Managing Forested Watersheds for Hydrogeomorphic Risks on Fans

LAND MANAGEMENT HANDBOOK

6 1

2 0 0 9

Ministry of Forests and Range Forest Science Program The Best Place on Earth Managing Forested Watersheds for Hydrogeomorphic Risks on Fans

D.J. Wilford, M.E. Sakals, W.W. Grainger, T.H. Millard, and T.R. Giles

Ministry of Forests and Range Forest Science Program The Best Place on Earth The use of trade, �rm, or corporation names in this publication is for the information and convenience of the reader. Such use does not constitute an oﬃcial endorsement or approval by the Government of British Columbia of any product or service to the exclusion of any others that may also be suitable. Contents of this report are presented as information only. Funding assistance does not imply endorsement of any statements or information contained herein by the Government of British Columbia. Uniform Resource Locators (URLs), addresses, and contact information contained in this document are current at the time of printing unless otherwise noted.

Library and Archives Canada Cataloguing in Publication Data Managing forested watersheds for hydrogeomorphic risks on fans / D.J. Wilford ... [et al.]. Includes bibliographical references. ISBN 978-0-7726-6119-7 1. Mass-wasting--British Columbia--Forecasting. 2. Landslide hazard analysis--British Columbia. 3. Forests and forestry--Environmental aspects --British Columbia. 4. Forest management--British Columbia--Planning. 5. Forest hydrology--British Columbia. 6. Alluvial fans--British Columbia. 7. Colluvium--British Columbia. I. Wilford, D. J. (David J.), 1950- II. British Columbia. Ministry of Forests and Range III. British Columbia. Forest Science Program SD387.E58M36 2009 634.961 C2009-909966-7

Citation Wilford, D.J., M.E. Sakals, W.W. Grainger, T.H. Millard, and T.R. Giles. 2009. Managing forested watersheds for hydrogeomorphic risks on fans. B.C. Min. For. Range, For. Sci. Prog., Victoria, B.C. Land Manag. Handb. 61. www.for.gov.bc.ca/hfd/pubs/Docs/Lmh/Lmh61.htm

Prepared by D.J. Wilford T.H. Millard Ministry of Forests and Range Ministry of Forests and Range Smithers, BC Nanaimo, BC M.E. Sakals T.R. Giles Ministry of Forests and Range Ministry of Forests and Range Smithers, BC Kamloops, BC W.W. Granger Grainger and Associates Consulting Ltd. Salmon Arm, BC

Prepared for B.C. Ministry of Forests and Range Research Branch Victoria, BC © 2009 Province of British Columbia Copies of this report can be obtained from: Crown Publications, Queen’s Printer PO Box 9452 Stn Prov Govt 563 Superior Street, 2nd Flr Victoria, BC V8W 9V7 1 800 663-6105 www.crownpub.bc.ca

For more information on Forest Science Program publications, visit: www.for.gov.bc.ca/scripts/hfd/pubs/hfdcatalog/index.asp ABSTRACT

Fans are linked to their watersheds by hydrogeo- In British Columbia, forest harvesting and road morphic processes—�oods, debris �oods, and debris building is associated with increased hydrogeomor- �ows. These processes move water, sediment, and phic hazards. The downstream eﬀects of these for- debris from the hillslopes of a watershed through estry activities in source areas may be far-reaching channels to the fan. Fans in British Columbia are and extend beyond the scope of conventional site- often the site of residential developments, and trans- oriented planning. A �ve-step approach is presented portation and utility corridors, as well as high-value to assist land managers undertake risk analyses and habitat for �sh and high-productivity growing sites assessments that place their proposed developments for forests. Collectively, these features are termed within the watershed-fan system. The �ve steps are: “elements-at-risk” because they may be vulnerable 1) identify fans and delineate watersheds; 2) identify to watershed-generated hydrogeomorphic processes elements-at-risk on fans; 3) investigate fan processes; that issue onto the fan. These processes may be natu- 4) investigate watershed processes; and 5) analyze ral or result from land use activities, and can cause risks and develop plans. This scheme is applicable to the partial or total loss of some or all of the elements forested watersheds throughout British Columbia. on the fan.

ACKNOWLEDGEMENTS

The need for this handbook was raised by the For- sion has had the bene�t of very thorough reviews by estry Committee of the Council of Forest Industries Todd Redding, Robin Pike, Ian Smith, Steve Webb, and represents 3 years of collaborative research Rita Winkler, and David Maloney. This handbook involving many people throughout British Co- has bene�ted from a meticulous editorial review by lumbia. The concept for the �ve-step approach was Steve Smith. We are indebted to the Forest Invest- created by Bill Grainger and we are indebted to him ment Account–Forest Science Program, BC Timber for his foresight. This handbook has gone through Sales, and the B.C. Ministry of Forests and Range for many revisions over the past 3 years, incorporating �nancial support. suggestions from workshop participants; this ver-

iii CONTENTS

Abstract ...... iii Acknowledgements ...... iii Introduction ...... 1 The Five-step Method ...... 1 Step 1 Identify Fans And Delineate Watersheds ...... 2 Step 1.1 The Fan-watershed System ...... 2 Step 1.2 Fan Identi�cation ...... 4 Step 1.3 Watershed Delineation ...... 7 Step 2 Identify Elements-at-risk on Fans ...... 8 Step 2.1 Human Safety ...... 8 Step 2.2 Anthropogenic Features ...... 8 Step 2.3 Natural Features ...... 10 Step 3 Investigate Fan Processes ...... 10 Step 3.1 Hydrogeomorphic Processes ...... 10 Step 3.2 Event Frequency ...... 12 Step 3.3 Event Magnitude ...... 13 Step 4 Investigate Watershed Processes ...... 14 Step 4.1 Watershed-fan Process Linkages ...... 14 Step 4.2 Oﬃce Investigations ...... 16 Step 4.3 Field Investigations ...... 18 Step 4.4 Synthesis of Watershed Processes ...... 19 Step 5 Analyze Risks and Develop Plans ...... 20 Step 5.1 Understanding Risk Analysis and Risk Assessment ...... 20 Step 5.2 Consequence ...... 20 Step 5.3 Hazard ...... 21 Step 5.4 Risk Analysis ...... 23 Step 5.5 Assessing Risk and Making Management Decisions ...... 23 Step 5.6 Document, Monitor, Evaluate, and Report ...... 25 Case Studies ...... 25 Summary ...... 26 Literature Cited ...... 57 appendices 1 Wathl Creek case study ...... 27 2 Eagle Summit Creek case study ...... 33 3 Shale Creek case study ...... 42 4 Hummingbird Creek case study ...... 46

iv Tables 1 Characteristics of hydrogeomorphic process deposits ...... 11 2 Forest management focus for diﬀerent hydrogeomorphic processes ...... 15 3 Predictive models for dominant hydrogeomorphic processes using the relative relief number and watershed length ...... 17 4 Example of long-term probabilities ...... 22 5 Qualitative frequency de�nitions ...... 23 6 Example of qualitative hazard analysis matrix ...... 23 7 Qualitative risk analysis matrix ...... 23 A1.1 A matrix combining hazard and consequence to determine risk ...... 31 A1.2 Consequences, hazards, and risks of all the identi�ed hazards and elements-at-risk, except the coarse sediment loading hazard and the consequence of potential impacts to residences and human safety, which are dealt with separately ...... 31 A1.3 Coarse bedload sediment risk ...... 32 A2.1 Qualitative frequency de�nitions ...... 39 A2.2 A qualitative hazard matrix ...... 40 A2.3 A qualitative risk matrix ...... 40 A4.1 Relative relief numbers for the Swansea Point fan ...... 51 A4.2 Forest management focus for diﬀerent hydrogeomorphic processes ...... 55 A4.3 A matrix combining hazard and consequence to determine risk ...... 56

Figures 1 An aerial oblique of a fan—the clearcut harvested portion of the landscape ...... 3 2 An aerial photograph with notation indicating fan areas in red and potential �ow pathways in blue . . . . 3 3 A landform map with alluvial and colluvial fans highlighted ...... 5 4 An example of an actively growing fan resulting from high sediment transport from the watershed to the fan ...... 6 5 An example of a dissected fan resulting from abundant water but limited sediment transport from the watershed ...... 6 6 A topographically de�ned watershed above a fan ...... 7 7 A water intake structure at the apex of a fan ...... 9 8 This building was damaged by a debris �ow caused by a drainage diversion related to a forest road . . . . 9 9 This drainage structure is not designed to accommodate the water, sediment, and debris from naturally occurring hydrogeomorphic events ...... 10 10 A debris �ow initiated as a result of hydrophobic soils in the Kuskanook watershed, damaging homes and the highway on the fan ...... 18 A1.1 The Wathl Creek watershed and fan...... 27 A1.2 A view of the Wathl Creek fan and Kitamaat Village looking east into the watershed...... 28 A2.1 Eagle Summit Creek watershed ...... 34 A2.2 Gully and fan reaches of Eagle Summit Creek, looking south ...... 35 A2.3 A view of the drainage structure on the Trans-Canada Highway at Eagle Summit Creek ...... 36

v A2.4 A view of the upper steep reach in Upper Eagle Summit Creek ...... 37 A2.5 A view looking upstream in the lower reach of Eagle Summit Creek ...... 38 A3.1 Location map for Block SKI 101 ...... 42 A3.2 Shale Creek watershed and fan ...... 43 A3.3 Channel network on Shale Creek fan ...... 44 A3.4 Stream 2 �owing in an older and larger channel ...... 44 A4.1 Map of Hummingbird and Mara Creek watersheds, and the Swansea Point Fan ...... 47 A4.2 Distribution of new debris �ow sediment on the Swansea Point fan, mapped on July 14, 1997 ...... 48 A4.3 Looking downstream from the con�ned valley reach of Hummingbird Creek to the apex of the Swansea Point fan ...... 49 A4.4 The “Hinkelstein” at the apex of the Swansea Point fan ...... 49 A4.5 Aerial photograph BC2616: 76 of lower Hummingbird Creek, 1959, showing a recent landslide to the north on similar steep, northwest-aspect terrain, a possible relic debris slide colonized by a younger forest stand, and the location of the 1997 debris avalanche ...... 51 A4.6 Pre- and post-development runoﬀ-contributing areas to the culvert located above the debris avalanche ...... 53 A4.7 Aerial photograph of lower Hummingbird Creek watershed with the debris avalanche below the road in right centre ...... 53

vi INTRODUCTION

British Columbia’s mountainous landscape con- above fans be planned and undertaken with an un- tains numerous forested watersheds connected to derstanding of the fan-watershed system and a criti- fans that have human development or high natural cal consideration of the risks to downstream features resource values. Fans are the cone-shaped deposits on fans (Jakob et al. 2000). of sediment formed where stream channels leave This manuscript describes �ve steps used to the con�nes of mountain valleys (Bull 1977). Fans recognize and analyze risk in fan-watershed systems. are desirable sites for development because of their The information presented builds upon a body of gentle gradients and workable materials. Unfor- knowledge regarding fan-watershed systems from tunately, fans are often dynamic landforms. They British Columbia and beyond. Notable sources are run-out areas for hydrogeomorphic events (i.e., include a multi-year investigation of forested fans debris �ows, debris �oods, and �oods) originating across British Columbia (Wilford 2003; Wilford et in the watersheds. Fans commonly have uncon�ned al. 2002, 2003, 2004a, 2004b, 2005a, 2005b, 2006), stream channels, resulting in the broadcasting of Land Management Handbook 56: Landslide Risk water and sediment, and channel avulsions. Risks Case Studies in Forest Development Planning and from hydrogeomorphic events in�uencing a fan Operations (Wise et al. 2004), and Debris Flow must be carefully considered when natural changes Hazards and Related Phenomena (Jakob and Hungr occur in a watershed (e.g., wild�re or mountain pine 2005a). This handbook provides direction to those beetle) or prior to watershed development (e.g., forest who, during the course of their work, encounter harvesting or other developments). This is because of fans and need to analyze associated risks (e.g., forest the potential to change water and sediment regimes practitioners and planners, engineers and geoscien- that can lead to changes in the timing, magnitude, tists, regional planners). This manuscript attempts to and frequency of hydrogeomorphic events aﬀect- provide an understanding of watershed-fan pro- ing fans. Even without land use changes in water- cesses and the planning and assessment of watershed sheds, natural hydrogeomorphic events on fans have activities to help better manage potential environ- resulted in the loss of life and high �nancial costs mental, social, and economic risks. It also helps to throughout British Columbia (Septer and Schwab identify when it is prudent to engage hydrogeomor- 1995) and world-wide (Sidle et al. 1985). It is therefore phic specialists (i.e., hydrologists, terrain specialists, important that resource development in watersheds and others) in the process.

THE FIVE-STEP METHOD

Forest-land management in British Columbia has a Ministry of Environment 2001) presents an approach largely site-level approach to planning. While many to assessing fan-watershed systems, but it is ap- strong programs exist to address the importance of plicable only to coastal areas of British Columbia. site-level management issues, some situations call The Watershed Assessment Procedure Guidebooks for a wider perspective. For example, landslides and (BCMOF and BCMOE 1999a) indirectly address risks peak �ows in a watershed can lead to serious hy- on fans at a watershed scale. While the Mapping and drogeomorphic impacts on a fan located many kilo- Assessing Terrain Stability Guidebook (BCMOF and metres away, destabilizing the fan surface, affecting BCMOE 1999b) states that off-site downslope/down- infrastructure, and altering aquatic habitat. These stream elements-at-risk and possible consequences effects are a result of connections in the fan-water- should be considered, assessments have been gener- shed system, and risk analyses on fans should match ally conducted at the site level and often stop short the scale of these interactions between the fan and of a risk assessment by focussing primarily on the the upstream watershed. hazards (e.g., initiation zones) (Schwab and Geertse- The Gully Assessment Procedure (British Co- ma 2008). In 2006, the Province of British Columbia lumbia Ministry of Forests and British Columbia formalized landslide risk assessments for residential

1 developments (Association of Professional Engineers es in the watershed (i.e., either natural processes and Geoscientists of British Columbia 2006), which or those related to proposed management ac- should result in better management of new activities tions). on fans, given existing hydrogeomorphic risks. How- Step 3: Investigate Fan Processes ever, these formal guidelines do not address incre- • Identify the nature of hazardous hydrogeomor- mental risks to existing elements (e.g., homes, roads, phic processes (type, frequency, and disturbance drainage structures) on fans due to natural processes extent). or land use developments in the watershed. Step 4: Investigate Watershed Processes This handbook provides the �rst comprehensive, • Identify watershed features controlling hydrogeo- provincial-level framework for assessing risk on fans morphic processes (e.g., watershed hydrology, from upstream watershed activities. The following geomorphology, and the role of vegetation cover). �ve steps for analyzing risk in fan-watershed systems • Identify the potential for incremental hazards as- are used as a framework for this handbook: sociated with management activities. Step 5: Analyze Risks and Develop Plans Step 1: Identify Fans and Delineate Watersheds • Develop planning options for the watershed and • Identify the physical inter-connections that exist assess the associated risks. between areas of potential activities (watersheds) • Document the process and establish a plan to and areas of potential impacts (fans). monitor, evaluate, and report. Step 2: Identify Elements-at-risk on Fans • Recognize and inventory values on the fan that may be aﬀected by the hydrogeomorphic process-

STEP 1 IDENTIFY FANS AND DELINEATE WATERSHEDS

Step 1.1 The Fan-watershed System and fan requires a watershed-level perspective. Watersheds can be of various scales. For example, A fan is a cone-shaped deposit of sediment formed a small cutblock may comprise most of the run- where a stream channel leaves the con�nes of a oﬀ-contributing area of a culvert along a mainline mountain valley (Bull 1977) (Figure 1). It is an ex- logging road. The same cutblock may be only a very pression of its watershed; the fan is created by and small portion of the watershed contributing water, represents a summary of the hydrologic and geo- sediment, and debris to a fan containing a highway morphic processes in the watershed. The watershed, bridge-crossing located several kilometres from the or catchment, is the source area for water, sediment, block. Medium-sized watersheds (on the order of and woody debris, and the stream channels are the �fty to hundreds of square kilometres) may include transport zone. The local climate provides the water numerous tributary junction fans—each with its and solar energy; geology and biota provide the own watershed. Water and sediment leaving the transportable material and control the watershed tributary junction fans subsequently in�uence main- morphology. Water and organic and inorganic mate- stem channel conditions and transport regimes. In rials are moved from hillslopes, into stream chan- other situations, mid-slope fans may cause the domi- nels through various pathways. Materials eventually nant �ow pathway to change from one downstream move out of the watershed and onto the fan. This watershed to another as the stream naturally moves strong linkage between the watershed, channel, about on the fan surface (Figure 2).

2 Figure 1 An aerial oblique of a fan—the clearcut harvested portion of the landscape. This is a “classic” fan, radiating as a cone-shaped landform where the stream leaves the conﬁnes of the hillslope.

Figure 2 An aerial photograph with notation indicating fan areas in red and potential ﬂow pathways in blue. Depending on the dominant ﬂow pathway on the mid-slope fan, water and sediment can be delivered to any of the valley- bottom fans.

3 Step 1.2 Fan Identification been completed for many areas of British Colum- bia at 1:50 000 scale and generally identi�es larger A “classic” fan has a cone shape; however, the devel- fans in mapped areas (e.g., Runka 1972). Due to the opment of a fan is affected by the space into which relatively small map scale, many smaller fans will be it builds and the hydrogeomorphic processes that amalgamated into larger polygons and thus may not occur there. Many fans are bounded on one or both be represented. More recently, landform mapping sides by hillslopes, often resulting in an asymmetric has been done in association with terrain stability shape to the fan. Some fans build into a broad valley mapping, which is usually completed at 1:20 000 that is not affected by any other hydrogeomorphic scale and therefore is much better at identifying process while some fans build into a con�ned valley smaller fans (Figure 3). Much of this work is also with a valley-bottom river. The valley-bottom river available in digital format (available at: www.em.gov. may remove sediment from the toe of the fan, thus bc.ca/mining/geolsurv/terrain&soils/frbcguid.htm). limiting the size of the fan and potentially the spatial However, in some areas, low-gradient terrain units extent of sediment and debris deposits on the surface are combined into large polygons (e.g., colluvial and of the fan. Some fans build into lakes or the ocean morainal blankets with fans) and the location of in- (fan deltas), affecting both the spatial extent of the dividual fans may not be indicated. When using any fan and the internal structure of the fan. mapped information, it is important to understand Identifying fans that may be affected by manage- what the purpose of the work was, what the methods ment activities in and beyond the planning area were, and at what scale the original mapping was is a critical component in developing appropriate completed. management strategies and prescriptions. A given Attempts have been made to automatically iden- planning unit may encompass no fans but may tify fans using geographic information systems (GIS); belong in the contributing areas of many separate success was limited due to challenges in identifying fan-watershed systems. Conversely, a planning area the distal (outer) boundaries of fans. The boundaries may include little of the contributing areas, but frequently transition into �oodplains with very little many fans. Also, multiple scales of fan-watershed change in gradient. systems may be present, so it is important to identify Ground-truthing is validating remotely sensed fans, including their apexes and their boundaries. information through �eld checks. It is recommended Various methods are available, and any method that to ground-truth potential problem areas or to ran- identi�es the fans, particularly the fan apexes, is ap- domly validate mapping. Once in the �eld, fans can propriate. be difficult to identify because: they can be large, A common method is to use stereoscopic aerial making it difficult to conceive of the entire landform photographs. A range of aerial photograph scales without seeing it; distal portions may grade into wet- (e.g., 1:30 000 and 1:60 000) allows different vertical land areas, making �eldwork difficult or impossible; exaggerations and different levels of visible detail to and fan boundaries may grade into other low-gradi- be reviewed. A degree of aerial photographic in- ent surfaces such as �oodplains, making fan bound- terpretation skill is required, but can be developed ary delineation difficult. Not withstanding these through practice and by interpreting areas where challenges, �eldwork is essential to complete Steps 2 one is familiar with ground conditions. and 3 of a fan-watershed assessment. On topographic maps the typical “fan-shaped” There is considerable discussion in the literature morphology is often apparent where stream con�ne- regarding the temporal origins of fans. In some ment ends at valley-bottom �oodplains or lakes. Us- situations, �eld evidence (e.g., elevated fan surfaces) ing stereo aerial photographs and topographic maps supports the position that fan origin is linked to late- together will usually result in the most accurate fan glacial sedimentation episodes unrelated to modern delineation. Using either or both of these methods, conditions (Ryder 1971a, 1971b; Ritter et al. 1993). the size of fans that can be identi�ed will vary de- During these periods, considerable volumes of sedi- pending upon the scale of available photographs or ment were deposited as fans. As watersheds reveg- topographic maps. etated, sediment supply declined and streams from Soil, sur�cial geology, and landform mapping has the watersheds eroded into the fan surface, leaving

4 Figure 3 A landform map with alluvial (Ff in yellow) and colluvial (Cf in orange) fans highlighted. Of note: one fan was missed by the mapper but identiﬁed during ﬁeldwork—it was mapped as a complex of morainal blanket and morainal veneer (Mbv) and is located just below the centre of the map and highlighted in orange (a portion of polygon 100). elevated fan surfaces and lower surfaces that are 1970). Watersheds that have abundant water but in�uenced by contemporary hydrogeomorphic pro- limited sediment transport may have dissected fans cesses. Inactive elevated or “paraglacial” fan surfaces (Figure 5) (Hunt and Mabey 1966). Some fans are can be found on many fans in British Columbia, but considered to be in a steady-state, dynamic equilib- other types of fans are also found. Watersheds with rium with sediment supply and water (Denny 1965, abundant sediment transported to their fans may 1967; Hooke 1968). The �eldwork described in Steps 2 have fans that are actively growing (Figure 4) (Beaty and 3 will identify the nature of speci�c fans.

5 Figure 4 An example of an actively growing fan resulting from high sediment transport from the watershed to the fan.

Figure 5 An example of a dissected fan (A) resulting from abundant water but limited sediment transport from the watershed. The stream channel (B) is incised into the fan surface 6–10 m.

6 Step 1.3 Watershed Delineation boundaries on appropriately scaled paper or digital topographic maps, as described below and illustrated It is necessary to identify the watershed upstream of in Figure 6. each fan in the planning area, and the fans outside of the planning area if activities are planned in the 1. Determine the point of interest. In the case of fan watersheds of these fans. A watershed, catchment, or studies, the point of interest is usually the apex, or drainage basin is de�ned as the area where the water top, of the fan. arriving at the surface will drain to a point of inter- 2. Draw the boundary moving upslope from the est (Strahler and Strahler 1987). Properly delineat- point of interest, crossing contour lines perpen- ing a watershed is one of the most important steps dicularly. An exception is commonly required in toward understanding the fan-watershed system. the immediate vicinity of the apex where the scale Watershed boundaries can be determined though of the map is generally insuﬃcient to properly a variety of methods, ranging from a pencil and a characterize the surface topography. topographic map to advanced GIS. 3. Continue drawing until a closed polygon has been For individual or a small number of watersheds, created de�ning the catchment area above the fan. the best method is manually drawing watershed

Figure 6 A topographically deﬁned watershed above a fan.

7 In this way, topographic watershed boundaries Manual delineation can also be completed with are readily de�ned and reasonably accurate and thus GIS. The advantage of usingGIS is that not only can are the standard for hydrology, geomorphology, and the watershed boundary be easily altered but water- engineering studies. There may be errors, such as in shed area and other relevant watershed metrics can areas of low relief where surface water �ow path- be readily and accurately attained. A planimeter or ways may not follow the mapped surface expression. digitizer can be used to determine watershed area if Where groundwater �ow paths do not conform to the topographic information is in paper format. topographically de�ned watershed boundaries, errors can also be introduced.

STEP 2 IDENTIFY ELEMENTS-AT-RISK ON FANS

A key part of risk analysis is to identify what is at legal liability. A positive note is that humans are risk on fans, and these are referred to as “elements- mobile features that: 1) may not always be present in at-risk”—human safety, public and private property that location, and thus have limited temporal expo- (including building, structure, land, resources, sure to fan hazards; and 2) can vacate the area recreational site, and cultural heritage features), if given enough notice. transportation system/corridor, utility and utility corridor, domestic water supply, �sh habitat, wildlife Step 2.2 Anthropogenic Features (non-�sh) habitat and migration, visual resource, and timber (BCMOF and BCMOE 2002). While humans can be mobile, their improvements Once all the involved watersheds and fans to be frequently are not, and any damage may result in included in the analysis have been identi�ed, the litigation. Thus anthropogenic features must be elements-at-risk on the fans need to be inventoried. included as elements-at-risk. This refers to any Both natural and human-made features should be human-made features on the identi�ed fan(s). included. This step involves office work (aerial pho- Elements-at-risk on the higher surfaces should be tographs and maps) and �eldwork. Include all ele- described and their speci�c exposure to hydrogeo- ments that are on the fan surface as de�ned in Step 1 morphic events should be evaluated. Assessment and as re�ned in the �eld during this step. of human-made features requires an awareness not The information collected for each element should only of direct costs (e.g., repairing the railway fol- include: description, location, value (qualitative or lowing a debris �ow) but also the indirect costs (e.g., quantitative), a description of fan features in the loss of business revenues due to delays). Indirect vicinity of the element (e.g., immediately beside an costs can be very signi�cant, as in the case of trans- old channel), and a description of the exposure (e.g., continental rail lines, where one day of traffic closure summer residences only, or office building occupied may cost $10 million (Jacob and Hungr 2005b). 10 hrs/day). Identifying elements-at-risk early in the Common anthropogenic features are water intakes risk analysis is important so that the intensity of the (Figure 7), roads (including bridges), railways, remainder of the investigation can be adjusted ac- houses (Figure 8), institutional buildings, power cordingly. If the elements-at-risk are high value (e.g., transmission lines, and buried �bre optic lines. human safety), more effort should be applied than Some features such as drainage structures on the if the elements-at-risk are of lower value (e.g., forest fan may not just be elements-at-risk, but—if poorly stand values). designed to accommodate the water, sediment, and debris from naturally occurring hydrogeomorphic Step 2.1 Human Safety events—may increase the likelihood of negative impacts on the fan (Figure 9). If there is a history Potential injury or loss of human life is generally of elements affected by �oodwater or debris from considered the highest-value element in risk analyses hydrogeomorphic events, subsequent activities in (Wise et al. 2004). Therefore, the presence of humans the watershed may be blamed for any future damage. on the identi�ed fans creates a strong potential for Thus, documenting the fan history prior to proposed

8 Figure 7 A water intake structure at the apex of a fan.

Figure 8 This building was damaged by a debris ﬂow caused by a drainage diversion related to a forest road.

9 Figure 9 This drainage structure is not designed to accommodate the water, sediment, and debris from naturally occurring hydrogeomorphic events. This is an additional risk factor for forest management in the watershed. or existing watershed activities is critical, not only to tures may also exist and should be researched. With identify the hydrogeomorphic issues that need to be any feature, an understanding of why it exists on the addressed, but to provide a defence should impacts fan is important to correctly managing upslope areas occur that are unrelated to those watershed activi- to preserve the conditions. A common example is ties. where high-quality spawning gravels and �sh habitat are present in the stream channel on the fan. To pre- Step 2.3 Natural Features serve the natural conditions, management within the watershed should not lead to degradation of water Fish habitat and forest sites are common natural quality, changes in bedload movement, or signi�cant features on forested fans. Special habitats or rare fea- changes to peak stream�ows.

STEP 3 INVESTIGATE FAN PROCESSES

The fan surface is investigated from the oﬃce and in Step 3.1 Hydrogeomorphic Processes the �eld to identify the nature of hydrogeomorphic processes—their type, frequency, and disturbance The three principal hydrogeomorphic processes extent. This step can be undertaken while investigat- in�uencing fans—debris �ows, debris �oods, and ing the fan for elements-at-risk. Further information �oods—leave characteristic deposits on fan surfaces on the topics covered in this section can be found in (Table 1). Evidence of all three processes can be Wilford et al. (2005b) and Jakob and Hungr (2005a) found on some fans. For example, it is common for (particularly Chapters 2, 8, and 17). debris �ow fans to also experience debris �oods and

10 Table 1 Characteristics of hydrogeomorphic process deposits (after VanDine 1985; Smith 1986; Pierson and Costa 1987; Wells and Harvey 1987; Costa 1988)

Characteristics Flood Debris �ood Debris �ow Mode of deposition Grain-by-grain, dominated Rapid grain-by-grain En masse by traction processes aggradation from both (bouncing or rolling along suspension and traction the streambed) Strati�cation Massive or horizontal None or horizontal None strati�cation (with cross strati�cation strati�cation) Grading Variable: as a result of Frequently normal-graded None; reverse; reverse to sequential processes rather (coarse on bottom, �ne normal than a single process on top) Sediment characteristics Clast-supported with an Clast-supported, with Matrix-supported; rarely and texture open framework or predominantly coarse sand, clast-supported; very poor to distinctly �ner grained moderate to poorly sorted, extremely poor sorting;

matrix of in�ltrated sand; bmax typically <80 cm but extreme range of particle

rounded clasts; wide range may be larger sizes; bmax 60–230 cm, may of particle sizes; sorting contain megaclasts up to

from front to tail; bmax 400 cm <10 cm to >20 cm Orientation of clast Always perpendicular to Large cobbles to boulders are Variable, based on location long-axisa; �ow; usually well usually perpendicular to �ow; within �ow; parallel to �ow imbricationb imbricated pebbles to small cobbles are is most prominent; weak to usually parallel to �ow; weak no imbrication imbrication and collapse packing Landforms and deposits Bars, fans, sheets, and Similar to water �ood but Marginal levees, terminal splays; channels have large deeper deposits lobes, trapezoidal to width-to-depth ratio U-shaped channel a A clast (e.g., pebble, gravel, cobble) has three axes: “a” is the longest, “b” is the second longest, and “c” is the shortest. The “b” axis

determines the sieve size that the clast will pass through. The term “bmax” refers to the “b” axis of the largest clast that can be found in a deposit. b Imbrication is the shingling or overlapping of clasts with the upper edge of each clast inclined downstream, similar to a deck of tilting cards. Clasts are moved into this position by stream�ow.

�oods, and while �oods and debris �oods can occur (Costa 1984, 1988; VanDine 1985; Smith 1986; Pierson more frequently than debris �ows, debris �ows can and Costa 1987; Wells and Harvey 1987; Hungr et al. be far more damaging. Identifying hydrogeomor- 2001). Debris �ows in British Columbia commonly phic processes is a key element of the fan-watershed have peak discharges 2–50 times larger than the assessment because each process is the result of 200-year �ood (Jakob and Jordan 2001). Debris �ow watershed factors that need to be addressed when levees are steep-sided features that are often in paral- planning activities in the watersheds. lel pairs, marking the path of the debris �ow. Levees Debris �ows are homogeneous (single phase) are formed as a debris �ow traverses a fan—the core sediment–water mixtures similar to wet concrete, continues downslope while the slower-moving edges with sediment concentrations between 70 and 90% of the �ow come to rest as deposits of large-diameter by weight (or 47–77% by volume), and result in the sediment. Levee height is proportional to event �ow formation of marginal levees and terminal lobes depth, and is typically 1–2 m high.

11 Debris �oods, also referred to as hyperconcen- return period of events considered too dangerous trated �ows, occur when a greater volume of stream- for residential construction in British Columbia was bed materials is mobilized than in a �ood. As with 1 in 475 years or less (British Columbia Ministry of �oods, the mixing of sediment is not complete and Transportation and Highways 1996). there is a rapid increase in solids concentration toward the bed. Debris �oods have sediment concen- Step 3.2 Event Frequency trations between 40 and 70% by weight (or 20–47% by volume); sediment deposits are bars, fans, sheets, Investigations of hydrogeomorphic activity on fans and splays (Costa 1988; Hungr et al. 2001). While clearly indicate that events are neither rare nor debris �oods move more and larger sediment than extreme (Innes 1985; Jakob and Jordan 2001). To �oods, peak discharges are at most 2–3 times that illustrate this point, evidence of recent hydrogeo- of major �oods (e.g., 100-year �ood) (Hungr 2005). morphic activity was investigated with a random Debris �oods commonly have a maximum particle sample of 51 fans in the Bulkley Timber Supply Area, size of the same order as the peak depth of major an area with a mix of plateaus and mountainous �oods—for example, up to several tens of centime- terrain. There had been little or no human activ- tres (< 1.0 m) in the case of typical small mountain ity in the watersheds of the sample fans. While this streams (Hungr et al. 2001). area would not be characterized as overly unstable, Floods have sediment concentrations between hydrogeomorphic activity was observed on 41 fans, 1 and 40% by weight (or less than 20% by volume); indicating that 82% of the fans had disturbance sediment deposits are bars, fans, sheets, and splays occurring on at least a portion of the fan surface in (Wells and Harvey 1987; Costa 1988; Hungr et al. the last 50 years. A similar study was undertaken 2001). Characteristically the bars, sheets, and splay on the southern British Columbia coast, and recent deposits from �ood events are not as extensive as hydrogeomorphic activity was found on 49 of 55 fans debris �ood deposits, are better sorted, and have (89%) (Millard et al. 2006). On the other hand, in the a sediment size that is not as large. For example, drier southern interior of the province there are wa- a �ood in a 2 m wide stream channel may deposit tershed-fan systems that have recently experienced sediment several metres each side of the channel, hydrogeomorphic activity on their fans outside of while a debris �ood may deposit sediment 10 metres stream channels, after several hundred years of inac- each side of the same-sized channel. tivity (Giles et al. 2005). These longer return period Fans can be formed from one or a combination systems, with events often related to widespread of these three processes, any of which may no longer forest health issues or severe forest �res, can result in be actively forming the fan. For example, the major underestimation of the hydrogeomorphic hazards. portions of many fans in British Columbia were Diﬀerent methods can be used to determine the formed as the glacial ice was melting approximately frequency of events. In populated areas it is common 10 000 years ago, and there was abundant meltwater to �nd records of events in local archives (Septer and and exposed, unconsolidated sediment. These are Schwab 1995). Dendroecology—using scars, dating referred to as “paraglacial fans” (Ryder 1971a, 1971b), cohorts of trees, and analyzing tree ring widths for and sediment deposits on the fan surface may not abrupt, long-lasting growth changes—is a useful re�ect contemporary processes. For example, if a method of establishing a year, and in some cases a fan does not have recent evidence of debris �ow season, of events (Strunk 1997). However, if recent activity, but debris �ow levees are present on a fan events have disturbed more of the fan surface, it is surface with 500-year-old trees and well developed common to lose some or all of the “tree record” of soil horizons, it is likely that the deposits re�ect older events. This method requires a degree of prepa- processes that are not currently active in the fan’s ration of increment cores from trees—sanding and watershed. However, caution must be used when microscopic identi�cation of growth rings to identify deciding whether evidence is no longer re�ective of growth changes. For a more detailed description of contemporary hydrogeomorphic processes. Jakob dendroecology techniques refer to Dendroecology—A and Jordan (2001) identify return periods of up to Guide for Using Trees to Date Geomorphic and Hy- 500 years for some debris �ows, and, historically, the drologic Events (Wilford et al. 2005a).

12 Dendroecology can be as simple as determining situations the best one can do is determine the time a possible date for a hydrogeomorphic-event deposit since the last event—because evidence of previous on a fan by determining the stand age on top of the events has been removed by subsequent events. In deposit. In �re-dominated ecosystems, the lack of these situations, the time estimate is an estimate of evidence of a stand-establishing �re on a fan sur- the event return period (T), or the number of years face (e.g., burnt stumps and logs and ash in the soil) between event occurrences. suggests that the stand was established following a hydrogeomorphic event, and the stand age will cor- Step 3.3 Event Magnitude respond closely to the age of the event (Grainger and Wilford 2004). Where there is evidence of stand- The disturbance extent of hydrogeomorphic events is establishing �re on a fan surface, it is difficult to an indication of the power and volume of water and/ determine how many stand rotations may have or material that issues from a watershed. The distur- occurred on the deposit, and the deposit may be bance extent of hydrogeomorphic events on forested hundreds (if not thousands) of years old. fans can be determined from aerial photographs and Trees can be partially buried by sediment in the �eldwork. Evidence on aerial photographs can range hydrogeomorphic riparian zone, and will not die if from strips of bare sediment to strips of century-old the roots have sufficient oxygen and water. Charac- trees (cohorts). However, some evidence may not teristically these trees do not have a basal or “butt” be apparent on aerial photographs due to the nar- �are, but the �are can re-establish over a period of row swath cut through the forest, and some events time. While lack of butt �are and recent deposits of may not be powerful enough to clear forests—sedi- sediment are indications of hydrogeomorphic activ- ment is spread on the fan surface beneath trees, and ity, it is possible for trees to recover after a period �eldwork is necessary to identify these disturbances. of time (as little as 25 years) and re-establish basal A zone of potential disturbance outside the stream �are. Dendroecology is a useful means of establish- channel on a fan is de�ned as the hydrogeomorphic ing whether trees have in fact been buried and have riparian zone. This zone can range from a few metres recovered a “normal” form (e.g., there will be an wide to hundreds of metres wide. For a more de- abrupt reduction in radial growth rings immediately tailed discussion of identifying hazardous zones on following the event, and continuing reduced growth fans refer to Forest Management on Fans—Hydrogeo- for a period of 10 or more years, then a gradual re- morphic Hazards and General Prescriptions (Wilford lease as the tree establishes a new root system et al. 2005b). just below the soil surface). Incised stream channels on fans may give the On �ood fans, individual events may be less im- impression that hydrogeomorphic events will not portant than the overall �ood and sediment regime. in�uence the fan surface. However, depending upon The spatial extent and complexity of the channel the con�guration of the channel and banks, debris network on the fan can be used to assess factors such �ows can deposit up to 6 m of material in a single as sediment supply or the type of vegetation on the event, resulting in signi�cant disturbance beyond fan (Millard et al., in press) an incised channel (Osterkamp and Hupp 1987). The fan surfaces identi�ed in Step 1 may include Evidence of over-bank disturbance from hydrogeo- elevated surfaces that have not been in�uenced by morphic events and channel conditions should be contemporary hydrogeomorphic processes. On for- explored prior to assuming that elements are not at ested fans this can be evident in lower site produc- risk. Channels with a high likelihood of debris jam- tivity on the elevated surfaces of some fans (Wilford ming may require an extra degree of caution when et al. 2006); however, site productivity may not vary developing forest management recommendations. in some climatic regions of British Columbia. In The natural disturbance extent of hydrogeomor- those situations, differences in stand age and soil phic events can also be modi�ed by anthropogenic pro�le development may indicate a lack of contem- features and practices (e.g., riparian harvesting). porary hydrogeomorphic disturbance. Some features such as dikes tend to convey the The output from this part of the fan investigation disturbance downstream to lower positions on the should be the identi�cation of how often a particu- fan; however, if features are constructed of local lar type of event has occurred in the past. In some streambed materials without armouring, they are

13 more likely to be eroded and fail. On fans, both The outcome from this part of the investigation inadequate drainage structures and roads that climb should be an estimate of the probable magnitude, in elevation toward stream channels can increase the runout zone, power, and potential damaging nature extent of disturbance (Wilford et al. 2003). Removal of the event to pair with the frequency of occurrence of the hydrogeomorphic riparian zone through determined in Step 3.2. It is not uncommon that forest harvesting, or through agricultural or resi- more than one kind of event has occurred on the fan, dential clearing, increases the potential for greater or on different parts of the fan, and that these sepa- disturbance on the fan surface (Wilford et al. 2003). rate types have different frequencies, and different The fan investigation should document the poten- implications for management in the watershed (see tial hydrogeomorphic implications of any existing the Eagle Summit Creek case study in Appendix 2). anthropogenic in�uence.

STEP 4 INVESTIGATE WATERSHED PROCESSES

The fan investigations provide an understanding of hydrologic �oods, which are the most common type past hydrogeomorphic processes that formed the of �ood in British Columbia. Forests can play an fan, and thus identify the processes produced by the important role in hydrologic �oods, particularly in watershed. Investigating the watershed above the fan snowmelt-dominated watersheds where forest har- is necessary to understand watershed characteris- vesting or natural deforestation (e.g., mountain pine tics, how they are linked to the processes observed beetle or wild�re) can alter snow accumulation and on the fan, and how that linkage creates the exist- melt rate processes (Winkler et al. 2005), potentially ing hydrogeomorphic hazard. Only then one can resulting in increased peak �ows. Roads, trails, and determine the degree to which planned development soil compaction (e.g., landings and other compacted activities in the watershed could in�uence those areas) can also in�uence �ood generation (Harr et processes and potentially alter the existing hazard, al. 1979; Wemple and Jones 2003). Equivalent clearcut creating additional or incremental hydrogeomorphic area (ECA) and percent of the watershed in roads and hazards. compacted areas are two important watershed indica- A series of avenues are pursued when investigat- tors where �ood impacts to fans are possible (BCMOF ing a watershed—including both oﬃce work and and BCMOE 1999a). However, attention must be paid �eldwork. The watershed investigation is similar to to mass wasting because land use–related landslides that undertaken using the Forest Practices Code Wa- can increase sediment delivery to �ood fans, result- tershed Assessment Procedures (BCMOF and BCMOE ing in channel avulsions and fan destabilization (see 1999a) with a focus on geomorphic processes issuing Shale Creek Case Study presented in Appendix 3). from the watershed, particularly debris �ows and Dam-release �oods are caused by a rapid release debris �oods. of large volumes of stored water such as moraine, snow avalanche, or debris �ow dam ruptures, glacial Step 4.1 Watershed-fan Process Linkages lake outbursts, beaver dam failures, and downstream dilution of debris �ows and debris �oods. Dam-re- This section discusses how watershed-initiated lease �oods are relatively rare, but they do occur in hydrogeomorphic processes are connected to the British Columbia. There is a record of glacial out- fan and which watershed processes could lead to burst �oods occurring in northern British Columbia increased or incremental hydrogeomorphic hazards. (e.g., Alsek River, Tulsequah Lake, Salmon, and Subsequently, this identi�es an appropriate manage- Bear Rivers near Stewart, and the Kitimat River) ment focus (presented in Table 2 and in italics in the (Septer and Schwab1995); however, with the melt- descriptions of the hydrogeomorphic processes). ing of alpine glaciers, the occurrence could increase (Geertsema and Clague 2005). Landslides that block Flood stream channels and subsequently release rapidly There are two basic types of �oods: hydrologic and have been documented throughout British Columbia dam-release. Rainfall and/or snowmelt generate (Jakob and Jordan 2001). Beaver dams may fail as the

14 Table 2 Forest management focus for different hydrogeomorphic processes. Key issues are bolded.

Process Initiation Management focus Flood Runoff ECA – harvest, �re, forest health Road density Fire – reduced soil in�ltration Landslides – increasing sediment load to channels on the fan (in�lling channels) Dam rupture Dam rupture – beavers, road drainage structures Landslide dams – harvest and roads (coast), gentle-over-steep (interior) Debris �ood Runoff ECA – harvest, �re, forest health Road density Fire – reduced soil in�ltration Landslide Harvest and roads on unstable terrain and gentle-over-steep in the interior Dam rupture Dam rupture – beavers, road drainage structures Landslide dams – harvest and roads (coast), gentle-over-steep (interior) Debris �ow Landslide Harvest and roads on unstable terrain and gentle-over-steep in the interior Runoff—in-channel ECA – harvest, �re, forest health initiation Road density Fire – reduced soil in�ltration result of a variety of processes, including high-inten- Debris �ood sity precipitation, rapid snowmelt, animals burrow- There is a continuum from water �oods through ing through the dam, human destruction of portions debris �ows, with debris �oods being an intermedi- of dams, lack of maintenance by beaver, and collapse ary process. Debris �oods should be viewed as an of upstream dams (Hillman 1998; Butler and Malan- extraordinary form of �ood in which an extremely son 2005). A dam-release �ood can exceed a design large volume of sediment with a wide range in grain stream�ow �ood (e.g., 100-year return period) by size is moved (essentially the whole streambed) and a factor of up to 100 (Jakob and Jordan 2001). The deposited in a short period of time (Smith 1986). potential for dam-release �oods should be explored, Where debris �ows have been observed closely, some particularly where landslide-prone terrain has previ- evolve into debris �oods as they proceed down- ously, or could potentially, produce a landslide dam. stream due to addition of water from tributaries In addition to local knowledge regarding past dam- (Pierson and Scott 1985). Thus the factors that lead to release �oods, forest cover on a fan may contain both debris �ows and �oods should be explored in de- dendroecological evidence of past events—cohorts bris �ood watersheds—increase in sediment delivery of trees and damage to older trees (Wilford et al. to streams and the potential to increase peak stream- 2005a). �ows, and sediment and debris loading in stream channels and gullies.

15 Debris �ow Step 4.2 Office Investigations Valley-con�ned debris �ows in humid temperate forests characteristically have three forms of initia- The �rst phase of office investigations is to compile tion (Takahashi 1981). The most common in British information related to watershed processes, to gain Columbia is by impulse loading of channels, where an overview of the watershed, and to focus subse- an open-slope failure (debris slide or debris ava- quent �eldwork. lanche) that enters a steep stream channel destabi- lizes sediment in the channel and proceeds down Morphometrics channel by the force of gravity as a debris �ow (Sidle In mountainous or “graded” watersheds, topo- et al. 1985; Jordan 1994; Millard 1999). This highlights graphic maps or GIS analyses can be used to generate the need to identify landslide-prone terrain that is several basic topographic measurements that can be tributary to steep gullies and stream channels and used to provide a �rst approximation of the hy- to manage the hazards appropriately. The two other drogeomorphic processes. The relative relief number forms of initiation occur during a major runoff event is watershed relief divided by the square root of wa- with the mobilization, or bulking, of sediment and tershed area (Melton 1965). The relative relief number debris stored in the stream channel, or the rupture has been used in various locations to differentiate of a debris jam. Particularly in snow-dominated wa- debris �ow, debris �ood, and �ood watersheds (Table tersheds, in-channel mobilization of sediment can be 3). There are some cautions when using this ap- affected by changes in peak �ow runoff due to changes proach. First, it is only an approximation and should to the forest canopy—as a result of wild�res, insect not be used to guide management decisions in the infestations, or forest management activities includ- watershed unless it is supported by �eld observations ing road density. and the detailed fan investigation undertaken in Step All three forms of initiation may occur following 3. Second, research shows regional variation within extreme climatic inputs, which are assumed to occur British Columbia, and so this approach should be randomly. While climatic events are essential (Septer used cautiously, particularly in areas not previously and Schwab 1995), other conditions must be present studied. Third, the method is designed for “graded” for either an open-slope failure or mobilization of mountain watersheds and does not apply to water- channel sediments (Miles and Kellerhals 1981; Dagg sheds in plateau terrain. For example, the Yard Creek 1987). For example, if there is little or no sediment watershed near Sicamous is 100 km2 with a large and debris load in a channel, an in-channel debris low-gradient upland plateau. The relative relief num- �ow is less likely to occur. A recent debris �ow may ber predicts that the dominant hydrogeomorphic have removed sediment and debris from a channel, process will be �ooding. However, there are exten- and the time to the next signi�cant event generally sive debris �ow deposits at the fan apex, originating depends on how rapidly the sediment load in the from a small steep-gradient tributary just above the channel is recharged. The exception is where very fan, which cannot be identi�ed except through aerial large open-slope landslides enter stream channels photograph interpretation and �eldwork. While the and have enough material to develop into signi�- large fan is formed predominantly by �oods, forest cant debris �ows regardless of the rate of sediment management planners must be aware that debris entrainment along the pathway. The rate of sedi- �ows can originate from a particular part of the ment delivery to channels varies widely between watershed and that these hazards must be managed watersheds, depending on geology and vegetative accordingly. cover. Some channels experience rapid sediment recharge and debris �ows can occur more than once Sediment production and movement: surface a year, while others require centuries or millennia erosion, landslides, and stream channels after an event to recharge channel sediment enough The general location and nature of sediment sources to experience another debris �ow. A key factor in in a watershed can be determined from aerial photo- determining debris �ow potential is to examine the graphs, topographic maps, and interpretative maps. sediment and debris loading in stream channels and Speci�c details on sediment production are explored gullies that is available for entrainment by an event, during �eldwork (e.g., sediment textures, connectiv- and to understand the hillslope-channel conditions in ity to stream channels, causes of landslide initiation). a watershed.

16 Table 3 Predictive models for dominant hydrogeomorphic processes using the relative relief number (RRN) and watershed length (the distance from the fan apex to the most distant point on the watershed boundary)

Hydrogeomorphic process Class limits Source

Flood RRN < 0.30 Jackson et al. 1987 Wilford et al. 2004b Millard et al. 2006 Debris �ood and debris �ow RRN > 0.30 Jackson et al. 1987 Debris �ood RRN > 0.30 and Length > 2.7 km Wilford et al. 2004b Debris �ood RRN 0.30–0.60 Millard et al. 2006 Debris �ow RRN > 0.60 and Length < 2.7 km Wilford et al. 2004b Debris �ow RRN > 0.52 Bovis and Jakob 1999 Debris �ow RRN > 0.60 Millard et al. 2006

Aerial photographs provide a perspective on both runoﬀ (Scott and Van Wyk 1990; Moody and where landslides have occurred and the connectivity Martin 2001) and debris �ows (Jordan et al. 2004) of the landslide-prone terrain to stream channels. (Figure 10). The implications of hydrogeomorphic Comparison of recent and historic aerial photo- impacts to fans can be signi�cant. graphs is useful, particularly in situations where Naturally unstable areas in the lower portion of a forest growth obscures the evidence of landslides. watershed have been found to more strongly in�u- Topographic maps are used to identify speci�c ence the nature of hydrogeomorphic processes (e.g., steep slopes and gentle-over-steep situations, and to more sediment can be delivered to fans during a provide indications of connectivity of hillslopes to hydrogeomorphic event) than unstable areas that are stream channels. GIS analysis can be used to identify located further from the fan in the headwaters of the steep slopes that could be prone to landslide activity; watershed (Wilford et al. 2005b). however, this approach is strengthened when other Aerial photographs are used to identify stream factors such as slope water movement and landslide reaches. The Channel Assessment Procedure Guide- history are included in the modelling analysis (Pack book (BCMOF and BCMOE 1996a) is a useful refer- et al. 1998). ence. If the channels are wide enough it is possible Sur�cial geology and terrain maps provide to identify the general nature of the bed materials information on the nature of sur�cial materials in and channel form (e.g., Hogan 2008 notes 30 m wide the watershed. This is particularly useful in areas channels with 1:50 000 aerial photographs down to with lacustrine deposits that have a high potential to 10 m wide channels with 1:5 000 aerial photographs). degrade water quality—either in the natural setting Aerial photographs are also used to identify gullied or as a result of land use activities. Terrain stability terrain and determine the degree of connectivity maps identify landslide hazards, but are generally between the gullies and stream channels. The ability restricted to either operating areas or the timber of stream channels to transport sediment is a key harvesting land base (i.e., the maps do not identify factor in determining the downstream consequences slope stability issues in alpine areas, which in some of sediment that enters a stream channel (Hogan and cases can be signi�cant, as described by Geertsema Wilford 1989). et al. 2006). Forest �re severity maps of burned areas are valu- Flood generation: forest cover and road density able in identifying areas with reduced soil in�ltra- Forest cover maps are used to identify the extent and tion—water repellency, removal of surface cover, and location of natural and anthropogenic disturbance, soil sealing (Larsen et al., in press). These soil condi- including wild�res, forest health issues, and past for- tions have resulted in signi�cant increases in rainfall est harvesting. These maps are also used to identify

17 Figure 10 A debris flow initiated as a result of hydrophobic soils in the Kuskanook watershed, damaging homes and the highway on the fan. (Source: D. Boyer, B.C. Ministry of Environment.) potential changes to forest cover, such as the spread and skid trails). Research has identi�ed that as little of mountain pine beetle. Speci�c issues relating to as 5% of a watershed in roads and compacted areas forest health should be explored, as the implications can in�uence peak �ows (Harr et al. 1979; King and to forest cover can be signi�cant in both immature Tennyson 1984). The value determined in the office and mature stands (Woods et al. 2005). The Veg- may be an overestimate that can be re�ned during etation Resource Inventory (VRI) is the source of �eldwork. For example, deactivated roads and spe- information on forest cover; however, data are gener- ci�c road segments may not be in�uencing hillslope ally not entered on the heights of regeneration until hydrology and peak �ow generation (Wemple and the openings have been declared free-to-grow—a Jones 2003). period of up to 20 years. Data on the extent of forest disturbances and the subsequent regeneration Step 4.3 Field Investigations are used to calculate the equivalent clearcut area (ECA)—a measure of hydrologic disturbance to the Fieldwork in the watershed is undertaken to explore forest cover (BCMOF and BCMOE 1999a). The location key factors identi�ed in the office work, particularly of hydrologically recovering stands is important in to verify data and collect information that cannot be snowmelt-dominated watersheds due to the more generated in the office. Fieldwork is undertaken at rapid snowmelt in openings and immature stands both the watershed scale and the proposed develop- (Winkler and Roach 2005). Snowmelt from higher- ment scale. The watershed scale explores the key elevation disturbed areas can synchronize with processes identi�ed in the previous �eld and office lower-elevation melt, increasing peak stream�ows. work. A �eld review of the area under plan (pro- Conversely, accelerated low-elevation snowmelt posed roads, harvesting, and other developments) could desynchronize runoff, tending to decrease should focus on areas identi�ed during the office peak stream�ows. work as potentially problematic—surface erosion Road density can be determined using GIS. Ad- and landslide-prone terrain in particular. This will ditional areas should be included in the calculation verify whether the situations are problematic and of compacted areas in the watershed (e.g., landings lead to the identi�cation of management options.

18 Sediment production and movement: stream identi�ed as a factor for hydrogeomorphic processes. channels, gullies, landslides, drainage structures, It is likely that the VRI information is not current, and surface erosion and this can have signi�cant implications for the An overview of the channel network is undertaken calculation of equivalent clearcut area. to describe key features in the reaches identi- Forest health issues identi�ed in the office should �ed from aerial photographs. Descriptions should be examined in the �eld from two perspectives. It is include bed and bank materials, riparian condition, important to integrate the current hydrogeomorphic and an assessment of sediment and debris loading. A implications of forest health issues as well as the useful aid is the Channel Assessment Procedure Field potential future issues. This is particularly relevant Guidebook (BCMOF and BCMOE 1996b). The ripar- in watersheds that have a high potential impact to ian zone can be explored during this �eldwork for the forest cover (e.g., mountain pine beetle). In wa- evidence of hydrogeomorphic activity (e.g., dendro- tersheds subject to signi�cant forest health issues, it ecology). is likely that there will be hydrogeomorphic conse- Gullies identi�ed during the office work should be quences, regardless of human activities. examined to identify sediment and debris loading, Areas that have had intense forest �res in the past evidence of debris �ow activity, and connectivity to 5 years should be examined to determine the degree the stream system. While the Gully Assessment Pro- of reduced soil in�ltration and explored for evidence cedure (BCMOF and BCMOE 2001) was developed for of erosion or overland �ow. The issue of hydrophobic coastal British Columbia gully systems, the approach soils has only recently become apparent in British is worthy of consideration for assessing gullies in Columbia, but the hydrogeomorphic consequences other areas of the province. have been signi�cant (Jordan et al. 2004). An overview of the road system, including non- A �eld examination of the road system identi�es status roads, is needed to address linkages to all the degree to which roads are in�uencing hillslope three hydrogeomorphic processes. The age of the hydrology and peak �ow generation. Emphasis in the road system could provide useful information, as �eld should be on identifying the degree of deactiva- construction standards changed signi�cantly with tion of non-active roads (i.e., determine if drainage the introduction of the Forest Practices Code in 1995 structures are still in place, and whether roads have (Horel 2006). It is necessary to identify past and compacted surfaces that are in�uencing water move- potential landslide risks from road systems, and the ment) and the quality of road cross-drainage (i.e., potential linkage to stream channels. identifying if ditchlines are collecting subsurface Drainage structures should be examined across �ows and delivering the water to streams rather than stream channels or gullies where hydrogeomorphic back onto the forest �oor). The office determination hazards have been identi�ed. The issue to assess is of percentage of the watershed in roads and com- whether the structures will pass or block potential pacted areas should be re�ned with this �eldwork. events, and determine the consequences. If an element-at-risk in a watershed is domes- Step 4.4 Synthesis of Watershed Processes tic water production, �eldwork should explore for evidence of surface erosion issues. Selective sampling Compiling the �eld and office information will give for surface erosion may be advisable (e.g., Carson et an overview of the current and projected hydrogeo- al. 2007). A challenge to be aware of is that evidence morphic situation in a watershed that could affect of past surface erosion can be masked by revegeta- the watershed-fan system and potentially increase tion, road reconstruction, or grading and surface the watershed hazard. maintenance. The Wathl Creek Case Study (Ap- The next step is to develop or review plans in light pendix 1) also explored other issues associated with of the potential for proposed activities to in�uence drinking water—contamination from wood leachate, those watershed processes, and assess the potential hazardous materials, and wildlife. risks from those activities to elements on the fan. This would include updatingECA calculations, Flood generation: forest cover and roads reviewing harvest locations for the potential to The status of regeneration should be examined, par- synchronize or desynchronize snowmelt runoff, ticularly in watersheds where peak �ows have been identifying road and harvesting location relative to

19 landslide-prone terrain, projecting road density, and cover in a watershed, it is important to develop a exploring any other activities that could increase forecast of background changes that can be antici- identi�ed hazards in the watershed. If forest health pated regardless of any new planning initiatives. or other natural issues are in�uencing the forest

STEP 5 ANALYZE RISKS AND DEVELOP PLANS

In Step 5, all the previously gathered information is this means understanding both the existing risks integrated in an analysis of existing and potential and the incremental risks as a result of proposed or incremental risks in the watershed-fan system. This imposed forest management activities in a water- understanding is then used to guide management shed. decisions concerning natural impacts to the forest Risk assessment includes evaluating the results cover (�re, forest health) or proposed (harvesting, of the risk analysis and determining whether that road) activities in the watershed. Risk de�nitions risk is acceptable. Generally this entails compar- and procedures in this section are consistent with ing the risk analysis results with public, corporate, Land Management Handbook 56, Landslide Risk and/or legislative standards of acceptable risk. Risk Case Studies in Forest Development Planning and assessment can include hydrogeomorphic hazard Operation (Wise et al. 2004), which can be refer- specialists and other experts, but the �nal decision enced for further discussion of these topics. regarding risk acceptability is generally the responsibility of land managers and regulatory decision Step 5.1 Understanding Risk Analysis makers. Under the British Columbia Forest and and Risk Assessment Range Practices Act (2006), the determination of acceptable risk, and the decision to proceed with any Risk is a measure of the probability of a speci�c forest management activities on the basis of that risk event occurring, and the consequence, or adverse assessment, is the responsibility of forest licensees eﬀects, of that event on speci�c elements, such as and BC Timber Sales. human health, property, or the environment. Simply If the risk assessment determines that the estimat- stated, risk is the product of hazard and conse- ed risks are unacceptable or intolerable, two options quence. are available: cancel further plans, or identify risk Risk analysis includes determining hazards and control and mitigation measures to reduce the haz- consequences. Hazards are the potentially damaging ards and/or reduce the consequences. Identi�cation events and the hazard analysis includes information of measures involves hydrogeomorphic hazard spe- such as the process, magnitude (including spatial cialists, who can determine the eﬀectiveness of risk extent), and likelihood of occurrence. Consequence control and mitigation measures. With this informa- refers to the likelihood of damage or losses to an tion, land managers can revisit the risk assessment element-at-risk in the event of a speci�c hazard; in light of the control and mitigation measures and analysis of the consequence includes the spatial their associated costs. and temporal exposure and the vulnerability of the The �ve-step procedure presented in this hand- elements-at-risk. Consequence can also include the book addresses risk analysis and risk control meas- value of losses or damage to elements. The concepts ures. For a further discussion of risk assessment and of risk, hazard, and consequence are useful tools for acceptable risk, the reader is referred to Cave (1992), land managers to employ, but completing hydrogeo- Fell (1994), Wise et al. (2004), and APEGBC (2006). morphic risk analyses is the realm of quali�ed professionals. Step 5.2 Consequence Risk analysis includes using some process to combine the hazards and consequences to estimate the In risk analysis it is common to analyze hazards risk to elements. In the forest management context prior to analyzing consequences. However, within

20 the forestry context it is common for managers to For the risk analysis process described in this want to know what the potential consequences are handbook, the consequence value should be de- �rst, because the level of consequence frequently scribed qualitatively (low, moderate, high). Other identi�es the level of effort needed in all stages of more quantitative procedures are possible, as de- the risk analysis. This approach also ensures that the scribed by Wise et al. (2004). consequence, which can be far away from the area managed, is not overlooked or paid less attention Step 5.3 Hazard than it deserves. The development area is generally investigated for hazards, but the consequence area, if In this analysis, hazard is de�ned as the probability left until a later planning stage, can mean that work or likelihood of occurrence of a hydrogeomorphic in the development area is either wasted or misdi- event of a speci�c process and magnitude, and with rected. Thus, in keeping with the approach taken in a speci�c power and spatial disturbance extent. It the forest sector, consequences are discussed prior to can be expressed as a quantitative probability with hazards. a value between 0 and 1, or qualitatively a value be- Consequence considers the effects of the speci�c tween very low and very high. hazardous hydrogeomorphic event on the elements- The investigations described in Steps 3 and 4 at-risk on the fan (probability between 0 and 1, or should determine the type, magnitude (size, power), qualitatively low, moderate, high). This can include and runout or impact area of an expected event, and an analysis of the likelihood the element will be at the the frequency (how often the event is likely to occur site at the time the event occurs, which is sometimes over time). This describes the existing hydrogeomor- called the temporal exposure. For �xed elements, phic hazard, which is the result of natural climatic, such as houses, highways, or forests, this probability geologic, vegetation, and hydrologic watershed is always equal to 1 and does not need to be analyzed conditions, as well as any existing human land-use further. For highway and railway traffic, or human changes in the watershed. Its likelihood of occur- occupation of residences or institutions, there is gen- rence can be expressed as the return period T, or the erally a temporal exposure probability of < 1. frequency (annual probability) Pa, which is simply Vulnerability of a given element will vary de- 1/T. pending on the severity of the hydrogeomorphic Section 5 describes the watershed processes con- event and the sensitivity of the element. Damage can tributing to the various hydrogeomorphic hazards. be minor, partial, or total, and this can be expressed An understanding of the existing hazards is the �rst quantitatively (between 0 and 1) or qualitatively step in analyzing the potential hazards that could (low, moderate, high). This determination is gener- result from proposed or naturally imposed activities ally a joint effort involving hydrogeomorphic hazard in the watershed, as also described in Section 5 and specialists and other appropriate specialists such as summarized in Table 2. �sheries biologists, foresters, and highway or railway Following the procedures described there, one transportation experts, depending on the elements should arrive at a description of the development- potentially at risk. related, or incremental hydrogeomorphic hazard, Element exposure and vulnerability can be com- which should include event type, magnitude, and bined to further re�ne the consequence, as described frequency, as a result of a particular forest activity. in more detail by Wise et al. (2004). The information It is important to note, however, that the probability gathered in Steps 2 and 3 of this process should allow of occurrence of forest development-related hazards such further re�nements, if necessary. will be different than the existing natural hazard, Risk analysis can include estimating the worth of because forest development effects usually have a individual elements to produce a monetary conse- relatively limited time frame compared to natural quence value. This can be expressed quantitatively hazards. For example, the hydrological effects of as a numerical dollar value, or qualitatively as a low, harvesting, �res, or insect-related mortality decrease moderate, or high dollar value. A qualitative ap- over several decades as the stand re-grows (BCMOF proach may be more appropriate for some elements, and BCMOE 1999a); roads can have a speci�ed use such as species at risk. Human lives should be as- period after which they are deactivated, and poten- signed the highest possible value in any consequence tial hazards decrease relative to the level and efficacy rating scheme. of the deactivation; and hazards associated with

21 hydrophobic soils in severe burns generally decrease Table 4 shows the long-term probabilities signi�cantly after several years. A recent study of calculated using Equation 1, for a range of annual landslides on northern Vancouver Island found that probabilities of events and a range of life spans of the the greatest frequency of landslides from harvested watershed activities contributing to those events. cutblocks occurred in the �rst 5 years following har- Table 5 summarizes the ranges of quantitative vesting (Horel 2006). probabilities from Table 4, described by each qualita- The long-term probability of a hydrogeomorphic tive hazard term, and de�nes those qualitative terms. event (Px) resulting from a particular forest develop- Continuing the previous example, there is a prob- ment can be calculated if the frequency, or annual ability of 0.33, or a high likelihood, that a particular probability of occurrence (Pa) and the duration of 20-year road would initiate a debris �ow of a par- effect of the forest development (x) are known, from: ticular magnitude and runout extent within that 20-year period. This is the incremental (watershed x Px = 1-(1-(Pa)) Equation 1 activity–related) hazard at that site. The �nal step in the hazard analysis is to assign a For example, if previous experience suggests that qualitative value to the event magnitude, based on its a road in a particular landscape position is expected power, which is a measure of its likelihood of causing to result in one landslide every 50 years, and that damage, and its extent, which is a measure of how landslide could initiate a debris �ow in the channel large an area and what risk elements it could affect. it affects, the forest road–related debris �ow hazard Table 6 is an example of a hazard analysis matrix. has a return period (T) of 50 years and an annual Assuming the debris �ow in our example would frequency (Pa) of 1/50= 0.02. If the road will be deac- reach risk elements on the fan, one could conclude tivated to remove the landslide hazard after 20 years that, since debris �ows are generally high-power (x), the long-term probability of the forest road– destructive events and their extent includes risk related debris �ow (Px) is: elements, the event magnitude would be considered high. Therefore, with a high frequency and a high 20 P20 = 1-(1-(0.02)) = 0.33 Equation 2 magnitude, the debris �ow hazard in this case would be very high. Therefore, there is a probability of 0.33 that a debris �ow will be initiated during the 20-year life of the road.

Table 4 Example of long-term probabilities (adapted from Wise et al. 2004, Table A4.3)

Pa = Px, long-term probability of occurrence: Px = 1-(1-Pa)x annual x = life of watershed activity (years) prob 1 2 5 10 20 25 50 100 200 250 500 1000 2000 2500 1/1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1/2 0.50 0.75 0.97 1 1 1 1 1 1 1 1 1 1 1 1/5 0.20 0.36 0.67 0.89 0.99 1 1 1 1 1 1 1 1 1 1/10 0.10 0.19 0.41 0.65 0.88 0.88 0.93 0.99 1 1 1 1 1 1 1/20 0.05 0.10 0.23 0.40 0.64 0.72 0.92 0.99 1 1 1 1 1 1 1/50 0.02 0.04 0.10 0.18 0.33 0.40 0.64 0.87 0.98 0.99 1 1 1 1 1/100 0.01 0.02 0.05 0.10 0.18 0.22 0.39 0.63 0.87 0.92 0.99 1 1 1 VH 1/200 0.01 0.01 0.02 0.05 0.10 0.12 0.22 0.39 0.63 0.71 0.92 0.99 1 1 1/250 0 0.01 0.02 0.04 0.08 0.10 0.18 0.33 0.55 0.63 0.87 0.98 1 1 1/500 0 0 0.01 0.02 0.04 0.05 0.10 0.18 0.33 0.39 0.63 0.86 0.98 0.99 1/1000 0 0 0 0.01 0.02 0.02 0.05 0.10 0.18 0.22 0.39 0.63 0.86 0.92 1/2000 0 0 0 0 0.01 0.01 0.02 0.05 0.10 0.12 0.22 0.39 0.63 0.71 1/2500 0 0 0 0 0.01 0.01 0.02 0.04 0.08 0.10 0.18 0.33 0.55 0.63 1/5000 0 0 0 0 0 0 0.01 0.02 0.04 0.05 0.10 0.18 0.33 0.39 VL L M H

22 Table 5 Qualitative frequency deﬁnitions (adapted from BCMOF and BCMOE 2002, Table A10.2)

Quantitative Probability Qualitative Hazard Qualitative Description

Px>0.64 Very high Event is imminent after the watershed activity (i.e., any phase of forestry operations)

0.18

0.04

0.02

Px≤0.02 Very low Very remote likelihood of event during the lifetime of the watershed activity

Table 6 Example of qualitative hazard analysis matrix Table 7 Qualitative risk analysis matrix (adapted from Wise (adapted from Wise et al. 2004) et al. 2004)

Magnitude Consequence Frequency High Moderate Low Hazard High Moderate Low Very high Very high Very high High Very high Very high Very high High High Very high High Moderate High Very high High Moderate Moderate High Moderate Low Moderate High Moderate Low Low Moderate Low Very low Low Moderate Low Very low Very low Low Very low Very low Very low Low Very low Very low

Step 5.4 Risk Analysis Step 5.5 Assessing Risk and Making Management Decisions The �nal step in the risk analysis procedure is to combine the hazard and consequence to arrive at an With the results of the risk analysis and technical estimate of the hydrogeomorphic risks from speci�c support where necessary, it is the role of land manag- watershed activities to speci�c elements on the fan. ers to complete the risk assessment and determine if Table 7 is an example of a qualitative risk matrix; the estimated risks are acceptable or tolerable. Three other matrices are possible. In this way, the qualita- options are usually available: tive consequence and risk values determined in the previous sections are combined to yield a qualitative 1. the risks are considered to be unacceptable and risk value. the project is cancelled, Depending on whether or not consequence ex- 2. the risks are considered to be acceptable with the posure, vulnerability, or worth have been taken into planned forest development or if forest manage- account, the result could be an analysis of partial ment strategies are applied to reduce the potential risk, speci�c risk, or speci�c value of risk (Wise et al. incremental hazards, or 2004). In any case, this last step completes the input 3. the risks are considered to be acceptable if the from technical specialists—the risk analysis. consequences can be reduced by designing, constructing, and maintaining protective structures on the fan such as dikes or berms (VanDine 1996).

23 In some cases where the risks are considered to be guarantees that no incremental hazard will occur in acceptable, options 2 and 3 are both implemented. all situations. However, a series of factors should be The hydrogeomorphic hazards in a speci�c water- considered when establishing a conservative ECA, in- shed are identi�ed in Steps 3 and 4, and summarized cluding watershed-speci�c processes and character- in Table 2. This knowledge leads directly to manage- istics, past impacts to elements on the fan, the nature ment strategies to reduce the potential incremental of the elements-at-risk, and location of past and pro- hazards. In debris �ow and debris �ood watersheds posed forest harvesting or natural disturbance to the it is essential to avoid initiating landslides, particu- forest cover. Watershed-speci�c mitigative opportu- larly those that could directly affect the channel. nities may exist through desynchronizing snowmelt A key step in this direction is to complete terrain runoff; for example, by locating forest harvesting on stability mapping and develop plans that recog- mid- to low-elevation or southern-aspect slopes nize and adequately manage unstable terrain. Field There is ongoing research on cumulative forest investigation may indicate that avoidance of unstable harvesting and ECA effects on stream�ows. Differ- terrain is a preferred management option, but it may ences have been highlighted between watersheds also determine that reasonable options are available. dominated by rainfall, rain-on-snow, and spring In coastal British Columbia, approximately one-half snowmelt. Huggard (2006) describes the differ- of all landslides occur in clearcuts and are not road- ent hydrological effects over time of dead standing related (Jakob 2000); proposed harvesting must be timber and salvage harvesting at the stand level, assessed and managed accordingly. and the Forest Practices Board (2007) explored the It is widely acknowledged that forest roads are issue at the watershed level. The effect of speci�c a major factor in landslide initiation, due both to watershed characteristics on the harvesting-related poorly constructed roads and to their in�uence on stream�ow impacts were examined by Whitaker et hillslope hydrology. In a coastal logging–landslide al. (2002) and Schnorbus and Alila (2004). Land study, no difference in soil disturbance was found managers should involve experts with an appropriate between helicopter and conventional yarding, but knowledge of watershed processes, linkages between there was a higher incidence of open-slope landslides cumulative impacts and stream�ows, and local on the conventionally yarded areas—due primar- conditions. ily to the subtle but signi�cant in�uence of roads Road density can be a peak-�ow hazard factor on hillslope hydrology (Roberts et al. 2004). In the where inadequate cross drains allow long sections interior of British Columbia, most signi�cant land- of road to direct intercepted �ows into stream slides associated with forestry operations are related channels. Strategies to limit road density include to drainage diversion by forest roads and trails total-chance watershed-level planning to avoid located on gentle to moderately sloping terrain, some constructing unnecessary roads, to deactivate roads, distance upslope from steeper, more landslide-prone and to provide frequent cross drainage to limit the terrain (gentle-over-steep) (Grainger 2002). Speci�c potential for ditch lines becoming a part of the terrain stability �eld assessment techniques and stream channel network (Wemple et al. 1996). A strategies to manage these potential hazards have mitigative option would be to assess and deactivate been described by Grainger (2002). non-status roads in the watershed. Mitigation opportunities for landslides may exist Non-licensee stakeholders often have great res- in some watersheds. Addressing current or potential ervations regarding forest harvesting in watersheds. erosion from non-status roads may be an appropri- Often they will refer to the protective services of ate strategy. Alternative yarding systems should be those forests to various ecological and watershed explored, although unconventional systems such as functions. It is poorly understood that forests must helicopters may not result in a reduction in land- be managed to maintain high levels of protective slides, particularly in gullied terrain (Roberts et al. services because forests are not static entities (Sakals 2004). et al. 2006). Forest licensees may gain the support of Maintaining an appropriate ECA is important watershed stakeholders if they commit to a long- in �ood and debris �ood watersheds, and also in term management program to enhance the protec- watersheds where there is a potential for in-channel tive services of forests. debris �ow initiation. There is no speci�c ECA that

24 Both forest harvesting and lack of forest manage- made will allow for retrospective analysis as time ment (e.g., not undertaking post-wild�re rehabilita- unfolds. Documentation is critical because it could tion as described by Curran et al. 2006) within a be a long time before climatic events occur that watershed can have a high likelihood of affecting the trigger hydrogeomorphic responses from the wa- downstream fan. With an awareness of the total risk tershed. When such events occur it is essential that situation and appropriate accommodations for re- monitoring be undertaken to determine whether ducing those risks, forest harvesting and forest man- responses on the fan were “beyond the natural range agement will be able to continue in many watersheds of variability.” Information collected on the fan in with minimal, and potentially positive, effects. Step 3 will prove valuable for this assessment. The collection of additional watershed information may Step 5.6 Document, Monitor, Evaluate, be required to update Step 4. Results of the evalua- and Report tion should feed into an update of Step 5, and should be applied where appropriate to future fan-watershed A key element in watershed management is to evalu- risk analyses. Reporting results to interested parties ate the effectiveness of plans and practices. The ob- (e.g., the public or government agencies) is critical jective in applying the �ve-step approach is to limit to maintain support for the forest development, and the hydrogeomorphic in�uence on a fan as a result may be necessary in situations where hazards are of forest management in the associated watershed. identi�ed. Documenting information collected and decisions

CASE STUDIES

Four case studies from across British Columbia are able to pass all the material—impacts have occurred presented in the Appendices to illustrate the applica- to the highway. The �ve-step approach was applied to tion of the �ve-step approach for analyzing risk in provide forest development direction that will limit fan-watershed systems in diﬀerent geographic set- incremental hydrogeomorphic hazards. In the Shale tings and with diﬀerent management issues. Wathl Creek case study, the �ve-step approach was used Creek watershed is a community watershed with the in analyzing impacts from historical logging on the village of Kitamaat located on the fan. The watershed fan and in the watershed. This information was then has had no forest development to date, and the case used to provide guidance for harvesting second- study demonstrates the application of the �ve-step growth stands on the fan. The Hummingbird Creek approach in providing guidance to forest develop- case study uses the �rst four steps of the �ve-step ment planning. Since there was not a development approach to describe the circumstances leading to a plan to analyze, some aspects of the risk analy- debris �ow and the resulting impacts to infrastruc- sis could not be completed. Eagle Summit Creek ture on the fan. The �ve-step method is then used watershed has had past forest development, and the to provide guidance—based on what could have Trans-Canada Highway crosses the lower fan. The been determined prior to the debris �ow, along with watershed has produced debris �ows and debris knowledge that we now have regarding debris �ows �oods in the past, and the highway culverts are not in that terrain setting.

25 SUMMARY

The �ve-step approach for analyzing risk in fan- Step 2: Identify Elements-at-risk on Fans watershed systems is designed to move forest man- • Recognize and inventory values on the fan that agement from a site-level focus to a watershed-level may be aﬀected by the hydrogeomorphic pro- perspective. This broader perspective allows for the cesses in the watershed (e.g., either natural pro- recognition of the linkages between fans and their cesses or those related to proposed management watersheds, the hydrogeomorphic processes in�u- actions). encing the fans, and an assessment of incremental Step 3: Investigate Fan Processes risks associated with watershed development plans • Identify the nature of hazardous hydrogeomor- or natural factors in�uencing the forest cover. phic processes (e.g., type, frequency, and distur- Changes to the forest cover in a watershed, either bance extent). through forest harvesting or natural factors, can Step 4: Investigate Watershed Processes have signi�cant eﬀects on downstream fans. For this • Identify watershed features controlling hy- reason, when management decisions are to be made, drogeomorphic processes (e.g., includes water- we encourage the use of the fan-watershed system as shed hydrology, geomorphology and the role of a conceptual framework to be applied across a range vegetation cover). of spatial scales. • Identify the potential for incremental hazards The �ve steps for analyzing risk in fan-watershed associated with management activities. systems are: Step 5: Analyze Risks and Develop Plans • Develop planning options for the watershed and Step 1: Identify Fans and Delineate Watersheds assess the associated risks. • Characterize the physical inter-connections that • Document the process and establish a plan to exist between areas of potential activities (water- monitor, evaluate, and report. sheds) and areas of potential impacts (fans).

26 APPENDIX 1 Wathl Creek case study

Wathl Creek is located near the head of Kitimat Arm Step 1 Identify fans and delineate of Douglas Channel, and is a designated community watersheds watershed (Figure A1.1). The watershed-fan system was the subject of a hydrogeomorphic assessment The boundaries of Wathl Creek watershed and fan (Grainger and Wilford 2008), undertaken prior to were identi�ed manually using topographic maps any signi�cant forest development planning for the and aerial photographs, and mapped at a scale of watershed (one block had been harvested via a low 1:20 000. A series of small fans is present through- watershed divide). This case study demonstrates the out the watershed—primarily in the mid and upper application of the �ve-step approach to guide forest reaches. Individual tributary fan watersheds were development planning in an undeveloped watershed. not delineated, but the fans were identi�ed on a landform map (Denny Maynard and Associates 2001).

Figure A1.1 The Wathl Creek watershed and fan.

27 Step 2 Identify elements-at-risk on fans been constructed along the western stream channel bank on the fan in response to signi�cant �ooding Natural features – Salmon habitat is present on the between 1964 and the late 1970s. The dike has cut oﬀ fan; a waterfall in the lower reach of the watershed is several old channels that �owed across the western an impassable barrier. portion of the fan. The eastern portion of the fan is Anthropogenic features – Kitamaat Village is located elevated, well above any river �ood level. Given the on the Wathl Creek Fan (Figure A1.2). Structures evidence on the fan, we concluded that the dominant include a school, administrative buildings, and many hydrogeomorphic process in�uencing the fan was houses. Two groundwater wells on the fan provide �ood. The highest instantaneous peak �ows are gen- domestic water to the village. Low �ows in the creek erated by rain-on-snow or rain events in September, and groundwater have been an issue during the October, and November. The largest monthly stream summer, and in July 2003 one well temporarily ran discharges occur in June and July during the snow- dry. A bridge crosses Wathl Creek, providing access melt freshet. to the west side of the fan where most of the village Event frequency – The largest storms for this area is located. A long dike constructed of primarily local probably occurred in 1891 and 1917 and it is likely channel material is present along the full channel that the area has not experienced storms of this mag- length on the fan, providing protection for most of nitude since (Schwab 2000). Local residents reported the village. large �oods and bridge washouts on the fan in 1974, Human safety – Approximately 800 people live in 1978, and 1988, and these events were con�rmed by Kitamaat Village on the Wathl Creek Fan. Septer and Schwab (1995). The impacts on the Wathl Creek fan could have been related to inadequate con- Step 3 Investigate fan processes struction of the dike and bridge. Some of the largest �oods in regional stream discharge records occurred Hydrogeomorphic process – Much of the fan surface in 1991 and 1992. It may be that these �ows were not has been modi�ed by human activity (forest re- as great in Wathl Creek, or if they were as large, they moval, construction of roads and dike). A dike has were less memorable because the new bridge was

Figure A1.2 A view of the Wathl Creek fan and Kitamaat Village looking east into the watershed.

28 designed to handle large storm �ow and no bridge stability concerns, and this terrain is generally di- washout occurred. rectly connected to the stream channel. The terrain Event magnitude – Peak �ows are currently being stability maps accurately describe unstable slopes in con�ned by the dike, although it is possible for the the watershed. The stream channel in the watershed dike to be over-topped—surveys show that the has very limited storage capability and efficiently 200-year �ood level is higher than the existing conducts materials through to the fan. berm near the fan apex. Step 5 Analyze risks and develop plans Step 4 Investigate watershed processes Consequence – elements of concern Office investigations Water quality – The groundwater wells are located Morphometrics – The watershed has an area of 127 less than 20 m from Wathl Creek and are 18–20 m km2 and relief of 1.6 km, and is graded (i.e., moun- deep. The nature of the subsurface materials is not tainous). The relative relief number is 0.14, suggest- known. A conservative position is to assume that ing that Wathl Creek is a �ood watershed (Wilford the aquifer has coarse sediment and that there is a et al. 2004b). direct connection to the stream channel. Suspended Sediment production and movement – Landform, sediment would most likely be trapped; however, any terrain stability, and sediment transport maps were dissolved chemical or biological contaminants may available for the watershed. These maps and aerial or may not be removed. Currently, the stream rarely photographs were used to identify nine stream runs turbid, and since the wells have been installed reaches and divide the watershed into four sections there has not been an issue with suspended sediment based on different levels of sediment production in the water supply. (e.g., frequency of gullies, presence of stability class The clean water in Wathl Creek has provided a IV and V, and general slope angles). good environment for �sh and other organisms. Flood generation – Aside from one small block, the Fish species present include chum, coho, pink, and watershed has natural forest cover and no roads. The sockeye salmon. potentially operable forest land is below 900 m and Water supply – The oral record suggests that there is approximately 30% of the watershed. have been periods where groundwater levels are so low that supply is compromised. Field investigations Residences, infrastructure, and human safety – The watershed was inspected by helicopter with sev- Most of Kitamaat Village is located on the southwest eral stops to explore �oodplains, channels, and wood side of Wathl Creek, on the lower-elevation portion quality in the forest. of the fan. The village is protected by the dike; Sediment production and movement – Connectivity however, the dike may not be adequate to protect the between unstable terrain and stream channels was village from a 200-year �ood. If the dike is breached, inspected in the �eld. Most gullies had a direct con- �ow in the channel will be reduced, resulting in the nection to the stream channel. Zones of the water- deposition of bedload, channel in-�lling, and shed with unstable slopes have a very limited valley subsequently more �oodwater issuing forth across �at. Streambed, streambank, and sediment storage the fan surface. characteristics were described for each of the nine reaches. Eight reaches were identi�ed as non-alluvial Hazards transport zones (primarily bedrock), and only one Potential forest development – It was suggested that reach had an alluvial channel. it would be prudent to use the Community Watershed Flood generation – The logged area had very low Guidebook (BCMOF and BCMOE 1996c) as a mini- relief and represented only a small portion of the mum set of standards to be followed for any forest overall watershed. A signi�cant portion of the water- operations within the watershed. It was noted that shed is steep, with shallow soils, and limited forest the hydrologic and geomorphic impact of any forest cover. No forest health issues were identi�ed. development, regardless of the level of planning, is Synthesis of watershed processes – Office work and dependent on how operations are carried out in the �eldwork identi�ed signi�cant areas with slope watershed. It was suggested that the licensee, Haisla

29 Forestry Limited, which is owned by the residents The additional or incremental coarse bedload of Kitamaat Village, should keep the community sediment hazard due to forestry activities is a informed of any forestry activities in the watershed. function of two things: the likelihood of erosion Water quality – Several water quality issues were or landslides, and the likelihood of a signi�cant raised by band elders and councillors: amount of the mobilized sediment reaching Wathl Creek. A signi�cant amount of sediment Biological contamination by wild animals We means that it will make a noticeable difference in are not aware of any cases where forest develop- sediment loads, compared to the relatively high ment in a coastal setting has resulted in contami- natural background sediment levels already in nation of previous pristine waters supplies as a Wathl Creek. Given the terrain stability map- result of wildlife feces contamination. Based on ping, we are able to conclude that: there is a this, it is our opinion that there is a very low like- very high incremental sediment hazard from both lihood that forest development would result in an forest harvesting and road construction in the introduction of contamination of water supplies by Upper Mountain Valley area and in the gullies of wild animals. However, there are some uncer- the larger tributary streams in the Mid Mountain tainties and we recommend monitoring wild Valley and South Plateau areas. Given the limited animal populations in the watershed, particularly amount of unstable terrain and lack of connec- in the north and south Plateau areas. tivity to the stream channel in the lower chan- Contamination by wood leachate We are not nel reaches, there is a low incremental sediment aware of any cases where forest harvesting has hazard from harvesting and road construction caused noticeable contamination in downstream on the open slopes of the Mid Mountain Valley, in domestic water supplies by wood leachate. There the North Plateau area, and in the South Plateau are cases of contamination from wood sorts and area. log dumps where there is a large amount of wood concentrated in a small area. These activities Water quantity should not be undertaken in the watershed. We Peak �ows – Given that the operable forest land conclude that there is a very low likelihood that base occupies only 30% of the watershed, and that forest development would cause chemical contam- harvesting within this area will be constrained due ination by wood leachate. to terrain, riparian protection, old-growth manage- Contamination by hazardous material spills ment, and other harvesting restrictions, the actual The record of this type of contamination in for- area harvested will be less than 30% of the water- estry operations suggest that the likelihood of a shed. We conclude that the incremental peak �ow signi�cant spill is low. Steps can be taken to con- hazard is very low. siderably reduce this hazard. Following appropri- Low �ows – Research has shown that, except in cases ate procedures, there is a very low likelihood that where there has been signi�cant soil compaction forestry operations will result in contamination of or there is signi�cant fog drip, harvesting generally Wathl Creek by a hazardous materials spill. increases or has no measurable effect on low �ows. Sediment loading Given the granitic bedrock Given the limited operational area in the Wathl in the watershed, the sur�cial materials have a Creek watershed there is a very low likelihood of relatively low amount of �ne-grained soils. The development-related impacts to low �ows. landform maps identify the location of several Except for the coarse sediment loading hazard in �ned-grained glacio-lacustrine sediment depos- some portions of the watershed, all other identi�ed its, and these could become a �ne-grained sedi- incremental forestry-related hazards were judged to ment source if not managed properly. Overall, be low to very low. there is a low likelihood of �ne-grained sediment impacts to surface water quality and �sh habitat. Risk analysis and recommendations There is a very low likelihood of �ne-grained sedi- Risk is the product of hazard (or likelihood an event ment impacts to groundwater-sourced domestic occurring and affecting a value) and the conse- water supplies. quence (or the exposure, vulnerability, and worth

30 of the element-at-risk; for example, the high-quality Table A1.2 Consequences, hazards, and risks of all the domestic water supply). While we have suggested identiﬁed hazards and elements-at-risk, except consequence values in order to carry out the follow- the coarse sediment loading hazard and the con- ing risk analyses, it is worth repeating that the �nal sequence of potential impacts to residences and determination of the consequence values and the human safety, which are dealt with separately subsequent risk analysis are the responsibility of the licensee, regulatory authorities, and stakeholders, as Incremental is the determination of acceptable risk and the deci- hazard from sion to proceed with any development based on that forestry – analysis. To manage all potential hazards and associ- Issue Consequence likelihood Risk ated risks, two general recommendations are made: Domestic Low to Very low Very low water – suspended Moderate – The Community Watershed Guidebook (BCMOF sediment and BCMOE 1996c) should be used as a minimum Domestic High Very low Low set of standards. water – industrial – Water users should be kept informed of any chemicals forestry activities planned or happening in the watershed, and of how potential hazards and Domestic High Very low Low risks are being managed. water – biological Domestic High Very low Low For the risk analysis we use Table A1.1, and the water – wood consequences, hazards, and risks for a series of the leachates elements-at-risk are presented in Table A1.2. Fish – suspended Moderate Low Low The coarse bedload sediment risk from proposed sediment development in various portions of the watershed to residences, other buildings, and human safety on the Fish – low �ows High Very low Low fan is presented in Table A1.3. Peak �ows High Very low Low Low �ows High Very low Low Table A1.1 A matrix combining hazard and consequence to determine risk

Hydrologic Consequence hazard High Moderate Low Very high Very high Very high High High Very high High Moderate Moderate High Moderate Low Low Moderate Low Very low Very low Low Very low Very low

31 Table A1.3 Coarse bedload sediment risk

Consequence – impacts to Forest development Coarse sediment Area buildings and human safety sediment hazard risk Upper Mountain Valley High Very high Very high Large tributary gullies – High Very high Very high Mid Mountain Valley, South Plateau Mid Mountain Valley – High Low Moderate except large tributaries South Plateau – except High Very low Low large tributaries North Plateau – all High Very low Low

Development planning and mitigative strategies be accessed and harvested with an acceptable Recommendations include: level of risk to elements-at-risk on the fan.

• No harvesting or road building in the unstable A series of forest management planning scenarios Upper Mountain Valley portion of the watershed. may be developed using these recommendations • No harvesting in the large tributary gullies of the along with other considerations (e.g., timber qual- mid-watershed. ity and equipment availability). These scenarios, • Road building with caution through the large along with mitigative strategies, should be evaluated tributary gullies. Carry out detailed on-site against the risk analysis. In the case of Wathl Creek, assessments to determine where there may be it is likely that the residents of Kitamaat Village will acceptable gully crossing sites, use experienced have input to any �nal forest management plans. road layout personnel, thorough geotechni- We noted in the report that there is a potential cal assessment and engineering, and cautious risk to residences, infrastructure, and human safety construction and maintenance practices. All this in Kitamaat Village because of apparent problems increases costs and there will be locations where with the dike along the southwest side of Wathl gully crossings with an acceptable level of risk Creek. These problems exist whether any forest will not be feasible. development takes place in the watershed or not. It • On Mid Mountain Valley open slopes, with was recommended that the Band Council undertake good forestry practices through all planning and further investigation of the level of safety the dike implementation phases (including total chance is providing the village, with respect to the cur- planning based on thorough geotechnical and rent 200-year design �ood level, and with respect to hydrological assessments), much of the area can expected future storm magnitudes and sea-level rise resulting from ongoing global climate change.

32 APPENDIX 2 Eagle Summit Creek Case Study

Eagle Summit Creek is located in the Monashee north side of the Eagle River, adjacent to the toe of Mountains near the headwaters of the South Thomp- the Eagle Summit Creek fan. son River, upstream of Shuswap Lake and just east Human safety – Hydrogeomorphic impacts to the of the Thompson River–Columbia River drainage Trans-Canada Highway could compromise human divide, about 30 km west of Revelstoke, B.C. The safety and life. A small debris slide from the lower, watershed-fan system was the subject of a hydrogeo- steep, Eagle River valley walls caused a fatality in the morphic assessment (Grainger 2007). The report is spring of 2002 when a machine operator was swept summarized using the �ve-step approach. into Clanwilliam Lake and drowned, while clearing This case study demonstrates that there can be debris from a previous natural landslide. In Decem- different processes in a watershed producing differ- ber 1967 a debris �ow in Camp Creek, on the north ent hydrogeomorphic events with different occur- side of the Eagle River Valley, and approximately 15 rence intervals and magnitudes. Knowledge of the km west of Eagle Summit Creek destroyed the Camp different watershed processes and the different ways Creek Bridge over the TCH, resulting in two fatalities those processes can be affected by various forest to passengers in a car that entered the debris �ow management activities allows land managers to focus path due to loss of the bridge. their decisions where they will be most effective in managing risks. Step 3 Investigate fan processes

Step 1 Identify fans and delineate Debris �ow – There are well developed levees on watersheds the steep concave fan adjacent to the channel, with boulders up to 2.0 m in diameter. Fan gradients are The boundaries of Eagle Summit Creek watershed 50% near the apex, 30% through the mid-fan, and and fan were identi�ed manually using topographic decrease to 10% near the toe, downslope of the TCH. maps and aerial photographs and �eld reviews, The fan surface is irregular, with abandoned chan- and mapped at a scale of 1:20 000. The fan is rela- nels and large boulders and blocks underlying the tively small, about 240 m wide and 170 m long, but vegetated surface. it blocks the narrow Eagle River valley-bottom, There is a mature western redcedar, Douglas-�r, forming Clanwilliam Lake just upstream of the fan and western hemlock stand with trees from 0.45 (Figure A2.1). The watershed has some steep alpine to 0.55 m diameter, estimated to be 90 years old, mountains at its upper extent and a broad, gently to growing on the fan surface, including the boulder moderately sloping plateau forming the mid wa- levee. Except for a couple of trees growing within tershed. The stream is in a steep, 1600 m long gully a few metres of the channel, no scarring was noted dissecting the plateau (Figure A2.2). on mature trees, even those immediately adjacent to the channel at the fan apex. There are also burned Step 2 Identify elements-at-risk on fans stumps throughout the fan, thought to be from the stand-establishing �re for the current stand. If the Natural features – Timber on the fan. There are no stand had been established by a hydrogeomorphic recognized �sh values in Eagle Summit Creek or event any evidence of the previous stand would have Clanwilliam Lake. been obliterated. This indicates that there has been Anthropogenic features – An approximately 200-m almost no hydrogeomorphic disturbance outside section of the Trans-Canada Highway (TCH) crosses the channel in at least 250 years, and possibly for the lower fan. The highway crossing of Eagle Sum- much longer. It was concluded that there have been mit Creek consists of two 1.0 m high oval culverts large debris �ows in Eagle Summit Creek that have (Figure A2.3). Hydrogeomorphic impacts could extended down the fan to the present position of disrupt automobile and truck traﬃc on the main the TCH, but that none have occurred in at least 250 highway connection between Alberta and southern years. British Columbia. The CPR mainline runs along the

33 Figure A2.1 Eagle Summit Creek watershed (2001 airphoto). Proposed cutblocks are shaded.

34 Figure A2.2 Gully and fan reaches of Eagle Summit Creek, looking south.

35 Figure A2.3 A view of the drainage structure on the Trans-Canada Highway at Eagle Summit Creek.

Debris �ood – In the Eagle Summit Creek chan- the recent debris �oods have blocked the culverts nel on the fan, recently moved sediment is mainly and deposited sediment on the highway, disrupt- cobbles and small boulders < 0.75 m diameter. Some ing traffic. No injury or other damage was reported larger blocks have been introduced to the channel (Thurber 1987) from high failing cliffs immediately above it. As discussed in Step 3.1, debris �oods have peak A small group (cohort) of young western red- discharges that are, at most, two to three times that cedar trees occupies a small gravel sediment wedge of major �oods (Hungr 2005), and have maximum adjacent to the channel. Dendroecology analysis particle size of the same order as the peak depth of (Wilford et al. 2005a) of the trees indicated that they major �oods—up to several tens of centimetres were established in the early 1950s and show impact (< 1.0 m) in the case of typical small mountain scars dating from 1967 and 1982. The fact that these streams (Hungr et al. 2001), which describes the young trees were not destroyed indicates that it was �ow magnitude and sediment size observed in Eagle not a high-powered debris �ow that affected them. Summit Creek on the fan. It was concluded that Three hydrogeomorphic events are reported to have there is an existing debris �ood hazard that extends affected the TCH in the last 35 years: in 1967, in the to at least the TCH and has a return period (T) of early 1980s, and one at an unspeci�ed date (Thurber approximately T = 14 years (i.e., a 55-year period of 1987), which corresponds well with the dendroecol- record with four debris �ood events). ogy �ndings. The early 1950s cohort establishment suggests that at least three and possibly four events Step 4 Investigate watershed processes have occurred in the channel in the last 55 years. These events appear to be largely contained in the Office investigations channel and extend to the TCH. It is noted that the Morphometrics – The watershed has an area of capacity of the existing TCH crossing is very likely 6.4 km2 and relief of 1.68 km. The relative relief num- too small to pass a debris �ood. At least some of ber is 0.66, suggesting that Eagle Summit Creek is a

36 debris �ow watershed (Jackson et al. 1987; Wilford et existing equivalent clearcut area (ECA) at the time of al. 2004b). the assessment was approximately 21%. However, relative relief number is designed for “graded” or concave mountainous watersheds, and Field investigations there is a wide, moderately to gently sloping plateau Areas proposed for harvest and road building were in the middle of the watershed. We have observed investigated on the ground, as was Eagle Summit that morphometrics designed for graded watersheds Creek channel, from the plateau down to its con�u- often do not give the correct results for watersheds ence with the Eagle River. The channel and sur- with signi�cant upland plateaus. rounding areas were also observed from the air in Sediment production and movement – Terrain a helicopter �y-over. stability mapping and air photos show that Class IV Sediment production and movement – Most sedi- (potentially unstable) and Class V (unstable) ter- ment production and delivery to the channel oc- rain is ubiquitous along the steep Eagle River valley curred in the steeper stream reaches along the south walls, with debris slides, rock falls, and debris �ows. Eagle River valley wall. This section was subdivided On the upland plateau, terrain was largely Class I to into two reaches based on sediment regime charac- Class III (stable). Little sediment was being deliv- teristics (Figure A2.2). ered to the channel on the plateau, or transported The upper steep reach is about 1.0 km long with through it to the steep reach above the fan. 35–55% channel gradients. Gully sideslopes range Flood generation – Stream�ow records show that from moderate (30–50% gradient) to very steep peak �ows occur during the spring freshet snow- (< 90% gradient) with bedrock slab failures and talus melt in late May and most commonly in early June. slopes of large blocks. There were also lesser till and Detailed analysis of damaging �oods in Eagle colluvium slopes, with some soil slumping observed Summit Creek and debris �ow in a smaller tributary above the channel. The channel is generally 6–8 m stream in the region in 1967 and 1968 show that rain wide, with 2–3 m deep substrate of large blocks and on melting snow caused those large runoﬀ events boulders from 1.0 to 4.0 m diameter (Figure A2.4). (Thurber 1987). The lower reach has similar channel gradients, There had been extensive harvesting on the pla- with mixed gully sideslopes of bedrock slab failures teau area of the watershed throughout the 1980s. The and thick till and colluvium with numerous failing

Figure A2.4 A view of the upper steep reach in Upper Eagle Summit Creek.

37 gullies delivering sediment directly to the Eagle Synthesis of watershed processes – It was concluded Summit Creek mainstem from small tributary debris that the upper steep reach could be the source of a �ows. The channel has a mixed substrate of blocks debris �ow if a sizeable landslide should affect the and boulders, cobbles and gravel, with some sections channel. In-channel debris �ow or �ood initiation scoured to bedrock. Immediately upstream of the of the 1.0–4.0 m diameter sediment by high stream fan there is a 50% gradient reach with mainly grav- �ows is considered unlikely. From the channel load- els, cobbles, and small blocks (< 1.0 m diameter) and ing observed, it was estimated that 15–25 m3 of blocky abundant woody debris, partially derived from pre- sediment could be mobilized in a debris �ow from 1970s harvesting on steep gully sidewalls. In spite of each metre-long section of channel. Depending on the reported recent events there was a high sediment where along the channel a debris �ow originated, load in the channel immediately above the fan, and from 25 000 to 40 000 m3 of blocky sediment could also in other sections of the channel, particularly be mobilized, which would reach the fan and deposit near the con�uence of tributary debris �ows that are anywhere along the 200 m of the TCH crossing actively delivering sediment to the main Eagle Sum- the fan. This is an order of magnitude greater than mit Creek channel (Figure A2.5). events observed in the last 55 years. Debris �ows have not been observed in Eagle Summit Creek for at least 250 years. Thus there is an existing debris �ow hazard with a very low frequency and a very large magnitude (power and extent). Landslide risk is the watershed process that needs to be focussed on when considering forest activities that could have an impact on the upper steep reach of Eagle Summit Creek on the Eagle River valley wall. It was also concluded that the frequent debris �oods observed in the last 55 years all originate in the steep lower reach, where there is more abundant �ner-grained sediment (gravel to small boulders and blocks) in the channel. There appears to be a high rate of sediment recharge to the lower steep reach. It was concluded that, in the lower steep reach, debris �oods can be initiated by small tributary debris �ow damming of the channel or by in-channel initiation in the mainstem by high stream �ows. Because of the sediment loading and the low return period between events of approximately 14 years, the lower steep reach is near its debris �ood failure threshold most of the time. Therefore, there is an existing debris �ood hazard with a high frequency and a much smaller magnitude (power and extent within the channel), which does however affect the TCH. It is considered very unlikely that development on the plateau could initiate landslides that could affect the lower steep Figure A2.5 A view looking upstream in the lower reach of reach above the fan. Therefore, the watershed process Eagle Summit Creek. The small boulders and that needs to be addressed when considering forest fine sediment originate from sidewall debris activities is an increase in peak �ows from harvest- flows. This material is draped over the large ing on the plateau. block substrate of Eagle Summit Creek.

38 Step 5 Analyze risks and develop plans The power and destructive force of a 25 000– 40 000 m3 debris �ow comprised in large part of The debris �ow return period in Eagle Summit Creek large boulders and blocks would be very high, and is at least 250 years, and possibly more. Therefore, the this very high-powered event would extend to the frequency, or annual debris �ow probability, is esti- TCH anywhere along the 200 m where it is on the mated to be at least 1/250 or 0.004. Debris �oods in Eagle Summit Creek fan. This was considered a large Eagle Summit Creek have a return period of approxi- to very large magnitude event. From Table A2.2, with mately 14 years, a frequency of 1/14 or 0.07. These are a high incremental frequency and a high to very high the existing natural hazards at the site. magnitude, there is a very high incremental debris �ow hazard. Incremental landslide and debris �ow hazard This large, powerful debris �ow would likely bury and risk some of the TCH, disrupting traﬃc along one of the Proposed developments included some harvesting at three main car and truck routes between British the edge of the upland plateau, with some cable har- Columbia and the rest of Canada. The Eagle Summit vesting proposed on the upper slopes of the steeper Eagle River valley walls. Also, an old road along the Table A2.1 Qualitative frequency deﬁnitions (adapted from edge of the plateau would be upgraded to access this B.C. Ministry of Forests 2002, Table A10.2) block. Extensive harvesting on the plateau located up to 2.0 km from the steep Eagle Creek valley walls Quantitative could increase the existing watershed ECA by about long-term 50%, to nearly 30% of the watershed. frequency The harvesting and road were a concern, par- (annual Qualitative Qualitative ticularly from possible disturbance and diversion probability) frequency description of poorly incised ephemeral streams in the steeper cable harvest area, and from the potential for the > 0.64 Very high Event is imminent after road to intercept and divert hillslope drainage onto the watershed activity steeper downslope areas. > 0.18, < 0.65 High Event is probable during The road was proposed to be long term (20 years). the lifetime of the water- Given this time frame and the sensitivity of the shed activity gentle-over-steep terrain in this area to drainage > 0.04, < 0.19 Moderate Event is possible, but not concentration by roads, it is appropriate from a risk likely during the lifetime management perspective to include a margin of error of the watershed activity with respect to road drainage. It is possible that during the life of the road, a drainage diversion could > 0.01, < 0.05 Low Remote likelihood of occur that would elevate the likelihood of a landslide event during the lifetime reaching Eagle Summit Creek. An assumption was of the watershed activity made that the landslide frequency would increase < 0.02 Very low Very remote likelihood of from the historical value of 0.004 to 0.05 (1 in 20 event during the lifetime years). From Table A2.1 this would be considered a of the watershed activity high incremental frequency.

39 Table A2.2 A qualitative hazard matrix (adapted from Wise • On > 60% gradient gully wall slopes, the outer et al. 2004) 2 m of the existing road prism �ll should not be disturbed. Any widening of the road should take Magnitude place by removing cutslope material, which was Frequency High Moderate Low located on the moderate to gently sloping plateau. Very high Very high Very high High Excavated material should not be placed on the High Very high High Moderate outer 2 m of the existing prism, but used to raise Moderate High Moderate Low the grade or end-hauled. Low Moderate Low Very low • A drainage plan was prepared after a detailed Very low Low Very low Very low assessment of hillslope drainage was undertaken by a geotechnical professional during a spring freshet. Drainage control structures (culverts and Creek crossing would likely be seriously damaged ditch blocks) were sized, located, �agged, and and would require rebuilding or replacement. If a painted in the �eld, and their location mapped vehicle and its occupants were hit by the debris �ow and documented. it would probably result in a fatality or fatalities. If • Deactivate the road by removing all culverts and a vehicle ran into the deposit, injury and possibly a installing cross ditches at all culvert locations fatality could result. after harvesting and replanting. The managing foresters considered the possibil- • The geotechnical professional should conduct ity of more detailed consequence analysis, including �eld reviews during road construction and exposure and vulnerability of traﬃc and human following deactivation, or at any time during safety, but decided that it would not add value to operations if any unexpected sub-surface �ows the risk analysis. Their decision was that any serious were encountered that would require any change forestry-related impact to the TCH was a high to very to the drainage plan. high consequence. • The lower block area with easily disturbed, poor- From Table A2.3 it was concluded that with a high ly incised ephemeral streams should be deleted to very high incremental debris �ow hazard, and from the harvesting plan. high to very high consequences on the fan, there was • Any bladed skid and backspar trails in the a very high incremental risk from the harvesting remaining block should be deactivated by full and road upgrading being considered near the Eagle pullback and slope recontouring immediately Summit Creek gully. following yarding and prior to the next season’s To reduce the debris �ow hazard and thus the freshet. risk, risk control measures were adopted as part of the planning process. Regarding the road above the It was concluded that these measures would steep Eagle Summit Creek gully walls, and the pro- reduce the incremental landslide and debris �ow posed cable harvesting on upper gully wall slopes, hazard to low. a Terrain Stability Assessment (TSA) recommended the following: Incremental peak �ow and debris �ood hazard and risk The watershed and fan analyses concluded that Table A2.3 A qualitative risk matrix (adapted from Wise et existing peak �ows are likely close to the natural al. 2004) threshold stream �ow values required to initiate debris �ooding in the steep lower Eagle Summit Consequence Creek reach—a reach with abundant gravel to small Hazard High Moderate Low boulder sediment where debris �oods could be initi- Very high Very high Very high High ated. Fieldwork and reports of events indicate that High Very high High Moderate this threshold is exceeded on average every 14 years. Moderate High Moderate Low Therefore, any noticeable peak �ow increase due to Low Moderate Low Very low upslope forest development could push the system Very low Low Very low Very low over that threshold and initiate debris �oods more frequently.

40 Proposed clearcut harvesting on the gentle- Table A2.3 there was a potential high to very high gradient plateau extended up to 2 km upslope from debris �ood risk, which was considered unacceptable the steep Eagle Summit Creek gully on the Eagle by the land managers. River valley walls. Recent research has shown that Because of the potential for harvesting-related in the snowmelt-dominated watersheds of southern elevated peak �ows, the Terrain Stability Analysis interior British Columbia the hydrological effects recommended that a forest hydrology professional of clearcut harvesting can lead to increases in all review the proposed harvesting further for potential sizes of peak �ows (Schnorbus and Alila 2004). increased peak �ow effects. That review concluded Therefore, there was the potential for proposed that the proposed clearcut harvesting above the H60 clearcut harvesting to increase the existing debris line (the line above which contains 60% of watershed �ood frequency of 0.07 to some unknown higher area, see Figure A2.1) could lead to stream�ow effects frequency. This increased frequency would exist for that could increase the frequency of �ood and/or de- at least 20–30 years, when hydrological green-up bris �ood events on the fan. The hydrologists report would signi�cantly decrease the clearcut debris �ood recommended deleting the proposed harvest area hazard. Although the speci�c effect that upslope above the H60 line until either: harvesting could have on peak �ows was not known to any degree of certainty, it was concluded by the • the B.C. Ministry of Transportation and High- risk analysis team that a noticeable increase in peak ways are noti�ed that previous events have shown �ows could entail at least a moderate, and possibly a that the existing culverts at the TCH crossing are high, incremental debris �ood frequency. inadequate to pass the frequent debris �oods that Based on past occurrences, a debris �ood would occur in Eagle Summit Creek, and that will occur be largely contained in the channel, at least until at some interval in the future whether additional it reached the TCH crossing, which it would affect. forest harvesting occurs in the watershed or not, This was considered a moderate magnitude event. and a suitably designed span is installed over Therefore, from Table A2.2, with a potential high Eagle Summit Creek, or incremental �ood frequency and moderate magni- • hydrologic green-up of existing harvested areas tude, there was potentially a moderate, and possibly has achieved a stand height and canopy closure a high, debris �ood hazard. such that upper watershed ECA is reduced to an A debris �ood would affect the highway crossing, acceptable level. most likely by blocking the culverts and overtopping the highway and disrupting traffic. The event would Harvesting of proposed areas below the H60 probably not cause a fatality directly, but could cause elevation was not considered an increased peak �ow a vehicle pile-up that could result in damage to hazard. In fact, lower-elevation harvesting could vehicles, and possibly human injury. Again, the land advance snowmelt in those areas early in the spring managers decided that any impact to the TCH that freshet, desynchronizing melt from the lower-eleva- would disrupt traffic and potentially result in human tion area from peak �ow–generating snowmelt in the injury was a high consequence. Therefore, from upper watershed area. This could result in sustained periods of moderately high, but not extreme, �ows.

41 APPENDIX 3 Shale Creek Case Study

British Columbia Timber Sales (BCTS) is develop- Step 1 Identify fans and delineate ing harvesting plans for a cutblock located on the watersheds alluvial fan of Shale Creek, Moresby Island, Haida Gwaii Forest District (Cutblock SKI 101, Figure The Shale Creek watershed-fan system is located on A3.1). The fan was previously harvested in 1951, and the Skidegate Plateau area of Moresby Island. Figure harvesting within the watershed of Shale Creek com- A3.2 shows the watershed boundary of Shale Creek menced shortly after the fan was logged. Evaluation and the fan boundary. The watershed is approxi- of the watershed and fan processes, and the eﬀects of mately 10.6 km2 in area, with an elevation range the initial harvesting of the Shale Creek fan and its of 80–620 m (Figure A3.2). Shale Creek fan builds watershed, are important for planning appropriate into a 1 km wide valley that trends west–east, which cutblock layout and fan management. includes Skidegate Lake. To the immediate west of The evaluation used aerial photographs from the fan, the valley is con�ned by hillslopes project- 1964, 1977, and 1994, as well as recent satellite im- ing into the valley from the northwest and southeast, agery (circa 2007) to examine watershed and fan and the valley narrows to about 0.5 km in width. The conditions. Maps provided additional information. fan is limited in its ability to expand to the southwest Fieldwork on the fan examined the channel network by the hillslope con�nement, but extensive fan ex- and other fan features. pansion to the east is possible. The fan toe transitions

Figure A3.1 Location map for Block SKI 101.

42 Figure A3.2 Shale Creek watershed and fan. Mapped stream channels shown on the fan are not correct, as is often the case with fan channel networks. into a �oodplain that extends east to Skidegate Lake, et al. 2005). Two channels currently exist on the fan a distance of about 1.5 km (Figure A3.1). The fan toe in the central and western portions (Figure A3.3). is partially de�ned by wetter soils and a change in Stream 1 is the main channel that carries water �ows timber type. and sediment from the watershed and ranges from 10 to 25 m wide, with the greater widths near the apex Step 2 Identify elements-at-risk on fans of the fan. Upstream of about 0 + 700 m on Stream 1 there are abandoned or slightly active channels Located in a relatively remote area, there is no hu- that avulse from, and return to, Stream 1. Stream 2 man habitation on the fan. A mainline logging road is smaller, with channel widths typically less than and power line cross the fan. Streams on the fan and 5 m. This stream is inset into an older channel that their high-value �sheries are the main element-at- is 15–20 m wide, similar in width to the presently risk. active Stream 1 (Figure A3.4). Upstream of 0 + 349 m on Stream 2 there is a broad area of recent gravel Step 3 Investigate fan processes deposits (less than about 50 years in age) that have buried the well de�ned older channel edges visible The fan is low gradient (< 2%) and shows evidence of downstream of 0 + 349 m. This broad gravel area site-level, high-power, water �ood events (Wilford extends upstream to the left bank of Stream 1 near

43 Figure A3.3 Channel network on Shale Creek fan. Preliminary proposed harvesting areas are shown in pink.

Figure A3.4 Stream 2 ﬂowing in an older and larger channel. Old channel banks are outlined in yellow. Groundwater re-emergence results in limited ﬂow variability, little sediment transport, and previously active gravel bars becoming vegetated.

44 0 + 724 m (Stream 1 measurement). The blockage of above the surface of the fan and acts as a dike. No Stream 2 has isolated it from Shale Creek �ows; only channels are evident on the downstream side of a tributary source and groundwater re-emergence the mainline road along the section where Stream 1 results in channel �ow. parallels the road. The alignment of Stream 1 with the road, and the lack of channels below the road in Step 4 Investigate watershed processes the area of the stream alignment, indicates that the road was present prior to the existence of Stream 1. The relative relief number (Melton 1965) for Shale This evidence, in combination with the large gravel Creek watershed is 0.17, indicating that debris �ows deposits blocking the entrance of Stream 2 and the or debris �oods are not likely delivered to the fan presently small channel size of Stream 2 inset into (Millard et al. 2006). Although most of the water- an older, larger channel, all support the air photo shed has low likelihood of landslides, there are some evidence that Stream 2 was the only active chan- slopes that are subject to landslides. In particular, nel on the fan prior to watershed logging, and that the channel reach immediately upstream of the fan the blockage of Stream 2 resulted in the creation of apex is narrowly incised into steep unstable slopes, Stream 1 below 0 + 730 m. where most landslides deposit directly into Shale With the exceptions of Stream 1, Stream 2, and Creek. The channel gradient of Shale Creek imme- partially active channels near the apex of the fan, diately upstream of the fan for a distance of almost 1 only a few very faint remnants of channels were km is 2–4%, indicating that landslides that initiate in found on the Shale Creek fan. All other surfaces this area and deposit sediment into the creek will not appeared old, with no �uvial activity for at least cen- continue to the fan as debris �ows. Field investiga- turies, and possibly millennia. The lack of channels tion of the fan did not �nd any evidence for debris across the surface of the fan indicates high channel �oods or debris �ows, which con�rms that Shale stability. Increased sediment deposition on the fan, Creek is a �ood-only fan. such as that which occurred from numerous landslides after the 1950s and 1960s logging of the slopes History of forest management and geomorphic upstream of the fan, is likely the main cause of chan- response nel instability. The fan area was harvested in 1951. The 1964 air photos show Stream 2 as the active channel that carries Step 5 Analyze risks and develop plans water and sediment from the watershed. The 1964 air photos also show that much of the lower portion The primary element-at-risk on the fan is �sheries of Shale Creek watershed was harvested by 1964, habitat. The �sh habitat is likely best protected by including the steeply incised slopes in the channel avoiding avulsions, which appear to be rare on this reach directly upstream from the fan apex. By 1964, fan, and by avoiding introducing sediment into the at least 14 logging-related landslides had occurred on channels. Harvesting of almost all trees on the fan these valley slopes immediately upstream of the fan, in 1951 did not appear to have caused the avulsion on with additional slide areas visible in the 1977 photos. the fan, although streamside logging likely resulted Most of these landslides deposited sediment directly in damage to stream banks. Logging-related land- into Shale Creek. Although individual landslide slides within the watershed likely led to an avulsion. events were not transported directly to the fan as The history of the watershed-fan system indicates debris �ows or debris �oods, the channel would have that to avoid management-caused avulsions on the transported much of the landslide sediment to the fan, watershed management is at least as important fan through subsequent �ood events. It is likely that as fan management. the sediment from these landslides resulted in the BCTS has agreed to follow the Haida Gwaii extensive gravel deposits near 0 + 724 m on Stream 1. Strategic Land Use Agreement for Block SKI 101. Although Stream 2 appears to be the only active This agreement has the objective of retaining active channel in the 1964 photos, by 1977 it appears to have �uvial units by maintaining forests within 1.5 tree been blocked. lengths of the outer edge of the active �uvial unit. Stream 1 parallels the Moresby Mainline road for For fans, the active �uvial unit is de�ned as the hy- approximately 200 m (Figure A3.3), which is built drogeomorphic riparian zone.

45 The hydrogeomorphic riparian zone on Shale network on Shale Creek fan is sensitive to watershed Creek fan extends from the left (or northeast) chan- processes. The avulsion was likely caused by logging- nel bank of Stream 1 to the right (or southwest) related landslides that led to extensive sediment de- channel bank of Stream 2. Upstream of about posits on the fan, blocking the then-active channel. 0 + 700 m on Stream 1, the active �uvial unit in- These events demonstrate the sensitivity of the Shale cludes some abandoned or slightly active channels Creek fan to forestry practices within the watershed. to the northeast of Stream 1. This zone encompasses Future harvesting within the Shale Creek watershed both the currently active channel (Stream 1) and should consider the potential for fan destabilization Stream 2, which has become isolated from sedi- as a result of landslides that could lead to increased ment and water �ows of Shale Creek. Since this fan sediment deposition on the fan. appears to exist primarily as a single-thread channel network and not a complex of channels across the Conclusion surface of the fan, this likely represents an oversized The forest management history of the Shale Creek active �uvial unit. All of the SKI 101 block areas are watershed-fan system demonstrates that eﬀective at least 2 tree lengths from the active �uvial unit. fan management must include consideration of This exceeds the stated objective in the Haida Gwaii watershed management. Planning for the SKI 101 Strategic Land Use Agreement and should provide block incorporated fan and watershed assessments protection to bank stability. to understand how the channel network on the fan The avulsion of the fan channel sometime responds to fan and watershed disturbances. between 1964 and 1977 indicates that the channel

APPENDIX 4 HUMMINGBIRD CREEK CASE STUDY

Hummingbird Creek is located in the Hunters Range they will be most eﬀective in managing risks for a on the east side of Mara Lake near Sicamous in the downstream fan. southern interior of British Columbia (Figure A4.1). Mara Creek joins Hummingbird Creek just below Step 1 Identify fans and delineate the apex of the fan, and the joint fan is locally known watersheds as Swansea Point. On July 11, 1997, a debris avalanche initiated in the watershed, entered Hummingbird The Swansea Point fan is built out into the waters of Creek, and initiated a debris �ow. The debris �ow Mara Lake. The watershed boundaries of Humming- travelled down the channel to the apex of Swansea bird and Mara Creeks and the Swansea Point fan are Point fan where sediment deposition began. The illustrated in Figure A4.1. Hummingbird watershed volume of the debris �ow was estimated at 92 000 m3, is approximately 16.4 km2 in area. The lower half of which resulted in the destruction of two houses and Hummingbird Creek follows a well de�ned, south- damage to several other houses, as well as to High- west-trending faultline and then turns abruptly west way 97A and local roads. as it reaches the fan. The fan extends out into Mara This case study uses the �rst four steps of the Lake; the sub-aerial portion of the fan is approxi- �ve-step method to describe the Swansea Point fan mately 0.6 km2. The distance from the fan apex to the and Hummingbird watershed, the elements-at-risk, lakeshore is approximately 1.3 km. The fan gradient and the hydrogeomorphic processes that occur, ranges from approximately 18% at the apex to 5% including the 1997 debris �ow event. A retrospective along the lakeshore. application of step �ve describes how the use of the method might have prevented the 1997 debris �ow, Step 2 Identify elements-at-risk on fans highlighting the fact that relatively subtle changes in a watershed can have very signi�cant implications Human safety – Approximately 250 people live on the fan. Knowledge of the potential hydrogeo- permanently in the Swansea Point area (Statistics morphic processes that can occur in a watershed Canada 2007), and during the summer more come allows land managers to focus their decisions where to stay at resorts on the fan.

46 Figure A4.1 Map of Hummingbird and Mara Creek watersheds, and the Swansea Point Fan. The dashed line follows the 1500-m contour, which is the break between the gentle upland plateau and the steeper escarpment.

Anthropogenic features – Non-indigenous settle- tially problematic situation (Wilford et al. 2005). In ment began on Swansea Point fan in the early 1900s 1997, Hummingbird Creek passed under the highway and there are presently approximately 277 private through a 3-m elliptical multi-plate culvert. A trash dwellings, of which 122 are occupied year-round rack was in place to remove boulders immediately (Statistics Canada 2007). A large condominium upstream of the culvert. During the debris �ow event complex developed in 2007 was reviewed by the in 1997, the trash rack rapidly blocked with debris Columbia-Shuswap Regional District but the con- and diverted most material from the channel. The cerns of residents on the fan were primarily the sew- culvert was relatively undamaged during the event age disposal system, upgrading of the water system, and remains in use today. In the aftermath of the increased boat use on Mara Lake, highway access, debris �ow, the upper channel was widened and a population density, beach access, and noise—very levee was constructed and armoured with the largest little attention was paid to the hydrogeomorphic boulders available in the vicinity. The culvert outlet hazards. and downstream banks were armoured extensively Highway 97A, connecting Sicamous to the Okan- to promote channel stability. agan Valley, climbs the fan towards the apex and Swansea Road, the main access road on Swansea there are numerous local roads on the fan, a poten- Point running west from the highway down to Mara

47 Lake, was heavily damaged by the �ner-grained evidence of previous events, such as boulder lobes fraction of the debris �ow and required rebuilding. and levees. However, there are exposures of debris Natural features – As a result of the passage of the �ow deposits overlying �uvial sediments, indicating 1997 debris �ow and subsequent mitigation works, a previous history of debris �ows on the fan (Inter- there is little or no complexity to the fan channel agency Report 1997). (i.e., the channel is quite straight with predominant- The 1997 debris �ow – The Swansea Point fan can be ly bare sand, gravel, and boulders along the margins; divided into upper and lower sections by Highway there are no vegetated banks and very little wood 97A (Figure A4.2): the upper fan section is approxi- or organic material). Mara Lake is home to a wide mately 600 m long and the lower section is approxi- variety of �sh, including several types of salmon and mately 700 m long. trout. Fish are not present in the Hummingbird wa- The upper fan is con�ned by bedrock ridges, has tershed: the steady gradient of the lower channel pre- moderate gradients between 10 and 18%, and has cludes �sh passage, and upper portions have no lakes coarse sedimentary deposits. The active hydrogeo- or wetlands for �sh habitat. Rainbow trout are noted morphic riparian zone (Wilford et al. 2005) remains in the Mara Creek watershed, likely in the lakes on relatively con�ned, about 50 m wide, for the �rst the upper plateau. Fish passage into Mara Creek is 400 m below the apex, then widens to over 100 m obstructed at the con�uence with Hummingbird wide at the highway crossing. The typical particle Creek by a very steep section of channel. size near the fan apex is up to 1 m (b-axis). One exceptionally large boulder (the “Hinkelstein,” 9 m Step 3 Investigate fan processes tall, 8 m wide, and 6 m long) can be seen in Figures A4.3 and A4.4. Hydrogeomorphic processes – From �eld observa- The lower fan has gentler gradients and �ner tions, dendroecology, anecdotal reports, and re- sedimentary deposits. Most of the coarse material view of various sets of aerial photographs, there is (particle size > 0.4 m) was deposited on the upper abundant evidence that debris �ow and debris �ood fan, but �ner debris �ow material and water crossed events have occurred on the fan in the past, with both north and south of the highway culvert. To the several lines of evidence pointing to a large event in south, the debris �ow carved a new channel across 1935 (Jakob et al. 2000). Human activities on the fan the road and then returned most of the �ow back prior to 1997 have probably altered morphological into the lower channel. To the north, much of the

Figure A4.2 Distribution of new debris ﬂow sediment on the Swansea Point fan, mapped on July 14, 1997 (from Inter-agency Report 1997). The shaded area incor- rectly follows Mara and not Hummingbird Creek at the lower right.

48 Figure A4.3 Looking downstream from the conﬁned valley reach of Hummingbird Creek to the apex of the Swansea Point fan. Note presence of bedrock on channel margins, large boulders in the channel, and the “Hinkelstein” boulder in the centre.

Figure A4.4 The “Hinkelstein” at the apex of the Swansea Point fan. This huge boulder was transported during the 1997 debris ﬂow.

49 �ow was carried straight down Swansea Road or uti- Step 4 Investigate watershed processes lized other old channel segments on the fan surface, before reaching Mara Lake in several locations. Oﬃce investigations Above the apex of the fan, the creek �ows in an While none of the aerial photographs dating back incised, steep-sided, bedrock-cored channel 8–10 m to 1928 shows evidence of recent landslides on the wide, with a gradient between 18 and 21% (Figure steep valley sidewalls above the lower channel, there A4.3). The average gradient between where the debris is evidence of older vegetated linear features and avalanche entered the creek and the fan apex is 27%. possibly a large relic landslide. There is also a recent Event frequency – Two methods are used to recon- large debris slide on similar terrain just north of struct debris �ow occurrences on the Swansea Point Hummingbird Creek (Figure A4.5). fan: a collection of anecdotal information and an Morphometrics – The relative relief number RRN( ) aerial photograph review. (Melton 1965) has been found useful for determining Anecdotal information – One account, presumed the hydrogeomorphic event that will emanate from from the 1930s, indicates that the fan was covered by a watershed (Wilford et al. 2004b). It has been found a layer of boulders 1.2 m in diameter, with the largest that for RRN < 0.3 the dominant hydrogeomorphic boulder having dimensions of 2.4 by 3 m (Jakob et process is �ooding; for 0.3–0.6 and > 0.6 with water- al. 2000). The highway bridge across Hummingbird shed length > 2.7 km the process is debris �ood; for Creek was destroyed in this event. > 0.6 with watershed length < 2.7 km debris �ows Aerial photograph review – Aerial photographs can occur (Wilford et al. 2004b). However, the RRN show that the main channel has migrated in recent is designed for “graded watersheds”—watersheds history. At the beginning of the 20th century, the that have steep terrain in the headwaters and gentle main channel was located on Railway Belt surveys terrain in the lower reaches. Caution must be used as heading more westerly across the fan (Fuller when applying the RRN to reversely graded water- 2001). It remained in this position until after 1928 sheds and in de�nition of the area of a watershed. (as seen on aerial photograph A360: 28). In the 1928 There are two distinct watersheds feeding onto aerial photograph there are two road alignments, the the Swansea Point fan, Hummingbird and Mara. newer of the two being roughly the present align- The individual watersheds indicate that Humming- ment of Highway 97A. On the 1951 aerial photo- bird (RRN = 0.44) and Mara (RRN = 0.39) are debris graphs (BC1292: 32 and 33) the creek �ows west until �ood watersheds, but the combined area suggests it it passes under Highway 97A, then turns southwest is more likely a �ood watershed (RRN = 0.29) (Table and �ows into Mara Lake. This is likely the result of A4.1). a debris �ood or debris �ow event reaching the fan Hummingbird and Mara watersheds are clearly and blocking the channel, and water being diverted not graded; the headwaters have gentle slopes and onto the old road alignment. This remains the pres- the lower portions are much steeper—a situation ent channel location. that is referred to as “gentle-over-steep.” To com- pensate for this, subdivision of the watershed to Synthesis of fan processes isolate the steeper sections was completed (Figure Compilation of all evidence suggests that there was A4.1). For Hummingbird Creek, subdivision of the a relatively large event that occurred between 1928 watershed into the steeper section below the 1500 m and 1951, most likely in 1935. The classi�cation of contour gives an RRN of 0.51. This ratio indicates that the event as either a debris �ow or debris �ood is the steeper portion of the watershed, which is more unclear, but it apparently had suﬃcient material to susceptible to landslides, is likely to be characterized block the existing channel and cause diversion onto by debris �ood events moving through the channels an old roadbed, which has subsequently become the and reaching the fan. Given that the 1997 event was present-day channel. While there is some evidence of a debris �ow, a watershed RRN between 0.29 and 0.51 more frequent smaller events, there have likely been indicates �ood or debris �ood, which underestimates two large potentially damaging events in the last 75 the actual hydrogeomorphic hazard, and also sug- years. gests that the two watersheds should be treated as separate for the RRN analysis.

50 Figure A4.5 Aerial photograph BC2616: 76 of lower Hummingbird Creek, 1959, showing a recent landslide to the north on similar steep, northwest-aspect terrain, a possible relic debris slide colonized by a younger forest stand, and the location of the 1997 debris avalanche.

Table A4.1 Relative relief numbers for the Swansea Point fan rain is common along the steeper face above Mara Lake. On the upland plateau, the terrain was gener- Elevation Range ally Class I to Class III (stable) with shallow bedrock Area km2 (km) RRN covered by a thin veneer of sediment. Channels are Hummingbird 16.4 1.79 0.44 moderately incised into the plateau surface, and sediment production and delivery to streams is limited Mara 21.2 1.79 0.39 to the stream escarpment slopes. The steeper slopes Combined 37.4 1.79 0.29 on the Mara Lake face show more signs of sediment Hummingbird 4.61 1.10 0.51 delivery to the channels with thin linear strips visible < 1500m in the tree cover on aerial photographs (Figures A4.5, A4.6, and A4.7). Several other creeks along the Hunters Range Sediment production and movement – Terrain above Mara Lake, including Mara, Rodgers, and stability mapping of the Hunters Range completed Johnson Creeks, have widened channels and exposed after the Hummingbird debris �ow shows that Class sediments in the 1951 aerial photographs (BC1292: IV (potentially unstable) and Class V (unstable) ter- 31–35 and BC1285: 34–37). There is no evidence on the

51 aerial photographs of logging within the upper wa- Hummingbird Creek (Figures A4.6 and A4.7). The tersheds, and recent landslide scars are not apparent, debris avalanche then rapidly transformed into a suggesting that in-channel �oods or debris �oods debris �ow, which travelled 2.4 km down Humming- might have been the cause. Sparse or young timber bird Creek onto the fan (Figure A4.7). An estimated types across the top of the Hunters Range hint at the 25 000 m3 of wood and sediment entered the creek occurrence of widespread wild�re. from the debris avalanche. A further 67 000 m3 of Flood Generation – Snow packs of 2–3 m thickness debris, or 28 m3 of debris per linear m of channel, are typical in this area. Limited stream�ow records was estimated to be scoured from the creek channel. on smaller watersheds in the southern interior offer The total volume of the debris �ow was calculated at only partial information on timing and volumes of over 92 000 m3 (Jakob et al. 2000). Deposition was �ows from the Hunters Range. Chase Creek water- estimated as 49 000 m3 on the upper fan and 36 000 m3 shed (approximately 40 km west of Mara Lake) is the on the lower fan, and 7 000 m3 reached Mara Lake. closest gauged watershed that drains directly from Following the debris �ow, the channel has a reduced an upland area. Chase Creek experiences peak �ows amount of stored sediment, and a landslide entering generated from snowmelt between mid-April to the channel now would have less sediment to mid-June, but maximum elevations are 500 m lower. entrain. Coldstream Creek watershed (about 50 km south of Mara Lake) is the next closest station and has peak Synthesis of watershed processes �ows generated between April and mid-June, but The upper portion of Hummingbird Creek water- maximum elevations are 600 m lower. From these shed is unlikely to generate debris �ows or debris records, peak �ows in Hummingbird and Mara �oods. The relatively low channel gradients are less Creeks are estimated to occur between mid-April able transport signi�cant sediment loads to the lower through late June. half of the watershed. Floods may be generated from snowmelt in the upper watershed, but the timing of Field investigations known events in late June and early July suggests On July 11th, 1997, a debris avalanche initiated on an that snowmelt has less of an effect than rainfall. open slope directly below a culvert on the Skyline Lower sections of the watershed have higher prob- Forest Service Road (Inter-agency Report 1997). A abilities of generating events because there is direct skid trail in a 1994 cutblock diverted water, which connectivity of steep, potentially unstable slopes increased the drainage area leading to the culvert by adjacent to the channel, and the channel gradient approximately a factor of 3 (Figure A4.6). This inter- is relatively steep (Millard 1999). The lower channel ception and concentration of hillslope runoff to the has little or no �oodplain to arrest landslide debris, culvert is considered a “major contributing factor in and consequently landslides have the potential to initiating the debris avalanche” (Inter-agency Report transform into a debris �ood or debris �ow. The 1997 1997). debris �ow occurred as the result of intense long- Snowpacks across the North Okanagan-Shuswap duration rainfall on top of unusually high ante- were 25% above normal in 1997, and followed record cedent moisture conditions and thus there was an levels of precipitation in the area over the preceding increased amount of stream�ow to propagate the 9 months (October 1996 through June 1997), result- landslide sediments through the creek to the Swan- ing in antecedent soil moisture conditions that were sea Point fan. substantially higher than normal. Heavy precipitation recorded at local climate stations between July Step 5 Analyze risks and develop plans 5 and 12, (60- to 85-year return period) contributed more water to the already wet soils, and late on July It is useful to review the 1997 debris �ow from the 11 the debris avalanche was triggered. perspective of the �ve-step method for managing The slope gradient below the culvert was up to risks on fans from upstream watershed activities. 70% and had a cover of sur�cial sediment < 1 m thick In particular, to address the question “could the over glacially smoothed, benchy, granitic-gneiss method have in�uenced how forest development in bedrock. The debris avalanche widened gradually the watershed took place and possibly reduced the downslope over a total slope distance of 560 m, from likelihood that a debris �ow would have occurred?” 5 m at the headscarp to 110 m where it entered The following analysis uses the risk analysis pro-

52 Figure A4.6 Pre- and post-development runoff-contributing areas to the culvert located above the debris avalanche. (Source: R. Winkler, B.C. Ministry of Forests and Range and D. Anderson, B.C. Ministry of Environment.)

Figure A4.7 Aerial photograph of lower Hummingbird Creek watershed with the debris avalanche below the road in right centre (15BCC04021: 202, taken in 2004). The debris avalanche developed into a debris ﬂow that proceeded down the channel to the Swansea Point fan.

53 cess to review what was known, or could have been of entrainable material would have been found known, prior to the logging and road building in prior to the 1997 event. This would have informed the watershed that preceded the 1997 debris �ow. It investigators of the potential for a large event. should be noted that, while hindsight is useful, this The texture of the material could give an indica- analysis is not meant as a criticism of licensee or tion of whether mobilization of the channel bed Ministry of Forests practices in Hummingbird Creek would require a landslide impact or if in-channel watershed prior to the 1997 debris �ow, which in our initiation could occur from peak �ows. understanding were to the standards of the day (For- est Practices Board 2001). Investigation of fan and watershed processes would lead to the conclusion that a large, poten- Summary tially damaging hydrogeomorphic event had oc- Step 1 Identify fans and delineate watersheds curred on the fan circa 1935, and that a debris �ow This would bring to the attention of forest plan- had also occurred some time in the past. It was not ners the connection between the upper plateau clear whether the 1935 event was a debris �ow or areas of the watershed where forest activities debris �ood. Therefore the linkages to be considered were planned, and the fan several kilometres between watershed processes and forest development away from those activities. would include the production and delivery of both Step 2 Identify elements-at-risk on fans sediment and runoff (Table A4.2). The presence of widespread risk elements (high The potential for landslides that could initiate density of residences, major provincial highway debris �ows or debris �oods exists along the lower and other infrastructure) that could potentially stream reach. Given the state of knowledge prior suffer losses would be identi�ed. Damage to resi- to 1997 (Hungr et al. 1984; Van Dine 1985, 1996), it dences, infrastructure, and the highway would be would have been considered possible, but unlikely, considered a high consequence. Potential human that a landslide affecting the relatively low-gradi- injury or loss of life by residents or highway users ent (average 27%) channel would have initiated a would be considered a high to very high conse- debris �ow. Subsequent to the Hummingbird event, quence, depending on the risk matrix employed. and several other events in southern interior Brit- These high consequences would require that the ish Columbia, there is an increased awareness of the remainder of the investigation and risk analysis potential of large debris �ows in relatively low-gradi- be carried out very thoroughly. ent channels. Step 3 Investigate fan processes There remains the possibility that a large landslide This would reveal the physical presence of debris could dam the relatively narrow V-shaped channel, �ow deposits on the fan. With the potential for causing stream�ow to back up until the dam was high consequences, a thorough review of past breached, initiating a large debris �ood. Manage- events, by various methods, would be appropri- ment of incremental landslide hazards on the lower ate. A review of historical aerial photographs steep slopes would be identi�ed as a concern. would show that a major channel avulsion had Landslide hazards could be from operations on occurred—further evidence of a signi�cant hy- the lower steep slopes, or from operations on the drogeomorphic event. A review of published or moderate to gently sloping plateau some distance oral history in the area, as was done subsequent above those slopes. Prior to the Hummingbird to the 1997 event, would con�rm the historic event, management of road and trail drainage on occurrence of at least one large damaging event “gentle-over-steep” plateau terrain was not widely over half a century ago. From this it could be recognized as a serious hazard. Subsequent to the concluded that Hummingbird Creek could pro- 1997 debris �ow, and partially because of it, man- duce a large damaging event, and that it had been agement of forest development in gentle-over-steep some time since one occurred. terrain has been addressed through research and the Step 4 Investigate watershed processes development of best practices (Grainger 2002; Jordan This would include a traverse along the channel 2002). above the fan to assess sediment and woody de- Investigation of channel-sediment texture in the bris loads available to be mobilized. High levels steeper reach could give some indication of whether

54 Table a4.2 Forest management focus for different hydrogeomorphic processes. Key issues are bolded.

Process Initiation Management focus Flood Runoff ECA – harvest, �re, forest health Road density Fire – reduced soil in�ltration Landslides – increasing sediment load to channels on the fan (in�lling channels) Dam rupture Dam rupture – beavers, road drainage structures Landslide dams – harvest and roads (coast), gentle-over-steep (interior) Debris �ood Runoff ECA – harvest, �re, forest health Road density Fire – reduced soil in�ltration Landslide Harvest and roads on unstable terrain and gentle-over-steep in the interior Dam rupture Dam rupture – beavers, road drainage structures Landslide dams – harvest and roads (coast), gentle-over-steep (interior) Debris �ow Landslide Harvest and roads on unstable terrain and gentle-over-steep in the interior Runoff—in-channel ECA – harvest, �re, forest health initiation Road density Fire – reduced soil in�ltration in-channel debris �ow or debris �ood initiation With the high to very high consequences existing due to peak �ows could occur, and whether ECA on the fan, obtained from standard risk matrices and other peak �ow watershed linkages would need (Table A4.3), there would be a high to very high management attention. incremental risk due to proposed forest development on the upland plateau, and further risk Step 5 Analyze risks and develop plans management would be appropriate in any sub- If one were to complete this step today for a simi- sequent plans. From a forestry perspective, the lar set of conditions as existed at Hummingbird most feasible course of action would be to reduce Creek prior to 1997, and with the knowledge of the incremental development-related hazards. forest development and watershed processes we now have, the available evidence would indicate With the hindsight we now have regarding man- that there was a moderate to high likelihood of a agement of gentle-over-steep terrain and resulting landslide affecting the channel and resulting in landslide hazards (Grainger 2002), it would be recog- a hydrogeomorphic event that would reach the nized as appropriate to fully deactivate the roads fan. Given the uncertainties in the analysis, this and trails within the cutblock immediately follow- would be considered a moderate to high hazard. ing harvesting to re-establish or maintain natural

55 Table A4.3 A matrix combining hazard and consequence to after road building and harvesting, to assess site determine risk conditions during operations and recommend changes to prescriptions or designs if necessary, Consequence and to ensure that operations are or were carried Hazard High Moderate Low out in conformance with recommendations or Very high Very high Very high High designs (APEGBC Quality Management Bylaw High Very high High Moderate 14(b)(4)). Moderate High Moderate Low Low Moderate Low Very low If a steep area downslope of a potential develop- Very low Low Very low Very low ment is judged to have a moderate or higher natural landslide potential, the prudent decision may be that the development should not occur within a speci- drainage paths. This deactivation would likely have �ed distance of the marginally stable slope. This was prevented enlargement of the runoﬀ-contributing probably not the case at Hummingbird Creek prior area to the culvert above the slide. Other actions to to 1997 (Inter-agency Report 1997). Although we can reduce hillslope runoﬀ due to harvesting include: never know for sure what the outcome of various management options would have been, it is likely • Road construction methods should minimize that if some or all of the above risk-mitigation steps subsurface �ow interception in high cut slopes, had been taken, harvesting and road building on the such as decreasing the road width or using a upland plateau above Hummingbird Creek could wider �ll to reduce cut slope height. have occurred without causing a major debris �ow. • Drainage redirection should be avoided by Gentle-over-steep terrain similar to the Hum- designing roads with no ditches or cross-drain mingbird case study is not unique. This case study culverts. Road cross-drainage can be maintained presents a sobering example of what can happen by out-sloping the road surface or by road con- when apparently subtle changes in a watershed result struction with engineered continuous sub-grade in signi�cant impacts to the downstream fan. Ap- drainage. plication of the �ve-step method may not always pre- • In high-risk situations, it is professionally vent such impacts, but it does provide a framework responsible for the geoscientist or engineer to rigorously explore hydrogeomorphic hazards, undertaking terrain assessment or road engi- consequences, and risks. neering design to prescribe site visits during and

56 LITERATURE CITED

Association of Professional Engineers and Geosci- Bull, W.B. 1977. The alluvial-fan environment. Prog- entists of British Columbia (APEGBC). 2006. ress Phys. Geog. 1:222–270. Guidelines for legislated landslide assessments for proposed residential development in British Butler, D.R. and G.P. Malanson. 2005. The geomor- Columbia. www.apeg.bc.ca/ppractice/docu- phic in�uences of beaver dams and failures of ments/ppguidelines/guidelineslegislatedland- beaver dams. Geomorphology 71:48–60. slide.pdf Carson, B., D. Maloney, S. Chatwin, M. Carver, and B.C. Ministry of Forests. 2002. Forest Road Engi- P. Beaudry. 2007. Protocol for evaluating the neering Guidebook. For. Prac. Br., B.C. Min. potential impact of foresty and range use on For., Victoria, B.C. B.C. Forest Practices Code water quality (Water Quality Routine Eﬀective- of British Columbia Guidebook. ness Evaluation). Forest and Range Evaluation Program, B.C. Min. For. Range and B.C. Min. B.C. Ministry of Forests (BCMOF) and B.C. Minis- Environ., Victoria, B.C. www.for.gov.bc.ca/hfp/ try of Environment (BCMOE). 1996a. Channel frep/indicators/table.htm#water Assessment Procedures. Forest Practices Code Guidebook. Cave, P.W. 1992. Hazard acceptability thresholds for development approvals by local government. ______. 1996b. Channel Assessment Procedure In: Geologic hazards in British Columbia. B.C. Field Guidebook. Forest Practices Code Guide- Geological Survey Branch, Open File 1992- book. 15:15–26. ______. 1996c. Community Watershed Guidebook. Chatwin, S.C., D.E. Howes, J.W. Schwab, and D.N. Forest Practices Code Guidebook. Swanston (editors). 1991. Management of landslide prone terrain. B.C. Min. For., Victo- ______. 1999a. Watershed Assessment Procedure ria, B.C. Land Manag. Handb. 18. www.for.gov. Guidebook. 2nd ed., version 2.1. Watershed bc.ca/hfd/pubs/Docs/Lmh/Lmh18.htm Assessment Procedures. Forest Practices Code Guidebook. Costa, J.E. 1984. Physical geomorphology of debris �ows. In: Developments and applications of ______. 1999b. Mapping and Assessing Terrain geomorphology. J.E. Costa and J.P. Fleisher. Stability Guidebook. 2nd ed. Forest Practices (editors). Springer-Verlag, New York, N.Y., Code Guidebook. pp. 268–317. ______. 2001. Gully Assessment Procedure. 4th ed. ______. 1988. Rheologic, geomorphic, and sedimen- Forest Practices Code Guidebook. tologic diﬀerentiation of water �oods, hyper- B.C. Ministry of Transportation and Highways concentrated �ows, and debris �ows. In: Flood (BCMOTH). 1996. Natural hazards in British geomorphology. V.R. Baker, R.C. Kochel, and Columbia. B.C. Min. Trans. Highways, Victo- P.C. Patton (editors). John Wiley, New York, ria, B.C. N.Y., pp. 113–122. Beaty, C.B. 1970. Age and estimated rate of accumu- Curran, M.P., B. Chapman, G.D. Hope, and D. Scott. lation of an alluvial fan. White Mountains, 2006. Large-scale erosion and �ooding after California. Amer. J. Sci. 268:50–77. wild�res: understanding the soil conditions. B.C. Min. For. and Range, Res. Br., Victoria, Bovis, M.J. and M. Jakob. 1999. The role of debris B.C. Tech. Rep. 030. www.for.gov.bc.ca/hfd/ supply conditions in predicting debris �ow pubs/Docs/Tr/Tr030.htm activity. Earth Surf. Process. Landforms 24:1039–1054.

57 Dagg, B.R. 1987. Debris supply to torrent-prone Giles, T.R., W. Grainger, N. Skermer, and K.P. Turn- channels on the east side of Howe Sound, er. 2005. The Stemwinder Creek Mud Flow: an British Columbia. MSc thesis. Univ. British unusual event? In: Landslide Risk Management Columbia, Vancouver, B.C. (Supplementary Volume CD), Proc. Int. Conf. on Landslide Risk Management, Vancouver, Denny, C.S. 1965. Alluvial fans in the Death Val- B.C. O. Hungr, R. Fell, R. Couture, and E. Eb- ley region, California and Nevada. U.S. Geol. erhardt (editors). Balkema, London, U.K. Survey Prof. Paper 466. Grainger, W.W. 2002. Terrain stability �eld assess- ______. 1967. Fans and pediments. Am. J. Sci. ments in the “gentle-over-steep” terrain of the 265:81–105. southern interior of B.C. In: Terrain stability Denny Maynard and Associates Ltd. (DMA). 2001. and forest management in the interior of Brit- Terrain classi�cation, terrain stability, surface ish Columbia. P. Jordan and J. Orban (editors). erosion potential and sediment delivery poten- B.C. Min. For. Res. Prog., Victoria, B.C. Tech. tial of lower Kitimat-Chist and Hirsch-Wathl Report 003. www.for.gov.bc.ca/hfd/pubs/Docs/ Planning Units, TFL 41. Skeena Sawmills, A Tr/Tr003.htm Division of West Fraser Mills Ltd., Terrace, ______. 2007. Terrain Stability Assessment, CP 943, B.C. Block 1 to 7, Englishman Creek. Report for Fell, R. 1994. Landslide risk assessment and accept- Federated Cooperatives Ltd. able risk. Can. Geotech. J. 31:261–272. Grainger, W.W. and D.J. Wilford. 2004. The eﬀects Forest Practices Board. 2001. Forest practices and of �re on geomorphic processes. Streamline the Hummingbird debris �ow, Complaint Watershed Manag. Bull. 7(4):16. investigation 990189. Summary Rep. ______. 2008. Wathl Creek Hydrological Assess- ______. 2007. The eﬀects of mountain pine beetle ment. B.C. Min. For. and Range, Smithers, B.C. attack and salvage harvesting on stream�ows. Unpubl. rep. Spec. Investig. Rep. 16. Harr, R.D., R.L. Fredriksen, and J. Rothacher. 1979. Fuller, T. 2001. Sediment coring at Swansea Point fan Changes in stream�ow following timber har- delta, Mara Lake, British Columbia - vest in southwestern Oregon. USDA For. Serv. Application of a coring method to determine Res. Pap. PNW 249. historical debris �ow events. In: Terrain stabil- Hillman, G.R. 1998. Flood wave attenuation by a ity and forest management in the interior of wetland following a beaver-dam failure on a British Columbia. P. Jordan and J. Orban (edi- second order boreal stream. Wetlands 18:21–34. tors). B.C. Min. For. Res. Prog., Victoria, B.C. Tech. Rep. 003. www.for.gov.bc.ca/hfd/pubs/ Hogan, D. [2008]. Channel measures. In: Compen- Docs/Tr/Tr003.htm dium of forest hydrology and geomorphology in British Columbia. R.G. Pike et al. (editors). Geertsema, M. and J.J. Clague. 2005. Jokuilhaups at B.C. Min. For. Range Res. Br., Victoria, B.C. the Tulsequah Glacier, northwestern British and FORREX, Kamloops, B.C. Land Manag. Columbia, Canada. The Holocene 15(2):310–316. Handb. Chap. 15. In press. www.forrex.org/pro- Geertsema, M., J.J. Clague, J.W. Schwab, and S.G. gram/water/compendium.asp Evans. 2006. An overview of recent large cata- Hogan, D.L. and D.J. Wilford. 1989. A sediment strophic landslides in northern British Colum- transfer hazard classi�cation system: linking bia, Canada. Eng. Geol. 83(2006):120–143. erosion to �sh habitat. In: Proc. of Watershed ’89. E.B. Alexander (editor). USDA For. Serv., Alaska Region. R10-MB-77, pp. 143–155.

58 Hooke, R. LeB. 1968. Steady-state relationships on Jakob, M.J., D. Anderson, T. Fuller, O. Hungr, and arid-region alluvial fans. J. Geol. 75:438–460. D. Ayotte. 2000. An unusually large debris �ow at Hummingbird Creek, Mara Lake, British Horel, G. 2006. Summary of landslide occurrence Columbia. Can. Geotech. J. 37:1109–1125. on northern Vancouver Island. Streamline 10(1):1–9. Jakob, M. and O. Hungr. 2005a. Debris-�ow hazards and related phenomena. Springer-Praxis, Ber- Huggard, D. 2006. Effects of salvage options for lin, Heidelberg, New York. beetle-killed pine stands on ECA. B.C. Min. Environ., Kamloops, B.C. Unpubl. rep. ______. 2005b. Introduction. In: Debris-�ow hazards and related phenomena. M. Jakob and Hungr O., 2005. Classi�cation and terminology. In: O. Hungr (editors) Springer-Praxis, Berlin, Debris-�ow hazards and related phenomena. Heidelberg, New York. M. Jakob and O. Hungr. (editors). Springer- Praxis, Berlin, Heidelberg, New York. Jakob M. and P. Jordan. 2001. Design �ood estimates in mountain streams – the need for a geomor- Hungr, O., S.G. Evans, M.J. Bovis, and J.N. Hutchin- phic approach. Can. J. Div. Eng. 28:425–439. son. 2001. A review of the classi�cation of landslides of the �ow type. Environ. and Eng. Jordan, P. 1994. Debris �ows in southern Coast Geosci. Vol. VIII, No. 3, pp. 221–238. Mountains. PhD thesis. Dep. Geog., Univ. Brit- ish Columbia, Vancouver, B.C. Hungr, O., G.C. Morgan, and R. Kellerhals. 1984. Quantitative analysis of debris torrent hazards ______. 2002. Landslide frequencies and terrain for design of remedial measures. Can. Geo- attributes in Arrow and Kootenay Lake Forest tech. J. 21:663–677. Districts. In: Terrain stability and forest management in the interior of British Columbia. Hunt, C.B. and D.R. Mabey. 1966. Stratigraphy and P. Jordan and J. Orban (editors). B.C. Min. For. structure. Death Valley, Calif. U.S. Geol. Sur., Res. Prog., Victoria, B.C. Tech. Rep. 003. www. Prof. Pap. 494-A. for.gov.bc.ca/hfd/pubs/Docs/Tr/Tr003.htm Innes, J.L. 1985. Magnitude-frequency relations of Jordan, R.P., M. Curran, and D. Nicol. 2004. Debris debris �ows in northwest Europe. Geograf. �ows caused by water repellent soils in recent Ann. 67A:23–32. burns in the Kootenays. Assoc. Prof. Eng. and Inter-agency Report. 1997. Hummingbird Creek Geosci., ASPECT 9(3): 4–9. debris event – July 11, 1997. B.C. Min. Water King, J.G. and L.C. Tennyson. 1984. Alteration of Land and Air Protection, B.C. Min. For., B.C. stream�ow characteristics following road Min. Transp. Highways, B.C. Min. Attorney construction in North Central Idaho. Water General, Victoria, B.C. Resourc. Res. 20(8):1159–1163. Jackson, L.E., R.A. Kostaschuk, and G.M. MacDon- Larsen, I.J., L.H. MacDonald, E. Brown, D. Rough, ald. 1987. Identi�cation of debris �ow hazard M.J. Welsh, J.H. Pietraszek, Z. Libohova, J. de on alluvial fans in the Canadian Rocky Moun- Dios Benavides-Solorio, and K. Schaffrath. tains. In: Debris �ows/avalanches: process, Causes of post-�re runoff and erosion: the roles recognition, and mitigation. J.E. Costa and of soil water repellency, surface cover, and soil G.F. Wieczorek (editors). Geol. Soc. Am., Rev. sealing. Soil Sci. Soc. Am J. In press. in Eng. Geol. Vol. VII. pp. 115–124. Melton, M.A. 1965. The geomorphic and paleoclimat- Jakob, M. 2000. The impacts of logging on landslide ic signi�cance of alluvial deposits in southern activity at Clayoquot Sound, British Columbia. Arizona. J. Geol. 73:1–38. Catena 38:279–300.

59 Miles, M.J. and R. Kellerhals. 1981. Some engineer- Pierson, T.C. and K.M. Scott. 1985. Downstream di- ing aspects of debris torrents. Proc. Can. Soc. lution of a lahar: transition from debris �ow to of Civil Eng., Fifth Can. Hydrotechnical Conf. hyperconcentrated stream�ow. Water Resourc. Fredericton, N.B., pp. 395–420. Res. 21:1511–1524. Millard, T.H. 1999. Debris �ow initiation in coastal Ritter, J.B., J.R. Miller, Y. Enzel, S.D. Howes, G. British Columbia gullies. B.C. Min. For., Nadon, M.D. Grubb, K.A. Hoover, T. Olsen, Vancouver For. Reg., Nanaimo, B.C. Tech. Rep. S.L. Reneau, D. Sack, C.L. Summa, I. Taylor, TR-002. www.for.gov.bc.ca/hfd/pubs/Docs/Tr/ K.C.N. Touysinhthiphonexay, E.G. Yodis, N.P. Tr002.htm Schneider, D.F. Ritter, and S.G. Wells. 1993. Quaternary evolution of Cedar Creek alluvial Millard, T.H., D.L. Hogan, D.J. Wilford, and B. fan, Montana. Geomorphology 8:287–304. Roberts. A method to assess fan channel networks, with a preliminary application to fans Roberts, B., B. Ward, and T. Rollerson. 2004. A com- in coastal British Columbia. Geomorphology. parison of landslide rates following helicopter Alluvial Fans 2007 Special Issue. In press. and conventional cable-based clear-cut logging operations in the Southwest Coast Mountains Millard, T.H., D.J. Wilford, and M. Oden. 2006. of British Columbia. Geomorphology Coastal fan destabilization and forest man- 61(3–4):337–346. agement. B.C. Min. For. Range, Coast Forest Region, Nanaimo, B.C. Tech. Rep. 034. Runka, G.G. 1972. Soil resources of the Smithers– Hazelton area. B.C. Dep. Agric. Soil Survey Moody, J.A. and D.A. Martin. 2001. Post-�re, rain- Div., Kelowna, B.C. fall intensity–peak discharge relations for three mountainous watersheds in the western USA. Ryder, J.M. 1971a. The stratigraphy and morphology Hydrol. Process. 15:2981–2993 of paraglacial alluvial fans in south-central British Columbia. Can. J. Earth Sci. 8:279–298. Osterkamp, W.R. and C.R. Hupp. 1987. Dating and interpretation of debris �ows by geologic and ______. 1971b. Some aspects of the morphometry of botanical methods at Whitney Creek Gorge, paraglacial alluvial fans in south-central Brit- Mount Shasta, California. In: Debris �ows/ ish Columbia. Can. J. Earth Sci. 8:1252–1264. avalanches: process, recognition, and mitigation. J.E. Costa and G.F. Wieczorek (editors). Sakals, M.E., J.L. Innes, D.J. Wilford, R.C. Sidle, Rev. in Eng. Geol. Vol. VII. Geol. Soc. Am., pp. and G.E. Grant. 2006. The role of forests in 157–163. reducing hydrogeomorphic hazards. For. Snow Landsc. Res. 80(1):11–22. Pack, R.T., D.G. Tarboton, and C.N. Goodwin. 1998. TheSINMAP approach to terrain stability map- Schnorbus, M.A. and Y. Alila. 2004. Forest harvest- ping. In: Proc. Eighth Int. Cong. Int. Assoc. ing impacts on the peak �ow regime in the Eng. Geol. and Environ. D. Moore and Columbia Mountains of southeastern British O. Hungr (editors). Balkema, Rotterdam, Columbia: an investigation using long-term Netherlands. numerical modelling. Water Resourc. Res. 40, W05205, doi:10.1029/2003WR002918. Pierson, T.C. and J.E. Costa. 1987. A rheologic classi�cation of subaerial sediment-water �ows. In: Schwab, J.W. 2000. The next big storm or �ood. B.C. Debris �ows/avalanches: process, recognition, Min. For. Range, Northern Interior For. Reg. and mitigation. J.E. Costa and G.F. Wieczorek Exten. Note 41. (editors). Geol. Soc. Am. Rev. in Eng. Geol. Schwab, J.W. and M. Geerstema. 2008. Terrain Vol. VII, pp. 1–12. stability mapping on British Columbia forest

60 lands: a historical perspective. In: Fourth Can. Wells, S.G. and A.M. Harvey. 1987. Sedimentologic Conf. on Geohazards: from cause to manage- and geomorphic variations in storm-generated ment. May 20–24, 2008. J. Locat, D. Perret, D. alluvial fans, Howgill Fells, northwest England. Demers, and S. Leroueil (editors). Presse de Geol. Soc. Am. Bull. 98:182–198. l’Université Laval, Québec, Que., pp. 477–484. Wemple, B.C., J.A. Jones, and G.E. Grant. 1996. Scott, D.F. and D.B. Van Wyk. 1990. The effects of Channel network extension by logging roads in wild�re on soil wettability and hydrological two basins, Western Cascades, Oregon. J. Am. behaviour of an afforested catchment. J. Hy- Water Resourc. Assoc. 32(6):1195–1207. drol. 121:239–256. Wemple, B.C. and J.J. Jones. 2003. Runoff production Septer, D. and J.W. Schwab. 1995. Rainstorm and on forest roads in a steep, mountain catchment. �ood damage: northwest British Columbia Water Resourc. Res. 39 (8), 1220, doi:10.1029/ 1891–1991. B.C. Min. For., Res. Br., Victoria, 2002WR001744. B.C. Land Manag. Handb. 31. www.for.gov. bc.ca/hfd/pubs/Docs/Lmh/Lmh31.htm Whitaker, A., Y. Alila, and J. Beckers. 2002. Evalu- ating peak �ow sensitivity to clear-cutting in Sidle, R.C., A.J. Pearce, and C.L. O’Loughlin. 1985. different elevation bands of a snowmelt- Hillslope stability and land use. (Vol. II). Am. dominated mountainous catchment. Water Geophys. Union, Washington, D.C. Resourc. Res. 38 (9), 1172, doi:10.1029/2001 WR000514. Smith, G.A. 1986. Coarse-grained nonmarine vol- caniclastic sediment: terminology and deposi- Wilford, D.J. 2003. Forest stand characteristics as in- tional process. Geol. Soc. Am. Bull. 97:1–10. dicators of hydrogeomorphic activity on fans. PhD thesis. Univ. British Columbia, Vancouver, Statistics Canada. 2007. 2006 Census Pro�le, B.C. Swansea Point UNP. Census of Population and Housing. Wilford, D.J., P. Cherubini, and M.E. Sakals. 2005a. Dendroecology: a guide for using trees to Strahler, A.N. and A.H. Strahler. 1987. Modern date geomorphic and hydrologic events. B.C. physical geography. 3rd ed. John Wiley and Min. For., Res. Br., Victoria, B.C. Land Manag. Sons, New York, N.Y., p. 191. Handb. 58. www.for.gov.bc.ca/hfd/pubs/Docs/ Strunk, H. 1997. Dating of geomorphological pro- Lmh/Lmh58.htm cesses using dendro-geomorphological meth- Wilford, D.J., M.E. Sakals, and J.L. Innes. 2002. ods. Catena 31:137–151. Forestry on fans: identifying hydrogeomorphic Takahashi, T. 1981. Debris �ows. Annu. Rev. Fluid hazards. In: Terrain stability and forest man- Mech. 13:57–77. agement in the interior of British Columbia. P. Jordan and J. Orban (editors). B.C. Min. For., Thurber Consultants Ltd. 1987. Debris torrent haz- Res. Br., Victoria, B.C. Tech. Rep. 003. www.for. ards along Highway 1: Sicamous to Revelstoke. gov.bc.ca/hfd/pubs/Docs/Tr/Tr003.htm Report to B.C. Min. Transp. Highways. ______. 2003. Forestry on fans: a problem analysis. VanDine, D.F. 1985. Debris �ows and debris torrents For. Chron. 79(2):291–296. in the southern Canadian Cordillera. Can. Geotech. J. 22:44–68. ______. 2005b. Forest management on fans: hydrogeomorphic hazards and general prescrip- ______. 1996. Debris �ow control structures for tions. B.C. Min. For., Res. Br., Victoria, B.C. forest engineering. B.C. Min. For., Res. Br., Land Manag. Handb. 57. www.for.gov.bc.ca/ Victoria, B.C. Work. Pap. 22. www.for.gov. hfd/pubs/Docs/Lmh/Lmh57.htm bc.ca/hfd/pubs/Docs/Wp/Wp22.htm

61 ______. 2006. Fans with forests. In: Alluvial fans: Winkler, R.P., D.L. Spittlehouse, and D.L. Golding. geomorphology, sedimentology, dynamics. 2005. Measured diﬀerences in snow accumula- A.M. Harvey, A.E. Mather, and M. Stokes (edi- tion and melt among clearcut, juvenile, and tors). Geol. Soc. of London, London, U.K. mature forests in southern British Columbia. Hydrol. Process. 19:51–62. Wilford, D.J., M.E. Sakals, J.L. Innes, and D. Rip- meester. 2004a. Kitsequecla fan case study: Wise, M.P., G.D. Moore, and D.F. VanDine (edi- speci�c risk analysis. In: Landslide risk case tors). 2004. Landslide risk case studies in forest studies in forest development planning and development planning and operations. B.C. operations. M.P. Wise, G.D. Moore, and D.F. Min. For., Res. Br., Victoria, B.C. Land Manag. VanDine (editors). B.C. Min. For., Res. Br., Handb. 56. www.for.gov.bc.ca/hfd/pubs/Docs/ Victoria, B.C. Land Manag. Handb. 56. www. Lmh/Lmh56.htm for.gov.bc.ca/hfd/pubs/Docs/Lmh/Lmh56.htm Woods, A., K.D. Coates, and A. Hamann. 2005. Is Wilford, D.J., M.E. Sakals, J.L. Innes, R.C. Sidle, and an unprecedented dothistroma needle blight W.A. Bergerud. 2004b. Recognition of debris epidemic related to climate change? BioScience �ow, debris �ood and �ood hazard through 55(9):761–769. watershed morphometrics. Landslides 1(1):61–66. Winkler, R. and J. Roach. 2005. Snow accumulation in B.C.’s southern interior forests. Streamline Watershed Manag. Bull. 9(1):1–5. www.forrex. org/publications/streamline/iss30/streamline_ vol9_no1.pdf