Denudation and landslides in coastal mountain watersheds : 10,000 years of erosion

Nutzungsbedingungen Die ETH-Bibliothek ist Anbieterin der digitalisierten Zeitschriften. Sie besitzt keine Urheberrechte an den Inhalten der Zeitschriften. Die Rechte liegen in der Regel bei den Herausgebern. Die auf der Plattform e-periodica veröffentlichten Dokumente stehen für nicht-kommerzielle Zwecke in Lehre und Forschung sowie für die private Nutzung frei zur Verfügung. Einzelne Dateien oder Ausdrucke aus diesem Angebot können zusammen mit diesen Nutzungsbedingungen und den korrekten Herkunftsbezeichnungen weitergegeben werden. Das Veröffentlichen von Bildern in Printund Online-Publikationen ist nur mit vorheriger Genehmigung der Rechteinhaber erlaubt. Die systematische Speicherung von Teilen des elektronischen Angebots auf anderen Servern bedarf ebenfalls des schriftlichen Einverständnisses der Rechteinhaber.


Introduction
Landslides are primary denuders of the landscape since they directly transport Sediment from upslope sources to both stream networks and lower more stable positions. Precipitation and earthquake triggered landslides in coastal British Columbia, Canada, annually erode the surrounding landscape concurrent with other dynamic modes of erosion such as stream incision and runoff. Here, we present a conceptual model of landslide-induced denudation for coastal mountain watersheds spanning 10,000 years of environmental change. Given that climate has varied substantially during the Holocene from warm-dry to cool-wet, the model fosters important insight into the interaction between climate and landslide-induced denudation. Further, the model considers recent and deleterious anthropogenic activity, mainly logging, and provides a framework by which human-induced denudation rates can be contrasted to those of the Holocene.
Setting. Vancouver Island is located off the southwest coast of British Columbia, Canada ( Figure 1). The island is comprised of 31,788 km2 of highly variable terrain, with the interior of the island containing the steep and rugged volcanic and intrusive Vancouver Island Ranges (Guthrie 2005a;Massey et al. 2003aMassey et al. , 2003bYorath & Nasmith 1995).The largest mountain peaks attain elevations of c. 2,200 m. Average annual precipitation varies longjtudinally across the island, with eastern rain shadow areas receiving as little as 700 mm of annual rainfall compared to >3,500 mm of rainfall on the oceanic west coast (Environment Canada 1993. The moist and müd cümate Supports widespread temperate rainforest in the lowlands.
At high elevations, cooler temperatures coincide with alpine forest and tundra.
The island is located near the surface trace of the Cascadia subduction zone and is tectonically active (Adams 1984;Clague & Iames 2002;Dragert 1987). At least two earthquakes of sufficient magnitude to cause landsüdes occurred in the last Century (Cassidy et al. 1988;Keefer 1984;Mathews 1979;Rogers 1980). Further, on a longer (semi-miüennial) timescale, Vancouver Island is subject to large earthquakes of >8 magnitude (Atwater 1987;Leonard et al. 2004;Satake 1995). In addition to tectonic activity, Vancouver Island has been significantly modified by Pleistocene glaciation (Clague & Iames 2002;Ryder & Clague 1989), resulting in steep U-shaped Valleys in the mountainous regions. On the mid and lower slopes, tiü and glaciofluvial deposits blanket the landscape.
Subsequent post-glacial erosion and denudation has fostered the development of widespread shallow colluvium.
The aforementioned geologjc, physiographic, climatic and tectonic regjmes of Vancouver Island have combined to produce a steep, youthful terrain that is generally prone to mass wasting. Landslide types typical to Vancouver Island include slides, slumps, flows, falls and topples in debris and rock according to the Varnes (1978) Classification. Debris slides and flows ( Figure 2A) are most common and almost 20 times more frequent than rock falls in the forested areas (Guthrie 2005b). Debris slides and flows are defined as extremely rapid, shallow mass movements of unconsolidated material that usually begin as translational failures. These movements typically break up as velocity or water content increases, ultimately forming an avalanche (dry) or flow (wet). Herein, the term «landslide» refers to events of this category. Channelized debris flows (CDFs) occur when a debris flow enters a confined Channel CDFs usually travel considerable distances and are common in coastal British Columbia ( Figure 2B). It is likely that CDFs are under-represented in air photograph interpretations as smaller events of this type tend to have a short persistence time in the landscape. There is not always a clear and objective distinction between channelized and unchannellized events and they are undifferentiated in the following discussions. Finally, rock falls constitute an extremely rapid displacement of rock from a steep surface, usually characterized by some component of falling through the air, bouncing and/or rolüng of material. Rock falls often break up on impact and continue down slope with fluid behaviour, typically referred to as an avalanche or Sturzstrom. Rock falls ränge in size from smaü (<1 m3) to large (>1 Mm3), with the largest rock fall avalanches initiating in the alpine zone (Guthrie 2005b).
Glaciers retreated rapidly at the end of the Pleistocene on Vancouver Island, with most upland areas free of ice by 13,000 radiocarbon years before present (14C y BP, Alley & Chatwin 1979). At this time, the combination of isostatic uplift, exposed bedrock, uncon-  Guthrie (2005b) indicates that the modern alpine zone produces approximately 4 times as many landslides as the wet west coast of Vancouver Island, including a larger number of rock falls and avalanches. Large events such as rock fall-avalanches were likely more common around 13,000 14C y BP as the newly debuttressed landscape sought to establish equüib-were similar to those of present-day, namely precipitation-induced debris südes and debris flows (Figures 2 and 3) with rock falls playing a less significant role.
It is also conceivable that large rock fall-avalanches were relatively inconsequential in the geomorphological development of the post-glacial landscape as they occurred too infrequently below 800 m (Guthrie & Evans 2007).

Methods
Overwhelming evidence reveals that Vancouver Island was ice free, vegetated and dry by 11,700 calendar years before present (y BP), foüowing a rapid rise in temperature (Alley & Chatwin 1979;Brown et al. 2006;Carlson 1979;Hay et al. 2007;Hebda 1983). Foüowing the warm-dryxerothermicinterval (11,700-7,000 y BP), an increase in precipitation coincides with the Start of the wetter mesothermic interval (7,000-4,000 y BP).
At this time, western hemlock (Tsuga het-erophylld) expanded on the island and the vegetated landscape began to resemble that of today. During the last several mülennia, precipitation has remained relatively stable. Consequently, it is possible to surmise that the type and form of landsüdes throughout the Holocene 2.1 Determining Holocene landslide rates Previously, a landslide potential map of Vancouver Island divided the island into four major categories based predominantly on the slope and climatic regime ( Photo: R.H. Guthrie exposed small outcrops, precipitation between 1600-2600 mm-y4 falling mostly in winter months. Landslides are typically debris slides and flows with minor numbers of rock falls. Zone III -The moderately dry east coast, characterized by more exposed bedrock and lower rainfall (<1600 mm-y4), increased urbanization and rural development and shallower slope gradients. One quarter of all landslides identified were rock falls.
Zone TV -The alpine zone, characterized by high elevation steep cüffs and plateaus, exposed bedrock, ponded water, steep gorges and sparse Vegetation, most of the precipitation falling as snow in the winter months. Landslides commonly include rock falls, rock avalanches, debris slides and debris flows and regularly result in the accumulation of coalescing talus slopes. Snow avalanches are similarly common and often related to land instabüity Consistent differentiation of landslides in the alpine zone is problematic due to their relatively high frequency and overlapping distribution. Through time, however, the frequency of landslides in the alpine zone is likely to fluctuate in response to changing biogeoclimatic conditions. For example, during the Little Ice Age, a drop in temperature of about 1°C caused alpine glaciers on Vancouver Island to advance several hundred meters in mountain Valleys (Lewis & Smith 2004;Smith & Laroque 1996). Fewer landslides are expected when slope walls are buttressed by ice, even if there is an increase in glacial erosion, whereas during and after glacial retreat the number of landslides is likely to increase as over-steepened, eroded and weathered slopes become exposed. Ultimately, however, the variabüity in climate related landsüde potential is expected to be minor in the alpine zone compared to the overaü high incidence of landslides that characterize this steep and sparsely vegetated In contrast to the alpine zone, the three non-alpine landsüde zones I-III (Figure 1), appear to be highly sensitive to changes in cümate, particularly precipitation (Guthrie 2005b 1000 year intervals for the Holocene and showed that overall, temporal changes in precipitation were generaüy subtle, though a notable increase in precipitation is observed at the end of the Holocene dry period. In the development of a conceptual mass wasting model, we incorporated the strong direct linkage between the incidence of landslides and precipitation (Guthrie 2005b). Present-day precipitation isopleths were overlain on the mass wasting zones of Guthrie (2005b), revealing that the precipitation isopleths are in broad agreement with the mass wasting map ( Figure 4). Subsequently, each 1000 year interval was assigned an ordinal category of either drier than present, modern (similar to present), or wetter than present based on the variability of precipitation compared to present-day. The mass wasting potential maps for the wet and dry climatic intervals (Figure 5) were established by shifting zones I-III in proportion to the modelled climatic changes, calibrated against the modern analog and taking into account elevation effects. Landslide frequencies were then estimated using the established rates for each zone (Table 1). As previously discussed, the alpine zone was excluded from the analysis. In addition, a physiographic region of low relief and low gradients on the east side of Vancouver Island, the Nanaimo Lowlands, was also excluded since it generally does not contain unstable terrain.
2.2 Determining 20* Century landslide rates Recent logging and road building have significantly altered landslide frequencies in coastal British Columbia, with reported increases ranging between 3-34 times the modern natural rates (Chatwin 2005;Guthrie 2002;Guthrie 2005b;Iakob 2000;Iordan 2003;Rood 1984;Schwab 1983). In addition to directly removing forest cover and changing the hydrologic regime, secondary forest harvest activities such as road building can also intercept, concentrate and reroute water to new locations on the hill slope. Guthrie (2005b) suggests that an order of magnitude increase in landslide frequencies reasonably reflects the impact of logging.

Widespread commercial logging began on Vancouver
Island at the onset of the 20th Century and continues today. The rate of logging was estabüshed using forest cover maps during two 50-year periods. To determine a landslide frequency that accounts for logging, the natural landslide rates were calculated to incorporate the accumulated area logged each year. Harvested cut-blocks were given a hydrologic recovery time of 25 years, representing the time required for a new forest to establish and exceed 10 m in height. Recovery was limited to 90% of the harvest to account for residual hazards. The total harvested area was subtracted from the unlogged area for each year in each zone, and the frequencies calculated by increasing the numbers of landsüdes in the area harvested by an order of magni-  (Brown et al. 2006) and mass wasting potential (Guthrie 2005b), revealing broad agreement. Zones I-III refer to the masswasting potential zones.
The activities of these early people on landsüde potential was likely negligible. A more pronounced impact by people, particularly on steep slopes, is clearly evident during the last 100 years as a result of humaninduced landscape modification. Approximately 56% of Vancouver Island has been altered in some way by agricultural activity, urbanization or logging. As the most prominent, logging has been both widespread and extensive, with some plantations in their third rotation. As a consequence of these activities, much of the remaining old growth forest is located in protected parks and reserves as well as in other areas with uneconomic timber. At high elevation, some old growth forest remains in areas with difficult access, typically on slopes steeper than 30°.
By 50 years ago, the island wide total landslide rate was an estimated 303 landslides-y4, with the rate increasing to modern levels of about 402 landslides-y4 or 0.015 landsüdes-y4-km2. These figures reveal that the average landslide rate (below the alpine zone) in the last 50 years is close to twice the highest average landsüde rate in the last 10,000 years. Thus, the impact of modern human action, such as logging, must be recognised as having a significant, and perhaps deleterious, affect that may exceed all previous Variation in natural landslide rates. movement potential zones for different climatic regimes, including a wetter regime present in the middle Holocene, a modern climate, and a drier climate present in the early Holocene Dark grey indicates zone I, zone II is medium grey and zone III hght grey. Excluded from landslide calculations are the alpine zone (white) and the flat Nanaimo Lowlands (white).
Zonen der potentiellen Erdmassenbewegungen in verschiedenen Klima-Typen: ein feuchtes Klima während des mittleren Holozäns, gegenwärtiges Klima und trockenes Klima während des frühen Holozäns Zones de mouvements de masses potentiels pour differents regimes climatiques, y compris un regime plus humide present au milieu de l'Holocene, un climat moderne et un climat plus sec present au debut de l'Holocene Cartography: R.H. Guthrie Given the nature of the human impact compared to past climatic shifts, one can argue that an improvement in logging practices is perhaps the Single most effective way to adapt to any future climate change scenarios.

Sediment yield
Magnitude-frequency characteristics of debris slides and debris flows on Vancouver Island were derived previously (Guthrie 2005b;Guthrie & Evans 2004, 2005, enabling an estimation of the total landslide impact in terms of area affected and volume of Sediment delivered. The mean total area of debris slides and debris flows is about 9,500 m2 on Vancouver Island, though ranging between 7,500-11,500 m2. where V landslide volume in m3 and A total area in m2. Multiplying the annual frequency by the mean total area and volume, and summing over the last 10,000 years yields a total of 1.75xl010 m2 and 6.2xl09 m3 respectively, approximately 1.2xl08 m3 of material eroded from the slopes in the last 100 years alone. Further, a total of 9,385 km2 on Vancouver Island is susceptible to landslides, given that 37% of Vancouver Island is designated as having landslide potential (Guthrie 2005b). An estimate of total down-wasting by landslides can be calculated by dividing the total estimated volume by the area avaüable for landsüde initiation, yielding 0.7 m of down-wasting on the steep slopes of Vancouver Island during the past 10,000 years. The reader is reminded, however, that this is an averaged result, and that the landscape wiü erode preferentially on steep sites with sufficient avaüable Sediment.

Conclusions
Landslides in coastal British Columbia are dominated by precipitation-induced shallow debris slides and flows. These events are responsible for much of the primary erosion by slope failure and play a significant role in shaping the landscape by transporting Sediment from upslope sources to lower more stable positions or into stream networks where the material can be removed.
Landslide frequencies were estimated for the past 10,000 years on Vancouver Island by examining climatic shifts in the vegetative record and, using modern conditions and associated frequencies as an analog, comparing those shifts to expected changes in landslide potential. The results suggest an initially low incidence of landslides in the early Holocene, followed by a substantial increase in landslide frequency between the Holocene dry period and the warm wet mesothermic interval in the mid Holocene. Thereafter, there is a slight reduction in landslide frequency at 3000 y BP after which landslide frequency remains relatively constant until recent human action drastically altered the landslide regime. Landslide rates varied between 0.005-0.008 landsüdes-y4-km2 during that time.
The impact of logging during the last 100 years is unambiguous as landslide frequency increased to 0.015 landsüdes-y x-km2. This increase reveals that the impact of logging outpaces that of climatic change. Thus, improving logging practices will help offset any potential increase in landslide incidence induced by cümate change.
Based on a mean landslide size it is estimated that debris slides and flows eroded an average of 0.7 m-nr2 across the Vancouver Island during the last 10,000 years.