Predisposition to Mass Movements on Railway Slopes: Insights from Field Data on Geotechnical and Pluviometric Influences

Abstract

Monitoring natural slopes, embankments, and unstable slopes is crucial to reducing predisposition to mass movements, especially in areas with geotechnical instability and high rainfall. This study proposes a methodology to identify geotechnical and pluviometric triggers of mass movements in railway slopes. It involves registering slopes and embankments along the railroad, recording accumulated rainfall indices, and documenting associated accidents. The experimental program included a cadastral survey at a pilot site on the MRS company’s railway network in the Paraopeba branch, Minas Gerais, Brazil. Surface and subsurface drainage conditions, anthropic interventions, and modifications affecting slope stability were also examined. Additionally, the history of accidents involving geotechnical and regional rainfall indices were incorporated to identify potential triggering events for mass movements. The study found a good correlation between landslide records and geotechnical risk mapping but a low correlation between landslide records and rainfall isohyets. The latter result is attributed to the low density and poor distribution of rainfall data and active pluviometers in the region. Overall, understanding the geological–geotechnical characteristics of slopes and the correlation between accidents and rainfall indices provides valuable insights for predicting potential landslide occurrences.

1. Introduction

Railways are critical infrastructures that facilitate efficient transportation of goods and passengers across vast distances [1]. However, the integrity and reliability of rail networks are frequently compromised by the instability of adjacent slopes. Slopes bordering railway tracks are particularly susceptible to soil and rock movements, which can lead to significant disruptions, including transport losses, speed restrictions, and traffic interruptions [2,3,4,5,6,7]. These mass movements can be classified based on their failure mechanisms, as well as their geological and geotechnical properties.

Effective management of these challenges requires a thorough understanding of the geotechnical triggers for mass movements. This involves distinguishing between predisposing factors, which create conditions conducive to failure, and effective agents, that directly cause slope failures [8,9]. By accurately identifying and analyzing these triggers, it is possible to develop more robust strategies for mitigating the risks associated with slope instability along railway corridors.

Predisposing agents refer to intrinsic characteristics inherent to the slope, independent of human actions. These include geological–geotechnical properties, local stratigraphy, natural slope geometry, and regional environmental conditions [10]. The geotechnical characteristics of natural soil slopes are often directly associated with the underlying bedrock, except in sediment slopes. Thus, the mechanical behavior of a slope is influenced by inherited bedrock structures and soil formation processes.

Effective agents are external elements that trigger mass movements, such as anthropic activities, earthquakes, erosion, and climatic factors. Rainfall is widely recognized as a major trigger for slope instability [4,8,11,12,13,14]. Rain infiltration increases soil saturation, which reduces suction and consequently decreases shear strength. When the slope is fully saturated, additional infiltration primarily increases pore water pressure, which does not directly affect shear strength but rather influences vertical stress. This change in vertical stress reduces effective stress between soil grains and directs the stress path toward failure, ultimately reducing the slope’s safety factor by increasing weight and shear stress [15,16,17].

According to van Westen et al. [18], mass movement risk can be expressed as the probability of a potentially harmful mass movement in a specified period of time and in a given area. In this sense, two main approaches have been widely used to identify geotechnical triggers based on the evaluation of the temporal probability of the future occurrence of mass movements. The first of these is based on the analysis of possible slope failures and the second is the statistical treatment of previous mass movement events [19,20]. The first approach analyzes the current conditions of the slope and assesses its potential for instability [13,21,22]. This approach, however, is less suitable for large areas [23]. The second focuses on analyzing the frequency of previous mass movement events [24,25], and can be performed directly, using historical records of mass movements, or indirectly, using information from rainfall-induced mass movement events [26]. Despite numerous studies on geotechnical triggers, many models are limited to their specific study areas, making it challenging to generalize findings due to differences in climate, land use, geological–geotechnical properties, and geomorphological configurations [27].

However, a significant gap remains in the application of advanced digital tools for identifying and managing geotechnical risks on railway slopes. The integration of Geographic Information Systems (GIS) with geological–geotechnical data presents a promising avenue for enhancing the precision and scalability of these analyses. In the context of advancing transportation infrastructure management in the digital and intelligent era, this study proposes a GIS-based methodology for mapping geotechnical triggers of mass movements on railway slopes. The study focuses on the Paraopeba extension, part of the “Linha do Centro” railway in Minas Gerais, Brazil. By integrating GIS technology, the methodology aims to map geotechnical risks and construct isohyets based on the region’s rainfall history, correlating these with mass movement occurrences.

The relevance of this research lies in its potential to provide a scalable and precise tool for infrastructure administrators, enabling more effective management of geotechnical risks. By addressing the limitations of current models and leveraging emerging GIS technologies, this study contributes to the optimization and maintenance of railway infrastructures in a digital and intelligent transportation era. Similar methodologies have been successfully applied in other contexts, such as mapping urban ground collapse susceptibility in Hangzhou, China [28], and modeling landslide susceptibility zonation in Vietnam using a combination of objective and subjective weighting approaches [29]. Additionally, GIS-based spatial interpolation methods have been effectively used for carbon footprint mapping in Turkey, illustrating the versatility and efficacy of GIS in environmental and geotechnical studies [30]. Furthermore, recent studies highlight the increasing use of machine learning and hybrid models in landslide susceptibility mapping, with approaches like deep learning and physics-informed data-driven models improving accuracy and interpretability. These methods, applied in diverse settings, demonstrate the potential of advanced data-driven techniques to enhance landslide risk assessment and mitigation strategies [31,32,33].

The structure of this article is as follows: the second section presents the experimental program, detailing the case study, historical evaluation of mass movements, geological–geotechnical cadastral survey, and assessment of rainfall incidence. The third section discusses the results, and the fourth section provides the final considerations.

2. Methods and Data Preparation

2.1. Case Study

The Paraopeba branch corresponds to a 160 km railway line operated by the company MRS Logística, which connects the Central Line in Congonhas/MG to the municipality of Belo Horizonte/MG in the state of Minas Gerais, Brazil. The route of the railway is arranged in the north direction, with a total of 12 active stations, covering the municipalities of Congonhas/MG, Jeceaba/MG, Belo Vale/MG, Moeda/MG, Marinhos/MG, Fêcho do Funil/MG, Sarzedo/MG, Ibirité/MG, Brumadinho/MG, and Belo Horizonte/MG [34]. In this context, the case study included a 112 km stretch of the Paraopeba branch, with an initial kilometric mark at km 504.00 and a final kilometric mark at km 616.00, as presented in Figure 1.

The regional relief of the area where the railway section fits is marked by enclaves (Figure 1b) and some reliefs of higher altitudes (Figure 1c), which have poorly developed pedological covers that closely border the railway line. In pedological terms (Figure 1d), the extension of the railway stretch runs through the majority pedogenetics of the Haplic Cambisols (km 504.00–584.00), with occurrences of Litholic Neosols (km 587.37–588.49) and Red-Yellow Argisols (km 588.49–616.00). According to the Brazilian Agricultural Research Corporation—EMBRAPA [35]—Argisols represent 26.9% of the Brazilian territory and are characterized by presenting clay accumulations in the B horizon; on the other hand, Cambisols present poorly developed horizons, with an incipient B horizon, which appear close to strongly undulating or mountainous reliefs. Additionally, it is highlighted that, when manifested on slopes, Cambisols are more susceptible to erosion processes [35,36].

Thus, the mapping and identification of geotechnical predisposing factors to mass movements become even more relevant for the studied railway stretch, since the occurrence of mass movement can present a more critical impact due to its distance from the line axis (between 1.50 m and 10 m, with an average value of 2.50 m), impacting the operation such as interruption or restriction of traffic. In Figure 2a, it is possible to observe the restricted box-like section of the railway stretch at km 508.111; additionally, in Figure 2b, a mass movement occurred in the Paraopeba branch in 2017 at km 599.124, in which there is interference of mass movement on the railway line.

It is important to note that future works will incorporate a hillshade map and specific geotechnical data, such as soil resistivity and strength parameters, to provide a more detailed slope stability analysis. However, for the scope of the present study, the current approach is sufficient to identify high-risk areas and guide initial mitigation strategies.

2.2. Geotechnical Cadastral Survey

The incident records database was compiled by the railway operator and presents all the events that impacted the operation of the railway line, including various causes such as equipment, superstructure or infrastructure failures, derailments, and even extreme road invasion events. Thus, the mass movement inventory was elaborated from the survey of all occurrences recorded in the railway stretch from 2015 to 2018. The incident records database details the date, type of incident, and the initial and final portions of the mass movement occurrence, which is the essential information for validating geotechnical risk maps. For the selection of valid records, the following criteria were applied: (1) records with a localization accuracy of less than 500 m were excluded; and (2) incidents not related to geotechnical events, such as falling trees, were also not considered. The mass movements analyzed in this study are classified according to the updated methodology by Hungr et al. [37], which provides an enhancement over the Cruden and Varnes [38] classification. The categories used include rockfalls, debris flows, earth slides, and complex landslides, allowing for a more detailed and precise analysis of instability types, thus aiding in the identification of high-risk areas. Nineteen mass movement events were recorded in the stretch during the inventory period (2015–2018), primarily involving barrier falls and rock block falls along the marginal slopes. Twelve of these events resulted in the total interruption of traffic (

Table 1. Events recorded in the railway stretch between km 504.00 and km 616.00 from 2015 to 2018, and the respective impacts on the line’s operation.

Date	Incident *	Impact on the Line’s Operation	Initial (km)	Final (km)
04/12/2017	Barrier or rock block fall on the railway line	Interruption	504.11	507.54
20/03/2015	Barrier fall	Interruption	509.21	517.04
23/03/2015	Barrier fall	Restriction	509.21	517.04
05/12/2017	Barrier or rock block fall on the railway line	Restriction	523.64	531.56
08/02/2015	Barrier fall	Interruption	531.56	533.80
08/03/2015	Barrier fall	Interruption	531.56	533.80
20/01/2016	Barrier fall	Restriction	531.56	533.80
13/12/2016	Barrier fall	Restriction	539.51	541.42
25/11/2016	Barrier fall	Restriction	548.13	556.21
14/12/2016	Barrier fall	Interruption	548.13	556.21
19/01/2016	Barrier fall	Restriction	563.31	567.81
19/01/2016	Barrier fall	Restriction	569.66	576.35
19/01/2016	Barrier fall	Restriction	569.66	576.35
08/03/2018	Barrier or rock block fall on the railway line	Interruption	579.90	582.30
05/10/2016	Barrier fall	Interruption	587.02	589.45
28/02/2018	Barrier or rock block fall on the railway line	Interruption	587.02	589.45
14/01/2017	Barrier fall	Interruption	595.56	597.48
26/01/2016	Barrier fall	Restriction	610.31	614.38
08/12/2016	Barrier fall	Interruption	614.38	616.70

* Events with significant impact on the railway operation.

).

Then, the data presented in Table 1 were exported to a GIS (Geographic Information System) environment and georeferenced in the free software QuantumGIS (https://qgis.org/pt/site/, accessed on 20 December 2023), based on the kilometric marks of the incidents, since the geographic coordinates were unavailable. Slope morphometry and soil characteristics vary locally with lithology, morphology, climate, and geological history [39]. Soil thickness, vegetation cover and its contribution to soil resistance, and local infiltration conditions are peculiar to a geographic location and may induce variable stability conditions in response to rainfall. The geotechnical–geological cadastral survey was based on field visits and the evaluation of the geological complex (petrographic nature, weathering alteration state, tectonic accidents, layer attitude, orientation and diving, stratigraphic forms, diaclasation intensity, etc.), morphological complex (surface slope, mass, relief form), climatic–hydrological complex (climate, meteoric and groundwater regime), gravity, temperature and original vegetation type of marginal railway slopes.

The study included the analysis of data from technical visits for the in situ evaluation of local geology and the geotechnical characteristics of existing slopes, rocky outcrops and erosion processes. Visual inspections were carried out by the railway operator’s technical team on the marginal slopes, with an attempt to identify possible rupture surfaces, the presence of discontinuities, traction cracks and open joints in the cuts and embankments of marginal slopes. The inspection takes place annually along the route of the railway network, in which the slopes are classified into five geotechnical risk groups, sequentially ordered according to the stability conditions verified by the cadastral survey in the field:

• R1: soils with low cohesion and prone to surface erosions and/or poorly fractured rock, with schistosity favorable to falling, and few blocks;
• R2: soils without cohesion and prone to surface erosions and/or medium-fractured rock, with schistosity favorable to falling, and few blocks;
• R3: soils with identification of imminent erosion or wedge and/or very fractured rock, with schistosity favorable to falling, with small blocks;
• R4: soils present some previous mass movements, with median erosion process; when rocks, they are very fractured, with schistosity favorable to falling, and with small blocks;
• R5: soils with visible wedges, advanced erosion process and/or very fractured rock, with schistosity favorable to falling, large blocks, and well-defined soil–rock contact.

The classification of predisposing factors to mass movements was carried out based on a qualitative analysis that considered the historical frequency of mass movement events and the geotechnical conditions observed in the field. This approach was chosen to facilitate a preliminary assessment over a large area. The classification of predisposing factors for past years is essential for understanding the frequency and distribution of mass movements over time. This historical perspective provides a baseline that informs future assessments by identifying trends and patterns. Therefore, the historical classification of predisposing factors serves as a critical step towards comprehensively understanding geotechnical conditions along the railway stretch. Future studies should incorporate these detailed geotechnical parameters to enhance the analysis, but their absence in the present study does not compromise its effectiveness. The study’s current methodology is sufficient for identifying high predisposition to mass movements and guiding initial mitigation strategies, which are crucial for immediate infrastructure safety planning.

In summary, the data originated from the geological–geotechnical cadastral survey for the analyzed railway stretch of the Paraopeba branch were evaluated in a time interval from 2017 to 2020.

Table 2. Classification of geotechnical predisposing factors to mass movements for the years 2017 to 2020.

Classification of Geotechnical Predisposing Factors to Mass Movements Classification	2017	2018	2019	2020
R1	4	4	7	0
R2	14	14	9	2
R3	7	5	14	3
R4	3	2	11	1
R5	6	2	2	3

presents a summary of the number of points surveyed (inspection in kilometric marks) according to the classification of geotechnical predisposing factors to mass movements R1, R2, R3, R4, and R5 for the analysis period.

The data obtained from the geological–geotechnical cadastral survey, as well as the points of the inventory of occurrences, were treated and georeferenced. Then, the geotechnical predisposing factors mapping was based on the Kernel density estimator (heat map) to identify the areas with the highest geotechnical susceptibility to mass movements according to the data collected.

2.3. Assessment of Rainfall Incidence

The identification of areas of greater susceptibility to intense rainfall is an important source of information for the evaluation of spatial patterns of precipitation that trigger mass movements. Each physiographic area is subject to a different rainfall regime, with typical intensities, durations, and patterns [40,41,42,43]. A better knowledge of precipitation variation, through forecast mappings of its spatiotemporal distribution, may represent a significant progress in the definition of extreme hydrogeomorphological event warning systems [44,45]. However, the spatial modeling of this variable largely depends on the quality and availability of surface observation data series [41].

The historical precipitation series are provided by the HidroWeb Portal (http://www.hidroweb.ana.gov.br, accessed on 12 December 2023) of the National Agency for Water and Basic Sanitation (ANA), as well as the geographical coordinates of the stations available in the region of analysis. The rain gauges were selected within a radius of 50 km of the railway line, as shown in Figure 3, in which 26 stations remain active and with complete data records, 9 stations are active and with incomplete historical series, and 134 stations are inactive (corresponding to 79.29% of the stations in the region).

Finally, the rainfall data were spatialized and interpolated according to the Inverse Distance Weighting (IDW) method [46], to create the isohyets of the analyzed region. The weight (${\lambda }_i$) of each station can be calculated according to Equation (1):

$${\lambda }_i={\displaystyle \frac{\frac{1}{|D_i{|}^d}}{{\sum }_{i=1}^{ns}\frac{1}{|D_i{|}^d}}},d>0$$

(1)

where $D_i$ is the distance between sampled and unsampled points, parameter d is cited as a geometric shape for weight and $ns$ is the total number of rain gauges.

3. Results

The Paraoapeba branch corresponds to a railway section under MRS operation subject to the recurrent occurrence of mass movements, affecting the operation and maintenance of the line. Therefore, this work aimed to meet a need to deepen the study on the events of mass movements in this railway section.

Thus, it is important to assess the geological–geotechnical predisposing factors to mass movements by regionalizing data from a specific location on alert and for the mass movement risk assessment. First, a geological–geotechnical cadastral survey was carried out to diagnose the predisposing factors of railway slopes and the mapping of scenarios according to the 2017–2020 time range, as shown in Figure 4. Geotechnical predisposing factors maps are tools used to identify areas susceptible to the mass movement occurrences, allowing the delimitation of these regions with the help of field visits, digital terrain models (via GIS), geotechnical data, historical rainfall data and/or numerical modeling.

The validation of the presented predisposing factors to mass movements was conducted through a regression that correlated geotechnical and accumulated rainfall indices with the recorded mass movements. This reinforces the importance of considering both geotechnical and rainfall data in planning preventive measures.

According to Figure 4, it was possible to evaluate the evolution of geotechnical predisposing factors to mass movements for the railway stretch (km 504.00–616.00). In the mapping for the year 2017 (Figure 4a), different classifications of geotechnical predisposing factors to railway slopes were observed along the railway stretch and, thus, it was subdivided into two stretches mostly of higher susceptibility according to the cadastral survey carried out in the field. The first stretch extends from km 507.00 to km 512.00, near the municipalities of Congonhas/MG and Jeceaba/MG, classified as R4; the second stretch, located in the municipality of Brumadinho/MG between km 560.00 and km 583.00, was mapped as level R5.

Thereafter, points were added for each year in geotechnical predisposing factors mapping. In 2018 (Figure 4b), surveys were carried out at points beyond those already performed in 2017 and/or redundant (in the same position as the previous ones) to increase the scope of geotechnical predisposing factors mapping and rectify (and/or ratify) some points previously analyzed. The same was conducted for 2019 and 2020, complementing and rectifying (and/or ratifying) some points of the 2018 and 2019 geotechnical maps, respectively. Thus, cumulatively in the future scenarios (2018, 2019, and 2020), it is observed from the addition of new registration data (represented by the points in Figure 4) the densification in the stretches between km 570.15–579.87 and km 509.17–507.46 in the 2018 scenarios, and the inclusion of new points registered between km 532.19–534.39, km 544.10–544.80, and km 603.81–616.37 with classification R5 in the 2019 and 2020 maps.

Figure 4 shows, therefore, the importance of mapping the railway geotechnical predisposing factors to mass movements and the need to complement information along the entire stretch and the ratification and/or rectification of the points already analyzed. This mapping serves to improve the railway operation management and to assist in the definition of future points to be analyzed (to densify the information and increase the coverage of data over the entire stretch). These maps are therefore fundamental for decision-making and the implementation of alert systems, as they allow managers and authorities to be informed of the risk of mass movements.

Additionally, once the geotechnical predisposing factors mapping of the studied railway stretch was performed, the rainfall data in the region were analyzed to seek a correlation between the incident occurrences with the historical rainfall indices of the region. Therefore, rainfall data available in the region (see Figure 3) during the months of December, January, and February were used due to higher rainfall and greater mass movement occurrences recorded in the railway section (Table 1). Based on a historical inventory from 1977 to 2006, the Inverse Distance Weighting (IDW) method was used to create the isohyets of the analyzed region.

Figure 5 shows the isohyets generated for the region under study, considering the December–January–February quarters, the points where mass movements were recorded affecting the railway line’s operation from 2015 to 2018 (see Table 1), and the resulting geotechnical predisposing factors to mass movements mapping in 2020. Figure 5 presents a good correlation between the locations of mass movement occurrences recorded along the studied railway stretch and the geotechnical risk mapping. In addition, it is important to verify that most of these occurrences (between km 504.00 and 584.00) occurred in a region of Cambisols, which present poorly developed horizons and high susceptibility to erosion [35].

Regarding the isohyets presented, there was a good correlation between the mass movement occurrence recorded further North (at the end of the stretch) with the highest rainfall indices for the region. However, although some studies point to an important association between heavy rainfall events with greater susceptibility to the occurrence of mass movements [8,14], it was found that the other recorded incidents are located in regions with historically lower rainfall indices.

It is worth noting, however, that Campos and Paz [47] verified the importance of assessing spatial variability of rainfall data, mainly for studies of extreme events. And, in the case of this study, as shown in Figure 3, the spatial distribution of active rain gauges in this region is too sparse (especially in the vicinity of the railway, where it presented only one active rain gauge). It is therefore necessary to optimize the collection of these rainfall data, either by increasing the density of the spatial distribution of the rain gauges or through the use of weather radars. This is a study that can be carried out in the future, enabling the complementation of the information obtained in the field (as demonstrated by geotechnical risk mapping) to optimize the prevention and reduction of mass movement risks in the region.

4. Conclusions

The monitoring of natural slopes, as well as embankment bodies and cutting slopes, is an important tool for reducing impacts and preventing mass movements during events of geotechnical instability, especially in seasons with higher rainfall incidence. Thus, it is noteworthy that mass movements are very diverse, characterized in terms of aspects such as materials involved and rupture form. They can lead to obstruction of the railway lines, with impacts on the ’stop hour train index’ (which is the sum of stopped train hours on the entire railway line or a given section), on losses of the transported product, and even on lives.

For this reason, the risk assessment of mass movements has become an urgent task that can help authorities reduce damage caused by these incidents through proper land use management for infrastructure development and environmental protection. This knowledge is crucial to develop better warning strategies to mitigate geohydrological risk and reduce socioeconomic damage. This article contributes to the development of better warning strategies to mitigate risks related to mass movements.

The case study of the Paraopeba branch, a railway section under MRS operation in Minas Gerais, Brazil, was used to analyze the elaboration of geotechnical triggers of mass movements on slopes due to the large number of occurrences of such incidents, causing impacts on the operation of the railway. Through the geological–geotechnical cadastral inventory of marginal slopes, obtaining a history of rainfall indices in the region, and records of mass movements causing accidents on the studied railway stretch, we sought to perform correlations in order to enable the prediction/prevention of potential occurrences of mass movements. The results of this study identified a good association between the mass movement occurrence records and the geotechnical risk mapping, but a questionable association between the same records and the isohyets generated based on the rainfall history of the region (due to the low density of rainfall data and the poor distribution of active rain gauges in the region). Therefore, this work demonstrates that the identification of geotechnical triggers help managers in decision-making and allow the adoption of more efficient train traffic operational measures to local conditions.

Finally, to enhance the prediction of future risks, the integration of predictive models based on climate projections and probabilistic analyses is suggested. This approach can offer a more accurate estimate of risks associated with extreme events.