Regional-Scale Evaluation of Landslide Distribution and Its Relation to Climate in Southern Alberta, Canada

Abstract

This work illustrates a semi-quantitative approach to evaluate changes in regional landslide distribution as a consequence of forecasted climate change, which can be adopted at other regions. We evaluated the relationship between climate conditions and landslide distribution at a regional scale. In this study, landslides on parts of the Battle, Red Deer, and Bow Rivers that are located within the Bearpaw Formation in Southern Alberta, Canada, were mapped, and their characteristics were compared. In order to find a relationship between the climate conditions and the mapped landslides, 30-year annual precipitation and other factors, such as slope aspect and geology, were compared between the river valleys. The results show that climatic conditions and the size and shape of the landslides are different in the Battle River area compared to the Red Deer and Bow Rivers regions. The weak Bearpaw overconsolidated shale and the bentonite layers throughout the region are sensitive to moisture and create favorable conditions for landslides in the river valleys. Further investigations into the long-term impact of climate on the formation of river valleys and the Bearpaw Formation support the argument that climate is one of the main factors in causing variations in landslide distribution across the study areas. These findings provide insight into possible changes in regional landslide distribution as a consequence of climate change.

1. Introduction

Slope processes and the distribution of landslides can be attributed to several main factors. These factors include the type of rock or soil (lithology), the prevailing climate conditions, the extent of weathering, the presence and movement of groundwater, the type and density of vegetation covering the slopes, the history and evolution of the valley, and the erosion occurring at the base of the slope (toe erosion). Although understanding the direct relationship between climate and slope processes can be problematic, it is undeniable that climate plays a significant role in shaping the Earth’s landforms [1,2].

Investigating the effects of climate on landslides has consistently remained an interesting and complex subject in geologic sciences and engineering, and its results provide valuable insights into managing the risks associated with landslide hazards. In this regard, numerous studies, including statistical analysis, numerical modelling, laboratory experiments, and qualitative investigations, have been conducted around the world in the regions susceptible to landslides. As explained by Cruden and Varnes [3], weather-related factors—notably, precipitation, freeze–thaw cycles, and weathering—are recognized as primary factors in landslide occurrences. In this regard, many researchers have investigated the effect of weather-related factors on landslide activity [4,5,6,7,8,9,10,11,12,13].

The challenges associated with regional-scale studies are different from local-scale studies. Among the differences, the lack of information regarding the exact time of landslide occurrences makes it impossible to investigate the impact of factors such as antecedent cumulative rainfall or freeze–thaw cycles on each individual landslide. Moreover, due to the limited availability of weather stations within a large geographical area that are continuously operational over a large period of time, it is necessary to rely on modern and paleoclimate models to simulate conditions for different time periods. Despite these challenges, the study of climate impact on landslides at the regional scale remains feasible, even if in a qualitative manner.

In this paper, we examine the influence of climate conditions at a regional scale on landslide occurrences within three regions located in Southern Alberta, Canada. To achieve this, we systematically collected relevant climate and geological data and employed robust analytical methods, which are detailed in the following section. The study area covers parts of the Battle, Red Deer, and Bow Rivers, with a total length of 182 km. To ensure that the geological factors have the least effect on the results, the study area that was chosen is entirely within the Bearpaw Formation, a geological formation known for its vulnerability to landslides due to its structurally weak nature [14].

2. Study Area

2.1. Location

Large parts of Western Canada are made up of rock formations that were formed during the Late Cretaceous period, around 75 million years ago. One of these formations is called the Bearpaw Formation, which is a sedimentary deposit composed primarily of silt and clay particles that have been subjected to consolidation loads more than those provided by the present overburden. The interparticle bonds of these materials can break when they come into contact with water [15].

The Bearpaw Formation covers large regions of Alberta and Saskatchewan in Canada, as well as the state of Montana in the United States. The extent of the Bearpaw Formation in Alberta is shown in Figure 1. Three rivers in the southern part of the province pass through this structure.

As a river stream gradually erodes the surrounding land, it carves out a valley and exposes the underlying geological structures. This makes river valleys an ideal location for studying geological formations. The Bearpaw Formation is particularly weak and prone to landslides. The combination of this vulnerable geological structure and the presence of river valleys creates favorable conditions for landslides to occur.

This study focuses on the parts of the Battle, Red Deer, and Bow Rivers, along with parts of their tributary channels, which lie entirely within the Bearpaw Formation to make sure that all of the mapped landslides are located within the same geological context. Figure 2 shows the study area along each river. In total, 54.5 km of the Battle River, 70.6 km of the Red Deer River, and 56.9 km of the Bow River were selected for this study. The geometrical characteristics of the study areas are presented in

Table 1. Overview of the rivers’ characteristics. All parameters were calculated for the parts of the rivers that are located inside the study areas. The rivers’ mean discharge rates were calculated for the period of 2011–2020 from the available historical hydrometric data [17].

	Length through the Bearpaw Fm. (km)	Total Valley Area (km2)	Mean Valley Slope (deg.)	Valley Width (km)	Valley Height (m)	Mean Riverbed Slope (deg.)	10-Year Mean Discharge in Spring (m3/s)	10-Year Mean Discharge in Summer (m3/s)
Battle River	54.5	86.6	11.3	1.0–3.0	45–100	3.2	16.9	6.6
Red Deer River	70.6	113.1	17.8	1.2–3.9	40–155	0.6	87.6	99.5
Bow River	56.9	47.7	15.7	0.3–2.2	20–100	0.2	109.6	167.5

Note: Spring months are March, April, and May. Summer months are June, July, and August.

.

The mean valley slope in the Red Deer River and Bow River regions is comparable and notably higher compared to the Battle River area. The Battle River has a lower flow compared to the Red Deer and Bow Rivers. The Red Deer River exhibits moderate river flow, while the Bow River has a higher flow. Geographically, the Red Deer River and Bow River are situated near each other and share several common characteristics.

According to Brooker and Scott [15], the current river valleys in the area were formed during the retreat phase of the latest continental glaciation, 12,000 to 25,000 years ago. In another study, Matheson and Thomson [18] showed that the deep, steep-sided channels which were frequently ice-marginal or glacial lake spillways, were rapidly eroded by the large volumes of glaciers meltwater and formed the current river valleys. They also stated that the existing river valleys have relatively steep valley walls, except where landslide activity has flattened them.

2.2. Geology

The geomorphology and surficial geology (to a depth of approximately 2 m) of the study areas are shown in Figure 3. As expected, all three river valleys are mostly covered superficially by fluvial and colluvial deposits. Generally, stagnant ice moraine is the most common material found in the region. However, there are some distinct differences in the surficial geology in the study areas. For example, although fluted moraine can be found widely right along the Battle River valley, it is not common in the Red Deer and Bow Rivers’ area. On the other hand, glaciolacustrine deposits are more commonly found in the Red Deer and Bow Rivers’ area compared to the Battle River area.

The geomorphology maps of the study areas show similar patterns in the Red Deer and Bow Rivers’ areas. Hummocky, ridged, plain, and gullied patterns are the most common patterns along these two rivers. The Battle River area is covered mostly with a reticulated pattern.

2.3. Ecoregion

The map of the ecoregions of Alberta is shown in Figure 4. According to this map, the study area along the Battle River lies within the Central Parkland ecoregion, but the study areas in the Red Deer and Bow Rivers are located within the Dry Mixed-Grass ecoregion. In comparison to the Dry Mixed-Grass subregion, which has hot summers, intense sunshine, high evaporation, and long, cold winters with low snow cover, Central Parkland lies between the cold, snowy northern forests and the warm, dry southern prairies and receives more rainfall during summer [20].

The surficial geology and geomorphology data with the information associated with the ecoregions, as well as the vegetation type and cover on the river valleys and the surrounding area, show that the climate regime in the Battle River area is different from the Red Deer and Bow Rivers’ areas.

3. The Bearpaw Formation

The Bearpaw Formation was named after the Bearpaw Mountains in Montana, USA, where marine shales of this formation were first identified [22]. The Bearpaw Formation extends throughout Central and Southern Alberta, with varying thicknesses in different areas. According to Allan and Sanderson [23], the formation is estimated to be between 120 and 150 m thick along the Red Deer River east of Drumheller. In Southern Alberta, Russell [24] measured 223 m of the formation along St. Mary River, south of Lethbridge.

Around 75 million years ago, during the Late Cretaceous Epoch, sediments were deposited in a broad, shallow sea called Bearpaw Sea, which was bounded by the Cordillera highlands on the west and by the Canadian Shield on the east [15]. Most of the sediment that constitutes the Bearpaw Formation came from the Cordillera highlands during a period when the sea level was fluctuating but gradually retreating [15]. According to Reeside [25], Bearpaw sediments accumulated slowly in calm waters. Radiometric dating of sedimentary sequences in the Alberta and Peace River basins suggests that the sediment accumulated at a rate of approximately 30 cm every 7000 years [26].

During the time when the sediments were being deposited, volcanic activity was taking place in what is now Southwestern Montana, USA. This volcanic activity led to the deposition of layers of volcanic ash within the sedimentary sequence of the Bearpaw Formation, which, over time, resulted in the formation of bentonite layers throughout the formation [15]. According to the stratigraphy sections provided in Lines [27], thin seams of bentonite are distributed in different levels throughout the Bearpaw Formation, resulting in bentonitic shale and sandstone zones. Bentonite and bentonitic layers, which will be explained in the Results and Discussion section, are a common geological cause of deep-seated landslides in Southern Alberta.

Millions of years later, during the Pleistocene Epoch (2.6 million—11,700 cal yr B.P.), which was the Earth’s most recent period of glaciations, Southern Alberta was covered by a thick ice sheet called Laurentide, reaching approximately 1220 m in the Red Deer Valley, east of Drumheller, and 670 m in the Lethbridge area [28]. Generally, a glacier can affect the area in three ways: erosion and reshaping the landscape as it moves, deposition of new materials as it melts or recedes, and crustal depression under the influence of the ice load. The sediments of the Bearpaw Formation experienced loading from the deposition of younger sediments and unloading through erosion during uplift periods, followed by further cycles of loading and unloading due to glaciation. This process led to the formation of overconsolidated shales and widespread drifts, including till, throughout Alberta. It should be noted that, according to Brooker and Scott [15], the preglacial load in Alberta was not enough to create a permanent cohesive bonding in the shale. As a result, the shale can still disaggregate when exposed to water.

As Fulton [29] stated, clay shale in the region appears to have undergone prolonged and deep weathering during the Tertiary period. Because of the presence of smectite clay minerals in the Bearpaw Formation [30], the bentonite-rich sediment is prone to softening if sheared [29]. These shearing processes occurred during glacial periods and continued as the postglacial cutting of deep valleys released earth pressure. Present-day instability further contributes to pressure release along the backscarps of slides and results in a progressive reduction in strength [29].

Figure 5 shows an outcrop section of the Bearpaw Formation near the hamlet of Dorothy, which is located inside the study area and along the Red Deer River. Dorothy is known for its 10–13 m thick bentonite layer, which can be traced for 20 km along the Red Deer River [31]. According to Brooker and Scott [15], landslides in Western Canada often occur in thin seams of bentonitic material or other geological discontinuities. The Bearpaw Formation shown in Figure 5 is approximately 180 m thick and consists mostly of mudstone and shale, with layers of sandstone.

4. Climate

According to ClimateData.ca [33], the highest levels of rainfall in the region happen between May and September and lead to the erosion of the riverbanks and increased water flow. The 30-year average annual precipitation from 1991 to 2020 in the Battle, Red Deer, and Bow Rivers’ areas are 399, 330, and 333 mm, respectively. Harsh winters, with daily lows reaching −30 °C, are also a climatic characteristic of Alberta. According to the 30-year mean monthly temperatures between 1991 and 2020, the mean temperature can be as low as −13.9 °C in the Battle River area, −12.7 °C in the Red Deer River area, and −11.2 °C in the Bow River area [33]. The total annual precipitation and mean monthly temperatures were calculated for a 30-year time period and do not reflect the extreme recorded values.

Frost may occur provided that all conditions are present. These conditions include moisture content, ambient air temperature, and thermal properties of soil [34]. According to Brooker and Scott [15], frost can penetrate the ground to a depth of about 3 m in the region, with the maximum penetration usually happening in late March or early April. Frost penetration and its impact on fractures are other factors that contribute primarily to shallow landslides in overconsolidated shales.

Slopes are influenced not only by present climate conditions but also by past climatic regimes that have shaped the landscape over long periods of time. Western Canada has experienced changes in precipitation and temperature due to climatic variations since the last deglaciation. However, it is unlikely that the precipitation gradient over relatively short north–south distances in the study area has been significantly altered. This is because the topographic influence of the Rocky Mountains has remained constant over time [2]. Therefore, the northern part of the study area (Battle River) would have been experiencing more precipitation than the southern part of the study area (Red Deer and Bow Rivers) since the last glaciation [35].

Moreover, Campbell and Campbell [35] stated that the long deglaciation period which occurred approximately 20,000–12,000 cal yr B.P. is the source of the emergence of several major rivers in Central and Southern Alberta, all of which originate from the Rocky Mountains or the foothills. According to the radiocarbon dates of proglacial lakes in the region, deglaciation itself was probably complete prior to 15,000 cal yr B.P. However, the runoff originated from the mountains continued to shape the landscape until approximately 13,000–12,000 cal yr B.P. The final postglacial period was associated with landslide formation in the region, extending from ca. 10,000 cal yr B.P. to the present [35].

Vance et al. [36] conducted an extensive paleoecological study in the Canadian Prairies to compare the climatic conditions throughout the region approximately 6000 cal yr B.P. with today’s climatic conditions and found that the mean annual temperature was 0.5 °C to 1.5 °C higher than today’s, with the summer temperature up to 3 °C higher. Also, the mean annual precipitation was reduced by 65 mm (summer precipitation was reduced by 50 mm) compared to today. This suggests that (1) it is reasonable to use the current modern climate models to investigate the differences in the climate conditions between the study areas, as their relative differences would have been similar in the past; and (2) because the climate has not changed significantly since the onset of post-glacial slope instability, it can be concluded that the difference between the climatic conditions at the study areas has remained approximately the same, and therefore, it is reasonable to study the effect of climate on landslides and compare the landslide distributions among the study areas.

5. Methods and Materials

5.1. Climate Data

We collected the annual precipitation from the ClimateData.ca [33] database to determine the climate conditions from 1950 to 2100, considering the RCP4.5 emission scenario. We selected the periods 1950–1979, 2010–2039, and 2070–2099 to represent distinct climatic periods: historical (mid-20th century), near-future (early 21st century), and long-term future (late 21st century). This selection allows for a comprehensive comparison across different timeframes, providing insights into how climatic conditions have changed and are projected to change.

The average 30-year annual precipitation was calculated for every 10 km by 10 km grid of Southern Alberta. Ultimately, three 30-year weather databases were created for 1950–1979, 2010–2039, and 2070–2099 by interpolating the weather data, using the Empirical Bayesian Kriging (EBK) method in ArcGIS Pro v2.8.0 software [37].

5.2. Landslides

The geometric characteristics of river valleys and landslides, including size, aspect, and the degree of slope, were calculated using the digital elevation models generated from lidar data in ArcGIS Pro v2.8.0 software. The resolution of the lidar data used in this study is 15 m, which makes it difficult to locate landslides smaller than 15 m long. Therefore, the location of landslides on the river valley and its tributary channels was mapped through the interpretation of aerial photos, satellite images, and the hillshades derived from the digital elevation models.

To determine the extent of the study area in each region, the general river path was marked with a simple polyline, and two parallel lines 2 km away from the river path on each side were drawn. Subsequently, the locations of landslides on the river valley and its tributary channels were mapped. It should be noted that since different failure mechanisms are involved in landslides occurring in the floodplains, only landslides on the river valleys (i.e., between valley bluff line and valley ridge line) were mapped in this study.

Since most of the landslides in the area are prehistoric to several hundred years in age, and no records of their occurrence are available, identifying the crown and main scarp of the landslides is challenging due to the erosion processes over time. Similarly, observation of the displaced materials is commonly not possible, as they have been eroded or washed away over time. To create a database that is as comprehensive as possible, we performed a multipass search on the river valleys, and in each pass, we looked for identifiable characteristics of landslides, including crown, main scarp, transverse cracks, flanks, and displaced material on aerial photos, satellite images, and hillshades simultaneously. Once a justifiable landslide property was found, the extent of the landslide was delineated in the ArcGIS v2.8.0 software.

The final landslide database was partially validated through direct comparisons with databases provided by Liang [2] and Pawley et al. [38].

5.3. Geological Data

Geological data at the regional scale were gathered from scientific articles, as referenced in the previous sections, and publicly accessible databases published by the governments of Alberta and Canada [16,19,20,21]. The local geological data were extracted from water wells and coal test drilling reports near each landslide [39]. It is worth noting that most of these wells were initially drilled for water access purposes, and therefore, the reported geological information may not be entirely accurate. However, the closest boreholes with the most complete and reliable data were used, and their data were checked for consistency with the other boreholes in the vicinity. Additionally, due to the lack of denser geological information and detailed geological context in the area, we assumed horizontal stratigraphy for the analysis, as it is consistent with the nature of the Bearpaw Formation. All study areas are assumed to be geologically homogenous within the Bearpaw Formation to minimize variations due to differing geological factors.

5.4. Relationship between Climate Data and Landslides

The relationship between climate and mapped landslides were analyzed semi-qualitatively in ArcGIS. This analysis was performed by overlaying landslide-prone areas with climate data and making a direct comparison of the average annual climate data with the location and geometrical properties of the river valleys and mapped landslides. The choice of the software platform allowed for the efficient handling and visualization of geographic and numerical data. It is important to note that, due to the nature of the study and limitations in the available input data, conducting a complex statistical analysis was not practical.

6. Results and Discussion

The findings from this study indicate a correlation between climate conditions and landslide distribution. This section presents the results and provides a discussion based on the methodologies outlined previously.

6.1. Climate

The average annual precipitation in the region for three 30-year time periods is shown in Figure 6. This figure shows how measured and projected precipitation changes over time throughout the region as a result of climate change. According to

Table 2. Average annual precipitation (mm) in each river valley in different time periods. The percentage changes relative to 1950–1979 are shown in parentheses.

	1950–1979	2010–2039	2070–2099
Battle River	377	386 (+2.4%)	402 (+6.6%)
Red Deer River	309	318 (+2.9%)	335 (+8.4%)
Bow River	319	329 (+3.1%)	345 (+8.2%)

, which shows the average annual precipitation within each river valley in the Bearpaw Formation, annual precipitation relative to the base period of 1950–1979 is expected to increase over time. The percentage changes relative to 1950–1979 vary across the regions. The Red Deer and Bow Rivers are projected to experience greater changes compared to the Battle River, with up to 3.1% in 2010–2039 and 8.4% in 2070–2099. Although annual precipitation is estimated to increase, the difference in the precipitation between the areas will remain approximately the same over time, meaning that the Battle River will continue to receive more precipitation than the Red Deer and Bow Rivers. Therefore, although climate change may impact the current rate of landslide activity in the river valleys, the difference in landslide activity between the river valleys could remain unchanged.

6.2. Landslides

Figure 7 shows the mapped landslides in the three regions. Contrasting landslide sizes are observed in the Battle River area in comparison to the Red Deer and Bow Rivers. The differences in the land cover and density of agricultural activities are also visible in the satellite images. These variations come from the different ecoregion where the Battle River is located and show the differences in the moisture content between the study areas.

As previously mentioned, the resolution limitation of the lidar data and satellite images made it difficult to detect small landslides. The smallest landslide recorded in this study is 28 m wide and 10.5 m long and is located along the Bow River.

Figure 8 and Figure 9 show the distributions of the area and mean degree of slope of the mapped landslides, respectively. The mapped landslides’ statistics are presented in

Table 3. Landslides’ information.

	Number	Landslide Area (m2)		Landslide Slope (Degree)		Total Area of Landslides per Unit Length of the River (km2/km)	Number of Landslides per Unit Length of the River (1/km)	Total Area of Landslides/Total Area of the River Valley
		Mean	Median	Mean	Median
Battle River	46	714,332	284,383	11.5	10.9	0.60	0.84	0.38
Red Deer River	126	70,836	24,546	16.8	15.6	0.13	1.78	0.08
Bow River	94	30,766	13,721	23.9	22.5	0.05	1.65	0.06

.

Based on the mean area values, the landslides along the Battle River are approximately 10 times larger than those in the Red Deer River area and 23 times larger than the landslides mapped along the Bow River. According to Table 3, The median area values in all three rivers are less than half of the mean values. This indicates that a few larger landslides in each river valley are causing a positive skew in the dataset. However, this will not change the general interpretation of the landslide areas, as the median landslide area in the Battle River is still significantly higher than that in the Red Deer and Bow Rivers. Moreover, while the Battle River has the lowest degree of slope, with an average of 11.5 degrees, landslides in the Bow River area are steeper than the other two regions, with an average of 23.9 degrees.

Table 3 also shows a higher landslide distribution in the Battle River area, with 0.6 km² landslide per one kilometer length of the river, which is 4.6 and 12 times higher than the landslide distribution along the Red Deer and Bow Rivers, respectively. This is consistent with the findings of Liang [2]. In terms of number of landslides per unit length of the river, due to the smaller size of the landslides, the Red Deer and Bow Rivers show a similar distribution rate, with 1.78 and 1.65 landslides per one kilometer of the river, which is almost 2 times the rate in the Battle River area.

The ratios of total landslide area to total valley area show a higher value in the Battle River area, with 38% of the total river valley within the Bearpaw Formation involved in landslides. In comparison, only 8% and 6% of the Red Deer and Bow valleys are involved in landslides, respectively. According to Thomson and Morgenstern [14], the proximity of the preglacial bedrock channels to the southern part of the study area in the Bow River resulted in a lower groundwater level, thereby reducing the landslide activity historically.

Figure 10 illustrates the hillshade samples along each river, providing enhanced visualization of the valleys’ shape, the surrounding terrain, and the typical landslides. These hillshades, generated from the digital elevation models, effectively highlight the bare earth topography and landslides, aiding in a more precise identification of landslide locations, especially in vegetated regions like the Battle River valley.

As shown in Figure 10, there are clear differences in the shape of the valleys, channels, and morphology in the Red Deer and Bow Rivers regions compared to the Battle River area. Landslides on the Red Deer and Bow Rivers valleys are similar in size and type and are different from the landslides on the Battle River.

Figure 11, Figure 12 and Figure 13 show landslide examples, along with the geological data retrieved from the nearby boreholes. Note that all the landslides shown in these figures are sub-horizontal, and the vertical scale was set 7 times the horizontal scale only to obtain a better visualization of the slopes and landslides. Shale and sandstone are the most common material found in the region and make up most of the landslide bodies. The geological sections shown in the figures are consistent with the geological characteristics of the Bearpaw Formation. Although there might be some spatial differences in the formation at the location of the three rivers, their depositional environment suggests that these differences would be minor. Any variations in the water-well logs could be attributed to simplified descriptions, as these boreholes were not intended to provide detailed geological descriptions.

A majority of the landslides in the region are categorized as translational and compound landslides, which are intermediate between rotational and translational slide types. Translational slides in the area are controlled by sub-horizontal weak layers composed of bentonite and coal. These weak layers, as described by Lines [27], occur in the form of thin seams at various levels between the sandstone and shale of the Bearpaw Formation.

The landslide process often begins with rotational movement until it encounters a weak sub-horizontal surface. Once this surface is found, the landslide transitions into a translational slide, continuing to slide on this weaker layer. This is consistent with the previous studies in the area by Thomson and Morgenstern [14], Thomson and Morgenstern [40], Liang [2], and Biagini et al. [41]. The possible slip surfaces and the direction of movement of the mapped landslides are shown in Figure 11, Figure 12 and Figure 13.

According to Fulton [29], landslides in the Canadian Prairies move down the slope at a slow to very slow rate [3], and as the slope toe advances, there is subsidence of individual blocks of material behind. Every few decades, sufficient subsidence and translation occur in front of the backscarp, causing a new block of material to fail and the landslide’s crown to retrogress. The retrogression of landslides along the Battle and Red Deer Rivers can be seen in the hillshades of Figure 10.

The orientation of the slope with respect to the North is called the aspect. The aspect is measured in degrees from north (0°) in a clockwise direction. The slope aspect plays an important role in local microclimate and, therefore, in plant cover and soil stability [29]. The percentage distributions of the mean aspect of the mapped landslides are shown in Figure 14. The results indicate that 50% of the landslides along the Red Deer River and 50% along the Bow River face north, northeast, or northwest. In contrast, 61% of the landslides along the Battle River face southeast, south, or southwest. Since north-facing slopes receive less direct solar radiation than south-facing slopes, it is expected that the north-facing slopes are wetter and subject to less frequent and intense episodes of drying and wetting [42].

7. Effect of Climate on the Landslides in the Bearpaw Formation

One objective of this study was to examine the regional impact of climatic conditions on landslides. As indicated previously, there are notable differences in the geometric attributes of landslides, such as size, aspect, and average slope between the Battle River region and the Red Deer and Bow Rivers regions. We also showed that the study area along the Battle River lies within the Central Parkland ecoregion, with higher average annual precipitation than the study areas along the Red Deer and Bow Rivers, which are located in the Dry Mixed-Grass ecoregion. These findings, in conjunction with previous research on long-term climatic conditions in the region, provide evidence of contrasting climatic conditions between the Battle River and the other two rivers which highlights the influence of climatic conditions on the formation of river valleys and the occurrence of landslides in the region.

Although the regional geological maps show some differences in the surficial materials throughout the study areas, the sections in Figure 11, Figure 12 and Figure 13 show that the landslides are deep-seated and primarily occur within the Bearpaw Formation. Consequently, considering the consistency in the geological structure observed within the landslide bodies, and taking into account the fact that all three rivers originate from the Rocky Mountains in the west with a similar formation history, it can be concluded that one of the main drivers in the difference in the current shape of the valleys, including the slope, height, and width of the valleys, can be considered as a result of the difference in climatic conditions and their prolonged impact on the valley’s formation. It should be noted that the valley bottom’s width would have been primarily controlled by discharge in the meltwater channels during deglaciation, while the width at the top of the valleys is controlled by valley height and the slope material.

In simple terms, the long-term impact of regional climatic conditions on landslide distribution in the study areas can be observed in three ways:

• More precipitation and moisture within a particular region in the upstream catchments alter the shape of the river valley. In other words, as the water level rises and the flow velocity increases, potential floods intensify toe erosion and wash away slope materials. On a timescale of hundreds to thousands of years, the downcutting and, in particular, lateral migration of river channels both lead to slope steepening and are major causes of slope instability [29]. Figure 15 compares the average annual precipitation with the landslide distribution in each study area for 1950–1979, 2010–2039, and 2070–2099. The average annual precipitation and the ratio of total landslide area over total valley area are approximately the same in the Red Deer and Bow Rivers, as they are both located in the same ecoregion and have similar climate conditions. On the other hand, the landslide distribution in the Battle River is significantly higher compared to that of the Red Deer and Bow Rivers. This is while the average annual precipitation is only about 20% higher in the Battle River area. This could be a sign of how climate can impact landslide distribution in the long term.

Climate models predict more precipitation in all three rivers until 2100. However, according to Table 2, the amount of increase is less than the 20% gap in the average annual precipitation between the Battle River and the Red Deer and Bow Rivers, showing that the landslide distribution in the Red Deer and Bow Rivers is not expected to transition to those characteristics in the Battle River in the short term. Therefore, although more landslide activity is expected in all three river valleys until 2100 because of the increase in the average annual precipitation (especially in the case of intense seasonal rainfalls), the difference between the river valleys is expected to remain the same.

Predicting changes in the landslide distribution in the long term, however, is more complicated and depends not only on the climatic conditions but also human activity and residential development.

2.

Frost penetration and precipitation infiltration into the ground and the valley walls through opened tension cracks lead to the erosion and weathering of materials. In the case of moisture-sensitive materials like the Bearpaw overconsolidated shale, increased moisture content results in a higher degree of disintegration. The amount and rate of infiltration of precipitation into the surrounding soils at a regional scale are unknown, as they are dependent on factors such as surficial permeability, soil porosity, air temperature, etc. However, even small amounts of infiltration into overconsolidated shales may lead to the generation of high swelling, along with consequent decreases in the shear resistance of the soil mass [15].

3.

Aspect plays a significant role in determining the amount of solar radiation received by the valley walls. In the northern hemisphere, surfaces facing south (south, southeast, and southwest) generally receive more solar energy compared to those facing north (north, northeast, and northwest). As a result, moisture content and, therefore, landslide activity and distribution are expected to be lower on the south-facing slopes. We showed in Figure 14 that a higher percentage of recorded landslides in the Red Deer and Bow areas face north, while 61% of recorded landslides along the Battle River face south. According to Fulton [29], south-facing slopes are more susceptible to soil erosion due to the presence of scattered shrubs and bunchgrass compared to north-facing slopes, which have a moister environment and a denser plant cover. They also noted that this distinction is lost in the moister Cordilleran valleys, where precipitation is greater than about 500 mm annually. Although the annual precipitation in the Battle River area is less than 500 mm, it is possible that, due to the higher moisture, the effect of solar radiation on landslides is negligible.

It is worth mentioning that, according to the 30-year climate data, the annual precipitation is not constant throughout each river valley, with some parts of each river receiving up to 2.5% more precipitation annually. However, this difference was not considered significant and did not show direct correlation with higher landslide distribution along a given river valley.

8. Conclusions

This work illustrates a semi-quantitative approach to evaluate changes in regional landslide distribution as a consequence of forecasted climate change, which can be adopted at other regions. This paper presents the characterization of landslides along three river valleys in Alberta, Canada, within the same geologic formation. It is observed that Battle River has less steep valleys than the Red Deer and Bow Rivers valleys with landslides that are larger in size, and the mean degree of slope of the landslides is lower in the Battle River region. The higher historical landslide activity along Battle River has flattened the valley in the long term.

According to previous studies, Battle River has been experiencing more precipitation than Red Deer and Bow Rivers since the slope-instability period in the last deglaciation. Our study shows that the current annual precipitation is higher in the Battle River and will remain higher until at least 2100. Therefore, climate and its long-term effects are one of the main factors for the different landslide distributions across the study regions. Based on the climate predictions, we expect more landslide activity in all three river valleys in the short term (i.e., until 2100) because of the increase in the average annual precipitation. However, the landslide-distribution difference between the river valleys is expected to remain the same at least until 2100.

The Bearpaw overconsolidated shale and the bentonitic layer throughout the region are weak and prone to landslides. These materials can soften and lose their strength when sheared. This is why retrogressive landslides on gentle slopes along Battle River are more common than in the other two regions. Moreover, due to the sensitivity to moisture, the Bearpaw shale can disaggregate when exposed to water.

Aspect plays an important role in local climate and, therefore, soil stability, as the north-facing slopes are wetter and subject to less frequent and intense episodes of drying and wetting. We showed that 50% of the landslides along the Red Deer River and 50% along the Bow River face north, northeast, or northwest. In contrast, 61% of the landslides along the Battle River face southeast, south, or southwest. It is possible that, due to the higher moisture in the Battle River region, the effect of aspect and, therefore, solar radiation on landslides is negligible.

These findings, in conjunction with previous research on long-term climatic conditions in the region, provide evidence of contrasting climatic conditions between the Battle River and the other two rivers, highlighting the influence of climatic conditions on the formation of river valleys and the occurrence of landslides in the region. On a regional scale, long-term climate conditions appear to be one main factor in landslide distribution where the geologic formation is the same. Forecasted changes in climate, however, indicate that the landslide distribution in the Red Deer River and the Bow River would not reach the levels (in terms of aerial extent) mapped at the Battle River.