Time-series Analysis of Slope Displacements using UAV Photogrammetry and its Relationship with Rainfall Intensity

doi:10.21203/rs.3.rs-2981888/v1

Download PDF

Research Article

Time-series Analysis of Slope Displacements using UAV Photogrammetry and its Relationship with Rainfall Intensity

https://doi.org/10.21203/rs.3.rs-2981888/v1

This work is licensed under a CC BY 4.0 License

Journal Publication

published 04 Apr, 2024

Read the published version in Landslides →

Version 1

posted

You are reading this latest preprint version

In this study, UAV (Unmanned Aerial Vehicle) photogrammetry was used to monitor the ground displacement on the slope below a coal waste-rock stockyard, and to investigate the role of rainfall on its displacement. The study area is a mountainous region in Korea, where coal mining is active, and coal waste-rock is stored on the slopes of the mountain. In this area, material removal work was undertaken to reduce the steepness of the slope and driving forces in order to prevent disasters, but the strategy requires continuous monitoring to confirm the stabilization of the slope. For slope monitoring, a total of six UAV photogrammetry campaigns were conducted between April 2019 and August 2020. As a result of data processing, an orthoimage and DSM (Digital Surface Model) were generated. The ground displacement was estimated through four steps: slope extraction, displacement area evaluation, horizontal displacement, and vertical displacement analysis. During the study period, the maximum vertical displacement was 3.3 m and the maximum horizontal displacement was 3.5 m. The horizontal displacement was effectively evaluated through the measurement of the movement of the drainage system using orthoimages captured with a periodic survey. The effects of rainfall on ground displacement were also investigated. A very clear linear relationship between rainfall intensity higher than 20mm/d and ground displacement was identified. Accumulated amount of rainfall also showed good correlation with slope movement, but the frequency of rainfall intensity below 20 mm/d showed relatively low correlation with ground displacements.

Slope displacement

UAV

DSM

Rainfall intensity

Steep hills constitute about 64% of the total area of South Korea. These hills have a high distribution ratio of weathered granite masses with low cohesive strength, increasing the risk of disasters such as landslides and debris flows. Large-scale mountain soil and sediment erosion occurred during Typhoon Rusa in 2002 and Typhoon Maemi in 2003, resulting in landslides and debris flows (Moon et al. 2016; Lee and Kim 2017). The Erosion Control Work Act (ECWA) of South Korea, which was enacted in 1973, is a regulatory framework that aims to prevent soil erosion and mitigate the damage caused by floods and landslides. The ECWA defines erosion control works as any project or activity carried out to prevent or mitigate soil erosion, such as retaining wall construction, vegetation management, and drainage system implementation. Since 2011, the ECWA has also been responsible for identifying landslide and debris flow-prone areas and implementing a control system to prevent such occurrences. However, measures for soil creep are still not in place due to underreported cases and a lack of studies and techniques to manage them (Park 2015; Kim 2015). Soil creep has been found in about 35 locations in Korea and this is expected to increase as more are being identified. However, due to the presence of large clods of soil covering these areas and the very slow nature of soil creep, accurately monitoring and predicting hazardous conditions are challenging (Loew et al. 2017; Crosetto et al. 2018).

There are numerous different methods and approaches to identify ground deformation by analyzing the characteristics of areas prone to soil creep. Remote sensing methods have been used for the monitoring of ground deformation or landslides. Optical remote sensing images offer detection of horizontal displacement in wide areas (Stumpf et al. 2017; Mazzanti et al. 2020). Many applications concerning landslide monitoring have been conducted by remote sensing techniques such as InSAR (Interferometric Synthetic Aperture Radar) (Bayer et al. 2017; Dong et al. 2019; Soltanieh and Macciotta 2022; Macciotta and Hendry 2021; Woods et al. 2020(A); Woods et al. 2020(B); Journault et al. 2018). Global Navigation Satellite System (GNSS) enables low-cost and efficient landslide monitoring (Hastaoglu et al. 2018; Notti et al. 2020; Macciotta et al. 2016; Rodrigue et al. 2021). Terrestrial laser scanning (TLS) provides 3-D information with high density point cloud data and can generate DSM (Digital Surface Model) having very high resolution (Barbarella et al. 2015; Jun 2016; Carla et al. 2019; Macciotta and Hendry 2021). Some commonly used research methods include the assessment of risks by recording the locations where a mass movement occurred and building a database of activity (Vařilová et al. 2015; Pertuz-Paz et al. 2020), numerical modelling for the prediction of the range and impulse of the landslide and debris flow by using a dynamic models (Mast et al. 2014; Rickenmann et al. 2006; Nikooei and Manzari 2020), and experiments reproducing the actual phenomena (Iverson 2015); all these which can be enhanced by information gathered through remote sensing techniques.

High-resolution topographic data obtained by a UAV (Unmanned Aerial Vehicle) photogrammetry is verified by using various methods such as a comparison with satellite images, confirmation through the generation of digital topographic maps and inspection (e.g. ground control points). UAV can further be deployed using a low-cost camera to capture the area of interest, and comparing the accuracies obtained from fixed-wing and rotary-wing UAVs with given minimum requirements for the geohazard being analyzed, therefore decreasing monitoring costs (Balek and Blahut 2017). Recent technological advances allowed researchers to obtain data from hard-to-reach areas, resulting in utilizing UAV in mountainous regions and regions with mass movements in a safe manner and with enhanced detail (Clapuyt et al. 2017; Macciotta et al. 2020; Rodrigue et al. 2020). Importantly, topographical changes in landslides can detect complex displacement patterns using subsequent DTMs (Digital Terrain Model) (Conforti et al. 2021). Other researchers have used the technique to calculate surface displacements caused by mining activities and earthquakes, which can be effectively monitored using UAVs (Deffontaines et al. 2017; Cwiakała et al. 2020). Even when the ground collapses, this technique has been used to monitor ongoing, post-collapse surface displacements (Confuorto et al. 2017).

Time series data of DTMs, particularly point clouds obtained through LiDAR or photogrammetry, can be utilized to identify changes that have occurred over time. This technique compares the distance between point clouds in the normal direction to the surfaces in a statistical manner (Lague et al. 2013). This technique provides useful information, however slope displacements can be far from perpendicular to the slope surface, and other complementary techniques could aid in interpreting the true slope displacement vectors.

This research describes the possibilities for using UAV photogrammetry and techniques that could be used as complementary to Change Detection, to monitor slope deformations. The approaches are illustrated for a slope with coal waste-rock stockyard. The slope of the study area is adjacent to a residential area, and landslides are occurring continuously with a possibility of damage to the population from large-scale landslides. To prevent large-scale damage, investigations using sensors such as GPS receivers, inclinometers, and optical sensors have been conducted previously, but UAVs are considered a very cost-effective alternative for monitoring (Ahmad et al. 2021). One methodology proposed here determines vertical and horizontal terrain displacements without any additional sensors.

It has been well documented that rainfall can trigger shear deformation of slopes mainly through infiltration (Sasahara 2017; Martins et al. 2020). Some investigations in the lowering area indicate that rainfall is a determinant of severe lowering (Bru et al. 2017; Yang et al. 2019; Tafreshi et al. 2020). The time variation of shear deformation is essential for risk assessments, and slope deformation needs to be monitored appropriately. In previous studies, ground displacements due to rainfall was monitored using sensors, but the survey through the sensor only proved ground deformation at a discrete location. In this study, the effect of rainfall characteristics such as intensity and cumulative rainfall on ground displacement is quantitatively investigated. If the effect of rainfall on landslides is clearly investigated, it can be used as an important factor for early warning of landslides, and is a consideration in the installation of structures in preparation for landslides.

Figure 1 shows the work flow for this research. An essential part of this study is the estimation of ground displacements and the investigation of the effect of rainfall on ground deformation. UAV photogrammetry and geographic information system (GIS) were effectively used to estimate the ground displacements. The UAV used was a DJI's Phantom 4 pro, and the aerial surveyed source data were post-processed through the Pix-4D program. The data acquisition step also establishes a UAV flight plan for obtaining point cloud data, a set of data points in space that produces a digital elevation model (DEM) or a digital surface model (DSM). As GPS sensors become increasingly unreliable due to UAV instability, UAV speed, and wind; the site was surveyed to compensate for errors in the survey data and ground control points (GCP) were surveyed, that included the entire study location. The photogrammetry data were obtained from April 3, 2019, to August 14, 2020, using the UAV to analyze the topographic changes at the research location (Table 1). Comparisons between the 3D topographic data sets during 16 months, including the rainy season obtained by the UAV, were possible with these datasets.

Data collection would be more effective if the research location is within range of the GCPs. 6–9 GCPs were installed in the study area (Table 2). After obtaining the GCP data, preparations for aerial photogrammetry were made, including flight altitude, flight time, range, camera resolution, and image overlay. Orthoimages and DSM are generated as the result of post-processing, and the ground displacement was estimated using these data. Orthoimages are geometrically corrected to remove distortion caused by terrain relief and sensor position, resulting in a map-like image that can be used for visualization and measurement purposes. By comparing two or more DSMs acquired at different times, the displacement of individual points can be calculated and mapped.

Displacement analysis consists of four types: slope extraction, displacement area evaluation, horizontal displacement, and vertical displacement analysis. Orthoimages were used for horizontal displacement analysis, and the drainage system installed on the slope for surface water outflow was used as a reference for horizontal displacement analysis. The drainage system on the slope is easy to identify by UAV photogrammetry, and since it is installed at a regular distance, the displacement of the entire area of the slope can be measured. This method provides high accuracy of horizontal displacement. DSM is used to analyze the amount of vertical displacements. Based on the DSM generated for each period, the amount of vertical displacement is determined by analyzing the surface elevation change to the preset baseline surface. Several baselines were used to show the most representative cross-section in the study area. The effects of rainfall on ground displacement were investigated using daily rainfall intensity and cumulative rainfall. Rainfall data were obtained from a Dogye gauge station located at latitude 37.22379, longitude 129.09609, and altitude 817m, which is 4.5 km away from the study area. The daily rainfall intensity was evaluated by dividing it into rainfall higher than 20mm/d and rainfall lower than 20mm/d. The effect of rainfall on ground deformation was then quantitatively evaluated.

The research area is located in Samcheok city, Kangwon Province, Korea. The geographical coordinates are 129°2'40'' east longitude and 37°13'20'' north latitude (Fig. 2). The surface geology of the slope, in which the research area is located, corresponds to the Early Triassic Pyeongan Supergroup, and the main layer has a thickness of about 150 to 250 m, where milky white coarse-grained sandstone and coarse sandstone dominate, and dark gray sandstone and shale are included. Due to potential slope hazards, field investigations have been conducted at this location in the past, which reported extensive tensile cracking and ground subsidence in the lower area of the Dogye Coal Waste Landfill (Cho et al. 2011). Slope stabilization was attempted after the collapse of a small portion of the slope in the past. Blocking and cascade drains were installed, however, ground deformation at this location was still persisting, causing tension cracks, loss of topsoil, and dislocation of drainage pipes, leading to a massive collapse of the slope. The Coal waste storage is located 150m above the research site. The load imposed by this stock pile was expected to be the cause of the ground displacement. The load compresses the ground and causes deformation on the lower slope. The angle of the coal waste deposit is 34°, the distance between the bottom of the deposit and the top is approximately 70 m, and the horizontal extent is approximately 120 m. A cut slope below the coal waste deposit had a width of 100 m, an angle of 26°, and a surface area of 7,000 m². A sudden mass movement can cause massive damage to residential areas located below the study site, unless the slope remains stable.

Two types of equipment were used to survey the research area (Fig. 3). Figure 3(A) shows the set of equipment used to measure the GCPs composed of an RTK receiver and a controller pole. It received approximately 30 satellite signals, using 25 for RTK positioning. The accuracy of the RTK is 8 mm horizontally and 15 mm vertically, with the maximum root mean square deviation (RMSE) of 15 mm. Figure 3(B) shows the remote-controlled UAV with a camera used for topographic surveys. The data obtained from photographic images were used for spatial analysis. The UAV weighs 1.4 kg, has a maximum flight altitude of 6,000 m, a maximum speed of 72 km/h, a maximum range of 7 km from the controller, and a maximum flight time of 30 minutes.

Between April 3, 2019, and August 14, 2020, a total of 6 UAV photogrammetric surveys were conducted on the study area (Fig. 4). The red dot represents the points where photos were taken during the flight path. Atmospheric conditions for each flight were different, and flight conditions were set differently accordingly. Moreover, if necessary, the survey area was set differently (Table 1). The overlap rate and flight altitude of a UAV can affect the number of images. A higher overlap rate and lower flight altitude can improve image resolution and accuracy but also increase image redundancy and processing time (Elhadary et al., 2022). In this study, UAV photogrammetry was conducted at different times and under varying environmental conditions. Consequently, the flight altitude and overlap rate were applied differently, resulting in a variation in the number of images obtained.

Table 1

Summary of the UAV photogrammetric flight.
	April 3, 2019	October 13, 2019	December 22, 2019	April 10, 2020	July 16, 2020	August 14, 2020
Flight height (m)	50	60	60	65	70	90
Flight speed (m/sec)	1.9	3.2	1.4	2.8	3.9	3.4
Photogrammetry area (km²)	0.042	0.029	0.032	0.075	0.072	0.080
Overlap rate (%)	90	80	87	87	90	90
Number of images	305	104	213	395	477	278
Ground Sampling Distance (cm)	2.12	1.96	1.64	1.76	2.13	2.09

Surveying the GCPs was necessary to generate coordinates for the photographic data obtained by the UAV. The GCPs are the reference points surveyed directly from the research location, and they serve as the reference for the horizontal location (X, Y) and height (Z) of the cut slope. The GCPs were selected based on the road around the cut slope and the corner of the drainage (Fig. 5) to accurately identify the location of the reference points. The number and location of GCPs in each period were adjusted according to the conditions of the investigation.

The conditions and Root-Mean Square Error (RMSE) of each GCP obtained at each period are summarized and shown in Table 2. If the survey area is large, more GCPs are required, and we tried to use the same GCP location as much as possible.

The RMSE values of April 3, 2019, were 0.034 m, 0.015 m, and 0.044 m in the X, Y, and Z directions, respectively, and the average RMSE is 0.031. In comparison, the RMSE values of October 13, 2019, were 0.032 m, 0.027 m, and 0.032 m in the X, Y, and Z directions, respectively, and the average RMSE is 0.030. At each ground point of the survey, the RMSE of the GCP was considered to satisfy the accuracy.

Through UAV photogrammetry and 3D data construction, orthoimages and DSMs of each period were generated (Fig. 6). In the figure, the colour of the line is displayed differently depending on the time of the survey. The resolutions of the generated orthoimages were less than 2 cm. The surveys were conducted six times, including two rainy seasons, which was considered sufficient for investigating ground displacement. As shown in Fig. 5, in some cases, a tarp was temporarily installed in the study area, but it did not interfere with the study's progress.

Table 2

RMSE of GCPs.
	April 3, 2019	October 13, 2019	December 22, 2019	April 10, 2020	July 16, 2020	August 14, 2020
N. of GCP	9	8	7	9	9	6
N. of Photos	305	104	213	395	477	278
Overlap rate (%)	90	80	87	87	90	90
X-RMSE (m)	0.034	0.032	0.034	0.040	0.020	0.019
Y-RMSE (m)	0.015	0.027	0.014	0.035	0.029	0.035
Z-RMSE (m)	0.044	0.032	0.008	0.029	0.035	0.038
Mean Error (m)	0.031	0.025	0.011	0.034	0.028	0.028

Terrain changes for 16 months from April 2019 to August 2020 were analyzed using orthoimages and DSM, which were generated by UAV photogrammetry. Only the cut slopes were extracted from the images and used for the time-series analysis of ground displacement, and the extracted DSM shows the change of the fully exposed surface well. The surface of the slope and the drainage system were revealed in the orthoimage, and the drainage system on the slope has been used as the baseline for the ground displacement analysis.

The first step in the ground displacement study has been to confirm the hazardous area that may cause the displacement. The area of ground displacement may change with time, and in particular, may change depending on the frequency and intensity of rainfall. The area of the ground displacement and the intensity of the deformation becomes essential data for the decision of the countermeasures for managing the risks associated with this slope.

Figure 7 shows the vertical displacement from April 2019 to August 2020. The difference was calculated by subtracting the DSM of each period as of April 2019. The results of the analyses were shown in blue-red gradation; blue and red indicate the rise and lowering of the ground, respectively. As the time series analysis results, it can be seen that the ground displacement area does not change significantly over time and remains constant. These results indicate two points about the ground displacement of the slope of the study area: One is that the cause of the ground displacement appears to be due to the rise of groundwater due to rainfall, not a load of the waste coal on the upper slope (see Fig. 2). If it would be due to the load on the top of the slope, the displacement range is expected to occur over the entire slope, particularly close to the location of the coal stockpile. Another is that the constant pattern of the ground displacement implies that the horizontal mass movement is the dominant status on the slope.

Following the area analysis, ground displacement over one year were analyzed using the orthoimage and DSM. Figure 8 depicts the seven distinct positions designated as the baseline for the ground displacement analysis, which were necessary to establish in advance. The baseline refers to a line that established as a reference in the analysis of topographical changes. The drainage system on a slope is similar to a straight line and tends to move sensitively when the ground surface changes, so it is suitable for use as a baseline. In Fig. 8, the peak of the slope is located near positions ①, ② and ③, while the surface flow moves towards directions ⑥ and ⑦. A drainage system is installed for surface drainage, and its location is consistent with No. ③, ④, ⑤, ⑥, and there is ⑦ on a retaining wall at the lower part of the slope. Figure 8 also shows the results of topographic changes in subsidence and uplift between April and October 2019. Since the period of six months includes the rainy season in Korea, it can be seen that the range of terrain change is relatively wide.

Figure 9 shows the profile of elevation change over 16 months from April 2019 to August 2020 at baseline ①. A value of 0 on the x-axis indicating the distance from the upper part of the slope, and the distance increases as it goes to the lower part of the slope. Figure 9(B) shows the changes from April to October 2019 and (C) from April 2019 to August 2020 more clearly at distances of 30–60 m. Therefore, Fig. 9 (B) reflects some of the changes in the overall slope shown in Fig. 8. Here, displacement by horizontal movement dominates the slope, and some subsidence sections are initially observed due to the movement of the horizontal area where drainage channels are installed. Baseline ① shows the longitudinal change of the slope, and it can be seen that the change is continuously progressing within the investigation period. In particular, it shows that the ground displacement is predominantly horizontal.

Figure 10 shows the change in elevation by subtracting the surface elevation at each period from the surface elevation in April 2019 at the baseline ①. In the figure, the red line corresponding to the 0 value on the y-axis represents the ground elevation difference in April 2019, which means the ground elevation before the occurrence of ground displacement. Moreover, based on this ground elevation, the change of ground displacement for each period is shown. The elevation change is not constant and is thought to be due to the influence of the drainage system on the slope. As the horizontal drainage system moves laterally, the net change increases significantly or results in close to zero.

In Fig. 10, the ground displacement is shown to increase, and if such continuous ground displacement is accumulated, it is expected that a large-scale landslide collapse could occur in the future. Ground displacement occurring at the landslide site is mainly caused by collapse, erosion, and deposition. However, the ground lowering at the top of the slope is relatively small, results would apparently show that the slope experienced an uplift process. It can be seen that this apparent uplift is zero at the x-axis distance of 18m and 45m. As shown in Fig. 9, there is a horizontal (flat) area with drainage channels in these locations. These results show that the ground displacement is caused by horizontal movement rather than vertical change, with the horizontal displacement leading to that apparent but misleading uplift. In addition, it can be seen that the displacement is more significant in the region where the distance is close to 0, and this is expected to be the result of accumulated displacement at the lower part of the slope. It was found that the change was larger at the lower part of the slope, and that soil arches also occurred. In addition, it can be seen that the change in each period is very different. The ground displacement is expected to be related to rainfall, and further investigation is needed.

Figure 11 shows the elevation change during the period as shown in Fig. 9 at the baseline ②, and the change is more clearly expressed in the distance of 0–30 m. Baseline ② is also constantly changing, but it can be seen that the change is small compared to baseline ①. In particular, the change in topography from April 2019 to August 2020 shows that the ground subsided. Unlike the horizontal displacement shown in Baseline ① in Fig. 9, the displacement in Baseline ② appears to be due to ground lowering. These results show that the displacement of the slope can be caused by a combination of horizontal displacement and lowering for certain locations.

This trend is well shown in Fig. 12 and shows that ground lowering occurs at the slope's upper side. In Fig. 10, the elevation change represents a waveform assuming a typical shape of elevation change caused by horizontal displacement. The elevation change pattern in baseline ② has a different shape from baseline ①.The lowering of the upper slope and the rising of the upper slope are typical features of ground movement that has a rotational component in its kinematics. In particular, the section where the ground lowering is a section where groundwater infiltration is active, and a countermeasure method to remove groundwater or prevent surface water infiltration could be effective.

From these results, it was considered that the landslide was progressing while the ground displacement showed the characteristics of soil creep. Considering that rainfall was an important cause of such ground displacement, an analysis was conducted on the progress of ground displacement caused by rainfall. The baselines ③, ④, ⑤, and ⑥ in Fig. 12 are for horizontal displacement analysis different from the baselines ① and ②. The horizontal displacement analysis process is shown in Fig. 13. For the analysis, the moving direction and the moving distance were measured based on the drainage channel locations by overlapping the orthoimages of each period.

Figure 14 shows the analysis results of horizontal displacement. Distance 0 means the starting point of each baseline indicated by a circle in Fig. 13. Figure 14, (a)-(d) represent the baselines ③, ④, ⑤, and ⑥. In (d), the drainage system was demolished during the investigation period, making continuous analysis impossible. For the other baselines, five ground displacements were measured using orthoimages. It can be seen that the displacement at each baseline is different, and the displacement is larger as it goes to the lower part of the slope. Excluding the baseline ⑥, the displacement was the largest at the baseline ⑤, and the displacement was 3.5m for 16 months.

Figure 15 shows the deformation of the retaining wall installed at the lower section of the slope. Figure 15(a) shows that the retaining wall is installed in a straight line without any deformation, but Fig. 15(b) shows that the retaining wall is significantly deformed after six months. Figure 15(c) is an overlapped orthoimage showing how far the retaining wall has been moved. The dotted line in Fig. 15(c) was where the retaining wall was in April 2019, and the solid line was where the retaining wall was located in October 2019. The 27.5 m earth retaining wall was moved 3.0 m southwest from the 8 m section on the left. The baseline ⑤ and ⑥ showed the drainpipe’s horizontal (southwest) movement, confirming the occurrence of mass movement on the slope over six months. The drainpipe and the earth retaining wall of the cut slope indicate a slope failure, and the movement observed can be understood as soil creep.

The primary purpose of this study is to analyze the ground movement and the effect of rainfall on slope movement. Figure 16 shows the daily rainfall data collected during the study period from the weather station closest to the study area. Rainfall occurs intermittently during the study period, and it can be seen that rainfall of more than 20mm/d frequently occurs between July and October, which is the summer period. It can also be seen that strong rainfall of more than 150mm/d occurred during the study period. Ground displacement is strongly influenced by rainfall. In this study, in order to analyze the sensitivity of ground displacement to rainfall intensity, the total rainfall cases were divided into strong rainfall exceeding 20 mm/d and rainfall cases less than 20 mm/d. For analysis, rainfall over 20 mm/d was marked with a red circle. Ground collapse due to strong rainfall of more than 100mm/d is well known, but the effect of medium-intensity rainfall of about 50mm/d on the ground displacement needs additional investigation. Therefore, in this study, rather than an analysis based on cumulative rainfall, daily rainfall was used to examine the effect of rainfall intensity on ground displacement. ①-⑤ in the upper part of Fig. 16 indicates the discontinuous UAV photogrammetry periods for the ground deformation monitoring. The 6 UAV photogrammetric surveys were divided into five representative time windows. Each of the five windows is shown by a bar of a different colour. A red circle indicates rainfall intensity exceeding 20mm/d, and rainfall not exceeding 20mm/d is indicated by a white circle.

Regression analysis results are shown in Fig. 17 using rainfall data and the ground displacement data. In Fig. 17(A), the x-axis represents the number of times a rainfall intensity during the rainfall that occurred during the analysis of the topography change of the five sections (above or below the 20 mm/d threshold). The y-axis represents the amount of ground displacement using the result of (c) among the ground displacements shown in Fig. 14. The result of (c) was used because it was considered that the baseline ⑤ was located at the lower section of the slope and could adequately represent the ground displacement. For better understanding, different colour of each symbol is arranged corresponding to the colour of each period in Fig. 16. The circle in Fig. 17(A) represents the displacement caused by the rainfall intensity exceeding 20mm/d, and the square represents the displacement caused by the rainfall intensity less than 20mm/d. If the rainfall intensity is expressed separately, the influence of high or low rainfall intensity on the ground displacement can be investigated. The rainfall intensity and the ground displacement showed a linear relationship. Although the number of times of each rainfall intensity and displacement showed a linear relationship, the R-squared value showed a higher correlation with high rainfall intensity. In Fig. 17(B), the effect of accumulated amount of rainfall on ground displacement was also investigated and showed a clear linear relationship to the ground displacement. These results show that the influence of rainfall is remarkable among the causes of displacement. It was thought that an appropriate drainage system for runoff of rainfall is essential for the stable management of slopes, and facilities for the exclusion of groundwater are also required. Furthermore, forecasting cumulative precipitation would allow forecasting landslide displacement with a significant accuracy, which can provide enhanced information for early warning systems.

In this study, UAV photogrammetry was adopted to monitor the ground displacement occurring on the slope below the coal waste-rock stockyard. The research area has applied slope stabilization and drainage system for surface water outflow after the collapse of the slope in the past. However, even the slope stabilization, the ground deformation of the cut part was still prevalent, resulting in a large-scale collapse of the slope, such as tensile cracks, loss of topsoil, and drainage pipes. In past studies, monitoring surveys using sensors have been conducted in this area, but more intensive investigations are needed. Six UAV photogrammetry was conducted between April 2019 and August 2020, and the ground displacement on the slope below the coal waste-rock stockyard was confirmed. Orthoimages and DSM generated through post-processing provide a quantitative assessment of horizontal and vertical displacement on slope. Based on the DSM for each period, vertical displacement was evaluated through the surface elevation change in the preset baseline, and the maximum vertical displacement was 3.3 m during the study period. The horizontal displacement was evaluated through the measurement of the movement of the drainage system using orthoimages captured with a periodic survey, and the maximum horizontal displacement was 3.5 m during the study period. The effects of rainfall on ground displacement were also investigated with daily rainfall intensity, dividing it into higher than 20mm/d and lower than 20mm/d and accumulated amount of rainfall. Higher rainfall intensity over 20mm/d and accumulated rainfall showed a clear linear relationship with horizontal displacement. This study confirmed the effect of rainfall on the ground displacement. These results imply that proper drainage of the surface runoff water and groundwater discharging is important in determining a slope stabilization method. The UAV photogrammetry was very effective for monitoring ground displacement because it could extract moving area and assess the vertical and horizontal displacement with high precision. Furthermore, this study presented displacement measurement approaches for estimating horizontal and vertical deformation using point clouds and DSMs that would complement deformation information as obtained from Change Detection methods, as well as illustrates a method to evaluate the impact of rainfall intensity and cumulative rainfall on landslide displacement. This last one can be used to forecast landslide displacements based on weather forecasts.

Conflicts of Interest

The authors declare that there are no conflicts of interest regarding the publication of this paper

Funding Statement

This research was supported by a grant(2022-MOIS61-001) of Development Risk Prediction Technology of Storm and Flood for Climate Change based on Artificial Intelligence funded by Ministry of Interior and Safety (MOIS, Korea) and grant of Kangwon National University in 2022.

Ahmad N, Shafique M, Hussain L (2021) Landslide assessment by using multi-temporal UAV datasets: a case study in northern Pakistan, Arabian Journal of Geosciences 14(18):1855–1864
Balek J, Blahut J (2017) A critical evaluation of the use of an inexpensive camera mounted on a recreational unmanned aerial vehicle as a tool for landslide research. Landslides 14(3):1217–1224
Barbarella M, Fiani M, Lugli A (2015) Landslide monitoring using multitemporal terrestrial laser scanning for ground displacement analysis, Geomatics Nat Hazards Risk 6(5–7):398–418
Bayer B, Simoni A, Schmidt D, Bertello L (2017) Using advanced InSAR techniques to monitor landslide deformations induced by tunneling in the Northern Apennines, Italy. Eng Geol 226:20–32
Bru G, González PJ, Mateos RM, Roldán FJ, Herrera G, Béjar-Pizarro M, Fernández J (2017) A-DInSAR monitoring of landslide and subsidence activity: A case of urban damage in Arcos de la Frontera, Spain. Remote Sens 9(8):787–804
Carla T, Tofani V, Lombardi L, Raspini F, Bianchini S, Bertolo D, Thuegaz P, Casagli N (2019) Combination of GNSS, satellite InSAR, and GBInSAR remote sensing monitoring to improve the understanding of a large landslide in high alpine environment. Geomorphology 335:62–75
Cho YC, Song YS, Kim KU (2011) Investigation and analysis of ground deformation at a coal waste depot in Dokye. J Eng Geol 21(3):199–212
Clapuyt F, Vanacker V, Schlunegger F, Oost KV (2017) Unravelling earth flow dynamics with 3-D time series derived from UAV-SfM models. Earth Surf Dyn 5(4):791–806
Conforti M, Mercuri M, Borrelli L (2021) Morphological changes detection of a large earthflow using archived images, Lidar-derived DTM, and UAV-based remote sensing. Remote Sens 13(1):120–145
Confuorto P, Martire D, Centolanza G (2017) Post-failure evolution analysis of a rainfall-triggered landslide by multi-temporal interferometry SAR approaches integrated with geotechnical analysis. Remote Sens Environ 188:51–72
Crosetto M, Copons R, Cuevas-González M, Devanthéry N, Monserrat O (2018) Monitoring soil creep landsliding in an urban area using persistent scatterer interferometry (El Papiol, Catalonia, Spain). Landslides, 15(7): 1317–1329
Cwiakała P, Gruszczynski PW, Stoch T (2020) UAV applications for determination of land deformations caused by underground mining. Remote Sens 12(11):1733–1758
Deffontaines B, Chang K, Champenois J (2017) Active interseismic shallow deformation of the Pingting terraces (Longitudinal Valley - Eastern Taiwan) from UAV high-resolution topographic data combined with InSAR time series. Geomatics Nat Hazards Risk 8(1):120–136
Dong J, Zhang L, Liao M, Gong J (2019) Improved correction of seasonal tropospheric delay in InSAR observations for landslide deformation monitoring. Remote Sens Environ 233:1–18 15.
Elhadary A, Rabah M, Ghanim E, Mohie R, Taha A (2022) The in-fluence of flight height and overlap on UAV imagery over feature-less surfaces and constructing formulas predicting the geometrical accuracy. NRIAG Journal of Astronomy and Geophysics 11(1):210–223.
Hastaoglu KO, Poyraz F, Turk T, Kocbulut F, Demirel M, Sanli U, Duman H, Sanli FB (2018) Investigation of the success of monitoring slow motion landslides using Persistent Scatterer Interferometry and GNSS methods. Surv Rev 50(363):475–486
Iverson RM (2015) Scaling and design of landslide and debris-flow experiments. Geomorphology 244:9–20
Journault J, Macciotta R, Hendry M, Charbonneau F, Huntley D, Bobrowsky P (2018) Measuring displacements of the Thompson River valley landslides, south of Ashcroft, BC, Canada, using satellite InSAR. Landslides 15(4):621–636
Jun BH (2016) Numerical simulation of the topographical change in Korea mountain area by intense rainfall and consequential debris flow. Adv Meteorol 2016: 9363675
Kim HG (2004) Slope stability and characteristics of shallow landslide occurred in granite hillslopes. Seoul. KyungHee University.
Lague D, Nicolas B, Jérôme L (2013) Accurate 3D comparison of complex topography with terrestrial laser scanner: Application to the Rangitikei Canyon (N-Z). ISPRS J Photogramm Remote Sens 82:10–26
Lee JS, Kim YT (2017) Development of optimum rainfall threshold to predict of rainfall-induced landslides occurrence. J Korean Soc Hazard Mitig 17(6):333–340
Loew S, Gschwind S, Gischig V, Keller-Signer A, Valenti G (2017) Monitoring and early warning of the 2012 Preonzo catastrophic rockslope failure. Landslides, 14(1):141–154
Macciotta R, Hendry M (2021) Remote sensing applications for landslide monitoring and investigation in Western Canada. Remote Sens 13(3): 366–389
Macciotta, R, Hendry M, Martin CD (2016) Developing an Early Warning System for a Very Slow Landslide Based on Displacement Monitoring. Nat Hazards 81(2): 887–907
Macciotta R, Gräpel C, Skirrow R (A) (2020) Fragmented rockfall volume distribution from photogrammetry-based structural mapping and discrete fracture networks. Appl Sci 10:6977–6995
Mazzanti P, Caporossi P, Muzi R (B) (2020) Sliding Time Master Digital Image Correlation Analyses of CubeSat Images for landslide Monitoring: The Rattlesnake Hills Landslide (USA). Remote Sens 12(4):592–607
Martins BH, Suzuki M, Yastika PE, Shimizu N (2020) Ground surface deformation detection in complex landslide area-bobonaro, timor-leste-using SBAS DInSAR, UAV photogrammetry, and field observations. Geosciences 10(6):245–271
Mast CM, Arduino P, Miller GR, Mackenzie HP (2014) Avalanche and landslide simulation using the material point method: flow dynamics and force interaction with structures. Comput Geosci 18(5):817–830
Moon YH, Lee SE, Kim MS, Baek JR (2016) Analyzing types of urban areas at high risk to landslide hazard based on the GIS approach. J Korean Soc Hazard Mitig 16(3):47–53
Nikooei M, Manzari MT (2020) Studying effect of entrainment on dynamics of debris flows using numerical simulation. Comput Geosci 134:1–13
Notti D, Cina A, Manzino A, Colombo A, Bendea IH, Mollo P, Giordan D (2020) Low-Cost GNSS Solution for Continuous Monitoring of Slope Instabilities Applied to Madonna Del Sasso Sanctuary (NW Italy). Sens 20(1):289–312
Park JH (2015) Analysis on the characteristics of the landslide - with a special reference on geo-topographical characteristics. J Korean For Soc 104(4):588–597
Pertuz-Paz A, Monsalve G, Loaiza-Úsuga JC, Caballero-Acosta JH, Agudelo-Vélez LI, Sidle RC (2020) Linking soil hydrology and creep: A northern Andes case. Geosci 10(11):472–492
Rickenmann D, Laigle D, McArdell BW, Hubl J (2006) Comparison of 2D debris-flow simulation models with field events. Comput Geosci 10(2):241–264
Sasahara K (2017) Prediction of the shear deformation of a sandy model slope generated by rainfall based on the monitoring of the shear strain and the pore pressure in the slope. Eng Geol 224(22):75–86
Soltanieh A, Macciotta R (2022) Updated Understanding of the Thompson River Valley Landslides Kinematics Using Satellite InSAR. Geosciences 12(8):298–322
Stumpf A, Malet JP, Delacourt C (2017) Correlation of satellite image time-series for the detection and monitoring of slow-moving landslides. Remote Sens Environ 189:40–55
Tafreshi GM, Nakhaei M, Lak R (2020) A GIS-based comparative study of hybrid fuzzy-gene expression programming and hybrid fuzzy-artificial neural network for land subsidence susceptibility modeling. Stoch Environ Res Risk Assess 34(7):1059–1087
Vařilová Z, Kropáček J, Zvelebil J, Šťastný M, Vilímek V (2015) Reactivation of mass movements in Dessie graben, the example of an active landslide area in the Ethiopian Highlands. Landslides 12(5):985–996
Woods A, Hendry M, Macciotta R, Stewart T, Marsh J (A) (2020) GB-InSAR monitoring of vegetated and snow-covered slopes in remote mountainous environments. Landslides 17(7):1713–1726
Woods A, Macciotta R, Hendry M, Stewart T, Marsh J (B) (2020) Updated understanding of the deformation characteristics of the Checkerboard Creek rock slope through GB-InSAR monitoring. Eng Geol 281:105974
Yang YJ, Hwang C, Hung WC, Fuhrmann T, Chen YA, Wei SH (2019) Surface deformation from sentinel-1A InSAR: relation to seasonal groundwater extraction and rainfall in central Taiwan. Remote Sens 11(23):2817–2840

No competing interests reported.

Download PDF

Journal Publication

published 04 Apr, 2024

Read the published version in Landslides →

Version 1

posted

You are reading this latest preprint version

Time-series Analysis of Slope Displacements using UAV Photogrammetry and its Relationship with Rainfall Intensity

Status:

Journal Publication

Version 1

Abstract

Figures

1 Introduction

2 Materials and Methods

3 Study Area and Field Surveys

4 Results and Discussion

5 Conclusions

Declarations

Conflicts of Interest

Funding Statement

References

Additional Declarations

Status:

Journal Publication

Version 1