A DIGITAL INVENTORY OF LANDSLIDES IN CHAMPLAIN CLAY Pete Quinn, D. Jean Hutchinson, Mark S. Diederichs, R. Kerry Rowe, Robin Harrap and Jackie Alvarez GeoEngineering Centre at Queen’s-RMC – Queen’s University, Kingston, Ontario, Canada ABSTRACT This paper describes development of a digital inventory of landslides in Champlain clay as part of ongoing efforts to map and communicate landslide hazards. The goal of this work is to support methodical landslide hazard mapping, which will be used for assessing and managing landslide risk in the marine sediments in the Saint Lawrence Lowlands. This paper describes project methodology and presents the outcomes of preliminary mapping via literature review, interpretation of a coarse digital elevation model, and air photo review. The paper closes with a discussion of future work and applications for the inventory. RÉSUMÉ Cette contribution décrit le développement d’un inventaire de glissements de terrains dans la glaise Champlain, formant une partie de l’effort envers la localisation et la communication de dangers dus aux glissements de terrains. Le but de ces travaux est le support de localisation de dangers dus aux glissements de terrain, qui sera utilisé pour déterminer et gérer le risque du aux glissements de terrains dans les sédiments marins de la région du fleuve St. Laurent. Ce papier décrit la méthodologie et les résultats d’une localisation préliminaire à partir d’un inventaire de la littérature, l’interprétation d’un model d’élévation numérique, et une revue de photos aériennes. Le papier finit avec une discussion de futurs travaux et les applications de l’inventaire. 1 INTRODUCTION Landslide hazard mapping can benefit from the availability of a detailed landslide inventory within a geographical information system (GIS). No detailed digital inventory presently exists for landslides in the sensitive Champlain clays of eastern Canada. This paper documents preliminary efforts to create such an inventory, with a discussion of challenges and limitations affecting the work, project methodology, and early findings from the work. The paper closes with suggestions for further work and discussion of applications for the inventory.

2 CHALLENGES AND LIMITATIONS The work to compile a comprehensive landslide inventory for manipulation and analysis in GIS is affected by a number of factors, including:

- Size and diversity of geology and physiography of the study area;

- Availability of appropriate data; - Resolution of available visual data; - Consistency of available data; and - Limited historical record and age information.

The post-glacial Champlain Sea that existed from about 11,000 to 9,000 years before present (Karrow, 1965) occupied a substantial area in eastern Canada. The

approximate marine limit is illustrated in Figure 1. Due to the size of the study area, it is a time consuming process to fully document the existence or absence of landslides.

Figure 1. Project study area As the subsequent sections will illustrate, the development of a landslide inventory for this study area relies on various data sources, including the literature, aerial photographs, satellite imagery, and digital elevation models (DEMs).

OttawaGeo2007/OttawaGéo2007

713

A large number of individual landslides and groups of landslides in Champlain clay have been documented in the literature, and this information serves as a useful starting point. However, the nature of the development of the literature database is such that only selected areas have been studied, and certain areas where there have been numerous landslides have been described in the literature. Therefore, while there are numerous landslides documented across the Champlain Sea study area, the record may not reflect a true snapshot of the actual extent of landslides. The literature is therefore an excellent starting point for this database, but is not necessarily representative of average or typical conditions across the study area, due to systematic censoring given the nature of its accumulation. Air photo coverage, as available through the National Air Photo Library, is excellent for many parts of the study area; however, there is less coverage (e.g. fewer years with air photos) in many remote areas. It is therefore possible that landslide identification by air photo interpretation will be less thorough in less built up parts of the study area. This may yield a bias in the dataset. The data available for visual interpretation include air photos, satellite imagery and representations of DEMs. The resolution of these sources limits the size of resolvable features. The elevation of air photo flights, and associated scale and resolution, varies across the study area similar to availability as described previously. This may produce another bias due to systematic censoring of smaller features in areas where only smaller scale photos are available. Similarly, available DEMs, at 30 m and 90 m resolution, while useful for visualizing terrain, are of limited use for identifying existing landslides, except for the largest ones. Features with a minimum dimension smaller than about 100 m are not resolvable in available small scale DEMs. Some data, while available across the full study area, are not consistent across the study area. The Canadian Digital Elevation Data (2000) Level 1 DEM (CDED1) at 30 m resolution has been prepared by interpolating elevations from contour lines on topographic maps, resulting in an averaging, or smoothing, of true elevation data, and thus eliminating some finer topographic detail. Nevertheless, this DEM has been of considerable use in the present work. However, there are subtle differences across the study area in the way the original data were interpreted to create the topographic contours, and there are therefore subtle differences in the relationship between actual topography and that represented by the DEM. There may therefore be systematic biases across the study area in any interpretation that relies on the DEMs. Very little information is available regarding the age of observed landslides. A relatively large number of historic landslides are known to have occurred on a certain date, and a smaller but still substantial number of pre-historic landslides have radiocarbon ages or have been dated by other techniques (e.g. tree ring dating or pollen analysis).

However, the number of landslides with known ages remains a very small proportion of the total number of landslides, most of which occurred prior to European settlement in the area, or occurred in sparsely populated areas, making observation less likely. In spite of these challenges, and the resulting limitations in applicability of the database, the construction of the digital database will represent a significant advance, and will support more sophisticated geo-spatial analysis of relationships between terrain features and landslide occurrence, thus supporting the development of a better approach to hazard mapping in this area. 3 REVIEW OF THE LITERATURE Hodgson (1927) presented the earliest inventory of historic landslides in Quebec, documenting nine events that occurred between 1840 and 1925. Fransham and Gadd (1977) prepared seventeen 1:50,000 geological maps showing large landslides in the Ottawa area. Chagnon (1968) presented a summary landslide inventory for Quebec, and LaRochelle et al. (1970) provided detailed mapping of three small selected regions in Quebec described more generally by Chagnon (1968). A large number of other papers document individual landslide events (e.g. Tavenas et al. 1972) or describe terrain that has been modified by landsliding (e.g. Karrow, 1972). Landslide data from all identified literature sources, which consists of approximately 50 journal papers and several maps, have been transcribed into a GIS where they can be visualized and their occurrence or absence can be associated numerically with other geospatial features (e.g. slope angle, surficial geology). In some papers, landslide locations have been identified only as a point on a map. In others, landslide footprints have been illustrated as polygons. In most cases, the landslide scarps have been illustrated, so that the feature is represented by a polyline. The growing digital database therefore includes point, line and polygon landslide features. Figure 2 shows the distribution, across the study area, of landslide data available from the literature. Figure 3 illustrates the data available from several different sources in one selected area. These landslide data in Figure 3 have been obtained from Demers et al. (1999), Karrow (1972), Lajoie (1974), and Desjardins (1980). Additional landslide features interpreted from several air photos (not intended as a thorough representation of slides visible in air photos in the area) are also shown. The very large scalloped series of scarps around Lac Saint Pierre were suggested by Lajoie (1974) to be landslide scarps. While this interpretation has not been corroborated by others, the authors of the present paper believe it to be plausible.

OttawaGeo2007/OttawaGéo2007

714

Figure 2. Landslide data interpreted from available literature

Figure 3. Landslides near Lac St. Pierre, Shawinigan and Trois Rivières, from various sources

OttawaGeo2007/OttawaGéo2007

715

4 COARSE PRELIMINARY INVENTORY FROM A DIGITAL ELEVATION MODEL

An initial appreciation of physiography can be obtained rapidly from various interpretations (e.g. hillshade or slope angle) of DEM elevation data. Some landslides, namely those with pronounced scarps and a footprint sufficiently large in relation to the DEM’s resolution, can be visualized with the DEM. A DEM that covers the entire study area may therefore be used to identify large landslide features on a regional basis. It is important to note, however, that most landslides in Champlain clay are too small to be resolved by interpretation of available DEMs. This approach is therefore only suitable for locating very large features. Two DEMs have been used in this work: the CDED1 DEM, and the Shuttle Radar Topography Mission (2000) DEM (SRTM). As described previously, the CDED1 DEM is available at 30 m pixel size. The SRTM DEM is available at 90 m pixel size; however, each pixel represents a true average elevation, and thus some

topographical nuances that have been lost in the Level 1 DEM may be seen in this DEM despite lower resolution. Both of these DEMs have therefore been used to complement each other in this work. The two DEMs have been used to compile a preliminary coarse inventory of possible large landslide features. Figure 4 illustrates large landslides along the South Nation River, eastern Ontario, as visualized in a slope angle model of the Level 1 DEM. Figure 5 illustrates the general distribution of possible large landslide features throughout the study area. It is important to note that landslides with a minimum dimension of about 100 m or less are systematically eliminated from this coarse inventory, so the results should be interpreted with caution. It is also noted that the coarse inventory includes all large features that resemble large landslide scarps, and may include non-landslide features resulting from wave erosion, as an example. Additional ground truthing via air photo analysis or site reconnaissance is required to confirm the validity of this coarse inventory.

Figure 4. Visualization of individual landslide features from the Level 1 DEM

OttawaGeo2007/OttawaGéo2007

716

Figure 5. Approximate distribution of large possible landslide features throughout the study area

Figure 6. Landslides in the Yamaska River area of Quebec as determined from air photo analysis

OttawaGeo2007/OttawaGéo2007

717

5 DETAILED INVENTORY FROM AERIAL PHOTO

ANALYSIS The final inventory will draw most heavily from a methodical review of available aerial photographs. The distribution of landslides along the Yamaska River in Quebec, as determined from air photo analysis, is illustrated in Figure 6. This interpretation compares well with a similar interpretation by La Rochelle et al. (1970).By contrast, the landslides mapped by La Rochelle et al. in the Saint Vallier and Desbiens areas of Quebec represent only a small proportion of the landslides evident in the air photo record, as illustrated in Figures 7 and 8, which show the literature and air photo records, respectively, of large landslides near Saint-Vallier. The air photo inventory work is in the initial stages. It is intended that air photo analysis will be conducted for the entire study area.

Figure 7. Landslides near Saint-Vallier, Quebec per La Rochelle et al. (1970)

Figure 8. Landslides near Saint-Vallier from air photos

6 LANDSLIDE CLASSIFICATION SYSTEM AND DESCRIPTIONS

Additional contextual information can be linked to each landslide feature in the GIS. Basic geometric data such as approximate area, length and width will be associated with each feature obtained from the literature or identified through air photo interpretation. Additionally, the source of the data (e.g. air photo number or reference paper) will be identified. Finally, landslides will be classified based on aspect ratio (e.g. L/W – see Figure 9).

Figure 9. Landslide dimensions and aspect ratio classification system Class 1 retrogressive landslides have an aspect ratio L/W < 1. Class 2 and 3 landslides have aspect ratios L/W ~ 1 and L/W > 1, respectively. Class 4 landslides are classified separately, and have an outlet with smaller width than the maximum width of the landslide scar. These landslides are of the “bottleneck” type. Examples of Type 1 include Saint-Boniface (Begin et al., 1996) and Rigaud (Carson, 1979). The South Nation River Slide of 1971 (Eden et al., 1971) is a Type 2 landslide. The Lemieux Slide of 1993 (Evans and Brooks, 1994) is an example of a Type 3 landslide. The Saint-Jean-Vianney (Tavenas et al., 1971) and Saint-Thuribe (Hodgson, 1927) slides are typical Type 4, bottleneck slides. The proposed classification system is an extension to that proposed by Carson (1977), which related the maximum width of the landslide scar to the width at the outlet of the landslide. That system reflects the degree to which the landslide has a “bottleneck” shape. The present system is a more general shape classification, and is believed to relate to the degree of flow in a retrogressive landslide in sensitive Champlain clay. Landslide features obtained from the literature will be linked to text files which summarize the referring paper, so that a user can quickly explore the context of the original information. In cases where the data come from case histories of individual slides, these summaries generally include descriptions of subsurface data, geometric data describing the shape of the landslide, and

OttawaGeo2007/OttawaGéo2007

718

a description of landsliding processes, whether observed or interpreted. 7 LANDSLIDE FREQUENCY DATA The preceding sections have described an analysis of the spatial distribution of large landslides in Champlain clay. Temporal distribution is another significant component of the hazard, and an understanding of frequency of occurrence of past landslides is a key to understanding the future frequency, and associated seasonal and annual probability of occurrence. Large landslides in Champlain clay in Quebec occur throughout the year, being most common in April/May, with a secondary peak in activity in October/November (Lebuis et al., 1983). These authors’ analysis of available data in relation to observed river levels and calculated infiltration and evapotranspiration suggests a strong link between river erosion and landslide incidence during these two periods of pronounced activity. The interpretation of historic and pre-historic landslide occurrence to interpret landslide frequency is hampered due to factors discussed previously. However, Lebuis et al. (1983) suggest that, based on available records from historical observations and the limited dates of pre-historic events, large landslides tend to be clustered in time. For example, a number of very large ancient landslides east of Ottawa have radiocarbon dates close to 4500 years before present, and have been suggested by Aylsworth and Lawrence (2003) to have been triggered by a single large earthquake. Similarly, several large landslides have been connected to the 1663 Charlevoix earthquake. These include very large scars around the 1971 Saint-Jean-Vianney slide (Leggett and Lasalle, 1978), very large scars near Shawinigan (Desjardins, 1980), a series of landslide scarps along the Rivière du Gouffre (Filion et al., 1991), and a large landslide at Saint-Joseph-de-la-Rive (Quilliam and Allard, 1989).

8 FURTHER WORK Initial exploratory efforts have been made to use remote sensing to improve the quality of the emerging landslide database. Spectral analysis of LANDSAT TM or other multi-spectral satellite imagery can be used to interpret soil moisture and vegetation classifications. Existing landslides tend to support different types of vegetation than the adjacent undisturbed clay plains (Lajoie, 1974), suggesting that old landslides might be distinguishable from undisturbed terrain. An advantage of using LANDSAT imagery for this purpose is that the process can be automated, thus eliminating the possibility of human error, and accelerating the work. The remote sensing component of this work may be extended to assist with landslide susceptibility mapping. Gagnon (1972) suggested that areas with elevated moisture relative to their surroundings are more likely to

suffer large retrogressive landslides. In fact, Gagnon (1972) suggested that the Lemieux area along the South Nation River was at risk of a large landslide based on observed soil moisture evident in air photos. Gagnon (1972) made this observation two decades before the 1993 Lemieux slide occurred. This represents the only known prediction in the literature of a large landslide in Champlain clay in eastern Canada. Efforts will be made to map soil moisture and to use this interpretation in support of landslide susceptibility mapping.

9 APPLICATIONS FOR THE INVENTORY The ultimate purpose for this landslide inventory is to support the development of landslide hazard maps that can be used by land and infrastructure managers to assess and manage landslide risk. Earlier work (Quinn et al. 2007) presented a number of patterns linking geospatial features with the absence or occurrence of landslides in Champlain clay based on preliminary air photo analysis. Those patterns can be used in GIS to develop a preliminary appraisal of landslide hazards across the study area. The outcome of the present work will be used for more robust and methodical analysis of geospatial patterns via raster arithmetic in GIS. This work will support a more methodical and comprehensive approach to hazard mapping. ACKNOWLEDGEMENTS The writers appreciate the assistance of Jan Aylsworth of Natural Resources Canada for suggesting literature sources, particularly related to dating of landslides in Champlain clay. This work was supported financially by the GEOIDE network and the Railway Ground Hazards Research Program, funded by NSERC, Transport Canada, CN Rail and CP Rail. The work benefited substantially from the leadership and interest of Mario Ruel at CN Rail. REFERENCES Aylsworth, J.M., Lawrence, D.E. 2003. Earthquake-

induced landsliding east of Ottawa: A contribution to the Ottawa Valley Landslide Project. Proceedings of Geohazards 2003, 57-64.

Bégin, C., Evans, S.G., Parent, M., Demers, D., Grondin, G., Lawrence, D.E., Aylsworth, J.M., Michaud, Y., Brooks, G.R., and Couture, R. 1996. Le glissement de terrain d’avril 1996 à Saint-Boniface-de-Shawinigan, Québec: observations et donnés preliminaries. Current Research 1996-E, Natural Resources Canada, 215-223.

Canadian Digital Elevation Data. Level 1 [Digital Elevation Model]. 2000. Natural Resources Canada, Sherbrooke, Quebec, Canada. Available www.geobase.ca. [Accessed September 2006 to April 2007].

OttawaGeo2007/OttawaGéo2007

719

Carson, M.A. 1977. On the retrogression of landslides in sensitive muddy sediments. Canadian Geotechnical Journal, 14: 582-602.

Carson, M.A. 1979. Le glissement de Rigaud (Québec) du 3 mai 1978: Une interpretation du mode de rupture d’après la morphologie de la cicatrice. Géographie physique Quaternaire, 33 (1): 63-92.

Chagnon, J.-Y. 1968. Les coulees d’argile dans la province de Québec. Naturaliste Canadienne, 95: 1327-1343.

Demers, D., Leroueil, S. and D’astous, J. 1999. Investigation of a landslide in Maskinonge, Quebec. Canadian Geotechnical Journal, 36: 1001-1014.

Desjardins, R. 1980. Tremblements de terre et glissements de terrain: Corrélation entre des datations au 14C et des données historiques à Shawinigan, Québec. Géographie physique Quaternaire, 34 (3) 359-362.

Eden, W.J., Fletcher, E.B., and Mitchell, R.J. 1971. South Nation River landslide, 16 May 1971. Canadian Geotechnical Journal, 8: 446-451.

Evans, S.G., Brooks, G.R. 1994. An earthflow in sensitive Champlain Sea sediments at Lemieux, Ontario, June 20, 1993, and its impact on the South Nation River. Canadian Geotechnical Journal, 31: 384-394.

Filion, L., Quinty, F. and Bégin, C. 1991. A chronology of landslide activity in the valley of Rivière du Gouffre, Charlevoix, Quebec. Canadian Journal of Earth Sciences, 28: 250-256.

Fransham, P.B., and Gadd, N.R. 1977. Geological and geomorphological controls of landslides in Ottawa Valley, Ontario. Canadian Geotechnical Journal, 14: 531-539.

Gagnon, H. 1972. La photo aérienne dans les etudes de glissement de terrain. Revue de Géographie de Montréal, XXVI (4): 381-406.

Hodgson, E.A. 1927. The marine clays of eastern Canada and their relation to earthquake hazards. The Journal of the Royal Astronomical Society of Canada, XXI (7): 257-264.

Karrow, P.F. 1965. The Champlain Sea and its sediments. In: Soils in Canada: Geological, Pedological and Engineering Studies, Edited by R.F. Legget, University of Toronto Press, 97-108.

Karrow, P.F. 1972. Earthflows in the Grondines and Trois Rivières areas, Québec. Canadian Journal of Earth Sciences, 9: 561-573.

Lajoie, P.-G. 1974. Les coulees d’argile des basses-terrasses de l’Outaouais, du Saint-Laurent et du Saguenay. Révue de Géographie de Montréal, 86 (4): 419-428.

La Rochelle, P., Chagnon, J.Y. and Lefebvre, G. 1970. Regional geology and landslides in the marine clay deposits of eastern Canada. Canadian Geotechnical Journal, (7): 145-156.

Lebuis, J., Robert, J.-M., and Rissmann, P. 1983. Regional mapping of landslide hazard in Quebec. In Symposium on Slopes on Soft Clays, Swedish Geotechnical Institute Report No. 17, Linkoping, 205-262.

Leggett, R.F., and LaSalle, P. 1978, Soil studies at Shipshaw, Quebec: 1941 and 1969. Canadian Geotechnical Journal, 15: 556-564.

Quilliam, L., and Allard, M. 1989. Évolution géomorphologique du glissement de terrain et du marais littoral de Saint-Joseph-de-la-rive, Charlevoix, Québec. Géographie physique et Quaternaire, 43 (3): 367-376.

Quinn, P.E., Hutchinson, D.J. and Rowe, R.K. 2007. Toward a risk management framework: sensitive clay landslide hazards affecting linear infrastructure in eastern Canada, 1

st North American Landslide

Conference, Vail, Colorado. Shuttle Radar Topography Mission. [Digital Elevation

Model]. 2000. United States Geological Survey. Available http://gsc.nrcan.gc.ca/urbgeo/ [Accessed September 2006].

Tavenas, F., Chagnon, J.-Y. and La Rochelle, P. 1971. The Saint-Jean-Vianney landslide: Observations and eyewitness accounts. Canadian Geotechnical Journal, (8): 463-478.

OttawaGeo2007/OttawaGéo2007

720