ABSTRACT: The record of disastrous landslides and geotechnical failures in Canada 1840-1999 ; implications for risk management.

CGRG Bibliography of Canadian Geomorphology
Search Results

Author :

Date :

Title :

Publication :

Issue :

Page(s) :

Abstract

Historical records of natural disasters are important in understanding disastrous processes (mechanics and causes), historical frequency and magnitude, and in pointing to ways in which they may be avoided or mitigated in future. They are also of considerable use in developing landslide risk criteria and can be used as reference levels in landslide risk assessment. As part of the Canadian contribution to the IDNDR a verified data base of disastrous landslides and geotechnical failures in the historical period has been assembled (Evans, 1997; 1999; Evans et al., 1997). The project is now considered to be complete. This paper presents additional data and a revised analysis as well as exploring the contribution of technological landslide risk to landslide risk in Canada. Events of concern in the present context are landslides in natural slopes or failures of artificial slopes, either built or excavated. The latter group of events are termed geotechnical failures. Evans (1997) defined a disastrous event in the Canadian context, as a single event failure which resulted, directly or indirectly, in the deaths of 3 or more people. Evans (1997, 1999) identified the two areas of Canada that are particularly vulnerable to damaging landslides and which have experienced the vast majority (91%) of the Nation’s disastrous landslides and geotechnical failures. To obtain some idea of the length of population exposure to landslide hazard, it is of interest to define the historical period. In addition, the nature of the historical landslide record for the two regions is examined, recognizing that landslides must have affected aboriginal peoples (Evans, 1997) in the pre-European period. St. Lawrence Lowlands - European settlements were not established in eastern Canada until the first decade of the seventeenth century culminating in the founding of Quebec by Champlain in 1608. The first European missionary arrived in New France in 1615 and the first settler in 1617 (Careless, 1970). The first reported landslides are those in Leda Clay associated with the great earthquake (M ~ 7.0) which struck New France on February 5, 1663. Although the region was sparsely populated, the direct effects of the earthquake are recorded in correspondence by religious orders active in the area at the time. Indirect evidence of the earthquake’s effects have been obtained by the radiocarbon dating of landslide debris (LaSalle and Chagnon, 1968; Leggett and LaSalle, 1978; Desjardins, 1980; Quilliam and Allard, 1989; Filion et al. 1991), dendrochronology (Filion et al. 1991), and the analysis of sediments in lakes and fiords (summarised in Syvitski and Schafer, 1996). It is evident that widespread landsliding took place throughout New France. We have not uncovered further reference to major landslides in the St. Lawrence Lowlands until 1771, the date of the earliest documented landslide at St-Pierre-de-la-Rivière-du-Sud, Quebec, in 1771. The landslideis described in The Annual Register (p. 164) for the year 1771 (Anonymous, 1779) and was a rapid retrogressive Leda Clay earthflow. One person was killed in the landslide when a farmhouse was buried in the debris. Although the historical period is ca. 384 years, systematic landslide data exists only after 1840 (Evans, 1997), a record of 161 years. Cordillera of western Canada - In western Canada the length of the historical period is considerably shorter. Alexander Mackenzie reached the Pacific at Bella Coola on July 22, 1793 only a month before Captain George Vancouver arrived on the coast by sea (Careless, 1970). This historic date gives the length of the historical period to 2000 of only 208 years. The first observations of a catastrophic landslide in the Cordillera are those made in 1858 describing the aftermath of the Rubble Creek landslide, southwest British Columbia, which occurred in the winter of 1855-56 (Hardy et al., 1978; Moore and Matthews, 1978). This timeline broadly corresponds to the beginning of scientific exploration in the Cordillera, began by The Palliser Expedition in 1857. The Cariboo gold rush of 1858 opened up the Interior of British Columbia, leading to the first newspaper report of a damaging landslide in the Cordillera at Barkerville, the centre of the Cariboo Gold Rush, in May 1866. This date marks the beginning of the systematic landslide record in the Cordillera giving a length of record to 2000 of only 135 years. Thus despite differences in the historical period, the length of the systematic landslide record in both regions is comparable. The sources of landslide disaster data are outlined by Evans (1997, 1999). Landslides and Geotechnical Failures; the historical record - A total of 43 events that met the nominal national disaster criterion defined above, occurred in Canada in a period of 160 years between 1840 and 1999 (Table 1). This is equivalent to a landslide disaster frequency of one every 3.7 years, or an annual frequency of 0.27. These disasters resulted in a minimum of 570 deaths. Assessment of historical record - Landslide type - A re-examination of the 1915 Jane Camp disaster (Evans, 2000) and the addition of a further three landslide disaster events, has led to an assessment of destructive landslide types somewhat different to the first approximation suggested by Evans (1997). Based on Table 1 the most destructive were small-scale rockslides and rockfalls involving volumes of less than 100,000 m3. These caused 27 % (155) of the deaths in seven events across Canada. Second, in terms of destructiveness were landslides in Leda Clay which caused 17.0% (98) of the deaths in 9 events in the St. Lawrence Lowlands of Quebec. The third most destructive type weregeotechnical failures involving the failure of man-made slopes, which accounted for 84 deaths, 15 % of the total. The fourth most destructive landslide types are rock avalanches involving volumes equal to or in excess of 100,000 m3. Rock avalanches caused 15 % (83) of the total deaths in only 3 disaster events in the Cordillera. Fifth, were debris flows and debris avalanches involving volumes of 60,000 m3 or less. These landslides caused 11% (65) of the disaster deaths in 13 events. 125 deaths (22 % of the total) were caused by rapid flowslide-type movements in Quaternary sediments (glaciolacustrine and glaciomarine) in 12 events in British Columbia and Quebec. It is noted that there is no simple correlation between deaths per event and event magnitude, nor is the record dominated by large magnitude events usually associated with landslide disasters. Indeed, the data suggests that the greatest number of landslide deaths have been caused by small magnitude-high frequency landslides. Secondary effects - Many of the landslide deaths were caused by the secondary effects of landslides and related geotechnical failure, i.e., landslide-generated displacement waves andoutburst floods. These caused 147 (26%) of the total number of deaths in only 5 events. The role of human forcing - technological landslide risk - Landslide disasters by their very nature occur in the vicinity of human activity which frequently has altered the natural condition of the failed slope. Changes in slope condition due to deforestation, irrigation, excavation in the surface or subsurface, are common. Fills or embankments may impart loading to slopes and block surface or subsurface drainage. Blasting may result in the dynamic loading of slopes. Thus, a listing of historical landslide disasters contains events which are partly, or largely, due to some element of human forcing and may thus be considered technological disasters. In the Canadian record 130 of the deaths (23%) are related to landslides caused directly or indirectly by human activity. These include deaths due to embankment failures, a landslide in an open pit mine that penetrated underground workings, and two deadly Leda Clay landslides caused by blasting. If the problematical 1903 Frank Slide, which may have been triggered by the effects of coal mining at the base of Turtle Mountain (Benko and Stead, 1998), and other landslides in which human activity was an important causal factor are included, this totalbecomes 261 or 46% of the total deaths. If landslide disasters found to be the result of human negligence (Evans, 1997) it could be argued that over 50% of the landslide deaths in the record are due directly and indirectly to human activity. A historical landslide disaster listing is therefore a composite record of events due to human and natural causes in varying proportions. Its use in the construction of F/N curves (Evans, 1997) must recognise that human activity has interacted with natural systems in complex ways generating hybrid events which blur the distinction between natural and man-made disasters. This is the case even though the failure of an artificial or modified natural slope system may be in response to an external natural trigger, most frequently an extreme meteorological event. A further point is that human activity not only increases risk by amplifying frequency but also increases risk by increasing vulnerability. In the Canadian case, the natural landslide magnitude and frequency signal is significantly distorted by human forcing over a wide range of landslide magnitudes. Part of this distortion consists of an amplification of frequency resulting in increased landslide risk. In Table 1 a rough partition has been made between natural landslides and those involving artificial slopes, built slopes, or landslides in natural slopes triggered by human activity. These data have been used to define, as a first approximation, the contribution of technological landslide risk to the Canada landslide risk envelope (Fig. 1). A significant increase in risk is seen above the risk defined by the natural envelope across the magnitude range of landslide disasters. Conclusions - Approximately 570 people died in disastrous landslides and geotechnical failures in the period 1840 to 2000. British Columbia and Quebec sustained most damage. The most destructive landslide type in terms of total number of deaths in that time are small frequent landslides as opposed to large infrequent events. Even so Canada’s largest single landslide disaster is the huge rock avalanche that buried part of the town of Frank in 1903. The Canada landslide risk envelope is largely defined by disastrous events that occurred before 1930 in what was essentially an unregulated pre-technical environment in a period of intense resource development. 432 (75% of the total) of the total fatalities in the Canadian landslide record took place before 1930. There is a strong suggestion that a major portion of the deaths in the record resulted from landslides, which were strongly influenced by human activity, and geotechnical failures. In the Canadian context, many disastrous landslides represent a technological failure triggered by an extreme meteorological event. Their presence in the record represents an element of added risk in the nationallandslide risk envelope. F/N curves constructed using historical data are used to define "acceptable" societal landslide risk; however, they contain significant elements of risk resulting from "unacceptable" technological failure. This is thought to be an important consideration in landslide risk management, identified as a key activity in the United Nations’ International Strategy for Natural Disaster Reduction.

Bibliography of Canadian Geomorphology