 JDay K800A K800B Kaien57A Kaien57B Lachmach1 Lachmach2 Lachmach79 Lachmach8 LogPull Max Minerva1 Minerva2 Minerva3 PorkNCheese PwrLineA PwrLineB RedBowl RedPine RenFace1 RenFace2 Rennell1 Return Ridley1 Ridley3 Shelley1 Shelley2 Shelley3 Silver1 skeenaView Skidegate1 Skidegate2  Lat NA 53.340155 53.338639 53.376067 34.096 34.096 53.392145 53.395091 53.395058 53.395058 NA NA NA NA NA NA NA 54.30152 54.30152 54.301646 54.300405 54.292339 54.292562 NA NA 54.286208 54.286208 54.286208 NA 54.294371 54.294371 NA 34.096 53.371899 53.372569 53.348847 NA NA NA 53.350187 53.350253 53.348738 54.408015 54.375737 53.330558 53.329553  Long NA -132.41416 -132.41328 -132.438499 135.73 135.73 -132.39843 -132.443934 -132.443934 -132.443934 NA NA NA NA NA NA NA -130.30293 -130.30293 -129.987878 -129.990479 -129.965865 -129.966967 NA NA -129.981468 -129.981468 -129.981468 NA -130.322283 -130.322283 NA 135.73 -132.487443 -132.483575 -132.43481 NA NA NA -132.369115 -132.37212 -132.378485 -130.201811 -128.86571 -132.139589 -132.140054  alt 400 297 176 337 NA 415 297 389 133 133 548 497 499 500 308 195 195 249 249 98 98 131 175 280 556 260 280 265 474 154 154 425 522 186 147 307 496 207 190 205 205 214 416 494 285 287  IRS 18 5 5 2 10 10 18 10 10 10 18 18 18 18 14 14 14 10 10 18 18 18 18 14 18 18 18 18 14 10 10 10 18 18 16 14 18 10 10 10 5 10 7 16 10 10 333  W 7 4 4 3 7 5 5 9 7 7 7 5 5 5 7 7 7 7 7 9 9 9 9 10 7 10 10 10 5 7 7 5 9 9 7 6 5 9 9 7 4 7 7 2 7 8  JS 15 8 8 8 15 15 8 21 12 12 10 8 10 8 15 8 8 15 15 21 21 24 24 15 10 28 28 28 10 15 15 15 21 15 15 8 10 12 12 15 8 15 18 10 15 15  JO NA 9 18 NA 18 20 14 9 5 5 20 5 5 5 9 9 9 18 18 9 9 5 5 5 NA 20 20 20 20 9 9 20 20 20 9 18 20 9 18 18 18 9 7 14 18 18  Table D.3. (Continued) name Somerset1 Somerset2 Stihl1 Stihl2 SwimPool Waka12-2A Waka12-2B Waka17 Whitebtm1 Whitebtm2 Yakoun1 YFin  Lat NA NA NA NA 34.096 34.096 34.096 34.096 54.371381 54.37156 53.365729 NA  Long NA NA NA NA 135.73 135.73 135.73 135.73 -128.85962 -128.859619 -132.30567 NA  alt 543 549 457 424 529 1136 1136 1070 584 579 388 NA  334  IRS 5 5 18 18 14 14 14 14 20 18 18 14  W 5 7 9 6 10 6 6 7 9 7 5 9  JS 8 8 10 10 21 10 10 11 25 18 8 10  JO 20 20 20 20 18 18 18 18 20 16 14 9  Table D.3. (Continued) name 209 BigHill1 BigHill2 CP144-6 Crest1 Firewood Gregory1 Gregory2 Gregory3A Gregory3B Hidden Home1 Home2 Home4 JDay K800A K800B Kaien57A Kaien57B Lachmach1 Lachmach2 Lachmach79 Lachmach8 LogPull Max Minerva1 Minerva2 Minerva3 PorkNCheese PwrLineA PwrLineB RedBowl RedPine RenFace1 RenFace2 Rennell1 Return Ridley1 Ridley3 Shelley1 Shelley2 Shelley3 Silver1 skeenaView Skidegate1 Skidegate2  JOMod NA 18 9 NA 9 5 14 18 20 20 5 20 20 20 18 18 18 9 9 18 18 20 20 20 NA 5 5 5 5 18 18 5 5 5 18 9 5 18 9 9 9 18 19 14 9 9  JW NA 6 5 NA 6 5 4 4 6 6 5 5 5 5 6 6 6 5 5 5 5 5 5 6 5 5 5 5 5 5 5 6 5 5 5 5 5 5 5 2 7 6 5 4 5 5  JC NA 3 4 NA 5 4 4 4 5 5 4 5 5 5 6 5 5 3 3 6 6 6 6 5 4 6 6 6 4 5 5 5 5 5 4 4 4 4 4 5 6 6 5 4 4 4  geology 0.67 0.37 0.47 0.22 0.65 0.63 0.56 0.61 0.48 0.48 0.68 0.49 0.51 0.49 0.61 0.52 0.52 0.62 0.62 0.72 0.72 0.71 0.71 0.59 0.59 0.93 0.93 0.93 0.62 0.54 0.54 0.65 0.83 0.77 0.60 0.59 0.66 0.52 0.62 0.61 0.51 0.56 0.52 0.53 0.63 0.64 335  geologyMod 0.67 0.47 0.37 0.22 0.55 0.47 0.56 0.66 0.64 0.64 0.52 0.65 0.67 0.65 0.70 0.62 0.62 0.52 0.52 0.68 0.68 0.87 0.87 0.74 0.59 0.74 0.74 0.74 0.46 0.64 0.64 0.49 0.65 0.61 0.69 0.49 0.50 0.62 0.52 0.51 0.41 0.66 0.60 0.53 0.53 0.54  scarLength 5.8 12.8 13 11.3 13.9 8 22 7.1 16.5 16.5 13 13 5.5 7 3.5 15 15 21 21 6 8.5 8 10 7.1 6 32 12 2.5 7.1 6 6 14 19.1 8.8 23.7 14 18.6 12 15 8.5 15.5 9.1 12.5 13 14 9.2  scarWidth 4.7 12.1 12.3 5.9 6.4 4.1 13.8 15.8 6.8 6.8 10.8 9 4 7 3.5 22 22 23.5 23.5 5 8.6 6 10 5.3 7 10 12 7 7.8 2.4 4.4 9 8.1 7.6 12.9 10.5 9.3 6.3 14 7.7 12.8 9.1 23 10.8 6.5 11.5  Table D.3. (Continued) name Somerset1 Somerset2 Stihl1 Stihl2 SwimPool Waka12-2A Waka12-2B Waka17 Whitebtm1 Whitebtm2 Yakoun1 YFin  JOMod 5 5 5 5 9 9 9 9 5 12 14 18  JW 5 5 5 5 5 5 5 5 4 2 5 5  JC 4 4 5 5 5 5 5 5 6 1 4 4  geology 0.50 0.52 0.71 0.68 0.78 0.62 0.62 0.64 0.89 0.66 0.57 0.54  336  geologyMod 0.34 0.36 0.55 0.52 0.68 0.52 0.52 0.54 0.70 0.58 0.57 0.64  scarLength 16.5 12 8 13 12.1 10 10 10 10.5 11.7 27 10  scarWidth 10.3 12.5 6 12 6.5 6 6 8 7.1 5.2 18 6.4  Table D.3. (Continued) name 209 BigHill1 BigHill2 CP144-6 Crest1 Firewood Gregory1 Gregory2 Gregory3A Gregory3B Hidden Home1 Home2 Home4 JDay K800A K800B Kaien57A Kaien57B Lachmach1 Lachmach2 Lachmach79 Lachmach8 LogPull Max Minerva1 Minerva2 Minerva3 PorkNCheese PwrLineA PwrLineB RedBowl RedPine RenFace1 RenFace2 Rennell1 Return Ridley1 Ridley3 Shelley1 Shelley2 Shelley3 Silver1 skeenaView Skidegate1 Skidegate2  axisBear 27 265 244 14 72 41 226 319 61 61 172 3 127 122 300 31 31 288 288 344 288 81 143 343 351 211 210 229 16 22 22 5 95 216 191 214 54 160 160 203 199 192 247 4 270 278  convergence 167 175 176 158 164 135 204 164 155 155 197 161 152 119 141 167 167 178 178 53 94 194 201 147 162 146 142 157 162 155 155 116 159 197 142 86 245 141 207 152 117 137 152 173 172 172  axisSlope 28 40 42 36 41 34 28 49 43 43 31 40 41 53 30 35 35 41 41 37 36 47 41 40 34 42 42 42 39 45 45 34 34 37 35 48 39 41 40 37 41 44 44 40 25 30  hsheight 35 NA NA NA NA NA NA NA 85 NA 70 134 172.5 45 40 95 95 NA NA 20 NA NA 30 20 NA NA NA NA 90 45 NA 50 NA NA NA NA 90 60 NA NA NA NA 25 40 NA NA  337  siteIndex 28.5 27 28.8 27.1 NA NA 27 27.2 29.2 29.2 24.5 27 27 27 26.4 25.2 25.2 28.5 28.5 28.6 29.2 22 22 27 27 15 15 15 25.7 29 29 27.5 NA 26.3 27.1 26 24 29 29 27 28.5 28.5 29.2 24.1 29.2 27.4  logged X X X X X X  X X  X  X  X X X X X X X  X X X X  failureType DS DS/DF DS DF DS/DF DS DS/DA DS/DF DS DS DS/DF DS/DF DF DF DS/DF DS/DF DS/DF DS DS DS/DF DF DS/DA DS/DA DS/DF DS/DF DS/DF DS/DF DS DS DF DF DS/DF DS/DF DS/DF DS/DF DS/DF DS/DF DS DS/DF DS DF DF DS/DF DS DS DS  Table D.3. (Continued) name Somerset1 Somerset2 Stihl1 Stihl2 SwimPool Waka12-2A Waka12-2B Waka17 Whitebtm1 Whitebtm2 Yakoun1 YFin  axisBear 36 39 74 100 48 152 152 250 49 38 295 53  convergence 152 120 96 142 124 142 142 173 170 191 161 162  axisSlope 36 42 49 46 24 40 40 45 35 41 29 52  hsheight 60 65 35 100 NA NA NA NA NA NA NA 65  338  siteIndex 25.2 25.2 27 24.8 NA NA NA NA 21 24.1 27.1 25.8  logged X X  X X X X  X  failureType DS DS DF DS/DF DS/DF DS/DF DS/DF DS/DF DF DF DS/DA DS  Table D.3. (Continued) name 209 BigHill1 BigHill2 CP144-6 Crest1 Firewood Gregory1 Gregory2 Gregory3A Gregory3B Hidden Home1 Home2 Home4 JDay K800A K800B Kaien57A Kaien57B Lachmach1 Lachmach2 Lachmach79 Lachmach8 LogPull Max Minerva1 Minerva2 Minerva3 PorkNCheese PwrLineA PwrLineB RedBowl RedPine RenFace1 RenFace2 Rennell1 Return Ridley1 Ridley3 Shelley1 Shelley2 Shelley3 Silver1 skeenaView Skidegate1 Skidegate2  PPT.9.11. 1150 1267 1270 1201 NA NA 1163 1171 1171 1171 1066 1041 1037 1035 1165 1056 1056 1049 1049 1142 1133 1168 1166 1205 1063 1137 1137 1137 1067 1066 1066 1169 NA 1114 1120 1175 1052 1023 1002 1267 1275 1289 1168 593 859 860  FWPPT 2702 2510 2516 2363 NA NA 2291 2298 2298 2298 2464 2428 2418 2413 2722 2466 2466 1955 1955 2141 2121 2194 2190 2812 2460 2129 2129 2129 2465 1995 1995 2755 NA 2191 2202 2318 2433 1910 1868 2492 2508 2537 2184 1193 1673 1676  scarPts 9 18 9 14 16 7 29 19 17 9 10 9 8 9 8 4 7 39 0 7 16 11 6 8 6 7 1 9 3 8 5 14 9 16 22 10 8 19 28 5 13 4 31 29 19 17  aPVNorm NA 0.006 0.003 0.004 0.000 0.001 0.002 0.000 0.031 0.000 NA NA NA NA NA NA NA 0.012 0.000 0.070 0.002 NA NA NA NA NA NA NA NA 0.014 0.002 NA 0.001 0.012 0.000 0.003 NA NA NA 0.000 0.003 0.000 0.008 0.000 0.001 0.000 339  allProcVol NA 1.000 0.480 0.238 0.031 0.029 0.470 0.030 3.497 0.000 NA NA NA NA NA NA NA 5.981 0.000 2.098 0.160 NA NA NA NA NA NA NA NA 0.200 0.065 NA 0.155 0.800 0.000 0.387 NA NA NA 0.000 0.692 0.000 2.410 0.067 0.095 0.000  LF NA 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 NA NA NA NA NA NA NA 0.010 0.000 0.000 0.000 NA NA NA NA NA NA NA NA 0.000 0.000 NA 0.000 0.000 0.000 0.001 NA NA NA 0.000 0.000 0.000 0.000 0.000 0.000 0.000  LS NA 0.200 0.000 0.018 0.000 0.000 0.338 0.000 1.663 0.000 NA NA NA NA NA NA NA 2.319 0.000 2.091 0.147 NA NA NA NA NA NA NA NA 0.118 0.063 NA 0.000 0.400 0.000 0.326 NA NA NA 0.000 0.112 0.000 2.106 0.066 0.001 0.000  NI NA 0.800 0.480 0.042 0.000 0.000 0.041 0.000 0.030 0.000 NA NA NA NA NA NA NA 0.000 0.000 0.000 0.000 NA NA NA NA NA NA NA NA 0.000 0.000 NA 0.000 0.000 0.000 0.000 NA NA NA 0.000 0.000 0.000 0.000 0.000 0.034 0.000  Table D.3. (Continued) name Somerset1 Somerset2 Stihl1 Stihl2 SwimPool Waka12-2A Waka12-2B Waka17 Whitebtm1 Whitebtm2 Yakoun1 YFin  PPT.9.11. 1067 1066 1036 1036 NA NA NA NA 633 633 910 1036  FWPPT 2465 2463 2415 2521 NA NA NA NA 1278 1278 1777 2415  scarPts 8 10 3 4 13 9 2 10 12 12 29 6  aPVNorm NA NA NA NA 0.000 0.003 0.000 0.000 0.000 0.000 0.003 NA  340  allProcVol NA NA NA NA 0.028 0.177 0.025 0.000 0.000 0.000 1.310 NA  LF NA NA NA NA 0.028 0.000 0.000 0.000 0.000 0.000 0.000 NA  LS NA NA NA NA 0.000 0.040 0.000 0.000 0.000 0.000 0.245 NA  NI NA NA NA NA 0.000 0.000 0.000 0.000 0.000 0.000 1.005 NA  Table D.3. (Continued) name 209 BigHill1 BigHill2 CP144-6 Crest1 Firewood Gregory1 Gregory2 Gregory3A Gregory3B Hidden Home1 Home2 Home4 JDay K800A K800B Kaien57A Kaien57B Lachmach1 Lachmach2 Lachmach79 Lachmach8 LogPull Max Minerva1 Minerva2 Minerva3 PorkNCheese PwrLineA PwrLineB RedBowl RedPine RenFace1 RenFace2 Rennell1 Return Ridley1 Ridley3 Shelley1 Shelley2 Shelley3 Silver1 skeenaView Skidegate1 Skidegate2  DR NA 0.000 0.000 0.000 0.000 0.000 0.005 0.000 0.000 0.000 NA NA NA NA NA NA NA 0.000 0.000 0.000 0.000 NA NA NA NA NA NA NA NA 0.000 0.000 NA 0.000 0.000 0.000 0.003 NA NA NA 0.000 0.000 0.000 0.000 0.000 0.000 0.000  SP NA 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.140 0.000 NA NA NA NA NA NA NA 0.005 0.000 0.000 0.000 NA NA NA NA NA NA NA NA 0.002 0.002 NA 0.000 0.000 0.000 0.039 NA NA NA 0.000 0.028 0.000 0.000 0.001 0.000 0.000  SW NA 0.000 0.000 0.160 0.029 0.029 0.086 0.030 1.664 0.000 NA NA NA NA NA NA NA 3.647 0.000 0.007 0.000 NA NA NA NA NA NA NA NA 0.080 0.000 NA 0.155 0.400 0.000 0.018 NA NA NA 0.000 0.552 0.000 0.304 0.000 0.060 0.000  W.1 NA 0.000 0.000 0.018 0.002 0.000 0.000 0.000 0.000 0.000 NA NA NA NA NA NA NA 0.000 0.000 0.000 0.013 NA NA NA NA NA NA NA NA 0.000 0.000 NA 0.000 0.000 0.000 0.000 NA NA NA 0.000 0.000 0.000 0.000 0.000 0.000 0.000 341  Xs9croscSD 0.000 0.010 0.009 0.011 0.001 0.156 0.012 0.029 0.001 0.001 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.030 0.081 0.106 0.051 NA NA 0.000 0.000 NA NA NA 0.000 0.236 0.000 0.000 0.034 0.047 0.001 0.138 0.000 0.000 0.000 0.100 0.038 0.046 0.037 0.065 0.004 0.042  Xs9longcSD 0.000 0.027 0.014 0.002 0.002 0.133 0.014 0.022 0.003 0.004 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.042 0.023 0.133 0.027 NA NA 0.000 0.000 NA NA NA 0.000 0.138 0.000 0.000 0.058 0.034 0.001 0.090 0.000 0.000 0.000 0.115 0.028 0.040 0.028 0.133 0.005 0.054  count_cccc NA 32 6 55 46 18 27 32 47 11 NA NA NA NA NA NA NA 42 1 22 30 NA NA NA NA NA NA NA NA 11 4 NA 17 20 64 19 NA NA NA 6 35 12 27 25 37 25  Table D.3. (Continued) name Somerset1 Somerset2 Stihl1 Stihl2 SwimPool Waka12-2A Waka12-2B Waka17 Whitebtm1 Whitebtm2 Yakoun1 YFin  DR NA NA NA NA 0.000 0.108 0.000 0.000 0.000 0.000 0.060 NA  SP NA NA NA NA 0.000 0.000 0.000 0.000 0.000 0.000 0.000 NA  SW NA NA NA NA 0.000 0.029 0.025 0.000 0.000 0.000 0.000 NA  W.1 NA NA NA NA 0.000 0.000 0.000 0.000 0.000 0.000 0.000 NA  342  Xs9croscSD Xs9longcSD count_cccc 0.000 0.000 NA 0.000 0.000 NA 0.000 0.000 NA 0.000 0.000 NA 0.013 0.040 31 0.035 0.023 23 0.057 0.007 9 0.044 0.076 8 0.001 0.001 52 0.063 0.014 0 0.006 0.003 69 0.000 0.000 NA  Table D.3. (Continued) name  active_cccc vol_cccc 209 NA NA BigHill1 2 1.000 BigHill2 0 0.000 CP144-6 3 0.238 Crest1 15 0.031 Firewood 5 0.012 Gregory1 19 0.461 Gregory2 2 0.030 Gregory3A 7 3.506 Gregory3B 0 0.000 Hidden NA NA Home1 NA NA Home2 NA NA Home4 NA NA JDay NA NA K800A NA NA K800B NA NA Kaien57A 7 3.676 Kaien57B 0 0.000 Lachmach1 6 1.930 Lachmach2 3 0.169 Lachmach79 NA NA Lachmach8 NA NA LogPull NA NA Max NA NA Minerva1 NA NA Minerva2 NA NA Minerva3 NA NA PorkNCheese NA NA PwrLineA 3 0.163 PwrLineB 1 0.014 RedBowl NA NA RedPine 5 0.070 RenFace1 1 0.800 RenFace2 0 0.000 Rennell1 8 0.093 Return NA NA Ridley1 NA NA Ridley3 NA NA Shelley1 0 0.000 Shelley2 7 0.693 Shelley3 0 0.000 Silver1 3 0.444 skeenaView 0 0.000 Skidegate1 8 0.092 Skidegate2 0 0.000  count_cccv NA 3 13 0 0 2 6 3 4 0 NA NA NA NA NA NA NA 15 1 8 4 NA NA NA NA NA NA NA NA 5 0 NA 0 13 0 15 NA NA NA 16 7 4 4 8 20 12  active_cccv vol_cccv NA NA 0 0.000 1 0.480 0 0.000 0 0.000 0 0.000 4 0.003 0 0.000 0 0.000 0 0.000 NA NA NA NA NA NA NA NA NA NA NA NA NA NA 0 0.000 0 0.000 1 0.168 0 0.000 NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA 1 0.015 0 0.000 NA NA 0 0.000 0 0.000 0 0.000 7 0.287 NA NA NA NA NA NA 0 0.000 0 0.000 0 0.000 1 0.075 0 0.000 0 0.000 0 0.000  343  count_cvcc NA  active_cvcc NA  3 0 8 0 8 11 16 0 0 NA NA NA NA NA NA NA  0 0 0 0 5 0 0 0 0 NA NA NA NA NA NA NA  3 1 3 6 NA NA NA NA NA NA NA NA  0 0 0 0 NA NA NA NA NA NA NA NA  8 2 NA  2 2 NA  6 13 0 4 NA NA NA  1 0 0 1 NA NA NA  4 1 1 14 12 0 9  0 0 0 1 3 0 0  Table D.3. (Continued) name Somerset1 Somerset2 Stihl1 Stihl2 SwimPool Waka12-2A Waka12-2B Waka17 Whitebtm1 Whitebtm2 Yakoun1 YFin  active_cccc NA NA NA NA 1 7 2 0 0 0 28 NA  vol_cccc NA NA NA NA 0.000 0.212 0.025 0.000 0.000 0.000 1.310 NA  count_cccv NA NA NA NA 17 10 0 11 0 25 1 NA  active_cccv NA NA NA NA 1 1 0 0 0 0 0 NA  344  vol_cccv NA NA NA NA 0.028 0.001 0.000 0.000 0.000 0.000 0.000 NA  count_cvcc NA NA NA NA  active_cvcc NA NA NA NA  0 2 0 11 0 2 0 NA  0 0 0 0 0 0 0 NA  Table D.3. (Continued) name  vol_cvcc 209 NA BigHill1 0.000 BigHill2 0.000 CP144-6 0.000 Crest1 0.000 Firewood 0.015 Gregory1 0.000 Gregory2 0.000 Gregory3A 0.000 Gregory3B 0.000 Hidden NA Home1 NA Home2 NA Home4 NA JDay NA K800A NA K800B NA Kaien57A 0.000 Kaien57B 0.000 Lachmach1 0.000 Lachmach2 0.000 Lachmach79 NA Lachmach8 NA LogPull NA Max NA Minerva1 NA Minerva2 NA Minerva3 NA PorkNCheese NA PwrLineA 0.022 PwrLineB 0.051 RedBowl NA RedPine 0.005 RenFace1 0.000 RenFace2 0.000 Rennell1 0.002 Return NA Ridley1 NA Ridley3 NA Shelley1 0.000 Shelley2 0.000 Shelley3 0.000 Silver1 1.500 skeenaView 0.067 Skidegate1 0.000 Skidegate2 0.000  count_cvcv NA 13 2 0 0 2 10 4 0 0 NA NA NA NA NA NA NA 20 2 6 13 NA NA NA NA NA NA NA NA 5 0 NA 4 10 0 9 NA NA NA 6 7 2 6 18 1 12  active_cvcv vol_cvcv NA NA 0 0.000 0 0.000 0 0.000 0 0.000 0 0.000 0 0.000 0 0.000 0 0.000 0 0.000 NA NA NA NA NA NA NA NA NA NA NA NA NA NA 2 2.005 0 0.000 0 0.000 0 0.000 NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA 0 0.000 0 0.000 NA NA 1 0.080 0 0.000 0 0.000 1 0.005 NA NA NA NA NA NA 0 0.000 0 0.000 0 0.000 1 0.077 0 0.000 0 0.000 0 0.000 345  Table D.3. (Continued) name Somerset1 Somerset2 Stihl1 Stihl2 SwimPool Waka12-2A Waka12-2B Waka17 Whitebtm1 Whitebtm2 Yakoun1 YFin  vol_cvcc NA NA NA NA 0.000 0.000 0.000 0.000 0.000 0.000 0.000 NA  count_cvcv NA NA NA NA  active_cvcv NA NA NA NA  1 0 0 16 0 5 0 NA  0 0 0 0 0 0 0 NA  vol_cvcv NA NA NA NA 0.000 0.000 0.000 0.000 0.000 0.000 0.000 NA  346  Table D.4. Point level data. name 209 209 209 209 209 209 209 209 209 209 209 209 209 209 209 209 209 209 209 209 209 209 209 209 209 209 209 209 209 209 209 209 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2  position RH RH AHS AAHS AHS AHS LH LH LH HS HS AHS RH RH RH RH BRS A BLS RH BRS A BLS LH LH LH LH S A S RH RH RH RH TRS RS BRS S S LH LH LH LH S S S  scar  X X  X X X X X X  X X X  X X X X  X X X  x 100 101.65 103.06 104.12 105.56 107.41 110.13 111.14 109.36 107.43 105.45 104.01 102.42 101.47 102.89 104.46 105.42 106.54 107.43 104.98 105.89 106.97 108.42 109.72 112.05 114.05 112.38 110.38 109.5 108.61 106.59 105.57 97.23 95.62 94.6 93.97 94.06 93.15 90.97 88.32 84.49 87.65 89.58 91.5 92.62 94.32  y 100 98.61 97.43 96.54 95.33 93.78 91.5 92.95 94.43 96.02 97.66 98.85 100.17 100.95 101.88 99.87 98.65 97.21 96.07 100.19 99.45 98.58 97.4 96.35 94.47 96.75 98.1 99.72 100.43 101.16 102.79 103.62 107.98 107.45 103.05 102.16 101.84 100.1 99.09 96.07 93.7 93.75 95.74 97.11 98.78 99.43 347  z 100 99.85 99.2 98.9 99.35 99.15 99.85 99.6 99.05 98.4 98.25 98.5 98.75 99.25 97.8 98.05 97.7 97.65 97.95 97.4 97.35 97.25 97.25 97.65 98.6 97.15 97 96.3 96.35 98.25 97.1 96.7 96.8 96.72 95.55 95.53 94.77 94.84 93.88 94.29 93.88 95.53 94.99 95.02 94.95 95.73  ssz 99.7 98.74 98.85 98.4 98.55 98.45 99.14 98.44 98.68 98 98.11 97.7 98.28 98.85 97.44 97.57 97.43 97.11 97.55 97.06 96.9 96.92 97.03 97.07 97.76 96.22 96.54 95.87 96.04 97.75 95.96 96.15 95.66 95.62 95.26 94.8 94.19 94.5 93.58 94.03 93.46 95.18 94.64 94.92 94.4 95.23  sd 0.3 1.11 0.35 0.5 0.8 0.7 0.71 1.16 0.37 0.4 0.14 0.8 0.47 0.4 0.36 0.48 0.27 0.54 0.4 0.34 0.45 0.33 0.22 0.58 0.84 0.93 0.46 0.43 0.31 0.5 1.14 0.55 1.14 1.1 0.29 0.73 0.58 0.34 0.3 0.26 0.42 0.35 0.35 0.1 0.55 0.5  sitePoint 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 3 4 5 6 7 8 9 11 12 13 14 15 16 17  Table D.4. (Continued) name Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2  position BRS RS TRS RH TRS RS BRS S S S LH LH AHS AHS BHS BHS BHS BHS AHS RH RH RH RH RH RH AHS BHS AHS AHS AHS HS AHS LH LH AHS AHS AHS AHS RH RH LH LH  scar X X  X X X X X  X X X X  X  X  x 94.96 96.01 96.93 99.44 98.21 96.89 96.19 95.09 93.61 92.57 90.7 89.35 91.09 92.44 92.48 94.34 95.95 96.8 97.59 100.24 101.77 104.59 102.82 100.44 99.1 97.91 89.83 96.91 96.27 95.72 94.42 94.16 88.52 94.05 97.99 99.81 102.86 104.45 106.64 105.44 106.56 113.35  y 100.43 101.24 104.16 104.06 102.46 100.45 99.88 98.38 97.37 96.56 94.89 92.49 93.92 94.43 95.49 96.2 97.16 99.2 99.39 102.47 104.32 106 100.58 98.21 99.2 97.99 92.76 96.21 94.91 95.55 95.14 93.84 90.84 91.19 94.69 95.86 101.81 103.4 106.37 105.58 105.34 106.91  348  z 95.7 97.01 98.1 99.46 99.19 97.58 96.79 96.5 96.02 96.01 96.71 97.68 97.82 98.19 96.93 97.08 97.66 97.54 98.51 99.42 101.13 102.21 102.59 100.73 99.31 99.01 93.62 99.54 99.31 98.57 97.94 99.18 99.52 101.85 101.1 101.9 102.18 104.25 108.69 107.81 107.03 107.94  ssz 95.13 95.61 97 98.87 97.69 96.51 96.26 96.2 95.69 95.96 95.98 97.08 97.43 97.65 96.88 96.83 97.48 96.85 98.35 98.57 100.26 101.38 102.02 100.03 98.69 98.37 93.52 98.84 99.09 98.47 97.65 98.74 98.71 101.06 100.65 101.1 101.94 103.78 107.46 106.51 106.3 107.46  sd 0.57 1.4 1.1 0.59 1.5 1.07 0.53 0.3 0.33 0.05 0.73 0.6 0.39 0.54 0.05 0.25 0.18 0.69 0.16 0.85 0.87 0.83 0.57 0.7 0.62 0.64 0.1 0.7 0.22 0.1 0.29 0.44 0.81 0.79 0.45 0.8 0.24 0.47 1.23 1.3 0.73 0.48  sitePoint 18 19 20 22 23 24 25 26 27 28 30 31 32 33 34 35 37 38 39 41 42 43 45 46 47 48 49 50 51 52 53 54 56 57 59 60 61 62 63 64 65 67  Table D.4. (Continued) name  expRod1 209 209 209 209 209 209 209 209 209 209 209 209 209 209 209 209 209 209 209 209 209 209 209 209 209 209 209 209 209 209 209 209  Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2  36 40 121 77  120 108 115 115 140 95 100  expRod2  KCP10 2 3 1 0 2 0 1 2 1 1 2 1 2 2 3 1 1 0 1 1 2 4 1 2 1 1 0 1 1 1 2 1 0 0 NA NA 1 2 NA 1 NA NA NA NA NA NA  KCP20 8 10 7 3 7 2 8 4 2 2 NA 3 4 9 16 3 7 9 6 6 2 13 5 5 3 2 0 4 3 2 6 3 2 0 NA NA 2 4 NA 2 NA NA NA NA NA NA 349  KCP30 28 14 9 8 12 9 17 4 6 12 NA 6 10 20 28 11 NA 32 27 8 12 27 NA 10 7 3 1 15 7 4 8 5 5 0 NA NA 4 11 NA NA NA NA NA NA NA NA  KCP40 NA 23 NA 26 16 29 29 4 NA 35 NA 15 20 29 NA 30 NA 46 58 NA 33 NA NA 16 12 6 4 34 NA 5 8 8 10 3 NA NA 8 NA NA NA NA NA NA NA NA NA  KCP50 NA 32 NA 50 18 50 42 7 NA NA NA 21 NA NA NA NA NA 50 NA NA NA NA NA 34 16 11 NA NA NA 18 8 10 14 NA NA NA 15 NA NA NA NA NA NA NA NA NA  KCP60 NA 43 NA NA 20 67 54 10 NA NA NA 25 NA NA NA NA NA NA NA NA NA NA NA NA 29 18 NA NA NA NA 8 NA 18 NA NA NA NA NA NA NA NA NA NA NA NA NA  Table D.4. (Continued) name Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2  expRod1 93 10 40  43 120 145 77 90  145 125 81 65 63 93 88 86 140  140 121 69 105 70  27 20  expRod2  KCP10 NA NA NA 1 1 NA 1 NA 1 NA NA NA 1 1 NA NA 1 NA 2 NA NA 1 NA 1 NA NA NA 1 1 NA NA 1 NA 1 1 NA 1 1 NA NA 1 1  KCP20 NA NA NA 1 1 NA 1 NA 1 NA NA NA 2 1 NA NA NA NA NA NA NA 2 NA 2 NA NA NA 1 1 NA NA 1 NA 1 2 NA 3 2 NA NA 1 4  350  KCP30 NA NA NA 1 1 NA 3 NA 4 NA NA NA 4 1 NA NA NA NA NA NA NA 2 NA 5 NA NA NA 2 NA NA NA 1 NA 2 3 NA NA 4 NA NA 2 8  KCP40 NA NA NA 2 1 NA 4 NA NA NA NA NA NA 1 NA NA NA NA NA NA NA 2 NA 12 NA NA NA 5 NA NA NA 3 NA 3 4 NA NA 17 NA NA 2 11  KCP50 NA NA NA 4 2 NA 10 NA NA NA NA NA NA 3 NA NA NA NA NA NA NA 2 NA 16 NA NA NA 12 NA NA NA NA NA 7 NA NA NA NA NA NA 2 NA  KCP60 NA NA NA NA 2 NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA 6 NA 19 NA NA NA 13 NA NA NA NA NA 10 NA NA NA NA NA NA 10 NA  Table D.4. (Continued) name 209 209 209 209 209 209 209 209 209 209 209 209 209 209 209 209 209 209 209 209 209 209 209 209 209 209 209 209 209 209 209 209 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2  KCP70 NA 60 NA NA 35 86 66 16 NA NA NA 31 NA NA NA NA NA NA NA NA NA NA NA NA 49 25 NA NA NA NA 9 NA 22 NA NA NA NA NA NA NA NA NA NA NA NA NA  KCP80 NA 73 NA NA 62 NA NA 20 NA NA NA 50 NA NA NA NA NA NA NA NA NA NA NA NA 57 28 NA NA NA NA 11 NA 25 NA NA NA NA NA NA NA NA NA NA NA NA NA  KCP90 NA 87 NA NA NA NA NA 22 NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA 38 NA NA NA NA 16 NA 27 NA NA NA NA NA NA NA NA NA NA NA NA NA  KCP100 NA 97 NA NA NA NA NA 26 NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA 20 NA 48 NA NA NA NA NA NA NA NA NA NA NA NA NA 351  KCP110 NA 105 NA NA NA NA NA 38 NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA 30 NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA  KCP120 NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA  KCP130 NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA  KCP140 NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA  Table D.4. (Continued) name Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2  KCP70 NA NA NA NA 3 NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA 7 NA NA NA NA NA 19 NA NA NA NA NA 11 NA NA NA NA NA NA 17 NA  KCP80 NA NA NA NA 8 NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA 13 NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA  KCP90 NA NA NA NA 17 NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA  KCP100 NA NA NA NA 27 NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA  352  KCP110 NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA  KCP120 NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA  KCP130 NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA  KCP140 NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA  Table D.4. (Continued) name 209 209 209 209 209 209 209 209 209 209 209 209 209 209 209 209 209 209 209 209 209 209 209 209 209 209 209 209 209 209 209 209 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2  KCP150 NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA  KCPstrikes KCPmax KCPnotes 28 30 Wood (Rx) 109 111 15 35 Rx 50 50 62 80 86 70 70 71 Rx 60 116 21 37 35 40 7 14 50 80 36 47 29 40 Rx 40 36 50 48 17 27 54 54 58 40 14 34 46 45 35 33 20 22 37 58 68 84 47 93 18 46 Rx? 42 43 18 31 Rx? 18 50 40 114 24 55 48 100 3 45 NA NA NA NA 20 58 Till 13 34 BRx NA NA 5 26 NA NA NA NA NA NA NA NA NA NA NA NA 353  procType  procL  F  LS  70  Table D.4. (Continued) name Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2  KCP150 NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA  KCPstrikes KCPmax KCPnotes NA NA NA NA NA NA 11 59 62 150 NA NA 12 53 NA NA 6 33 NA NA NA NA NA NA 7 39 8 54 NA NA NA NA 3 18 NA NA 4 16 NA NA NA NA 16 83 NA NA 30 70 NA NA NA NA NA NA 21 70 3 22 NA NA NA NA 7 44 NA NA 14 79 9 49 NA NA 7 24 30 47 NA NA NA NA 19 73 13 48 Stone  354  procType  procL  F  W F W  SW SW SW SW  R 20 50  Table D.4. (Continued) name 209 209 209 209 209 209 209 209 209 209 209 209 209 209 209 209 209 209 209 209 209 209 209 209 209 209 209 209 209 209 209 209 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2  procW NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA 60 NA NA  procD NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA 35 NA NA  procVol NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA 0.147 NA NA  LWDL 4 3 1.5 2 NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA 8 4 NA NA NA NA 4 NA 8 355  LWDD 15 10 15 20 NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA 60 50 NA NA NA NA 15 NA 30  LWDVol 0.071 0.024 0.026 0.047 NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA 2.261 0.785 NA NA NA NA 0.071 NA 0.565  LWDOrient 90 0 0 45 NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA 0 0 NA NA NA NA 45 NA 0  Table D.4. (Continued) name Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2  procW NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA 30 25 NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA  procD NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA 15 10 NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA  procVol NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA 0.009 0.013 NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA  LWDL 7 NA NA NA NA NA NA NA NA NA NA NA 10 NA NA NA NA 12 NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA  356  LWDD 40 NA NA NA NA NA NA NA NA NA NA NA 40 NA NA NA NA 20 NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA  LWDVol 0.879 NA NA NA NA NA NA NA NA NA NA NA 1.256 NA NA NA NA 0.377 NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA  LWDOrient 45 NA NA NA NA NA NA NA NA NA NA NA 45 NA NA NA NA 90 NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA  Table D.4. (Continued) name 209 209 209 209 209 209 209 209 209 209 209 209 209 209 209 209 209 209 209 209 209 209 209 209 209 209 209 209 209 209 209 209 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2  surfCover VEG VEG VEG VEG VEG VEG VEG VEG VEG VEG VEG VEG VEG VEG VEG VEG VEG VEG VEG VEG VEG VEG VEG VEG VEG WD WD VEG VEG VEG VEG VEG VEG VEG VEG VEG WD VEG WD WD WD WD WD VEG VEG VEG  aspect NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA 183.485 179.609 180.318 159.807 153.234 131.361 NA 124.868 144.37 125.594 145.872 159.084  slope NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA 39.4 37.212 37.241 31.807 30.735 35.569 NA 41.093 39.572 39.802 34.842 35.223  s3longc NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA 0 NA -0.078 -0.086 -0.061 0.05 0.037 -0.127 0 -0.009 -0.042 -0.158 -0.09 -0.054 357  s3crosc NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA 0 NA -0.006 -0.173 -0.037 0.033 0.258 -0.164 0 -0.012 0.104 -0.676 -0.165 -0.005  s9longc NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA 0 0 -0.041 -0.057 -0.064 -0.029 -0.077 -0.1 0 -0.118 -0.074 -0.065 -0.034 -0.064  s9crosc NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA 0 0 0.006 -0.099 -0.099 -0.1 -0.123 -0.08 0 -0.06 -0.005 -0.045 -0.126 -0.125  Table D.4. (Continued) name Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2  surfCover VEG VEG WD VEG VEG VEG VEG VEG VEG VEG VEG VEG WD VEG RX VEG VEG VEG VEG WD WD VEG VEG WD WD WD S WD WD VEG VEG VEG WD WD WD VEG WD WD WD VEG WD VEG  aspect 166.618 181.46 173.485 155.758 180.468 177.117 170.601 152.131 133.34 128.888 137.073 125.486 128.747 136.299 134.547 130.252 148.013 164.878 167.206 162.387 183.245 190.398 171.004 157.519 162.537 151.493 126.007 145.204 130.877 134.421 123.759 114.312 NA 122.304 139.588 145.421 180.179 192.118 204.376 192.953 209.509 305.426  slope 36.086 41.197 37.66 32.874 38.172 40.628 39.698 39.126 38.269 40.111 41.202 47.768 46.381 46.785 43.775 41.379 41.94 40.109 40.98 33.173 47.481 47.614 40.659 42.561 41.945 43.654 47.931 45.598 46.442 44.814 44.558 46.515 NA 48.286 45.69 43.749 46.294 44.421 43.042 47.806 41.218 41.379  s3longc -0.056 -0.075 -0.115 -0.247 -0.21 0.129 -0.012 -0.073 -0.145 -0.077 0.093 -0.033 -0.15 -0.064 0.006 -0.063 -0.078 -0.092 0.033 0.608 -0.705 -0.155 -0.092 0.01 0.023 0.028 -0.086 -0.097 -0.17 -0.097 -0.185 -0.317 -0.063 -0.07 0.091 0.013 -0.8 0.077 0.165 0.089 0.114 0.121  358  s3crosc -0.229 -0.139 -0.022 0.23 -0.014 0.006 -0.203 -0.069 -0.242 -0.073 0.015 -0.041 -0.233 0.057 0.46 -0.207 -0.097 -0.203 0.079 0.004 -0.416 0.106 -0.535 -0.112 0.017 -0.025 0.004 -0.25 -0.247 -0.207 -0.087 0.006 -0.024 -0.121 -0.026 -0.068 -0.075 0.075 0.156 0.127 0.187 0.088  s9longc -0.069 0.001 0.013 -0.034 0.024 0.004 -0.027 -0.08 -0.067 -0.052 -0.086 -0.084 -0.084 -0.071 -0.076 -0.079 -0.075 -0.018 -0.003 -0.026 -0.2 NA 0.016 -0.013 -0.013 -0.015 -0.076 -0.045 -0.036 -0.065 -0.069 -0.038 -0.093 -0.01 0.015 0.031 -0.03 0.141 0 NA 0.076 0  s9crosc -0.124 -0.075 0.064 0.054 0.043 -0.111 -0.129 -0.134 -0.132 -0.049 -0.024 -0.031 -0.053 0.045 0.07 -0.13 -0.125 -0.14 -0.132 0.052 0.005 NA -0.129 -0.067 -0.112 -0.089 -0.028 -0.112 -0.146 -0.153 -0.069 0.043 -0.038 0.062 -0.085 -0.082 -0.07 0.048 0 NA 0.161 0  Table D.4. (Continued) name 209 209 209 209 209 209 209 209 209 209 209 209 209 209 209 209 209 209 209 209 209 209 209 209 209 209 209 209 209 209 209 209 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2  ssaspect NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA 177.989 169.071 167.704 147.402 149.845 130.296 NA 129.702 141.919 126.595 139.348 148.049  ssslope NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA 35.838 33.567 33.564 29.581 33.064 37.705 NA 38.077 35.632 39.645 34.688 35.103  ss3longc NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA 0 NA 0.026 0.004 -0.097 0.021 -0.009 -0.07 0 0.091 -0.344 -0.083 -0.093 -0.103  ss3crosc NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA 0 NA 0.033 -0.299 0.124 0.078 -0.108 -0.049 0 0.074 -0.251 -0.414 -0.09 -0.167 359  ss9longc NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA 0 0 -0.034 -0.069 -0.066 -0.04 -0.064 -0.032 0 -0.048 -0.039 -0.044 -0.055 -0.073  ss9crosc NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA 0 0 -0.001 -0.076 -0.086 -0.087 -0.109 -0.06 0 -0.026 -0.007 -0.018 -0.08 -0.11  Table D.4. (Continued) name Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2 Lachmach2  ssaspect 158.715 173.093 171.305 163.925 174.554 164.195 158.724 143.431 131.254 129.571 135.671 127.588 129.765 136.614 138.568 129.224 141.706 152.675 152.184 173.415 171.602 180.038 165.742 153.684 157.095 143.138 128.304 141.949 128.396 133.541 125.071 134.486 NA 128.165 137.439 146.561 173.823 180.233 179.353 180.107 181.839 NA  ssslope 34.733 36.254 35.837 36.924 38.985 36.528 37.336 39.685 39.212 39.834 39.016 43.421 41.91 42.583 38.875 40.473 41.375 38.817 39.982 38.27 48.02 45.142 47.509 43.75 42.411 42.12 43.228 42.84 43.201 42.036 42.704 45.38 NA 46.27 43.727 43.291 48.043 51.486 39.756 44.135 39.582 NA  ss3longc -0.053 0.041 -0.005 0.345 -0.16 0.014 -0.059 -0.08 -0.046 -0.039 -0.025 -0.043 -0.105 -0.033 0.013 -0.052 -0.077 -0.014 -0.047 0.672 -0.083 0.064 0.021 -0.001 0.064 0.009 -0.038 -0.044 -0.051 -0.005 -0.16 0.018 -0.051 0.015 -0.106 0.063 -0.106 -0.109 0.108 0.075 0.106 NA  ss3crosc -0.175 -0.075 0.032 0.061 0.046 -0.117 -0.158 -0.068 -0.063 -0.052 -0.224 -0.016 -0.061 0.06 0.106 -0.087 -0.086 -0.097 -0.199 0.324 -0.153 0.015 0.068 -0.023 -0.133 -0.024 -0.015 -0.075 -0.231 -0.142 -0.136 0.344 -0.017 0.093 -0.28 -0.132 -0.15 0.377 0.025 0.017 0.072 NA  360  ss9longc -0.059 -0.016 -0.024 -0.046 -0.034 -0.032 -0.04 -0.072 -0.05 -0.035 -0.05 -0.056 -0.059 -0.061 -0.056 -0.043 -0.043 -0.045 -0.049 -0.059 -0.127 NA -0.101 -0.019 -0.045 -0.02 -0.052 -0.015 -0.018 -0.028 -0.042 -0.025 -0.061 -0.016 -0.008 0.004 -0.104 0.139 0 NA 0.103 0  ss9crosc -0.12 -0.075 0.019 -0.003 0.025 -0.128 -0.13 -0.093 -0.061 -0.007 -0.026 -0.018 -0.047 -0.013 0.028 -0.044 -0.076 -0.122 -0.129 0.032 -0.021 NA -0.034 -0.093 -0.126 -0.06 -0.015 -0.072 -0.087 -0.094 0.007 0.05 -0.023 0.018 -0.087 -0.063 -0.061 -0.042 0 NA 0.045 0  

Cite

Citation Scheme:

        

Citations by CSL (citeproc-js)

Usage Statistics

Share

Embed

Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                        
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            src="{[{embed.src}]}"
                            data-item="{[{embed.item}]}"
                            data-collection="{[{embed.collection}]}"
                            data-metadata="{[{embed.showMetadata}]}"
                            data-width="{[{embed.width}]}"
                            async >
                            </script>
                            </div>
                        
                    
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:
http://iiif.library.ubc.ca/presentation/dsp.24.1-0070976/manifest

Comment

Related Items