4.1. Developing the 3D model of the SJV area
a) Creating the database for the SJV area
A hydrogeological database has been developed for the Saguenay-Lac-Saint-Jean region as part of a groundwater data acquisition program—Programme d'acquisition de connaissance sur les eaux souterraines; CERM-PACES (2013). This regional database includes data from all regional boreholes and wells, other relevant information, and the interpretation of stratigraphic sections (Figure 4). The stratigraphic sections 511 & 514 from this CERM-PACES database indicate that a clay layer, averaging 30 m thick, covers the bedrock. An intercalated discontinuous till layer, 1 to 5 m thick, is also present.
Given that the objective of the PACES project was related to groundwater resources, the thick clay deposits of the SJV area were not fully described in the initial CERM-PACES (2013) project, apart from the abovementioned sections.
For topographic data, we relied on a recent LiDAR (light detection and ranging) survey covering the SJV area. We undertook an initial interpretation of the regional stratigraphy with this LiDAR data. We then selected points for which the stratigraphy was known and added this new data to the database.
b) Acquiring and interpreting the geophysical data across the SJV area
In 2019, we conducted a field campaign using TEM to improve the data coverage within the study area. We added 75 new data points (blue dots in Figure 5), resulting in 18 pseudo-sections of material electrical resistivity.
Control points represent log data where the stratigraphy is known. During our field work, we conducted TEM surveys near these points. We then compared the electrical resistivity values for each lithology, using the Palacky (1987) chart (Figure 2) as a reference, with the observed stratigraphy at the control points (Table 1).
Table 1
Comparison of electrical resistivity values of our 2019 TEM surveys with the lithology of the control points within the Saint-Jean-Vianney (SJV) area.
|
Clay
|
Sand
|
Gravel
|
Rock
|
Minimum
|
1.51
|
35.37
|
95.61
|
0
|
Maximum
|
72.50
|
254.60
|
402.10
|
0
|
Mean
|
25.97
|
143.85
|
228.29
|
0
|
Standard deviation
|
19.92
|
88.63
|
101.35
|
0
|
Number of values
|
60
|
14
|
18
|
0
|
Proposed range
|
0–50
|
50–200
|
200–1000
|
≥1000
|
* Value in Ω·m
|
** Comparison carried out on five TEM surveys with control points within 50 m.
|
The observed resistivity values for sand and gravel (Table 1) are much lower than the Palacky (1987) values. These large differences underline the importance of adapting the chart to the specific territory under study. We therefore proposed new ranges of electrical resistivity for each material. It must be noted that fieldwork occurred mostly in the spring when the regional soils are highly saturated.
In the Palacky (1987) chart, the resistivity ranges of the materials sometimes overlap; however, we preferred to differentiate each range of values and avoid any overlap. A given value fits only one material; this approach therefore facilitates interpreting the data in the subsequent steps. The minimum value for a given material was established as the average of the maximum of the lower category and the minimum of the following category. To separate clay from sand, for example, we obtained a value of 50 by rounding roughly the average of the values of 35.37 (minimum value for sand) and 72.5 (maximum value for clay). Because of a lack of values for rock, the Palacky (1987) chart set this range at a value of ≥1000.
Using this new chart and the pseudo-sections, we converted all TEM surveys into stratigraphy logs, and the results were entered into the database.
c) Interpreting the soil stratigraphy within the SJV area
We interpreted the stratigraphy between data points using ArcGIS (ESRI 2015) and the Arc Hydro Groundwater tools (Aquaveo 2019).
We constructed 20 stratigraphic sections (Figure 6) in the study area after accounting for the spatial distribution of the data points. Two sections rely exclusively on borehole logs and serve as references for the other 18 sections based on the TEM pseudo-section data. For each section, topography and the PACES-interpreted surface deposits (Rouleau and Daigneault 2013) are displayed, as well as all streams, intersections with other sections, TEM and borehole logs, and other structural elements. We used all this information to interpret each section and ensured that the intersections between sections were concordant.
During this process, we reinterpreted some TEM logs to provide a better fit to the surrounding data. As well, we found large differences in precision between the borehole and the TEM logs. Borehole data scans identify different layers of thickness at a meter-scale resolution and less, whereas TEM readings produce average values that cannot be interpreted at less than a meter scale. Given that the SJV area has been affected by at least two major landslides, these events resulted in many alterations of the thin clay and sand layers across the study area that the TEM surveys could not differentiate; we therefore had to simplify the description of some sections.
A limited number of borehole logs describe the till layer (12 of 218 boreholes), and it is difficult to differentiate till from sand or gravel on the basis of their respective electrical resistivities. Consequently, the till deposit has been integrated into the granular layers. Moreover, the low precision of the TEM survey at less than a meter-scale resolution and the complex stratigraphy of the SJV area restricted our interpretation to four layers. Ordered from the surface downward, the first is a granular alluvial sediment–dominated layer (Granular 2). The second layer is the clay-silt dominant surface layer (Clay), and the third is a granular glaciomarine sediment–dominated layer (Granular 1). The final layer is the bedrock. This simplified stratigraphy is considered sufficient for geotechnical purposes.
d) Creating the 3D model of the SJV area
To export data from ArcGIS to Leapfrog, we used some of the methodology developed by Chesnaux et al. (2011) for building a geodatabase to map the hydrogeological features and produce a 3D modeling of groundwater systems. This approach consists of creating virtual boreholes at 50 m intervals within the previously developed sections on the basis of the geological interpretation of these sections. The augmented data set was imported into Leapfrog Geo software (Seequent 2020). The resulting 3D model consists of, from the bottom to top, the bedrock, the Granular 1 layer, the clay layer, and then the Granular 2 layer. (Figure 7).
4.2. Assessing the values of geotechnical parameters across the SJV area
During fieldwork, we collected clay cores at three locations (labels 1–3 on Figure 8). We collected six Shelby tube cores, two at each location. The Shelby tubes were 700 mm long, having a 75 mm external diameter and 73 mm internal diameter. We then produced 31 samples, each approximately 10 cm long, from these Shelby cores. We used nine in the laboratory tests (Table 2), and the 22 samples were sealed in paraffin.
Table 2
Description of the samples collected in the Saint-Jean-Vianney (SJV) area for laboratory tests.
Location
|
Sample
|
Sediment
|
Color*
|
Odor
|
Observations
|
Log data (Sampling date; depth)
|
1
|
TM1-01
|
Silted clay
|
Greenish gray
(5Y 4/2)
|
None
|
Presence of silty layers and an orange layer (rust). A piece of plastic wrap was found in the sample.
|
2019-05-09; 0.00––0.12 m
|
1
|
TM2-04
|
Silted clay
|
Greenish gray
(5Y 3/2)
|
None
|
|
2019-05-09; 0.35–0.45 m
|
1
|
TM2-06
|
Silted clay
|
Greenish gray
(5Y 4/2)
|
None
|
Presence of a millimetric silt layer.
|
2019-05-09; 0.55–0.70 m
|
2
|
TM3-02
|
Silted clay
|
Gray (2,5Y 3/1)
|
None
|
2 mm thick silt layer
|
2019-10-01; 0.10–0.20 m
|
2
|
TM3-04
|
Silted clay
|
Gray (5Y 3/1)
|
Slight organic odor
|
Centimeter-thick silty horizon in the center of the sample.
|
2019-10-01; 0.30–0.42 m
|
2
|
TM4-04
|
Silted clay
|
Gray (5Y 3/2)
|
None
|
A silty horizon of approximately 1.5 cm in the center of the sample
|
2019-10-01; 0.30–0.42 m
|
3
|
TM6-01
|
Clay, silt
|
Gray (2,5Y 3/1)
|
None
|
Slightly oxidized on the surface.
|
2019-10-01; 0.00–0.10 m
|
3
|
TM6-03
|
Clay, silt
|
Gray (2,5Y 3/1)
|
None
|
|
2019-10-01; 0.19–0.30 m
|
3
|
TM6-04
|
Clay, silt
|
Gray (2,5Y 3/2)
|
None
|
A small earthy mass is present in the sample.
|
2019-10-01; 0.30–0.47 m
|
* Munsell H. (2009)
|
The nine samples were used for triaxial compression tests according to the ASTM standard (ASTM D2850-15, 2015). The results presented in Table 3 include the maximum principal stress at failure (σ1failure), the minimum principal stress at failure (σ3failure), and the undrained shear strength (cu). These results allow estimates of the cohesion (c) and the friction angle (𝜑) for each sample location (Table 4) (Holtz and Kovacs 1981). We applied the Mohr–Coulomb failure criteria to obtain the values presented in Table 4. These values show that the cohesion and internal friction angle of the soils are consistent for each location; therefore, we can state that the average cohesion (locations 1 and 3) of the clay is 55.37 kPa, and the average internal friction angle is 13.99°. Note that the test on sample TM4-04 from Location 2 does not show conclusive results; the sample had already fractured at two places prior to the test, which resulted in lower resistance to axial stress. All other tests are conclusive.
We used six of nine samples to estimate the plastic and the liquid limits according to BNQ-2501-090; BNQ (2019) (Table 5). The results are reliable, and average values have been computed for each location. We also carried out a sedimentation test (BNQ-2501-025; BNQ (2013) on TM2-06 and TM3-02; the results are presented in Table 6. Both samples used for the sedimentation test show near equal quantities of clay and silt (Table 6). The difference between these two samples relates to the presence of different numbers of thin silt beds. In general, the soil in the area can be defined as "CL" according to the Unified Soil Classification System (Holtz and Kovacs 1981), i.e., inorganic clays of low to medium plasticity.
Table 3
Triaxial compression test results for the samples collected from the Saint-Jean-Vianney (SJV) area.
Area
|
Sample
|
σ1failure (kPa)
|
σ3failure (kPa)
|
cu (kPa)
|
1
|
TM1-01
|
199.48
|
50
|
74.74
|
TM2-04
|
289.13
|
100
|
94.57
|
TM2-06
|
437.98
|
200
|
118.99
|
2
|
TM3-02
|
366.63
|
50
|
158.31
|
TM3-04
|
478.20
|
100
|
189.10
|
TM4-04
|
398.00
|
200
|
99.00
|
3
|
TM6-01
|
222.62
|
50
|
86.31
|
TM6-03
|
357.38
|
100
|
128.69
|
TM6-04
|
488.73
|
200
|
144.37
|
Inconclusive result
|
Table 4
Friction angle and cohesion at each location for the samples collected from the Saint-Jean-Vianney (SJV) area.
Location
|
φ (°)
|
c (kPa)
|
1
|
12.81
|
50.70
|
2
|
-
|
-
|
3
|
15.16
|
60.04
|
Average
|
13.99
|
55.37
|
Table 5
Liquid limit (Wl), plastic limit (Wp), and plasticity index (Ip) for six samples collected from the Saint-Jean-Vianney (SJV) area.
Location
|
Sample
|
Wl (%)
|
Wp (%)
|
Ip (%)
|
1
|
TM1-01
|
44.10
|
25.26
|
18.84
|
TM2-04
|
42.79
|
24.87
|
17.93
|
Average
|
43.45
|
25.06
|
18.39
|
2
|
TM3-04
|
39.36
|
21.96
|
17.40
|
TM4-04
|
40.03
|
21.97
|
18.06
|
Average
|
39.70
|
21.96
|
17.73
|
3
|
TM6-03
|
41.36
|
24.87
|
16.49
|
TM6-04
|
42.28
|
25.01
|
17.27
|
Average
|
41.82
|
24.94
|
16.88
|
Table 6
Sedimentation test results for two samples collected from the Saint-Jean-Vianney (SJV) area.
|
TM2-06
|
TM3-02
|
Clay
|
47.4%
|
55.0%
|
Silt
|
52.6%
|
45%
|
e) Back analysis of a past landslide in the SJV area |
For the back-analysis phase, we selected a small landslide along the Aux Vases River to determine soil geotechnical parameters at a large scale. This small slip, which occurred between 1971 and 1975, was identified by analyzing available aerial photos covering the last 50 years. Only two aerial photos are available for this location during this period, one in 1971 and one in 1975; thus, this landslide occurred after the 1971 SJV event and before the 1975 air photo survey.
Photo A (Figure 9A) shows the slope condition in 2015. Photo B (Figure 9B) shows the pre-landslide slope. The scar of an initial slip is delimited by a yellow line on Photo B. The apex of the initial slip is 40 m from the river, whereas that of the final slip is 120 m distant from the river. The elevation at the top and bottom of the slope of the initial slip is approximately 40 m and 7 m, respectively. The 40 m estimate is extrapolated from the heights of the northern and southern ends of the present-day landslide scar. The estimated pre-slide slope of the 1971 embankment is 39.5°. The estimated 1971 embankment geometry places the top of the slope at 40 m from the slope foot and a height of 33 m. The embankment height is then computed to be 49.76 m for the 120 m from the foot to the top of the slip (Figure 10).
We introduced the 1971 slope geometry (Figure 10) and the present-day slope profile into the Slide software. An analytical slope stability approach (Richer et al. 2020) can undertake a back analysis (Rocscience 2018). To estimate the shear strength parameter of the soil, we undertook several iterations of slope stability modeling using various input values of soil cohesion and analytical methods. The density (γ) of the clay soil was estimated at 18.6 kN/m3 from the triaxial tests. To determine the shear strength of the soil, we varied the values of one parameter at a time, either the cohesion value or the friction angle, as back analysis is not possible with two unknown variables. Slope stability is not sensitive to the friction angle of the clay; therefore, the laboratory estimated value (14°) was considered as the known value.
Figure 11 illustrates the model geometry used in the Slide software and presents the nearest approximation of the failure surface for the interpreted topography. We found that the best conceptual analysis method for representing the slip profile is the "Corps of Engineers #2"(U.S. Corps of Engineers 2003). For all failure geometries that we considered, a cohesion value of 60 kPa is required to reach a safety factor of 1, a cohesion value that lies within the range of values obtained using the triaxial tests (Table 5). Figure 12 shows that the slip profile having the minimum safety factor (SF = 1) is deeper than that interpreted according to the profile of the observed present slope (purple line). The resulting shape of the slip profile is very similar to the present-day slope profile. Therefore, a cohesion of 60 kPa and a friction angle of 14° using the "Corps of Engineers #2" conceptual analysis produced the best profile of the slip.
4.3. Developing a landslide zonation map for the SJV area
The Quebec Ministry of Transport has mapped most of Quebec's territory to identify areas at risk of landslides. These zones are delimited by taking into account the characteristics of the soil and conditions related to the land, such as the soil type, the embankment slope, and the history of landslides in the zone. The Saint-Jean-Vianney area is unfortunately not mapped.
The City of Saguenay has, however, produced its own stress maps to identify zones having embankments with a landslide risk (Figure 12). For these maps, embankment size was determined using the criteria of a height ≥5 m and a slope ≥14° (Ville de Saguenay 2007).
Using the historical aerial photos and the existing LiDAR topography, we undertook an analysis of landslides that have occurred along the Aux Vases and the Petit-Bras rivers (#1 and #2 in Figure 12). From this analysis, we could parameterize these landslides and, in particular, estimate retrogression distance. We studied 30 landslides, including the 1971 SJV landslide and the smaller slip used in the back analysis. We applied a third-order moving average (Turmel et al. 2018) to determine the average retrogression distance along both the eastern and western banks of the rivers. Thus, the resulting estimate varies with location along the river. The landslides along the western bank showed greater retrogression than those found along the eastern bank (Figure 13). Note that the 1971 landslide occurred on the western bank.
From this graph, we can map zones having a landslide risk, which takes into account the variable-width areas related to the retrogression distances of past landslides (see Figure 14). The results of this method are more realistic, as they take into account the actual characteristics of the geotechnical and hydrological environments. To design this map, we considered all ≥14° slopes having a height of ≥4 m, following the MTMDET (2017). The widths of the areas classified as at risk of landslide were variable; we observed a constraint zone of approximately 80 m wide for the western bank and about 40 m wide for the eastern bank (Figure 14).
4.4. Local slope stability analysis of the SJV area
The database, the 3D model, and the zonation map created for the SJV area facilitate further stability analysis of slopes across the area. A stratigraphic section can be created for a local study site. A back analysis can then be conducted at an existing landslide site to estimate the soil geotechnical parameters at a true scale and identify the appropriate analytical method to use for calculating slope stability. Finally, the mapping of the constraint zone will identify the main zones at risk of landslide, and any project proposed for the SJV area must first take into account this map of landslide risk (Figure 14). The site of planned infrastructure located in a constraint area should be investigated further through a local stability analysis. The database could then be used to generate a 3D model of the site, and this model could then be imported into modeling software to assess slope stability. The data acquired from the back analysis of past landslides helps assign realistic values to the geotechnical parameters used in the stability analysis.
As an example, we conducted a local stability analysis of the slope near the site of the planned lookout (Figure 15). Its projected location is at the top of an embankment characterized by a landslide risk. We obtained the topography of the section from a 3D model of the site that also indicates that the slope is composed entirely of clay.
Our modeling of the lookout slope, using the Slide software, assumed a uniform load weight, and we applied the "Corps of Engineers #2" method, as it was the most appropriate for this analysis on the basis of the earlier back analysis. The geotechnical soil property values were also determined through the back analysis: a fully saturated clay soil, γ = 18.6 kN/m3, φ = 14°, and cohesion = 60 kPa. The uniform weight of the lookout was estimated at 34.01 kPa. The modeling (Figure 15) produced a minimum safety factor of 2.764, which is greater than 1.5, a conservative value used for a construction at the top of a slope. Thus, the proposed lookout design is safe for this embankment.